Classical and Quantum Cosmology: Gianluca Calcagni

Graduate Texts in Physics
Gianluca Calcagni
Classical and
Quantum
Cosmology
Series editors
Kurt H. Becker, Polytechnic School of Engineering, Brooklyn, USA
Jean-Marc Di Meglio, Université Paris Diderot, Paris, France
Sadri Hassani, Illinois State University, Normal, USA
Bill Munro, NTT Basic Research Laboratories, Atsugi, Japan
Richard Needs, University of Cambridge, Cambridge, UK
William T. Rhodes, Florida Atlantic University, Boca Raton, USA
Susan Scott, Australian National University, Acton, Australia
H. Eugene Stanley, Boston University, Boston, USA
Martin Stutzmann, TU München, Garching, Germany
Andreas Wipf, Friedrich-Schiller-Univ Jena, Jena, Germany
Graduate Texts in Physics publishes core learning/teaching material for graduate-

and advanced-level undergraduate courses on topics of current and emerging fields
within physics, both pure and applied. These textbooks serve students at the
MS- or PhD-level and their instructors as comprehensive sources of principles,
definitions, derivations, experiments and applications (as relevant) for their mastery
and teaching, respectively. International in scope and relevance, the textbooks
correspond to course syllabi sufficiently to serve as required reading. Their didactic
style, comprehensiveness and coverage of fundamental material also make them
suitable as introductions or references for scientists entering, or requiring timely
knowledge of, a research field.
More information about this series at http://www.springer.com/series/8431

Gianluca Calcagni
Classical and Quantum

Cosmology
123
Gianluca Calcagni
Instituto de Estructura de la Materia
Consejo Superior de Investigaciones
Científicas (CSIC)
Madrid, Spain
ISSN 1868-4513 ISSN 1868-4521 (electronic)

ISBN 978-3-319-41125-5 ISBN 978-3-319-41127-9 (eBook)
DOI 10.1007/978-3-319-41127-9
Library of Congress Control Number: 2016950570
© Springer International Publishing Switzerland 2017

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of
the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation,
broadcasting, reproduction on microfilms or in any other physical way, and transmission or information
storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology
now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or
the editors give a warranty, express or implied, with respect to the material contained herein or for any
errors or omissions that may have been made.
Printed on acid-free paper
This Springer imprint is published by Springer Nature

The registered company is Springer International Publishing AG Switzerland
Acknowledgements
Several people helped me to improve this work through their careful reading of
its parts and, last but not least, through their encouragement. I thank Emanuele
Alesci, Raúl Carballo, Dario Francia, Steffen Gielen, Renate Loll, Johannes
Mosig, Giuseppe Nardelli, Daniele Oriti, Thanu Padmanabhan, Johannes Thürigen,
Aleksey Toporensky and, in particular, Claus Kiefer and Edward Wilson-Ewing for
valuable feedback. Special thanks go to Sachiko Kuroyanagi (gravitational waves
consulting), Claus Kiefer, my editor Angela Lahee and all those who believed in
this project. Angela’s patience with my delays in delivering the book has been, to
put it in one word, exemplary.
Chapters 2, 3, 4, 5, 6, 9 and 10 are partly based on three series of lectures
on quantum gravity and cosmology given at Penn State University during Spring
Term 2009 and at the University of Potsdam in Summer Semester 2010 and Winter
Semester 2011–2012.
v
Contents
1 Introduction .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 1
1.1 Micro from Macro.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 1
1.2 Outline of the Topics .. . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 4
1.3 About Citations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 6
1.4 Conventions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 6
1.5 Measure Units . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 7
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 8
2 Hot Big Bang Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 13
2.1 Cosmic Expansion and Cosmological Principle . . . . . . . . . . . . . . . . . . . . 14
2.1.1 The Universe at Large Scales . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 14
2.1.2 Friedmann–Lemaître–Robertson–Walker
Background .. . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 19
2.2 Einstein and Continuity Equations . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 24
2.2.1 Energy Conditions . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 26
2.3 Perfect Fluid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 27
2.3.1 Scalar Field . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 30
2.4 Friedmann Equations.. . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 31
2.5 Content of the Universe .. . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 33
2.5.1 Dust and Radiation .. . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 33
2.5.2 Hot Big Bang and the Big-Bang Problem.. . . . . . . . . . . . . . . 36
2.5.3 Dark Energy and the cosmological Constant
Problem .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 39
2.5.4 Spatial Curvature and Topology .. . . . .. . . . . . . . . . . . . . . . . . . . 43
2.6 An Obscure Big Picture . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 44
2.7 Problems and Solutions . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 46
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 58
3 Cosmological Perturbations . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 63
3.1 Metric Perturbations.. . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 64
3.1.1 Linearized Einstein Equations .. . . . . . .. . . . . . . . . . . . . . . . . . . . 64
3.1.2 Gauge Invariance and Gauge Fixing .. . . . . . . . . . . . . . . . . . . . 66
vii
viii Contents
3.1.3 Cosmological Horizons . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 66

3.1.4 Separate Universe Approach.. . . . . . . . .. . . . . . . . . . . . . . . . . . . . 68
3.2 Linear Tensor Perturbations . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 70
3.2.1 Transverse-Traceless Gauge . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 70
3.2.2 Equation of Motion . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 71
3.2.3 Mukhanov–Sasaki Equation and Solution . . . . . . . . . . . . . . . 71
3.2.4 Discovery of Gravitational Waves . . . .. . . . . . . . . . . . . . . . . . . . 74
3.3 Scalar Perturbations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 76
3.3.1 Non-linear Perturbations .. . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 76
3.3.2 Non-linear Perturbations at Large Scales . . . . . . . . . . . . . . . . 77
3.3.3 Linear Perturbations at Large Scales .. . . . . . . . . . . . . . . . . . . . 78
3.4 Gaussian Random Fields . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 81
3.4.1 Power Spectrum .. . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 82
3.4.2 Bispectrum and Trispectrum .. . . . . . . . .. . . . . . . . . . . . . . . . . . . . 83
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 88
4 Cosmic Microwave Background . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 91
4.1 Cosmic Background Radiation . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 93
4.1.1 Boltzmann Equation and Spectral Distortions . . . . . . . . . . . 94
4.1.2 Last-Scattering Surface . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 97
4.2 Temperature Anisotropies: Formalism . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 97
4.2.1 Spherical Harmonics . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 97
4.2.2 Gaussian Spectrum .. . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 99
4.2.3 Ergodic Hypothesis and Cosmic Variance . . . . . . . . . . . . . . . 102
4.3 Temperature Power Spectrum . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 104
4.3.1 What we Observe . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 104
4.3.2 Angular Scales . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 106
4.3.3 Sachs–Wolfe Plateau (` . 60) .. . . . . . .. . . . . . . . . . . . . . . . . . . . 110
4.3.4 Acoustic Peaks (60 . ` . 1000) .. . . .. . . . . . . . . . . . . . . . . . . . 114
4.3.5 Damping Tail (` & 1000) . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 117
4.3.6 Secondary Anisotropies .. . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 118
4.4 Cosmological Parameters and Observational Constraints .. . . . . . . . . 119
4.4.1 Shape of the Angular Spectrum and Parameters .. . . . . . . . 119
4.4.2 Primordial Spectra . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 120
4.5 Polarization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 126
4.5.1 Formalism . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 128
4.5.2 Spectra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 130
4.5.3 What we Observe . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 131
4.6 Non-Gaussianity.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 135
4.6.1 Bispectrum.. . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 135
4.6.2 Trispectrum .. . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 137
Contents ix
4.6.3 Physical Origin . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 138

4.6.4 Current Estimates . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 141
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 146
5 Inflation .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 153
5.1 Problems of the Hot Big Bang Model .. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 154
5.1.1 Planck and GUT Scale . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 155
5.1.2 Flatness Problem . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 156
5.1.3 Horizon Problem . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 157
5.1.4 Monopole Problem .. . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 158
5.1.5 Primordial Seeds Problem . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 159
5.2 Inflationary Mechanism .. . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 159
5.2.1 Solution of the Flatness Problem .. . . .. . . . . . . . . . . . . . . . . . . . 160
5.2.2 Solution of the Horizon Problem .. . . .. . . . . . . . . . . . . . . . . . . . 161
5.2.3 Solution of the Monopole Problem .. .. . . . . . . . . . . . . . . . . . . . 163
5.2.4 Solution of the Primordial Seeds Problem . . . . . . . . . . . . . . . 163
5.3 Cold Big Bang .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 163
5.3.1 Equation of State . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 164
5.3.2 Chaotic Inflation.. . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 165
5.3.3 Reheating . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 167
5.3.4 Observable Inflation .. . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 169
5.3.5 Timeline of the Early Universe . . . . . . .. . . . . . . . . . . . . . . . . . . . 170
5.4 Scalar Field: Background Dynamics . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 171
5.4.1 Hamilton–Jacobi Formalism .. . . . . . . . .. . . . . . . . . . . . . . . . . . . . 171
5.4.2 Slow-Roll Parameters . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 172
5.4.3 Inflationary Attractor .. . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 176
5.5 Models of Inflation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 178
5.5.1 Large-Field Models . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 179
5.5.2 Small-Field Models . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 183
5.5.3 Multi-field Inflation . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 185
5.6 First Glimpse of the Quantum Universe.. . . . . . . .. . . . . . . . . . . . . . . . . . . . 187
5.6.1 Decoherence .. . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 188
5.6.2 From Quantum Fields to Classical Spectra . . . . . . . . . . . . . . 189
5.6.3 Choice of Vacuum .. . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 191
5.6.4 Mukhanov–Sasaki Equation Revisited .. . . . . . . . . . . . . . . . . . 193
5.6.5 Eternal Inflation . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 197
5.7 Cosmological Spectra . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 198
5.7.1 Gaussianity . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 198
5.7.2 Linear Tensor Perturbations . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 199
5.7.3 Linear Scalar Perturbations . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 200
5.7.4 Consistency Relations and Lyth Bound . . . . . . . . . . . . . . . . . . 204
5.8 Non-Gaussianity.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 205
5.8.1 Stochastic Inflation .. . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 206
5.8.2 Multi-field Non-Gaussianity .. . . . . . . . .. . . . . . . . . . . . . . . . . . . . 213
x Contents
5.9 Observational Constraints on Inflation . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 213

5.9.1 Temperature Spectra . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 213
5.9.2 Polarization .. . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 216
5.9.3 Non-Gaussianity.. . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 217
5.10 Unsolved and New Problems . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 218
5.10.1 Graceful Entry Problem .. . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 218
5.10.2 Graceful Exit Problem . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 219
5.10.3 Trans-Planckian Problem . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 219
5.10.4 Naturalness or Model-Building Problem.. . . . . . . . . . . . . . . . 220
5.11 The Inflaton and Particle Physics .. . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 221
5.11.1 Not Only Scalars . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 221
5.11.2 Higgs Inflation.. . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 222
5.12 Supersymmetry and Supergravity . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 224
5.12.1 Global Supersymmetry.. . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 225
5.12.2 Supergravity . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 228
5.12.3 -problem .. . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 229
5.12.4 Inflation in Supergravity . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 230
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 239
6 Big-Bang Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 261
6.1 Spacetimes and Singularities . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 262
6.1.1 Globally Hyperbolic Spacetimes . . . . .. . . . . . . . . . . . . . . . . . . . 262
6.1.2 Focusing Theorems . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 265
6.1.3 Classifications of Singularities . . . . . . .. . . . . . . . . . . . . . . . . . . . 267
6.2 Singularity Theorems . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 269
6.2.1 Hawking–Penrose Theorems . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 269
6.2.2 Borde–Vilenkin Theorems .. . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 272
6.2.3 Borde–Guth–Vilenkin Theorem . . . . . .. . . . . . . . . . . . . . . . . . . . 273
6.2.4 An Undecided Issue . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 275
6.3 BKL Singularity .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 278
6.3.1 Tetrads and Bianchi Models . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 278
6.3.2 Kasner Metric . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 280
6.3.3 Generalized Kasner Metric . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 282
6.3.4 Mixmaster Dynamics . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 284
6.3.5 BKL Conjecture .. . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 291
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 295
7 Cosmological Constant Problem .. . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 301
7.1 The Problem in Field Theory . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 302
7.1.1 Spontaneous Symmetry Breaking
and Dynamical . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 302
7.1.2 Zero-Point Energy and Higher Loops .. . . . . . . . . . . . . . . . . . . 305
7.1.3 Supersymmetry and Supergravity . . . .. . . . . . . . . . . . . . . . . . . . 308
Contents xi
7.2 Other Versions of the Problem and Strategies . .. . . . . . . . . . . . . . . . . . . . 310

7.2.1 Broken Symmetries . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 310
7.2.2 The 4 Puzzle .. . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 311
7.2.3 UV or IR Problem? Strategies for a Solution . . . . . . . . . . . . 313
7.3 Quintessence.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 315
7.3.1 Tracking, Freezing and Thawing . . . . .. . . . . . . . . . . . . . . . . . . . 316
7.3.2 Periodic and Power-Law Potentials . .. . . . . . . . . . . . . . . . . . . . 321
7.3.3 Exponential and Hyperbolic Potentials . . . . . . . . . . . . . . . . . . 322
7.3.4 Inverse Power-Law Potential . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 323
7.3.5 Other Potentials . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 324
7.3.6 Quintessence and the Inflaton . . . . . . . .. . . . . . . . . . . . . . . . . . . . 325
7.3.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 326
7.4 Scalar-Tensor Theories .. . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 327
7.4.1 Motivations . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 328
7.4.2 Conformal Transformations . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 328
7.4.3 Perturbations, Quantum Theory
and Extended Inflation . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 332
7.4.4 Cosmological Constant Problem . . . . .. . . . . . . . . . . . . . . . . . . . 333
7.4.5 Experimental Bounds and Chameleon Mechanism.. . . . . 334
7.5 Higher-Order and Higher-Derivative Gravity Models .. . . . . . . . . . . . . 336
7.5.1 Motivation and Ghosts . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 336
7.5.2 General f .R/ Action . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 338
7.5.3 Palatini Formulation.. . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 340
7.5.4 Form of f .R/ . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 340
7.5.5 Horndeski Theory and Extensions.. . .. . . . . . . . . . . . . . . . . . . . 342
7.6 Other Approaches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 343
7.6.1 Varying Couplings . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 343
7.6.2 Void Models . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 344
7.6.3 Unimodular Gravity .. . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 346
7.6.4 Analogue Gravity and Condensates . .. . . . . . . . . . . . . . . . . . . . 348
7.7 Emergent Gravity.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 351
7.7.1 Rindler Observer and Null Congruences .. . . . . . . . . . . . . . . . 352
7.7.2 Dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 355
7.7.3 Holographic Equipartition . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 356
7.7.4 Cosmological Constant Problem . . . . .. . . . . . . . . . . . . . . . . . . . 360
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 369
8 The Problem of Quantum Gravity .. . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 389
8.1 Do We Need to Quantize Gravity? . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 390
8.2 Perturbative Quantum Gravity .. . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 392
8.2.1 Supergravity . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 397
8.2.2 Effective Field Theory . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 397
8.2.3 Resummed Quantum Gravity . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 398
8.3 Approaches to Quantum Gravity . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 400
xii Contents

References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 402
9 Canonical Quantum Gravity . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 407
9.1 Canonical Variables in General Relativity . . . . . .. . . . . . . . . . . . . . . . . . . . 408
9.1.1 First-Order Formalism and Parity . . . .. . . . . . . . . . . . . . . . . . . . 408
9.1.2 Hamiltonian Analysis . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 414
9.1.3 Ashtekar–Barbero Variables . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 424
9.1.4 ADM Variables .. . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 426
9.2 Wheeler–DeWitt Equation .. . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 428
9.2.1 Superspace and Quantization . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 429
9.2.2 Semi-classical States . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 432
9.2.3 Boundary Conditions . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 437
9.3 Some Features of Loop Quantum Gravity . . . . . .. . . . . . . . . . . . . . . . . . . . 439
9.4 Cosmological Constant Problem . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 442
9.4.1 Chern–Simons State. . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 443
9.4.2 as a Condensate? . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 447
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 459
10 Canonical Quantum Cosmology . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 467
10.1 Mini-superspace .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 468
10.1.1 Classical FLRW Hamiltonian . . . . . . . .. . . . . . . . . . . . . . . . . . . . 469
10.2 Wheeler–DeWitt Quantum Cosmology . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 471
10.2.1 de Sitter Solutions and Probability of Inflation . . . . . . . . . . 473
10.2.2 Massless Scalar Field and Group Averaging .. . . . . . . . . . . . 476
10.2.3 Quantum Singularity .. . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 480
10.2.4 Cosmological Constant and the Multiverse . . . . . . . . . . . . . . 482
10.2.5 Perturbations and Inflationary Observables .. . . . . . . . . . . . . 484
10.3 Loop Quantum Cosmology . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 489
10.3.1 Classical FLRW Variables and Constraints . . . . . . . . . . . . . . 490
10.3.2 Quantization and Inverse-Volume Spectrum .. . . . . . . . . . . . 493
10.3.3 Mini-superspace Parametrization .. . . .. . . . . . . . . . . . . . . . . . . . 495
10.3.4 Quantum Hamiltonian Constraint . . . .. . . . . . . . . . . . . . . . . . . . 496
10.3.5 Models with Curvature or a Cosmological Constant . . . . 500
10.3.6 Homogeneous Effective Dynamics .. .. . . . . . . . . . . . . . . . . . . . 501
10.3.7 Singularity Resolved? .. . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 508
10.3.8 Lattice Refinement: Quantum Corrections Revisited . . . . 510
10.3.9 Perturbations and Inflationary Observables .. . . . . . . . . . . . . 516
10.3.10 Inflation in Other Approaches . . . . . . . .. . . . . . . . . . . . . . . . . . . . 525
10.3.11 Is There a Bounce? .. . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 526
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 531
Contents xiii
11 Cosmology of Quantum Gravities . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 543

11.1 Hausdorff and spectral dimension .. . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 545
11.2 Asymptotic Safety .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 547
11.2.1 Framework .. . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 548
11.2.2 Cosmology . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 551
11.3 Causal Dynamical Triangulations . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 555
11.3.1 Framework .. . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 555
11.3.2 Cosmology . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 561
11.4 Spin Foams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 563
11.4.1 Framework .. . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 563
11.4.2 Cosmology . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 564
11.5 Group Field Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 569
11.5.1 Framework .. . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 569
11.5.2 Cosmology . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 572
11.6 Causal Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 584
11.6.1 Framework .. . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 584
11.6.2 Cosmology . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 588
11.7 Non-commutative Spacetimes . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 591
11.7.1 Framework .. . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 591
11.7.2 Cosmology . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 593
11.8 Non-local Gravity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 596
11.8.1 Non-locality . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 596
11.8.2 Framework .. . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 599
11.8.3 Cosmology . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 603
11.9 Comparison of Quantum-Gravity Models . . . . . .. . . . . . . . . . . . . . . . . . . . 604
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 607
12 String Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 625
12.1 Bosonic String .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 627
12.1.1 Classical Free Strings and Branes . . . .. . . . . . . . . . . . . . . . . . . . 627
12.1.2 D-Branes .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 631
12.1.3 Quantum Strings and Critical Dimension .. . . . . . . . . . . . . . . 632
12.1.4 Interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 637
12.1.5 Low-Energy Limit. . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 638
12.1.6 String Field Theory . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 640
12.2 Superstring.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 642
12.2.1 Action .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 643
12.2.2 Quantization . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 645
12.2.3 Type-I Superstring . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 646
12.2.4 Type-II Superstrings . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 647
12.2.5 Interactions and Anomaly Cancellation .. . . . . . . . . . . . . . . . . 648
12.2.6 Heterotic Superstrings . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 649
12.2.7 Massless Spectra and Low-Energy Limits . . . . . . . . . . . . . . . 651
12.2.8 Branes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 653
12.2.9 Superstring Field Theory . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 655
xiv Contents
12.3 Compactification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 656

12.3.1 T-Duality .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 656
12.3.2 Spontaneous Compactification . . . . . . .. . . . . . . . . . . . . . . . . . . . 658
12.3.3 Calabi–Yau Spaces and Orbifolds .. . .. . . . . . . . . . . . . . . . . . . . 658
12.3.4 Cycles and Fluxes . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 662
12.3.5 Moduli . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 663
12.3.6 Stacking Branes . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 668
12.3.7 Flux Compactification . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 669
12.3.8 String Theory and the Standard Model.. . . . . . . . . . . . . . . . . . 671
12.3.9 Anti-de Sitter Vacua .. . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 671
12.4 Dualities and M-Theory.. . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 678
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 683
13 String Cosmology .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 701
13.1 String Landscape . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 703
13.1.1 de Sitter Vacua.. . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 705
13.1.2 Cosmological Constant . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 709
13.1.3 Open Problems . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 714
13.2 Inflation in the Landscape.. . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 715
13.2.1 Single-Field Inflation . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 716
13.2.2 Large-Field Models and the Weak Gravity
Conjecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 718
13.2.3 Multi-field Inflation . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 719
13.2.4 Moduli Problem and -Problem .. . . . .. . . . . . . . . . . . . . . . . . . . 720
13.3 Size Moduli Inflation .. . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 720
13.3.1 Large-Volume Inflation . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 721
13.3.2 Volume-Modulus Inflation .. . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 725
13.3.3 Fluxless Inflation .. . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 727
13.4 Axion Inflation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 728
13.4.1 Racetrack Axion Inflation.. . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 729
13.4.2 Axion Valley .. . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 730
13.4.3 N-flation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 732
13.4.4 Aligned and Hierarchical Axion Inflation.. . . . . . . . . . . . . . . 734
13.4.5 Monodromy Inflation . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 735
13.4.6 Problems with Axion Inflation and Ways Out . . . . . . . . . . . 740
13.5 Slow-Roll D-Brane Inflation .. . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 742
13.5.1 Early Brane-Inflation Models .. . . . . . . .. . . . . . . . . . . . . . . . . . . . 742
13.5.2 Warped D-Brane Inflation and KLMT Model . . . . . . . . . . 744
13.5.3 Cosmological KLMT Dynamics . . . .. . . . . . . . . . . . . . . . . . . . 748
13.5.4 Refinements and Related Models. . . . .. . . . . . . . . . . . . . . . . . . . 750
13.5.5 Why the Tensor Spectrum Is Small . .. . . . . . . . . . . . . . . . . . . . 753
13.6 DBI Inflation.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 754
13.6.1 Setting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 754
13.6.2 UV Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 755
13.6.3 IR Model.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 758
Contents xv
13.7 Other Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 761

13.7.1 Braneworld . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 761
13.7.2 Cosmological Tachyon .. . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 765
13.7.3 Modified Gravity . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 767
13.7.4 Non-local Models . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 769
13.7.5 Pre-Big-Bang and Dilaton Cosmology .. . . . . . . . . . . . . . . . . . 770
13.7.6 String-Gas Cosmology .. . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 772
13.7.7 Cyclic Ekpyrotic Universe .. . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 776
13.8 Inflation and Alternatives: Compact Summary .. . . . . . . . . . . . . . . . . . . . 783
13.9 Big-Bang Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 786
13.9.1 Big Bang in String Theory .. . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 786
13.9.2 Classical Billiards: Living with the Singularity.. . . . . . . . . 788
13.9.3 Quantum Billiards: Avoiding the Singularity? .. . . . . . . . . . 792
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 795
14 Perspective.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 823
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 826
Index . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 829
Chapter 1
Introduction
Nel suo profondo vidi che s’interna

legato con amore in un volume,
ciò che per l’universo si squaderna;
sustanze e accidenti e lor costume,
quasi conflati insieme, per tal modo
che ciò ch’i’ dico è un semplice lume.
— Dante, Paradiso, XXXIII, 85–90
In its depths enclosed I saw
bound with love in a volume,
that which enfolds in the universe;
substances and accidents and their bonds,
almost united together, so that
what I say is a simple glimpse.
Contents
1.1 Micro from Macro. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Outline of the Topics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 About Citations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.4 Conventions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.5 Measure Units. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.1 Micro from Macro
According to modern data, the large-scale structure of the Universe and the
anisotropies of the cosmic microwave background can be explained by an early
stage of accelerated expansion, called inflation, driven by an effective cosmological
constant. The latter is often identified with a scalar field (generically dubbed
“inflaton”) slowly rolling down its potential. The same dynamical mechanism may
also provide an explanation for the present phase of acceleration.
© Springer International Publishing Switzerland 2017 1

G. Calcagni, Classical and Quantum Cosmology, Graduate Texts in Physics,
DOI 10.1007/978-3-319-41127-9_1
2 1 Introduction
Inflation was originally devised for solving a number of problems afflicting

the hot big bang model, in particular the flatness or entropy problem (Why is
the Universe flat? Why does it have such a high entropy?), the horizon problem
(Why are distant, causally disconnected regions in thermal equilibrium?) and the
monopole problem (Where are the topological defects we would expect emerging
from cosmological phase transitions?). However, the reasons of the success of
inflation rely on an aspect which is far more than a side bonus. In fact, an
immediate consequence of this scenario is that cosmological large-scale structures
were originated by the exponential dilatation of quantum fluctuations of the inflaton
up to macroscopic scales. In a coniunctio oppositorum, the study of the macrocosm
allows us to investigate the microcosm. Thus, cosmological observations are
complementary to those with ground-based particle accelerators and they clarify
both the composition and geometric structure of the Universe and the primordial role
of the known elementary interactions. Inflation is a promising playground whereon
to understand the high-energy, small-scale behaviour of Nature and, in particular,
gravity at quantum scales. And how does gravity behave at such scales? Is it a
quantum force or not?
During the last years, quantum gravity has been receiving a great amount of
attention from the community of theoreticians. The driving motivation, familiar to
anyone who has tried their fortune in this broad subject, is to realize a consistent,
predictive merging of general relativity with quantum mechanics. The programme
can be carried out in various forms, from ambitious theories of everything (such
as string theory) where all forces are unified, to more minimalistic approaches
aiming to quantize gravity alone, not to mention proposals where gravity emerges
as an effective phenomenon from more fundamental degrees of freedom.1 A
problem endemic in most of these scenarios is their difficulty in making contact
with observations. This stems from the highly technical nature of the theoretical
frameworks, where the notions of conventional geometry and matter, continuum
spacetime, general covariance and physical observables are typically deformed,
modified, or disappear altogether. The lack of experimental feedback makes it
quite difficult to discriminate among different models and to characterize them as
falsifiable. It is natural to turn to cosmology in an attempt to bridge this gap and
advance our knowledge.
The early Universe is not the only context wherein to accept these challenges:
also some unexplained facts concerning the composition of the late-time and
present Universe might require new physics. The state of the art in cosmology
has been experiencing an interesting phase. On one hand, there are a number of
1
Often, but not always, we will reserve the umbrella name “quantum gravity” to the second type
of approaches, where matter is introduced by hand. This is done only for convenience in certain
chapters but it not a clear-cut terminology. In a sense, string theory is a theory of quantum gravity
(and it was presented as such in many early papers), although it may be argued that it is also
a realization of emergent gravity. Group field theory will be included among quantum-gravity
theories (Chap. 11) but it is also a theory of everything, since it aims to derive both geometry and
matter degrees of freedom from a unified structure. And so on.
1.1 Micro from Macro 3
phenomenological models which have been formulated in great detail and fit the
available data. On the other hand, observational data obtained in ground-based
observatories and space missions (satellites and balloon probes), have the task and
possibility to discriminate among the many cosmic theories on the market. The
predictions of what we will come to know as the cold big bang model are in excellent
agreement with observations but, despite the progress in precision cosmology, we
know less than 5 % of the content of the present universe. The rest is divided into two
components which are dubbed “dark matter” and “dark energy” in homage to our
ignorance. While dark matter may be understood within models of particle physics,
the nature of dark energy is more elusive. Is it a cosmological constant? Why is
its energy density so small but non-zero? Why did the universe start accelerating
only in its recent history? Many believe that a true solution of the cosmological
constant and dark energy problems is deeply tied with a consistent formulation of
a quantum theory of gravity. In other words, the cosmological constant might be a
yet unrecognized manifestation of physics beyond the Standard Model. In turn, this
theory should also be able to address other questions: Can we avoid the big-bang
singularity and, more generally, spacetime singularities? How did inflation happen?
What is the inflaton?
It might as well be that most or perhaps all these questions are being formulated
in a wrong way. Progress in the knowledge of a unified quantum theory of Nature,
including both particle physics and general relativity in some effective limits, should
help answering these questions or, at least, asking them correctly. We are of course
working upon the hope, motivated by aesthetic arguments, that such a theory does
exist.
This book discusses models stemming from classical and quantum theories at the
frontier of modern cosmology. Applications of quantum mechanics and quantum
field theory range from the theory of inflationary perturbations to mini-superspace
quantizations, from the big-bang problem in loop quantum cosmology to inflation
in string theory, down to the relation between a quantum theory of gravity and
the cosmological constant problem. We will review how the early universe can be
described within candidate theories of quantum gravity, in particular string theory
and loop quantum gravity. However, we shall only touch the tip of the iceberg
and will not embark on an in-depth study of these theories. We will consider only
cosmological phenomenology and the main problems a complete model of the early
universe should be able to face. We forewarn the reader that none of the proposals
advanced to tackle issues such as the big-bang and the cosmological constant
problem is, for several reasons, commonly regarded as satisfactory. Nonetheless,
they constitute attempts of resolution of these problems and, as such, they deserve
attention.
The text is meant for advanced undergraduate and graduate students planning
to work on cosmology of the early universe, theories of quantum gravity or
string cosmology, but also for specialists in quantum gravity or string theory who
would not dislike a cosmological venture. On the other hand, the more traditional
cosmologist can find an updated non-technical overview of modern approaches
to inflation, the big-bang and the cosmological constant problems, theories of
4 1 Introduction
quantum gravity at large and their cosmology, string theory and string cosmology.
Researchers looking for a reference book with extensive bibliographic resources
will also find benefit in the reading. Most chapters are accompanied by partially or
fully solved problems.
For these reasons, this is not a standard textbook on cosmology. Many important
branches of cosmology will either be reviewed qualitatively (cosmic microwave
background anisotropies) or just mentioned briefly here and there (dark matter,
structure formation, observational and statistical techniques, astrophysical pro-
cesses). The reader can find further material in the main bibliography, although not
all themes are covered by extant books or review articles.
1.2 Outline of the Topics
1. Hot big bang model. Cosmological standard model; content of the universe;
thermal history. Textbooks: [1–3] (general relativity), [4] (Chaps. 2–3 on
general relativity), [5–8] (cosmology).
2. Cosmological perturbations. Linear and non-linear cosmological perturba-
tions; separate universe approach; Gaussian random fields. Textbooks: [5]
(linear perturbations, separate universe and ıN formalism). Reviews: [9, 10]
(linear perturbations).
3. Cosmic microwave background. CMB primer; Gaussian and non-
Gaussian spectra; polarization. Textbooks: [5, 11] (CMB). Reviews: [12–15]
(anisotropies), [16, 17] (polarization), [18–20] (non-Gaussianity).
4. Inflation. Problems of the cosmological standard model; standard scalar-
field models; classical and quantum dynamics; model building; spectra and
non-Gaussianity; open issues. Textbooks: [5, 6, 21] (basic and multi-field
inflation, perturbations), [22] (quantum field theory in curved spacetime), [23]
(eternal inflation). Reviews: [24] (particle-physics models of inflation), [25]
(inflationary perturbations), [26] (more updated review on inflation).
5. Big-bang problem. Globally hyperbolic spacetimes; classification of singulari-
ties; singularity theorems; mixmaster universe and BKL conjecture. Textbooks:
[27] (Chap. 8 on singularity theorems, Chap. 10 on big-bang singularity).
Reviews: [28] (BKL singularity).
6. Cosmological constant problem. The problem in quantum field theory;
quintessence; acceleration from modified gravity and alternative models. Text-
books: [29] (dark energy). Reviews: [30–34] ( problem), [35–38] (dark
energy), [39–42] (scalar-tensor, f .R/ and higher-order models).
7. The problem of quantum gravity. Do we need to quantize gravity?; perturba-
tive gravity; approaches to quantum gravity.
8. Canonical quantum gravity. Hamiltonian formalism in Ashtekar–Barbero and
ADM variables; Wheeler–DeWitt equation; cosmological constant problem in
canonical quantum gravity. Textbooks: [4] (Chap. 4 on first-order formalism,
Chaps. 6–9 on loop quantum gravity and spin foams), [43] (Chaps. 1 and
1.2 Outline of the Topics 5
2 on constrained Hamiltonian systems), [44] (mathematical introduction to

canonical gravity and loop quantum gravity), [45] (canonical quantum gravity).
Reviews: [46, 47] (loop quantum gravity), [48] (Wheeler–DeWitt equation).
9. Canonical quantum cosmology. Mini-superspace; Wheeler–DeWitt quantum
cosmology; loop quantum cosmology; big-bang and cosmological constant
problems revisited; inhomogeneities and inflation. Textbooks: [49]. Reviews:
[48, 50] (Wheeler–DeWitt cosmology), [51–53] (loop quantum cosmology).
10. Cosmology of quantum gravities. Asymptotic safety; causal dynamical tri-
angulation; spin foams; group field theory; causal sets; non-commutative
spacetimes; non-local gravity. Review references will be given at the beginning
of Chap. 11.
11. String theory. Overview of string theory; bosonic string; superstring; low-
energy limits; branes and fluxes; compactification; Calabi–Yau spaces and
orbifolds; moduli stabilization; dualities; M-theory. Textbooks: [54–57] (string
theory). Reviews: [58] (basics of string theory), [59] (open strings), [60, 61]
(string field theory), [62] (M-theory), [63–67] (flux compactification).
12. String cosmology. String landscape; de Sitter vacua; cosmological constant
problem; KKLT uplifting scenarios; large-volume uplifting scenarios; infla-
tion in string theory; moduli inflation; axion inflation; slow-roll D-brane
inflation; DBI inflation; braneworld cosmology; cosmological tachyon; higher-
order string gravity; non-local string cosmology; pre-big-bang cosmology;
string-gas cosmology; ekpyrotic universe; big-bang problem in string theory;
cosmological billiards. Books: [68] (string inflation). Reviews: [63–67] (string
landscape), [69–74] (moduli and D-brane inflation), [75] (cosmic strings),
[76] (pre-landscape perspective on the cosmological constant), [71] (big-bang
problem), [77] (cosmological billiards).
The list of cosmological models in Chaps. 7 and 11 is by no means complete.
Notable absentees include, among others, the Dvali–Gabadadze–Porrati scenario
[78–80], massive and bimetric gravity [81–89] and Hořava–Lifshitz gravity [90–95]
(although the latter is mentioned in a couple of places in Chap. 11).
An elementary knowledge of quantum field theory is a helpful pre-requisite
[96–99]. Important branches of theoretical physics such as supersymmetry and
supergravity will be touched upon in the context of inflation (Sect. 5.12), the
cosmological constant problem (Sect. 7.1.3), renormalization properties of pertur-
bative gravity (Sect. 8.2) and string theory (Chaps. 12 and 13). To get a deeper
insight into supersymmetry, the interested reader can consult specialized books
[100, 101] and reviews [102–106]. Applications of supersymmetry to canonical
quantum cosmology can be found in [107, 108].
The recent advances in the experimental determination of natural observables
have been spectacular. It is impressive to think back at the number of breakthroughs
we witnessed in physics during the years 2012–2016: the discovery of the Higgs
boson at the LHC, culminating a never-ending chain of achievements for the
Standard Model of particle interactions; the great improvement in our knowledge of
the cosmic microwave background temperature spectrum first by WMAP and then
6 1 Introduction
by PLANCK, and the ensuing restriction of the parameter space of inflation; a false
alarm about the detection of gravitational waves, followed some months later by the
actual discovery of such waves by Advanced LIGO, from the first black-hole merger
ever observed —not only a major success for Einstein’s theory of general relativity,
but also the first time we could literally hear the voice of gravity. Through this chain
of events, I found myself in the embarrassing situation of rewriting some parts of
the book time and again. All the above observations carry important consequences
for the discrimination of astro-particle and cosmological models and, hence, for our
investigation of quantum-gravity and string phenomenology. Most probably, and
hopefully, more of this ongoing story will have to be revised in the next few years.
1.3 About Citations
Whenever practical and up to exceptions indicated in the text, an exhaustive list of

references (updated November 2016) has been provided in order to offer ample
documentation for those readers seriously interested in learning more about a
subject to work on it actively. The simplified presentation of many models in the
book, especially in string cosmology, does not do justice to the tremendous mole of
work behind these models.
The main disadvantage of this choice is that the most important references
marking historical landmarks may be lost, at a first reading, in a sea of data quite
often not directly used in the book. The main tools to obviate to such a problem
have been, on one hand, to cite foundational papers and stress their role separately
from the others and, on the other hand, to order the references in each group by
submission date. For a few topics, the group of references is disproportionately large
with respect to the importance of such topics in the economy of the book. Perhaps
one factor that influenced this outcome is the child’s insatiable curiosity many of
us share when starting to learn about the scope and story of some research theme
we come across. A curiosity which makes us ask: How did it continue? How will it
end?
This system, with all its flaws, omissions and imprecisions the author may have
introduced, may be best suited for a reader who enjoys a critical exploration of the
literature.
1.4 Conventions
:D and D: define the symbol or function on the side of the colon: e.g., f .x/ :D
ln.1 C x/2 D: g.x/. The equality D is used in a standard way, when two things are
the same (logically, this includes also objects already defined): f .x/ D ln.1 C x/2 D
2 ln.1 C x/. Depending on the context, is an identity (1 1, f .x/ g.x/) or
an equivalence relation (a b). Except in Chap. 9, means approximation of a
1.5 Measure Units 7
function or a number to a number (e.g., 3:14, N 60 e-folds, or (1.2), (1.3)

and (1.4) below). / and denote proportionality between functions, respectively
exact and asymptotic, while ' means “asymptotically equal to” and is employed
between functions. For instance, when x 1 we have g.x/ f .x/ D ln.1 C x/2 '
x.2 x/ ' 2x / x 0. To make implicit use of the asymptotic expansion, we
would write f .x/ x. Sometimes, also appears in order-of-magnitude estimates,
e.g., if 3:2 102 we will say that D O.102 / or 102 . Finally, Š means
isomorphism between sets or groups. The mathematician will forgive me if I have
not been completely consistent with the above usage of symbols.
Greek indices ; ; : : : D 0; 1; : : : ; D 1 from the middle of the alphabet
denote spacetime directions, while Greek indices ˛; ˇ; : : : from the beginning of
the alphabet label spatial directions. Whenever the vielbein formalism is employed,
Latin indices a; b; : : : from the beginning of the alphabet denote directions in the
internal (tangent) space; we reserve the middle-alphabet letters i; j; : : : to the spatial
directions. In Chaps. 12 and 13, ; ; : : : are labels of 26- or 10-dimensional target
spacetime, reserving M; N; D 0; 1; : : : ; 10 for dimensions in M-theory. In the
same chapters, a; b; : : : indicate world-sheet and world-volume coordinates or, in
the case of compactifications, the coordinates in the non-compact four-dimensional
manifold. Directions in Calabi–Yau spaces are labelled by l; m; : : : . Exceptions to
these rules may occur, in which case awareness of the context will avoid confusion.
Unless stated otherwise,
P the Einstein summation notation for repeated indices is
employed, a b D a b .
We choose the “mostly plus” spacetime signature .; C; ; C/.
1.5 Measure Units
A length unit commonly used in cosmology is the megaparsec (Mpc):
1 Mpc D 3:26156 106 light years 3:0857 1022 m :
To describe phenomena at quantum scales, the Planck length is better suited. In units
of mass m, length l and time t, the scaling of the speed of light, gravitational constant
and Planck constant in D dimensions is, respectively, Œc D lt1 , ŒG D lD1 t2 m1
and Œ„ D l2 t1 m. The Planck length, time and mass are then
D2
1 D2
1 D2
1
„G „G „D3 c5D
lPl :D ; tPl :D ; mPl :D :
c3 cDC1 G
(1.1)
8 1 Introduction
In four dimensions,
r
„G
lPl D 1:6163 1035 m ; (1.2)
c3
r
„G
tPl D 5:3912 1044 s ; (1.3)
c5
r
„c
mPl D 1:2209 1019 GeVc2 ; (1.4)
G
p
to which we add the reduced Planck mass MPl :D mPl = 8 2:435 1018 GeV
and the Planck temperature
TPl 1:4 1032 K :
Here we used the conversion formula

TeV
TK D 1:1605 104 TeV ;
kB
where kB 8:6173 105 eV K1 is Boltzmann constant.

We shall adopt natural units in which the speed of light and the Planck constant
are dimensionless:
c D 1; „ D 1:
These units will be occasionally restored.
References
1. S. Weinberg, Gravitation and Cosmology (Wiley, New York, 1972)

2. C.W. Misner, K.S. Thorne, J.A. Wheeler, Gravitation (Freeman, New York, 1973)
3. L.D. Landau, E.M. Lifshitz, The Classical Theory of Fields (Butterworth–Heinemann,
London, 1980)
4. C. Rovelli, Quantum Gravity (Cambridge University Press, Cambridge, 2007)
5. D.H. Lyth, A.R. Liddle, The Primordial Density Perturbation (Cambridge University Press,
Cambridge, 2009)
6. S. Weinberg, Cosmology (Oxford University Press, Oxford, 2008)
7. V. Mukhanov, Physical Foundations of Cosmology (Cambridge University Press, Cambridge,
2005)
8. S. Dodelson, Modern Cosmology (Academic Press, San Diego, 2003)
9. H. Kodama, M. Sasaki, Cosmological perturbation theory. Prog. Theor. Phys. Suppl. 78, 1
(1984)
10. V.F. Mukhanov, H.A. Feldman, R.H. Brandenberger, Theory of cosmological perturbations.
Phys. Rep. 215, 203 (1992)
References 9
11. R. Durrer, The Cosmic Microwave Background (Cambridge University Press, Cambridge,
2008)
12. M.J. White, D. Scott, J. Silk, Anisotropies in the cosmic microwave background. Ann. Rev.
Astron. Astrophys. 32, 319 (1994)
13. W.T. Hu, Wandering in the Background: A CMB Explorer. Ph.D. thesis, UC Berkeley,
Berkeley (1995). [arXiv:astro-ph/9508126]
14. W. Hu, N. Sugiyama, J. Silk, The physics of microwave background anisotropies. Nature 386,
37 (1997). [arXiv:astro-ph/9604166]
15. W. Hu, S. Dodelson, Cosmic microwave background anisotropies. Ann. Rev. Astron.
Astrophys. 40, 171 (2002). [arXiv:astro-ph/0110414]
16. M. Kamionkowski, A. Kosowsky, A. Stebbins, Statistics of cosmic microwave background
polarization. Phys. Rev. D 55, 7368 (1997). [arXiv:astro-ph/9611125]
17. W. Hu, M.J. White, A CMB polarization primer. New Astron. 2, 323 (1997).
[arXiv:astro-ph/9706147]
18. N. Bartolo, E. Komatsu, S. Matarrese, A. Riotto, Non-Gaussianity from inflation: theory and
observations. Phys. Rep. 402, 103 (2004). [arXiv:astro-ph/0406398]
19. E. Komatsu et al., Non-Gaussianity as a probe of the physics of the primordial universe and
the astrophysics of the low redshift universe. arXiv:0902.4759
20. M. Liguori, E. Sefusatti, J.R. Fergusson, E.P.S. Shellard, Primordial non-Gaussianity and
bispectrum measurements in the cosmic microwave background and large-scale structure.
Adv. Astron. 2010, 980523 (2010). [arXiv:1001.4707]
21. A.D. Linde, Particle Physics and Inflationary Cosmology (Harwood, Chur, 1990).
[arXiv:hep-th/0503203]
22. N.D. Birrell, P.C.W. Davies, Quantum Fields in Curved Space (Cambridge University Press,
Cambridge, 1982)
23. S. Winitzki, Eternal Inflation (World Scientific, Singapore, 2009)
24. D.H. Lyth, A. Riotto, Particle physics models of inflation and the cosmological density
perturbation. Phys. Rep. 314, 1 (1999). [arXiv:hep-ph/9807278]
25. A. Riotto, Inflation and the theory of cosmological perturbations. arXiv:hep-ph/0210162
26. A.D. Linde, Inflationary cosmology. Lect. Notes Phys. 738, 1 (2008). [arXiv:0705.0164]
27. S.W. Hawking, G.F.R. Ellis, The Large Scale Structure of Space-Time (Cambridge University
Press, Cambridge, 1973)
28. G. Montani, M.V. Battisti, R. Benini, G. Imponente, Classical and quantum features of the
mixmaster singularity. Int. J. Mod. Phys. A 23, 2353 (2008). [arXiv:0712.3008]
29. L. Amendola, S. Tsujikawa, Dark Energy (Cambridge University Press, Cambridge, 2010)
30. S. Weinberg, The cosmological constant problem. Rev. Mod. Phys. 61, 1 (1989)
31. V. Sahni, A.A. Starobinsky, The case for a positive cosmological -term. Int. J. Mod. Phys.
D 9, 373 (2000). [arXiv:astro-ph/9904398]
32. S.M. Carroll, The cosmological constant. Living Rev. Relat. 4, 1 (2001)
33. S. Nobbenhuis, Categorizing different approaches to the cosmological constant problem.
Found. Phys. 36, 613 (2006). [arXiv:gr-qc/0411093]
34. J. Martin, Everything you always wanted to know about the cosmological constant problem
(but were afraid to ask). C. R. Phys. 13, 566 (2012). [arXiv:1205.3365]
35. T. Padmanabhan, Cosmological constant: the weight of the vacuum. Phys. Rep. 380, 235
(2003). [arXiv:hep-th/0212290]
36. P.J.E. Peebles, B. Ratra, The cosmological constant and dark energy. Rev. Mod. Phys. 75, 559
(2003). [arXiv:astro-ph/0207347]
37. E.J. Copeland, M. Sami, S. Tsujikawa, Dynamics of dark energy. Int. J. Mod. Phys. D 15,
1753 (2006). [arXiv:hep-th/0603057]
38. S. Tsujikawa, Quintessence: a review. Class. Quantum Grav. 30, 214003 (2013).
[arXiv:1304.1961]
39. T.P. Sotiriou, V. Faraoni, f .R/ theories of gravity. Rev. Mod. Phys. 82, 451 (2010).
[arXiv:0805.1726]
40. A. De Felice, S. Tsujikawa, f .R/ theories. Living Rev. Relat. 13, 3 (2010)
10 1 Introduction
41. S. Tsujikawa, Modified gravity models of dark energy. Lect. Notes Phys. 800, 99 (2010).
[arXiv:1101.0191]
42. T. Clifton, P.G. Ferreira, A. Padilla, C. Skordis, Modified gravity and cosmology. Phys. Rep.
513, 1 (2012). [arXiv:1106.2476]
43. M. Henneaux, C. Teitelboim, Quantization of Gauge Systems (Princeton University Press,
Princeton, 1994)
44. T. Thiemann, Modern Canonical Quantum General Relativity (Cambridge University
Press, Cambridge, 2007); Introduction to modern canonical quantum general relativity.
arXiv:gr-qc/0110034
45. C. Kiefer, Quantum Gravity (Oxford University Press, Oxford, 2012)
46. T. Thiemann, Quantum gravity: from theory to experimental search. Lect. Notes Phys. 631,
412003 (2003). [arXiv:gr-qc/0210094]
47. A. Ashtekar, J. Lewandowski, Background independent quantum gravity: a status report.
Class. Quantum Grav. 21, R53 (2004). [arXiv:gr-qc/0404018]
48. D.L. Wiltshire, An introduction to quantum cosmology, in Cosmology: The Physics of the
Universe, ed. by B. Robson, N. Visvanathan, W.S. Woolcock (World Scientific, Singapore,
1996). [arXiv:gr-qc/0101003]
49. M. Bojowald, Quantum cosmology. Lect. Notes Phys. 835, 1 (2011)
50. C. Kiefer, B. Sandhöfer, Quantum cosmology, in Beyond the Big Bang, ed. by R. Vaas
(Springer, Berlin, 2008). [arXiv:0804.0672]
51. M. Bojowald, Loop quantum cosmology. Living Rev. Relat. 11, 4 (2008)
52. A. Ashtekar, P. Singh, Loop quantum cosmology: a status report. Class. Quantum Grav. 28,
213001 (2011). [arXiv:1108.0893]
53. K. Banerjee, G. Calcagni, M. Martín-Benito, Introduction to loop quantum cosmology.
SIGMA 8, 016 (2012). [arXiv:1109.6801]
54. M.B. Green, J.H. Schwarz, E. Witten, Superstring Theory (Cambridge University Press,
Cambridge, 1987)
55. J. Polchinski, String Theory (Cambridge University Press, Cambridge, 1998)
56. K. Becker, M. Becker, J.H. Schwarz, String Theory and M-Theory (Cambridge University
Press, Cambrdige, 2007)
57. B. Zwiebach, A First Course in String Theory (Cambridge University Press, Cambridge,
2009)
58. D. Tong, String theory. arXiv:0908.0333
59. C. Angelantonj, A. Sagnotti, Open strings. Phys. Rep. 371, 1 (2002); Erratum-ibid. 376, 339
(2003). [arXiv:hep-th/0204089]
60. K. Ohmori, A review on tachyon condensation in open string field theories.
arXiv:hep-th/0102085
61. E. Fuchs, M. Kroyter, Analytical solutions of open string field theory. Phys. Rep. 502, 89
(2011). [arXiv:0807.4722]
62. M.J. Duff, M theory (the theory formerly known as strings). Int. J. Mod. Phys. A 11, 5623
(1996). [arXiv:hep-th/9608117]
63. M. Graña, Flux compactifications in string theory: a comprehensive review. Phys. Rep. 423,
91 (2006). [arXiv:hep-th/0509003]
64. M.R. Douglas, S. Kachru, Flux compactification. Rev. Mod. Phys. 79, 733 (2007).
65. R. Blumenhagen, B. Körs, D. Lüst, S. Stieberger, Four-dimensional string compactifications
with D-branes, orientifolds and fluxes. Phys. Rep. 445, 1 (2007). [arXiv:hep-th/0610327]
66. F. Denef, M.R. Douglas, S. Kachru, Physics of string flux compactifications. Ann. Rev. Nucl.
Part. Sci. 57, 119 (2007). [arXiv:hep-th/0701050]
67. F. Denef, Course 12 – Lectures on constructing string vacua. Les Houches 87, 483 (2008).
[arXiv:0803.1194]
68. D. Baumann, L. McAllister, Inflation and String Theory (Cambridge University Press,
Cambridge, 2015). [arXiv:1404.2601]
References 11
69. S.-H.H. Tye, Brane inflation: string theory viewed from the cosmos. Lect. Notes Phys. 737,
949 (2008). [arXiv:hep-th/0610221]
70. R. Kallosh, On inflation in string theory. Lect. Notes Phys. 738, 119 (2008).
71. L. McAllister, E. Silverstein, String cosmology: a review. Gen. Relat. Grav. 40, 565 (2008).
[arXiv:0710.2951]
72. M. Cicoli, F. Quevedo, String moduli inflation: an overview. Class. Quantum Grav. 28,
204001 (2011). [arXiv:1108.2659]
73. C.P. Burgess, L. McAllister, Challenges for string cosmology. Class. Quantum Grav. 28,
204002 (2011). [arXiv:1108.2660]
74. C.P. Burgess, M. Cicoli, F. Quevedo, String inflation after Planck 2013. JCAP 1311, 003
(2013). [arXiv:1306.3512]
75. E.J. Copeland, L. Pogosian, T. Vachaspati, Seeking string theory in the cosmos. Class.
Quantum Grav. 28, 204009 (2011). [arXiv:1105.0207]
76. E. Witten, The cosmological constant from the viewpoint of string theory.
arXiv:hep-ph/0002297
77. M. Henneaux, D. Persson, P. Spindel, Spacelike singularities and hidden symmetries of
gravity. Living Rev. Relat. 11, 1 (2008)
78. G.R. Dvali, G. Gabadadze, M. Porrati, 4D gravity on a brane in 5D Minkowski space. Phys.
Lett. B 485, 208 (2000). [arXiv:hep-th/0005016]
79. D. Gorbunov, K. Koyama, S. Sibiryakov, More on ghosts in DGP model. Phys. Rev. D 73,
044016 (2006). [arXiv:hep-th/0512097]
80. W. Fang, S. Wang, W. Hu, Z. Haiman, L. Hui, M. May, Challenges to the DGP model from
horizon-scale growth and geometry. Phys. Rev. D 78, 103509 (2008). [arXiv:0808.2208]
81. S. Deser, R. Jackiw, S. Templeton, Topologically massive gauge theories. Ann. Phys. (N.Y.)
140, 372 (1982)
82. E.A. Bergshoeff, O. Hohm, P.K. Townsend, Massive gravity in three dimensions. Phys. Rev.
Lett. 102, 201301 (2009). [arXiv:0901.1766]
83. C. de Rham, G. Gabadadze, Generalization of the Fierz–Pauli action. Phys. Rev. D 82, 044020
(2010). [arXiv:1007.0443]
84. C. de Rham, G. Gabadadze, A.J. Tolley, Resummation of massive gravity. Phys. Rev. Lett.
106, 231101 (2011). [arXiv:1011.1232]
85. S.F. Hassan, R.A. Rosen, On non-linear actions for massive gravity. JHEP 1107, 009 (2011).
[arXiv:1103.6055]
86. S.F. Hassan, R.A. Rosen, Resolving the ghost problem in non-linear massive gravity. Phys.
Rev. Lett. 108, 041101 (2012). [arXiv:1106.3344]
87. G. D’Amico, C. de Rham, S. Dubovsky, G. Gabadadze, D. Pirtskhalava, A.J. Tolley, Massive
cosmologies. Phys. Rev. D 84, 124046 (2011). [arXiv:1108.5231]
88. S.F. Hassan, R.A. Rosen, Bimetric gravity from ghost-free massive gravity. JHEP 1202, 126
(2012). [arXiv:1109.3515]
89. A. De Felice, A.E. Gümrükçüoğlu, C. Lin, S. Mukohyama, On the cosmology of massive
gravity. Class. Quantum Grav. 30, 184004 (2013). [arXiv:1304.0484]
90. P. Hořava, Quantum gravity at a Lifshitz point. Phys. Rev. D 79, 084008 (2009).
[arXiv:0901.3775]
91. G. Calcagni, Cosmology of the Lifshitz universe. JHEP 0909, 112 (2009). [arXiv:0904.0829]
92. E. Kiritsis, G. Kofinas, Hořava–Lifshitz cosmology. Nucl. Phys. B 821, 467 (2009).
[arXiv:0904.1334]
93. R. Iengo, J.G. Russo, M. Serone, Renormalization group in Lifshitz-type theories. JHEP
0911, 020 (2009). [arXiv:0906.3477]
94. D. Blas, O. Pujolas, S. Sibiryakov, Consistent extension of Hořava gravity. Phys. Rev. Lett.
104, 181302 (2010). [arXiv:0909.3525]
95. P. Hořava, C.M. Melby-Thompson, General covariance in quantum gravity at a Lifshitz point.
Phys. Rev. D 82, 064027 (2010). [arXiv:1007.2410]
96. P. Ramond, Field Theory: A Modern Primer (Westview Press, Boulder, 1997)
12 1 Introduction
97. S. Weinberg, The Quantum Theory of Fields, vol. I (Cambridge University Press, Cambridge,
1995)
98. S. Weinberg, The Quantum Theory of Fields, vol. II (Cambridge University Press, Cambridge,
1997)
99. M. Srednicki, Quantum Field Theory (Cambridge University Press, Cambridge, 2007)
100. P.C. West, Introduction to Supersymmetry and Supergravity (World Scientific, Singapore,
1990)
101. J. Wess, J. Bagger, Supersymmetry and Supergravity (Princeton University Press, Princeton,
1992)
102. P. Van Nieuwenhuizen, Supergravity. Phys. Rep. 68, 189 (1981)
103. S.J. Gates, M.T. Grisaru, M. Roček, W. Siegel, Superspace, or one thousand and one lessons
in supersymmetry. Front. Phys. 58, 1 (1983). [arXiv:hep-th/0108200]
104. J.D. Lykken, Introduction to supersymmetry. arXiv:hep-th/9612114
105. S.P. Martin, A supersymmetry primer, in Perspectives on Supersymmetry, ed. by G.L. Kane
(World Scientific, Singapore, 1998). [arXiv:hep-ph/9709356]
106. A. Van Proeyen, Tools for supersymmetry. arXiv:hep-th/9910030
107. P.D. D’Eath, Supersymmetric Quantum Cosmology (Cambridge University Press, Cambridge,
2005)
108. P. Vargas Moniz, Quantum Cosmology – The Supersymmetric Perspective. Lect. Notes Phys.
803, 1 (2010); Lect. Notes Phys. 804, 1 (2010)
Chapter 2
Hot Big Bang Model
This to attain, whether heav’n move or earth,

Imports not, if thou reckon right; the rest
From man or angel the great Architect
Did wisely to conceal, and not divulge
His secrets to be scanned by them who aught
Rather admire; or if they list to try
Conjecture, he his fabric of the heav’ns
Hath left to their disputes, perhaps to move
His laughter at their quaint opinions wide
Hereafter, when they come to model heav’n
And calculate the stars . . .
— Milton, Paradise Lost, VIII, 70–80
Contents
2.1 Cosmic Expansion and Cosmological Principle. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.1.1 The Universe at Large Scales. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Background. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.2 Einstein and Continuity Equations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.2.1 Energy Conditions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.3 Perfect Fluid. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.3.1 Scalar Field. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.4 Friedmann Equations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.5 Content of the Universe. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.5.1 Dust and Radiation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.5.2 Hot Big Bang and the Big-Bang Problem. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
2.5.3 Dark Energy and the cosmological Constant
Problem. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.5.4 Spatial Curvature and Topology. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
2.6 An Obscure Big Picture. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
2.7 Problems and Solutions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58

DOI 10.1007/978-3-319-41127-9_2
14 2 Hot Big Bang Model
2.1 Cosmic Expansion and Cosmological Principle
2.1.1 The Universe at Large Scales
Cosmology is the study of Nature at very large scales. Conventionally, “large scales”
span a range of about 1–8000 Mpc, from our local group of galaxies to the most
distant light we can possibly detect (Fig. 2.1). Phenomena occurring within a galaxy
(ours or another) and associated with galactic media, stars, supernovæ, black holes,
gamma-ray bursts and so on are also part of cosmology, in so far as they determine
relative distances, ages, composition and gravitational properties of local patches of
the universe, as well as important information on elementary particle physics and
gravity.
Fig. 2.1 Cosmological scales (Source: adaptation of a figure by C. Schimd)

2.1 Cosmic Expansion and Cosmological Principle 15
The large-scale structure of the universe has been studied by several experiments,
including the 2-degree Field Galaxy Redshift Survey (2dFGRS) [1], the 6-degree
Field Galaxy Survey (6dFGS) [2, 3] and the Sloan Digital Sky Survey (SDSS)
[4, 5], later combined with the Supernova Legacy Survey (SNLS) [6, 7]. The first
mapped about 4 % of the sky, recording over 220;000 galaxies and 100 quasars as
far away as 800 Mpc (see Problem 2.7). The 6dFGS covered a fraction of the sky ten
times larger, sampling 125;000 galaxies. The database of the SDSS-III survey covers
about one quarter of the sky, over 1;800;000 galaxies and over 300;000 quasars.
The galaxy distribution near the local group is rather irregular and characterized
by several super-clusters of galaxies where visible matter is much more concentrated
than in other almost-empty regions called giant vacua (Figs. 2.2 and 2.3). However,
at larger scales the galaxy distribution becomes more uniform in all directions
(Fig. 2.4). With good approximation, the large-scale matter distribution of the
universe is homogeneous (independent of the point at which the observation takes
place) and isotropic (independent of the direction of observation). This is all the
more apparent at even larger scales. In fact, the sky is also uniformly filled with
a radiation background of cosmic origin, called cosmic microwave background
(CMB). Its distribution has been mapped with great accuracy by the Wilkinson
Microwave Anisotropy Probe (WMAP) [10–12] and by the PLANCK satellite
[13, 14], and it shows a remarkable degree of isotropy (Fig. 2.5).
Fig. 2.2 The Sloan Great Wall in a DTFE reconstruction of the inner parts of the 2dF Galaxy
Redshift Survey. The most distant galaxies in the figure are at about one billion light years
(300 Mpc) from Earth (Source: [8])
Fig. 2.3 Slices through the SDSS 3-dimensional map of the distribution of galaxies. Earth is at
the centre and each point represents a galaxy, typically containing about 100 billion stars. Galaxies
are colored according to the ages of their stars, with the redder, more strongly clustered points
showing galaxies that are made of older stars. The outer circle is at a distance of two billion light
years. The region between the wedges was not mapped by the SDSS because dust in our own
Galaxy obscures the view of the distant universe in these directions (Credit: M. Blanton and the
Sloan Digital Sky Survey [4])
The CMB is regarded as a snapshot of the universe dated back to more than 13
billion years ago, when matter, a hot homogeneous plasma of baryons and other
particles, became transparent to radiation for the first time. Today, the CMB has
cooled down to a mean temperature of [15, 16]
T0 D .2:7255 ˙ 0:0006/ K : (2.1)
Looking in any direction, the observer measures the same value (isotropy). Since
there is no reason why we should occupy a privileged place in the universe, the
Fig. 2.4 2dFGRS Galaxy map (Credit: 2dF Galaxy Redshift Survey [9])
Fig. 2.5 Temperature anisotropy map of the cosmic microwave background radiation in the 2013
PLANCK data release. Colors indicate warmer (red) and cooler (blue) spots with respect to the
mean temperature (©ESA and the Planck Collaboration [13])
Fig. 2.6 Hubble diagram: recession velocity of extragalactic supernovæ as a function of distance.
©2004 National Academy of Sciences, USA (Source: [17])
same value should be measured also by observers positioned at any other point
(homogeneity). Deviations from T0 are of the order of micro-Kelvins, so that the
very distant (i.e., early) universe is, for any observer, homogeneous and isotropic up
to one part over 105 .
We also see other large-scale phenomena. The rotation curves of galaxies and
gravitational lensing effects strongly indicate the presence of more matter than
what we observe. Baryons and leptons are typically interested by electromagnetic
interactions, so that this extra component (called dark matter) is believed to be made
of exotic particles not yet detected in the laboratory.
Finally, galaxies are seen to recede away from us according to Hubble’s law.
This law states that the farther the galaxy, the greater the recession speed (Fig. 2.6).1
Observations of type Ia supernovæ (SNe Ia) show a deviation from the linear Hubble
law [18–23], thus indicating that the universe not only expands, but it does so with
increasing rate.
To summarize, observations establish the ingredients of the cosmological stan-
dard model known, for reasons which shall soon become clear, as hot big bang
(henceforth “standard model;” we shall use capital initials for the Standard Model
of particles):
(I) At sufficiently large scales, the universe is isotropic: its properties are inde-
pendent of the direction of observation.
(II) Copernican principle: our location is not special. Consequently, if the universe
is observed as isotropic from everywhere, it is also homogeneous: its thermal
1
The distance of a galaxy is determined by astrophysics standard candles such as variable stars and
supernovæ. For a historical introduction on Hubble law, see [17].
properties are the same at every point. By “point” we mean a sufficiently large
local patch.
(III) The universe is composed of radiation and baryonic as well as non-baryonic
matter. At early times, it was in thermal equilibrium.
(IV) The universe expands. At late times, the expansion is accelerated.
Points (I) and (II) go under the name of cosmological principle:
Cosmological principle. The universe does not possess a privileged point or
direction; it is therefore homogeneous and isotropic (approximately).
Background
The above ingredients can be formalized in Einstein’s theory of general relativity.

We assume spacetime is a “globally hyperbolic” Lorentzian manifold .M; g/
endowed with a metric g . Definitions of global hyperbolicity, space- and time-like
curves and geodesics will be given later in the technical Sect. 6.1.1. Here, we only
need to grasp the intuitive meaning of global hyperbolicity, namely, the possibility
to foliate spacetime into spatial sections stacked along a natural time direction.
Homogeneity and isotropy are realized by the Friedmann–Lemaître–Robertson–
Walker (FLRW) metric:
ds2 D g dx dx D dt2 C a2 .t/ ˛ˇ dx

˛
dxˇ ; (2.2a)
where t D x0 is synchronous (or proper) time, a.t/ is the expansion parameter or

scale factor, and
˛ dr2
˛ˇ dx dxˇ D C r2 d˝D2
2
(2.2b)
1 K r2
is the line element, in hyperspherical coordinates, of the maximally symmetric .D

1/-dimensional space ˙Q of constant sectional curvature K (intrinsic curvature). The
latter is equal to 1 for an open universe, 0 for a flat universe and C1 for a closed
universe with radius a (Fig. 2.7). In four dimensions (polar coordinates), d˝22 D
d 2 C sin2 d' 2 , where 0 6 6 and 0 6 ' 6 2.
The fact that the universe follows a homogeneous evolution allows us to use the
same universal clock at each point, so that physical properties which are equal in
different places imply synchronized clocks.2 Sometimes, the choice of a coordinate
system (equivalently, of a metric g ) is referred to as a gauge choice. The FLRW
2
This is true, of course, for large-scale observations, while at very small scales, where non-FLRW
metrics are better descriptions of the environment, the problem of synchronization of clocks
persists.
Fig. 2.7 Three possible geometries of the universe illustrated for two-dimensional spatial
sections: closed, open and flat from top to bottom, corresponding to a total density parameter
˝0 today respectively greater than, less than or equal to 1 (see (2.84)). In a closed or open .2 C 1/-
dimensional universe, the sum of the internal angles of a triangle is, respectively, greater than or
less than . Only in a universe with flat spatial curvature the sum is (Credit: NASA/WMAP
Science Team [10])
metric corresponds to synchronous gauge, which we will meet again later. Another
gauge makes use of conformal time , defined as
dt
d :D : (2.3)
a.t/
The line element (2.2) becomes
ds2 D a2 . /.d 2
C ˛ˇ dx
˛
dxˇ / :
For a free particle moving at the speed of light, the coordinate distance spanned in
a conformal time interval is c .
Since the time coordinate is not physical time, we can choose different universal
clocks according to the problem at hand. Apart from t, , or the redshift z below,
one can also use matter as relational time, provided it evolves monotonically.
For instance, in the presence of a classical scalar field one can make a time
reparametrization whenever P (dots denote differentiation with respect to syn-
chronous time t) does not change sign (non-singular Jacobian), so that the Hubble
parameter and other geometrical quantities are thought of as depending on .
The spatial part of the metric is homogeneous and isotropic by construction. The
coordinates x˛ are “glued” to the continuous fluid elements representing the matter
content. Therefore, a coordinate point x˛ in space represents a fluid element passing
at time t on the given point. This is why spatial coordinates are called comoving. It
is easy to relate comoving distances with proper distances:
.proper distance/ D a.t/ .comoving distance/ : (2.4)
Proper distances are also called dynamical or, more often, physical, but we will
not use these alternative names here (strictly speaking, comoving distances are also
physical and dynamical).
Observations tell us that the universe is expanding, so we assume that the scale
factor a.t/ > 0 increases, aP .t/ > 0. Cosmological distances (see Problem 2.7) are
better characterized by the redshift
a0
1 C z :D ; (2.5)
a
where a0 :D a.t0 / is the scale factor today. The scale factor is not a physical
observable and its normalization is arbitrary. Typically, one chooses a0 D 1. A
local observer O is at z D 0 and distant objects are at z > 0. For large redshift, 1=z
roughly indicates the size of a closed universe with respect to its radius today.
The name redshift comes from the fact that
0
zD ;

where is the wave-length of radiation emitted by a source and 0 is the observed

wave-length. To show this, consider two nearby objects separated by a small proper
distance dR, where dR D a.1 Kr2 /1=2 dr. The relative velocity of the two objects
is given by the Hubble law, dv D HdR. By Doppler law, the infinitesimal change in
wave-length is d= D dv=c D Hdt D da=a, where we used c D R. P Integrating,
we get
0
D : (2.6)
a a0
The momentum of a photon scales as a1 or, in other words, the comoving wave-
length of the signal is the same at the source and for the observer.
Comoving coordinates define a frame where the universe is isotropic and all
observers move along the Hubble flow. On the other hand, a proper observer sees
nearby galaxies receding but this does not mean that O is the “centre” of the
universe, since there is no privileged point. Redshift measurements and distance
statements such as (2.4) always refer to the relative position of the observed object
with respect to O. To use a simple analogy, one can imagine the two-dimensional
surface of a balloon (the universe) dotted with spots (galaxies). As long as the
balloon inflates, the distance between the spots, measured on the balloon surface
by a two-dimensional observer, increases.
The Hubble parameter
aP .t/
H.t/ :D (2.7)
a.t/
describes the expansion rate of the universe and defines the Hubble distance (or
radius or, improperly, horizon; c units temporarily restored)
c
RH .t/ :D ; (2.8)
H.t/
which is the distance between the observer at time t and an object moving
with the cosmological expansion at the speed of light. The Hubble horizon is a
crucial quantity in the theory of structure formation and in the notion of quantum
measurement in cosmology. It marks the boundary between the causal region
centered at the observer O and the external region and roughly corresponds to the
size of the observable universe (not to be confused with the physical dimensions of a
closed universe parametrized by a). Actually, the true causal horizon is the particle
horizon, defined as the sphere, centered at O, containing all the points which could
have interacted with O through light signals since the “beginning” at t D ti [24].
The radius of the sphere is
Z t Z a.t/
c c
Rp D a.t/ rp :D a.t/ dt0 D a.t/ da : (2.9)
ti a.t0 / ai Ha2
We will see that, in some cases, the particle and Hubble horizon are approximately
the same. Also, when the comoving particle horizon is zero at ti ,
rp D c : (2.10)
We will almost always use natural units, so that from now on we trade the symbol
rp with . The context will make it clear whether is a distance (comoving particle
horizon) or conformal time.3
The value of the Hubble parameter today is measured combining data on type Ia
supernovæ, galaxy distributions and CMB observations:
H0 D 100 h km s1 Mpc1

3:336 h 101 Gpc1
1:747 h 1061 mPl
2:133 h 1042 GeV : (2.11)
The PLANCK 2015 value combining information on the temperature spectrum, on

the polarization spectra at low multipoles and on lensing reconstruction (“PLANCK
TT+lowP+lensing” likelihood) is [14]
h D 0:678 ˙ 0:009 .68 % CL/ ; (2.12)
where the error is at the 68 % confidence level (CL) and includes statistical and
systematic uncertainties.4 Estimates change for different experiments and data sets
in the same experiments, but they generally agree (see [12] for WMAP and [14] for
other likelihood analyses within PLANCK).
Lower bounds on the age of the universe t0 can be obtained from estimates of
the age of astrophysical objects (e.g., globular clusters) or gamma-ray bursts, but a
more precise value is found from the Hubble parameter:
t0 ' H01 D 9:786h1 Gyr ; (2.13)
where Gyr is one billion years. With (2.12), one obtains t0 14:4 Gyr. An
O.1/ correction factor in the above formula, depending on the various energy
components, can refine this estimate (see Problem 2.1). According to the PLANCK
TT+lowP+lensing likelihood [14],
t0 D 13:799 ˙ 0:038 Gyr .68 % CL/ : (2.14)
3
Throughout this book, we shall call “universe” the spacetime region causally connected with the
observer, while storing the name “Universe” with capital U for the whole spacetime, inclusive of
the regions outside the horizon.
4
Although errors are bound to change as soon as new measurements become available, we report
them to give the reader an idea of the level of accuracy of modern observations. The student should
take these numbers with a critical attitude. Where do they come from? How do they change if one
varies the data samples and the prior constraints? Is a model still acceptable if it predicts numbers
outside the 1 experimental interval? (The answer to the last question is Yes. One should start
worrying, or getting excited, only when the model offshoots beyond the 3 level.)
2.2 Einstein and Continuity Equations
In this section, we introduce the classical total action and equations of motion of
general relativity in D dimensions. Our conventions for the Levi-Civita connection,
Riemann and Ricci tensors, and Ricci scalar are

:D 12 g @ g C @ g @ g ; (2.15)
R :D @

@ C ; (2.16)
R :D R ; R :D R g
: (2.17)
The Ansatz of general relativity is the Einstein–Hilbert gravitational action
Z
1 p
Sg D dD x g .R 2/ ; (2.18)
2 2
where g :D det.g / is the determinant of the dimensionless metric g , 2 D

8G is D-dimensional Newton’s constant, and is the cosmological constant. The
couplings have dimension
Œ 2 D 2 D ; Œ D 2 : (2.19)
In a spacetime with D D 2 dimensions, the Newton constant is dimensionless,

the Einstein–Hilbert action is a topological invariant and there are no dynamical
gravitational degrees of freedom. The Newton constant has negative energy dimen-
sionality for D > 2, a fact of utmost importance for the renormalization properties
of a quantum theory of gravity (Sect. 8.2).
The gravitational action is second order in spacetime derivatives. It can be
augmented by higher-order curvature terms (Sect. 7.5) but there is no immediate
need to do so because standard general relativity already explains most of the large-
scale observations.
The constant in (2.18) was originally introduced by Einstein in order to have
stationary-universe solutions. After the discovery of cosmic expansion by Hubble
in 1929, stationary cosmology was abandoned in favour of the big bang model
and was dropped. In the last years, however, the cosmological constant has
been regaining the interest of the community thanks also to the advancements in
astronomical and cosmological observations.
Assuming that matter is minimally coupled with gravity, the total action is
S D Sg C S@ C Sm ; (2.20)
R p
where Sg is (2.18) and Sm D dD x gLm is the matter action. The piece S@ , which
will be further discussed in Sect. 9.1.4, is the York–Gibbons–Hawking boundary
2.2 Einstein and Continuity Equations 25
term (first written by Einstein [25]) added for consistency with the variational
principle [26, 27]. In fact, the latter only requires that variations of the metric be
zero at the boundary (i.e., the geometry of the boundary is fixed) but, in general, the
normal first derivatives may be non-vanishing. In order to take this issue into account
and obtain the Einstein equations correctly, one must add a specific S@ to the total
action. Unless stated otherwise, throughout the book we will ignore the boundary
term either because we consider closed manifolds (which have no boundary) or just
for simplicity of presentation.
To find the equations of motion (Einstein equations) we need the variations (see
Sect. 3.1.1)
p p
ı g D 12 g g ıg ; (2.21)

ıR D .R C g r r / ıg ; (2.22)

where r V :D @ V V is the covariant derivative of a vector V and D

r r is the curved d’Alembertian or Laplace–Beltrami operator. From (2.20), the
Einstein equations ıS=ıg D 0 read
G C g D 2 T ; (2.23)
where
1
G :D R g R ; (2.24)
2
2 ıSm @Lm
T :D p
D 2 C g Lm : (2.25)
g ıg @g
Taking the trace of (2.23) gives

D
1 R C D D 2 T : (2.26)
2
T is the energy-momentum tensor (or stress-energy tensor) of matter. Its definition
determines the continuity equation. In fact, let
Z
1 p
ıSm D dD x g T ıg (2.27)
2
be the infinitesimal variation of the matter action with respect to the external field
ıg . For a constant infinitesimal coordinate transformation
x0 D x C ıx ; ıx D a ; (2.28)

one has
ıg D g @ a C g @ a C a @ g ; (2.29)
where we used the definition of the Lie derivative for rank-2 and rank-0 tensors (see
Chap. 3). Plugging (2.29) into (2.27) and integrating by parts, we get
Z
p 1p
ıSm D dD x a @ . gT / gT @ g : (2.30)
2
Invariance of the action under diffeomorphisms requires ıSm to vanish on shell (i.e.,
when the dynamical equations are satisfied). Using the properties of the Levi-Civita
connection (2.15) and the definition of the covariant derivative of a rank-2 tensor,
r T D @ T C

T T
1 p 1
D p @ gT .@ g /T ;
g 2
one finally obtains the continuity equation
r T D 0 : (2.31)
The energy-momentum tensor carries the contribution of all forms of energy in the
universe, except gravity.
The continuity and Einstein equations are not independent because of the
contracted Bianchi identities
2r R D r R : (2.32)
The divergence of (2.23) correctly reproduces (2.31).
2.2.1 Energy Conditions
The matter sector can be classified according to a set of covariant conditions [28].
Consider an arbitrary future-directed time-like vector t (t t < 0; see Sect. 6.1.1)
and a null vector n (n n D 0). The null energy condition (NEC) is
T n n > 0 8 n null : (2.33)

2.3 Perfect Fluid 27
When the Einstein equations (2.23) hold, this is equivalent to the null convergence
condition
R n n > 0 8 n null : (2.34)
We will use this form of the NEC in Chap. 6.

The weak energy condition (WEC) states that
T t t > 0 8 t time-like : (2.35)
In particular, given the unit time-like vector u (u u D 1), the matter energy
density
:D T u u (2.36)
is always non-negative. Most of the known matter fields obey the NEC and WEC.
Exceptions are condensates, which admit negative-energy states. Violations of the
WEC can also come from quantum effects (Casimir energy, squeezed states, and so
on).
The dominant energy condition (DEC) requires that T t is either future-
directed or null,
T t D Au C Bn ; (2.37)
where A; B > 0. In other words, momentum cannot (be observed to) flow faster than
light (causal flux). In particular, the DEC implies the WEC.
The time-like convergence condition requires that
R t t > 0 8 t time-like : (2.38)
If the Einstein equations hold, this is equivalent to the strong energy condition (SEC)

1 2
T g T C g 2 t t > 0 : (2.39)
D2 D2
2.3 Perfect Fluid
The energy-momentum tensor assumes a simple form when the only matter content
of the universe is a perfect fluid (zero heat flow and anisotropic stress). Its
definition is
T D . C P/ u u C P g ; (2.40)

where D T00 and P D T˛˛ =.D1/ are the energy density and pressure of the fluid,
u D dx =dt is the comoving D-velocity (unit time-like vector, u u D 1) tangent
to a fluid element’s world-line, and t is proper time along the fluid world-line. Since
we are in a globally hyperbolic spacetime, u also corresponds to the unit vector
normal to spatial Cauchy surfaces. The metric g induces a Riemannian metric h
defined by the first fundamental form
h D g C u u ; h D h h ; u h D 0 ; (2.41)
which is the metric spatial projection orthogonal to the comoving velocity.

At every point of spacetime, the local rest frame of a fluid is defined as the
coordinate system where the off-diagonal components of the fluid vanish, T 0˛ D 0,
and

u D ı0 D ..g00 /1=2 ; 0; : : : ; 0/ : (2.42)
In covariant formalism [29–31], the gradient of the four-velocity field is decom-

posed as
1
r u D C ! C h uP u : (2.43)
D1
Here,
1 1
:D r. u/ h C uP . u/ D h . r u/ h (2.44)
D1 D1
is the symmetric shear tensor,
! :D rŒ u C uP Œ u D h Œ r u (2.45)
is the anti-symmetric vorticity tensor,
:D r u (2.46)
is the volume expansion (or expansion of the family of time-like geodesics u ) and
a dot defines the proper-time derivative
P :D u r : (2.47)
Integrating the volume expansion along a world-line with respect to t, one defines
the number of e-foldings
Z
1 t
Na :D dt0 ; (2.48)
D1 ti
2.3 Perfect Fluid 29
where ti is some initial time. This definition is unique up to an integration constant

for each world-line, which can be fixed by choosing a reference hypersurface where
Na D 0.
The continuity equation (2.31) contracted with u is
P C . /. C p/ D 0 ; (2.49)

where D D u uP . This equation is exact and valid at all scales.
Contracting (2.31) with h , we get
D p C h uP . C p/ D 0 ; (2.50)
where
D :D h r D r C u u r (2.51)
is the spatial projection of the covariant derivative. Equation (2.50) can also be
written as
u PP C .Pu C u /. C P/ C r P D 0 : (2.52)
Defining the effective barotropic index
P
w :D ; (2.53)

the energy conditions of the previous section are readily translated into conditions
on (2.53). The NEC is
NEC: .1 C w/ > 0 ; (2.54)
and is implied by all the other conditions. In fact, one can write any time-like vector
of modulus jtj as t D jtju C n . Contracting t with (2.40) according to the
given condition (2.35), (2.37) or (2.39), and then sending jtj to zero, one always
reobtains (2.54). The WEC, DEC and SEC then are
WEC: > 0; w > 1 ; (2.55)

DEC: > jPj ; 1 6 w 6 1 ; (2.56)
2
SEC: Œ.D 3/ C .D 1/w > 0; (2.57)
2
while
D3
DEC C SEC. D 0/ W 6 w 6 1: (2.58)
D1
In D D 4, (2.58) reads 1=3 6 w 6 1.

Summarizing,
DEC H) WEC H) NEC (H SEC. D 0/ :
2.3.1 Scalar Field
If matter is a scalar field with potential V. /, its Lagrangian density is
1
Lm D L D @ @ V. / : (2.59)
2
Scaling dimension (or engineering, or canonical dimension; in momentum units) of
is
D2
Œ D ; (2.60)
2
while for a polynomial potential
X
N
V. / D n n
; (2.61)
nD0
the engineering dimension of the couplings is
n.D 2/
Œn D D : (2.62)
2
It is straightforward to see that the scalar equation of motion ıSm =ı D 0 is in

agreement with (2.31). Invariance of the total action under the infinitesimal shift
! Cı (2.63)
yields the equation of motion
@L d @L
0D : (2.64)
@ dx @.@ /
From (2.59), we get
T D ı L C @ @ (2.65)
2.4 Friedmann Equations 31
and
V; D 0 ; (2.66)
where V; D @V=@ and
1 p
D p @ . g@ / : (2.67)
g
The scalar field is a particular case of perfect fluid, with world-lines orthogonal to
D const hypersurfaces [32]:
@
u D : (2.68)
P
2
When V / , (2.66) is called Klein–Gordon equation.
2.4 Friedmann Equations
On an FLRW background, the only non-vanishing Levi-Civita and Ricci compo-

nents are
0 ˇ
˛ˇ D Hg˛ˇ ; ˛0 D Hı˛ˇ ;
˛ˇ
D ˛ˇ . ˛ˇ / ; (2.69)
and
R00 D .D 1/.H 2 C H/P D .D 1/ aR ; (2.70)

a

2K P g˛ˇ ;
R˛ˇ D C .D 1/H 2 C H (2.71)
a2

R D .D1/ R C .D 1/ DH 2 C 2H P

2K 2 P
D .D 1/ C DH C 2H ; (2.72)
a2
where we have exploited the symmetries of the space-like hypersurface ˙Q [33] and
.D1/
R is its Ricci scalar.
For simplicity we assume that the universe is filled by a perfect fluid, (2.40).
The FLRW shear and viscosity vanish, while the volume expansion and the Hubble
parameter coincide,
a
D .D 1/H ; Na D ln : (2.73)
ai
The 00 component of the Einstein equations (2.23) is

D 2 K
1 H2 D C 2; (2.74)
2 D1 D1 a
while combining that with the trace equation (2.26) one obtains
2
P C 2 D Œ.D 3/ C .D 1/P :
.D 2/ H 2 C H (2.75)
D1 D1
The continuity equation (2.49) has no shear and becomes
P C .D 1/H. C P/ D 0 ; (2.76)
while (2.52) is trivial in the fluid rest frame. For a barotropic fluid p D w, w is
constant and the continuity equation is solved by
D 0 a.D1/.1Cw/ ; (2.77)
where a0 D a.t0 / D 1. From (2.74) with D 0 D K, it follows that

.D1/.1Cw/
2
t
aD : (2.78)
t0
For a scalar field,
P2 P2
D CV; P DL D V; (2.79)
2 2
and (2.76) becomes (2.66),
R C .D 1/H P C V; D 0 : (2.80)
From now on we specialize to four dimensions, D D 4. In this case, we obtain

the first and second Friedmann equations
2 K
H2 D C 2 ; (2.81)
3 3 a
2.5 Content of the Universe 33
2
aR P D . C 3P/ C :
D H2 C H (2.82)
a 6 3
Equation (2.82) is the FLRW version of the Raychaudhuri equation (Problem 6.1).
2.5 Content of the Universe
Equation (2.82) can be rewritten in terms of the parameter ˝, defined as the ratio
between the total energy density and the critical density crit sufficient to stop the
expansion:
3H 2
˝ :D ; crit :D : (2.83)
crit 2
Equation (2.81) with D 0 becomes
˝ 1 D ˝K ; (2.84)
where
K
˝K :D (2.85)
a2 H 2
is the deviation from the critical density. If the universe is spatially flat, ˝ D 1.
Let us now take point (III) of the recipe of the universe into account (Sect. 2.1.1).
Apart from the intrinsic curvature term, the contributions to the total energy density
is typically divided into radiation (˝r , which includes photons), baryonic matter
and non-baryonic matter (˝m D ˝b C ˝nb C ˝ , where we include also massive
neutrinos).5 In the view of explaining point (IV), we also add an extra component
we call ˝ (which might or might not correspond to a non-vanishing in (2.81)):
˝ D ˝r C ˝m C ˝ C ˝ K : (2.86)
2.5.1 Dust and Radiation
Regarded as a perfect fluid, each contribution obeys an independent continuity

equation (2.76). The case w D 1=3 is radiation which redshifts away
5
In general relativity and quantum gravity, one calls “matter” everything which is not geometry.
We did so in Sect. 2.2 but here we use cosmology jargon and separate radiation from the rest of the
“matter.”
zeq zΛ 0
z
Fig. 2.8 Matter (thick curve), radiation (dashed curve), and cosmological constant (thin line)
energy densities as functions of redshift. Radiation-matter and matter-dark energy equality times
are indicated. The redshift scale is arbitrary
as (see (2.77))
r D r0 .1 C z/4 ; (2.87)
while w D 0 is pressureless (dust) matter,
m D m0 .1 C z/3 : (2.88)
The radiation density decreases faster than matter density during the expansion (z !
0). Therefore, the radiation component dominates over matter at early times, but
matter eventually takes over (Fig. 2.8).
The moment6 teq when the two densities coincide is called equality. When the
universe is filled only with radiation and dust, the total energy density evolves
as (2.87) when z > zeq , and as (2.88) otherwise. The redshift at equality is
constrained by CMB observations (from the ratio of the first peak to the third
peak of the power spectrum) combined with other observations. For the PLANCK
TT+lowP+lensing likelihood [14],
zeq D 3365 ˙ 44 .68 % CL/: (2.89)
Different data sets all converge to the rounded value
zeq 3400 ; (2.90)
6
In an approximate sense: the transition is smooth, not point-wise.
which we will use in Sect. 2.7. Once the redshift of an object is found, one can
calculate the corresponding age. In particular, radiation-matter equality happened
when the universe was less than 10;000 years old (see Problem 2.2).
From (2.78), one can see that
2
a t3 ; w D 0; (2.91)
1 1
a t2 ; wD : (2.92)
3
In these scenarios where the universe expands as a power law (p is a constant, not
to be confused with pressure),
2 p
a tp ; pD ; HD ; (2.93)
3.1 C w/ t
and when
1
w> .0 < p < 1/ (2.94)
3
holds, the particle horizon (2.9) is about the same as the Hubble horizon (2.8):
Z t
p t ti 1 p
Rp D t p dt0 t0 ' tD RH
ti 1p 1p
' RH : (2.95)
Several properties of the particle and Hubble horizons and the size at different
epochs can be found in Problems 2.3–2.6 (see also Sect. 3.1.3). Exact power-law
solutions are the subject of Problem 2.11.
Recent estimates of radiation and matter density today are (for the PLANCK
TT+lowP+lensing likelihood at 68 % CL) [14]
˝r0 h2 D 2:469 105 ; (2.96)

˝b0 h2 D 0:02226 ˙ 0:00023 ; (2.97)
2
˝c0 h D 0:1186 ˙ 0:0020 ; (2.98)
˝m0 h2 D 0:1415 ˙ 0:0019 : (2.99)
The difference between the total contribution of matter (measured from the dynam-
ics of galaxy clusters) and baryonic matter (from the visible galaxy distribution) is
ascribed to the presence of another non-baryonic component, dark matter (see [34]
for a review). Dark matter does not interact with photons and it can be observed
only indirectly. In (2.98), we indicated it as ˝c D ˝nb C ˝ . ˝m is a derived
quantity obtained from the sum of baryonic and dark matter densities. Evidence
is in favour of dark matter being made of particles moving at non-relativistic
speed and lying outside the Standard Model of particle physics. Because of the
weak or absent interaction with radiation, density fluctuations of non-baryonic
dark matter can start growing much earlier than for ordinary matter, thus having
time to tune their amplitudes at the level observed today in large-scale structures.
Among the particle candidates from minimal extensions (supersymmetric and
not) of the Standard Model are axions, sterile neutrinos and WIMPs (weakly
interactive massive particles) such as neutralinos; experiments can place bounds
on the abundances of these particles [35, 36]. There is also a possibility that gravity
itself, via some modified action, may account for dark-matter effects. A small part
of ˝c can be due to massive neutrinos, ˝0 h2 < 0:0025 (95 % CL) [14].
2.5.2 Hot Big Bang and the Big-Bang Problem
The quasi isotropy of the CMB temperature indicates that the early universe was in
thermal equilibrium. Therefore, we can express the energy density of radiation and
matter in terms of thermodynamical quantities such as the temperature T.
The most important processes through which radiation interacts with matter are
three: single and double Compton scattering and Bremsstrahlung. Single Compton
scattering describes the collision of a photon with a free electron,
./ C e ! . 0 / C e :
When the photon energy is small with respect to the rest mass of the electron, the
scattering process can be approximated by Thomson scattering.
When an electron at rest collides with a photon, a negative energy transfer
occurs from the electron to the photon, whose frequency red-shifts ( 0 < ).
The combination of this effect with the inverse scattering, where a photon gains
energy from a relativistic electron ( 0 > ), leads to thermal equilibrium. These are
elastic processes and the number of photons is conserved. In the double or inelastic
Compton scattering, on the other hand, photons are emitted or absorbed,
C e ! C C e :
In the Coulomb interaction between an electron and an atomic nucleus X, the

former is accelerated in the rest frame of the ion X and emits radiation. In this case,
one has thermal Bremsstrahlung,
e C X ! e C X C :
The inverse reaction may also happen, since charged particles can absorb photons.
The energy (temperature) of these processes determines which is the dominant

effect:
• Bremsstrahlung for 1 eV < T < 90 eV.
• Single Compton scattering for 90 eV < T < 1 keV.
• Double Compton scattering for T & 1 keV,
where T D kB TK is measured in eV. All these processes have a characteristic velocity
greater than the cosmic expansion and hence they modify the photon distribution of
the cosmological plasma.
At equilibrium, the photons involved in Compton scattering obey a Bose–
Einstein distribution
1
f .!/ D „!
; (2.100)
e T 1
where ! D 2 is the angular frequency and is the chemical potential, non-
vanishing if the total number of photons is conserved. However, the number of low-
frequency photons change due to inelastic Compton scattering and Bremsstrahlung,
thus sending effectively to zero. In the meanwhile, elastic Compton scattering
redistributes the photons over the spectrum. One then obtains a black-body spectrum
of intensity (in „ D 1 units)
!3 1 !3
I.!; T/ D f .!/ D ! : (2.101)
2 2 2 2 e T 1
Once equilibrium and the black-body spectrum are achieved, they are preserved
both by the above processes and by cosmic expansion, whose effect is to rescale
the spectrum while maintaining its form. We have seen in (2.6) that a, so that
! 1 C z and the ratio !=T in (2.101) is unchanged if T 1 C z. We can find
the proportionality constant as follows. The distribution (2.100) defines the number
density of bosons of the i-th species per unit volume, dni .!/ D d3 ! gi .2/3 f .!/,
where gi is the number of spin states (g D 2 for photons). Therefore, the energy
density of a species with black-body distribution is the integral of dn.!/ ! over all
frequencies (d3 ! D 4d! ! 2 ), which is the integrated intensity
Z C1 Z C1
i D dni .!/ ! D gi d! I.!; T/
0 0
Z C1
gi .T/ 4 x3 2
gi T 4 :
x D !=T
D T dx D
2 2 0 ex 1 30
One should sum over all species i in thermal equilibrium (i T), including
fermions. The latter obey a Fermi–Dirac distribution f .!/ D .e!=T C 1/1 ,
resulting in a contribution .7=8/gi for each species. The total energy density of
radiation is the Stefan–Boltzmann law with gi replaced by a total effective g .T/:
2
r D g .T/ T 4 : (2.102)
30
The coefficient g .T/ D O.1/ – O.10/ is temperature dependent because a species

with mass mi no longer participates in (2.102) when the temperature falls below mi .
Away from these decoupling events, g can be regarded as constant. For a radiation-
dominated universe, comparing (2.102) with (2.87) one gets the temperature as a
function of redshift:
1=4 1=4
30 r0 ˝r0 h2
T D .1 C z/ ' 3:96 .1 C z/ 103 eV
2 g g
2:79
.1 C z/ 104 eV
(2.96)
1=4
(2.103)
g
3:24
1=4
.1 C z/ K : (2.104)
g
Using (2.81) and (2.92), one has
mPl tPl
HD ; (2.105)
2t
so that, with (2.81),
1=2 2
45 m2Pl tPl 2:42 106 eV
tD 1=2 s: (2.106)
16 3 g T 2
g T
With these formulæ, one can calculate the temperature at a given redshift and the
age of the universe at a given temperature (see Problem 2.8).
Since T / 1 C z, one concludes that the universe was a high-temperature,
radiation-dominated plasma in early epochs. This is why the standard model is
called hot. As in the great majority of cosmological solutions, the power-law
profile (2.93) tends to a ! 0 in the past. At the big bang t D 0 (the earliest
instant ever), the metric is singular and general relativity breaks down. The Universe
reduces to a point of infinite temperature and energy density. Generic classical
cosmological solutions do possess a big-bang singularity and one cannot trust the
theory at the very beginning. This is clearly a problem of self-consistency which is
desirable to solve. We will talk about it much later (Chaps. 6, 10 and 13).
2.5.3 Dark Energy and the cosmological Constant Problem
The hot big bang model works very well from early ages until almost today. This
is the main reason why we assumed that only dust and radiation are present.
Other types of fluid, if added, should first prove themselves necessary to explain
experimental data, and then find their place in a field-theory model. For example,
an extra stiff matter component (w D 1) would scale as stiff D stiff0 .1 C z/6
and it would dominate at early times. Let zQeq be the redshift when stiff D r . The
hot big bang model is verified to a high degree of accuracy starting from the big-
bang nucleosynthesis (BBN; see Problem 2.8) at zBBN 109 , so that zQeq > zBBN .
From (2.96), one must have ˝stiff0 h2 . 1023 , and any such component would be
unobservable today. Therefore, there is no need to consider stiff matter.
On the other hand, from the positivity of , (2.82) and the strong energy
condition (2.57) with D 0, it follows that a universe filled only with matter
and radiation must decelerate, aR < 0. This is in contrast with point (IV) so that,
after all, we do have to add some other contribution to the total energy density.
So far, we have ignored the cosmological constant. Let us rewrite the Friedmann
equations (2.81)–(2.82):
aR 2
D Œ. C / C 3.P C P / ; (2.107)
a 6
2 K
H2 D . C / 2 ; (2.108)
3 a
where

D P :D : (2.109)
2
A particular solution of these Friedmann equations is the de Sitter universe [37], a
cosmological model without matter ( D 0) and flat spatial sections (K D 0):
r

HD ) a.t/ D a.0/ eHt : (2.110)
3
The Hubble rate is constant and the scale factor expands exponentially. It is the
prototypical inflationary background and the cosmology with the simplest analytic
properties (see Problem 2.10). It has an analogue also for a closed universe ( K D 1):
a.t/ D H 1 cosh.Ht/ ; (2.111)

p
where H D =3.
In a quantum field theory context, and p represent the energy density and
pressure of quantum fluctuations of the vacuum. At the semi-classical level, gravity
can be treated as classical, while Einstein equations take the form [38]
G D 2 hT i ; (2.112)
which can be argued to be valid when dispersion in the phase of matter wave-
functions is negligible [39]. In a local inertial frame, Lorentz invariance requires
that the vacuum expectation value of the energy-momentum tensor be proportional
to the Minkowski metric, hT i / . Therefore, in a general frame [40, 41]
hT i D vac g : (2.113)
This is a cosmological-constant contribution vac D = 2 D and the “vacuum

equation of state” is (2.109), corresponding to the barotropic index
w D 1 : (2.114)
The fact that the pressure P is negative should not worry the reader. In fact, in
gravity the quantity P loses its usual meaning of thermodynamical pressure; here, it
encodes a generic effect of “anti-gravitational repulsion.”
A notable feature of the de Sitter universe is that it accelerates,
aR
D H 2 ˝ > 0 : (2.115)
a
Measuring the redshift of type I supernovæ, in 1997 it was discovered that the
universe is indeed accelerating [18, 42, 43]. First estimates gave 0:3 6 ˝0 6
0:9. The advance in what is now called precision cosmology can be appreciated
by comparing this early constraint with the most recent one to date from CMB
observations [14] (PLANCK TT+lowP+lensing),

˝0 D D 0:692 ˙ 0:012 (68 % CL) ; (2.116)
3H02
or from the combination of data [14] about the CMB, supernovæ [6], baryon
acoustic oscillations (BAO) [5, 44–48] and H0 (“PLANCK TT+lowP+lensing+ext”
likelihood):
˝0 D 0:6935 ˙ 0:0072 (68 % CL) : (2.117)
Using (2.116),
3H02
D ˝0 5:4 10123 m4Pl .3:3 103 eV/4 : (2.118)
2
This is a very low density, about 1029 g cm3 , which would be extremely difficult to
detect in the laboratory. Nevertheless, it constitutes about 70 % of the total density,
since it uniformly fills the universe and all the regions otherwise empty of matter.
The experimental value w0 of w today varies according to the prior constraints
(zero or non-zero curvature, constant or varying w , and so on). For instance, for a
constant w and a flat universe [14] (PLANCK TT+lowP+lensing+ext),
w0 D 1:006C0:085
0:091 (95 % CL) : (2.119)
Somewhat weaker bounds come from redshift-space distortions [49–51]. Letting

a.t0 / D a0 D 1 be the present value of the scale factor, a two-parameter model of a
varying w is [52, 53]
w D: w0 C wa .a 1/ ; (2.120)
so that today w D w0 . In this case, the PLANCK TT+lowP+lensing+ext posteriors

are
1:2 < w0 < 0:6 ; 1:5 < wa < 0:6 (95 % CL) : (2.121)
A cosmological constant is the simplest explanation of late-time acceleration.

However, is a problematic object in many respects. Other “small” numbers do
appear in cosmology but, usually, they are described by known physics and are not
perceived as troublesome as (2.118) [54]. An instance is the energy density of matter
or radiation at the time of equality,
eq 2:4 10113 m4Pl ; (2.122)
which depends on the number density of photons and matter particles and can be
calculated within the framework of standard high-energy physics [55]. On the other
hand, the cosmological constant problem consists in the “unnatural” smallness of
the quantity (2.118). There is an issue of fine tuning both at the classical level
(via the ratio =eq 1010 ) and in an effective quantum field theory set-up.
Consider microscopic quantum degrees of freedom described by states with proper
momentum p < pmax , and assume a uniform distribution of the number of states per
unit volume in momentum space, d3 p n.p/ d3 p. Then, the vacuum energy density
due to these quantum states is (see Sect. 7.1)
Z pmax
vac vac;0 dp p3 p4max ; (2.123)
0
where p D jpj and pmax is a cut-off above which the effective theory breaks down.
We would expect that
pmax mPl 1019 GeV (2.124)

at the Planck scale, or that
pmax D MSUSY > 1016 mPl 103 GeV (2.125)
for supersymmetry. If, as suggested above, is the energy contribution of vacuum,

1=4 1=4
then observations impose that vac D 1031 mPl 103 eV, much smaller
than (2.124) or (2.125). Therefore, there must be a contribution V0 to the potential
of the effective theory cancelling vacuum effects almost exactly. A calculation in
the Standard Model of particles predicts too small a V0 , insufficient to meet the
experimental value (2.118). On the other hand, supersymmetry overshoots the target
and vacuum effects are cancelled exactly ( D 0); however, supersymmetry is
broken at low energy scales, thus reducing but not removing the fine tuning. This
is one formulation of the cosmological constant problem, which we will study in
greater detail in Chap. 7.
Also, the universe has been accelerating only since aR D 0, at (varying-w
parametrization [56])
zacc D 0:81 ˙ 0:30 (95 % CL) ; (2.126)
that is, 6:8 ˙ 1:4 billion years ago. For a constant D 0 , that epoch would
be marked as somewhat special in the history of the universe, because it would be
triggered by a non-thermodynamical effect. In other words, the model would depend
on the initial condition D 0 (coincidence problem).
Notice that zacc is larger than the redshift z at which m D : the onset of
acceleration is before the cosmological constant dominates over matter (see [56] for
a discussion). Figure 2.8 shows the evolution of the matter, radiation, and energy
densities.
There may be other candidates, cloaked under the mysterious name of dark
energy, driving acceleration and accounting for the contribution (2.116) (see
Problem 2.9). In the vacuum interpretation, the cosmological “constant” naturally
varies with time, since the vacuum energy is temperature dependent. A running .t/
better accommodates the coincidence problem and can be realized by a number
of mechanisms. One of the most popular is quintessence, a dynamical scalar field
operating like the inflaton but at different energy scales. The equation of state of
a single homogeneous scalar spans the whole range of the DEC and, in general,
models dominated by an energy component with w < 1=3 always accelerate:
aR 1
D H 2 .1 C 3w/˝ : (2.127)
a 2
However, due to the observational constraint w 1 the quintessence field must
be very similar to a cosmological constant and its dynamics is subject to a certain
amount of fine tuning.
2.5.4 Spatial Curvature and Topology
The curvature contribution ˝K can be constrained by combining CMB data

(mainly the position of the temperature power spectrum peaks, plus polariza-
tion) with distance indicators. In the so-called CDM concordance model (Cold
Dark Matter with a pure cosmological constant, w D 1), one has (PLANCK
TT+lowP+lensing+BAO [14])
˝K0 D 0:000 ˙ 0:005 (95 % CL) : (2.128)
The constraint changes according to data pools, but it is always compatible with flat
geometry.
From observations, one can also constrain the “shape of the Universe.” A non-
trivial cosmic topology [57–62] arises if the Universe is not simply connected. In
the latter case, it can have disconnected components or be multi-connected, i.e.,
some spacetime points are identified. The closed compact (hypersphere), flat and
open non-compact FLRW topologies are only three special cases among many
other possibilities, including some where spatial sections are flat and compact (3-
torus T3 ; e.g., [63]), flat and non-compact (some Clifford–Klein spaces [57, 64]) or
with positive curvature (some Clifford–Klein spaces or, e.g., Poincaré dodecahedral
space [65]).
Detecting topology is difficult because the dynamics is, in general, the same. For
example, the hypertorus T3 is locally isomorphic to three-dimensional Euclidean
space, so that spatial flat sections and the 3-torus share the same metric; the only
change is in the boundary conditions. If the characteristic curvature scale of the
topology is larger than the observed universe, then there is little or no hope to see
any effect. However, if the observed patch is larger than physical space, one could
observe multiple images of luminous sources in large-scale structures [66] and the
CMB [67].
The impact of topology on the cosmic microwave background and its interplay
with the inflationary mechanism have been studied extensively [65, 68–99]. A
total density parameter close to the critical value ˝ 1 implies that many
topologies are undetectable, while others are already excluded. For some compact
and semi-compact topologies, one can place bounds on their characteristic scale
[99]. Although, so far, there is no evidence for a non-trivial topology and the scale
of a compact topology is tightly constrained [100, 101], it may be important to keep
an eye on this direction. A Universe with non-trivial topology can arise, for instance,
via quantum creation in Wheeler–DeWitt quantum cosmology [63, 90, 102–104], in
string theory [105, 106] or in quantum gravity and quantum cosmology models of
“third quantization” (Sect. 10.2.4). The discovery of an imprint of a specific cosmic
topology could be relevant in the big quest of a quantum theory of Nature, since it
could justify, constrain or refine some of the models mentioned in this book.
2.6 An Obscure Big Picture
To summarize, the total density parameter measured today is
˝0 D ˝K0 C ˝r0 C .˝b0 C ˝c0 / C ˝0 : (2.129)
Using the estimate of the Hubble parameter h in (2.12), and neglecting radiation
and curvature contributions (which amount to less than 1 % of the total), one ends
up with the following picture:
˝b0 4:8 % ; ˝c0 26 % ; ˝m0 31 % ; ˝0 69 % :

(2.130)
We do not have striking evidence in favour of any specific model of cold dark matter
or dark energy over the others, so we could say that we know less than 5 % of
our world! The first cosmological models did not have dark components at all, but
observations forced themselves upon our conception, which was then changed to
make room for new ingredients. As Figs. 2.9 and 2.10 illustrate, our picture of the
universe has been evolving dramatically in the last fifty years. Science is a measure
of our ignorance.
To the best of our knowledge, the concordance model still fits all data without the
compelling need to invoke exotic physics such as supersymmetry, string theory or
quantized gravity. In this sense, cosmology does not really need quantum gravity at
large. However, certain ingredients create a sense of uneasiness in the researcher,
mainly for theoretical reasons. The cosmological constant problem, both in its
old version of Sect. 2.5.3 and in the new one (Why is ˝0 of the same order of
magnitude as ˝m0 ?) is one example and we will soon see others (the big-bang
Fig. 2.9 Evolution of our knowledge of the content of the universe from the 1970s (Credit:
NASA/WMAP Science Team)
2.6 An Obscure Big Picture 45
Fig. 2.10 Content of the universe as estimated by WMAP [12] (left pie chart) and PLANCK 2013
[107] (right) (©ESA and the Planck Collaboration [13])
problem and the inflation-related problems). So, in this other respect, quantum
gravity might need cosmology. One duty of modern theoretical physics beyond the
Standard Model is to be in agreement with observations, in particular cosmological.
Suppose, for instance, one has a model of matter and gravity at hand which can
explain a late stage of acceleration (this is a typical goal of phenomenology).
One starts from the Friedmann equations or their analogue in, e.g., synchronous
time, which is an unphysical parameter. Switching to expressions in redshift z such
as (2.87) and (2.88), rescaling the energy densities i ! ˝i and plugging the
measured values ˝i0 for all the components inserted “by hand” (e.g., matter and
radiation), one can follow the dynamical evolution from large redshift until today
or to the future, and constrain the parameters of the model (in this example, those
determining the dark energy density).
However, it is often difficult to falsify phenomenological models on the basis of
constraints placed a posteriori. It would be greatly desirable to construct models
with predictive power, i.e., with few or no free parameters, or such that its free
parameters determine the cosmological observables only within certain intervals.
Therefore, a second, tremendously challenging task of theoretical physics is to
enhance its predictive power by asking the right questions on one hand (What are
the physical observables? How can we formulate the cosmological constant and big-
bang problems in a non-misleading, operational way?) and looking towards the right
directions on the other hand: Does the model agree with observations? How are dark
energy and inflation realized? Can we build an arsenal of smoking guns to favour
one model against another?
2.7 Problems and Solutions
2.1 Age of the universe. Calculate the formula for the age of the universe
in the presence of matter and a cosmological constant. Compare it with the
estimate t0 H01 given in (2.13) (ignore error bars).
Solution From (2.81) and (2.88), it follows that

2 2
H D m C 2 D H 2 .˝m C ˝ / D H02 1 ˝m0 C ˝m0 a3 :
3
In the last step, we have used the fact that ˝0 D 1 ˝m0 . The above equation can
be integrated from the beginning (a.ti / D ai ) until today (a0 D 1). Since ti t0 ,
one then has ai 0 and
Z 1
1 da
t0 D p
H0
0 .1 ˝m0 /a2 C ˝m0 a1
p
1 2 1 C 1 ˝m0
D p ln p
H0 3 1 ˝m0 ˝m0
1
0:955
H0
13:79 Gyr ;
where we used (2.12), (2.13) and (2.99). This estimate is 96 % the crude one t0
H01 and is very close to (2.14). Adding also radiation does not change the numbers
because the radiation-dominated era is only a small fraction of the present age.
2.2 Equality and decoupling. An important epoch in the history of the

early universe is decoupling, when radiation last scattered with matter.
Observations center this period at
zdec 1091 : (2.131)
How old was the universe at zdec and zeq ? Use the experimental estimate of
the age of the universe today t0 and ignore the errors.
2.7 Problems and Solutions 47
Solution We must express the redshift z as a function of time t, and to do so we

have to choose a profile a.t/. For redshifts z & zeq , we can use (2.93):
a0 t p t0
0
z.t/ C 1 D D ) tD : (2.132)
a.t/ t Œz.t/ C 11=p
The universe is dominated by dust and p D 2=3. For the age given in (2.14), t0
13:8 Gyr, one obtains
tdec 3:83 105 yr : (2.133)
At decoupling time, the universe was about 380;000 years old. Equation (2.91) is
still an acceptable approximation near matter-radiation equality,
teq 7:0 104 yr : (2.134)
2.3 Cosmological horizons 1. Describe the behaviour of the proper and

comoving particle horizons for flat power-law cosmologies (2.93) with 0 1:
P D 3 H 2 .1 C w/ :
H (2.135)
2
Therefore, the Hubble horizon always increases. In particular, for w > 1=3 (0 <
p < 1) the particle horizon is well defined and follows the same evolution as RH .
From (2.95) and ti D 0,
t p 1
Rp D D : (2.136)
1p 1pH
Also the comoving particle horizon increases with time,
Rp t1p p
D D D rH ; (2.137)
a 1p 1p
but at a lower rate.

2.4 Cosmological horizons 2. Determine the recession speed (in c units) of

the particle horizon for power-law cosmologies (2.93) with 0 c: (2.138)
1p
This result does not violate special relativity, because the latter is valid only in
local inertial frames and does not apply to relative speeds of distant objects. On
the other hand, one requires that signals do not travel faster than light, but no signal
is interchanged between us and the horizon.
2.5 Horizons and distances 1. Determine the size of the particle and Hubble
horizons at zeq , zdec and today for a matter-dominated universe. How much
did the observable universe increase from decoupling to equality and from
equality until today?
Solution At these redshifts, (2.91) is a very good approximation of the cosmic

expansion. From (2.95) and (2.132), one has
p
H 1 1
zC1D ) D : (2.139)
H0 H H0 .z C 1/1=p
We know that
1
3h1 Gpc 4:42 Gpc : (2.140)
H0
Therefore, for p D 2=3 the size of the particle horizon is Rp D 2=H and
Rp .zeq / 45 kpc ; Rp .zdec / 245 kpc ; Rp .0/ 8:8 Gpc :
The Hubble horizon H 1 is simply half the particle horizon. The growth rate of the
horizons from redshift z1 to redshift z2 is
3=2
Rp .z2 / RH .z2 / 1 C z1
D D ; (2.141)
Rp .z1 / RH .z1 / 1 C z2
so that
Rp .zdec / Rp .0/ Rp .0/

5:5 ; 36 103 ; 2 105 :
Rp .zeq / Rp .zdec / Rp .zeq /
The comoving particle horizon is
1 1
.z/ D .1 C z/Rp .z/ D .z C 1/1 p : (2.142)
H0
It follows that 0 D Rp .0/, while
eq 152 Mpc ; dec 268 Mpc :
The growth ratios are
dec 0 0
1:8 ; 33 ; 58 :
eq dec eq
2.6 Horizons and distances 2. Repeat the previous exercise for the CDM
model with radiation included. Find analytic expressions for the correction
factors.
Solution A better estimate of particle horizons should also take radiation and the
cosmological constant into account:
Z a.z/
da
.z/ D
0 Ha2
Z .1Cz/1
1 da
D p : (2.143)
H0 0 .1 ˝m0 ˝r0 /a4 C ˝m0 a C ˝r0
The integral can be done exactly in certain regimes. At high redshifts 10 < z < 104 ,
one can ignore the cosmological constant and obtain
Z .1Cz/1
1 da
.z/ ' p
H0 0 ˝m0 a C ˝r0
p p
˝r0 .1 C z/ C ˝m0 ˝r0 .1 C z/
D CDM .z/ ; (2.144)
˝m0
where
2
CDM .z/ D p : (2.145)
H0 1 C z
In particular, the correction factors with respect to those found in the previous
exercise are
eq 0:89 CDM;eq ; dec 1:18 CDM;dec : (2.146)
As expected, the correction factors are close to 1.

At late times, dark energy is causing the recent expansion of the universe to be
greater than in the past. Therefore, we expect to obtain a higher value for 0 . In fact,
Z .1Cz/1
1 da
.z/ D p
H0 0 .1 ˝m0 /a4 C ˝m0 a

1 1 1 7 1 ˝m0
D CDM;0 p F ; I I ; (2.147)
˝m0 6 2 6 ˝m0 .1 C z/3
where F is the hypergeometric function

1
X .a C n/ .b C n/ .c/ xn
F.a; bI cI x/ :D : (2.148)
nD0
.a/ .b/ .c C n/ nŠ
Today, 0 1:63 CDM;0 . The effect, actually, turns out to be quite large because
the universe has been accelerating for about half its age. Including all contributions
(radiation, matter, ), (2.143) yields
0 1:61 CDM;0 : (2.149)
Using (2.143), we summarize the results in Table 2.1.

The growth ratios do not change much and are
dec 0 0
2:3 ; 45 ; 106 ; (2.150)
eq dec eq
Table 2.1 Comoving and

z Rp
proper particle horizons
zeq 134 Mpc 39 kpc
zdec 315 Mpc 289 kpc
0 14.2 Gpc
and
Rp .zdec / Rp .0/ Rp .0/

7:3 ; 5 104 ; 3:6 105 : (2.151)
Rp .zeq / Rp .zdec / Rp .zeq /
From matter-radiation equality until today, the linear size of the causal patch
(observable universe) has been increasing 360;000 times.
2.7 Distances. Determine the distance of luminous objects at redshift z.

Write an approximate formula for z 1. How far are sources at zdec in a
flat universe? And galaxies at z D 0:2?
Solution The comoving distance .t/ of an object at redshift z.t/ from us is the
distance light covered from time t until today. This is related to the particle horizon
via the photon geodesic equation. In conformal time (c D 1),

dr2
0 D ds2 D a2 . / d 2
C ; (2.152)
1 Kr2
where r is the spatial coordinate interval. Taking the square root and integrating,
Z 0
Z
0 dr
d D p : (2.153)
0 1 Kr2
Inverting with respect to ,

8 1=2
<jKj sinh jKj1=2 . 0 / ; K<0
. / D 0 ; KD0 : (2.154)
: 1=2
K sin K1=2 . 0 / ; K>0
In a flat or closed universe, signals emitted inside the particle horizon could not have
travelled a distance greater than the radius of the horizon today.
For high redshifts, the comoving distance in a flat universe is well approximated
by the particle horizon today. For instance, at z D zdec the comoving horizon is one
order of magnitude smaller that the horizon today 0 (see (2.150)) and the comoving
distance of the z D zdec surface (last-scattering surface) is
.zdec / ' 0 9:6h1 Gpc 14:2 Gpc ; (2.155)
where we presented the result also in the standard form with the h factor reinstated.
The proper distance
d.z/ :D a.z/.z/ D .1 C z/ .z/ (2.156)

Rp .0/
D Rp .z/ (2.157)
1Cz
is always smaller than . We will see in Chap. 4 that zdec corresponds to the time
when the CMB was originated. Since the CMB is isotropic only in the comoving
coordinate frame,7 statements about the “distance” of the last scattering surface
implicitly refer to the comoving distance .z/.
At small redshift, curvature effects can be neglected. The proper and comoving
distances are about the same and one can expand (2.157) to get
c ˇ
d.z/ D a0 .z/ C O.z2 / D @z a.z/ˇzD0 z C O.z2 /
H0 a 0
cz
D C O.z2 / 3z h1 Gpc : (2.158)
H0
Objects at z D 0:2 are 600 h1 Mpc away from us. The 2dFGRS survey covers
redshifts z . 0:3, so that the largest redshift observed by the survey corresponds to
a distance of about 1:3 Gpc. The SDSS main galaxy sample contains galaxies with
z . 0:4 and quasars as far as z 5.
Another distance of great interest in astrophysics is the luminosity distance,
which can be expressed either via the absolute magnitude Mabs of an object and
its apparent magnitude Mapp or via the luminosity L (in Watts) and the energy flux
F (Watts per area):
r
Mapp Mabs
5 L
dL :D 10 5 Mpc D : (2.159)
4F
The luminosity of certain objects such as type I supernovæ is known. For these
“standard candles,” the luminosity distance can be determined with a certain
accuracy. In these cases, dL D . / ' 0 and, via (2.156), one can extract
valuable information on the expansion properties of the universe.
2.8 Thermal history of the universe. Determine the temperature of radia-

tion at zeq , zdec and today. Calculate the age and the order of magnitude of
(continued)
7
Sometimes, this frame preference is perceived as a contradiction of general relativity. However,
global coordinate frames may be defined once we fix our metric. The observed CMB frame is a
solution of Einstein’s equations.
redshift corresponding to T 1 MeV and T 0:1MeV. Assume g D 1 and

round up the numbers from below, since g > 1.
Solution From (2.103) and (2.104),
Teq 1 eV 104 K ; (2.160)

3
Tdec 0:3 eV 3 10 K ; (2.161)
T0 3 104 eV 3 K : (2.162)
More precise numbers can be obtained by inserting realistic values for g [108].
In particular, (2.162) agrees with the measured temperature (2.1) of the microwave
background.
T 1 MeV 1010 K is the typical binding energy of nuclei. When the
temperature of the universe is lower than that, atomic nuclei are being synthesized.
From (2.103) and (2.106) at high redshift (z 104 T eV1 ), we get
t 2s; z 1010 : (2.163)
However, the inverse process also occurs and nuclei are destroyed until the universe
cools down enough. One can show that 4 He nuclei begin to form at about T
0:1 MeV. This temperature roughly marks the onset of the epoch known as big-bang
nucleosynthesis:
tBBN 200 s ; zBBN 109 : (2.164)
This era begins three minutes after the big bang and lasts about 17 minutes, after
which the nuclear fusion reaction rate drops off (T 20–50 keV). Measurements of
light elements have confirmed these calculations. Nucleosynthesis has been taking
place again in the core of stars since their formation, 100 million years after the big
bang.
We summarize the thermal history of the universe in Table 2.2. The particle
horizon before equivalence is found from (2.139) and (2.105),
1 1
Rp D D 1=2
6 108 z2 kpc 2 1025 z2 km ; (2.165)
H H0 ˝r0 .1 C z/ 2
which also correctly reproduces the large-z asymptotic limit of (2.143).

The reader should not take these values too seriously. Depending on both the
available experimental data and the details of the high-energy particle physics
involved during the early stages, the big-bang timeline can change from time to
time, and from book to book. However, it is still remarkable that we have been able
Table 2.2 Simplified thermal history of the universe from BBN until today. The first line
corresponds to the highest energy probed in ground-based laboratories, at the Large Hadron
Collider (LHC) [109]. The values of this table should be taken only as indicative
T (K) T (eV) t z Rp Event

1016 1012 1012 s 1016 0:2 mm Highest energy probed
in laboratory (LHC).
1010 105 2s 1010 105 km Formation and destruction
of nuclei begins.
109 104 200 s 109 107 km Nucleosynthesis of light
ions.
108 3 103 20 min 108 109 km Big-bang nucleosynthesis
ends.
104 1 7:0 104 yr 3400 39 kpc Radiation-matter equality.
Matter domination begins.
3000 0:3 3:8 105 yr 1090 289 kpc Decoupling of matter and
radiation. CMB forms.
Atoms form.
100 102 108 yr 25 110 Mpc First stars.
3 3 104 14 109 yr 0 14.2 Gpc Today.
to reconstruct the history of the early universe in such a detail from a handful of
formulæ. We collect them here again in an approximated fashion, for a radiation-
dominated universe:
2
4 TeV 106 eV
TK 10 K; t s;
1 eV TeV

4 TeV
z 10 ; Rp 1025 z2 km :
1 eV
2.9 Accelerating universe. Consider a D D 4 flat universe filled only with a

barotropic fluid. Discuss the meaning and behaviour of the parameter
d ln H HP aR
:D D 2 D1 : (2.166)
d ln a H aH 2
Can an accelerating universe be dominated by matter or radiation?
Solution Equation (2.166) is the definition of the so called first slow-roll parameter,
and is nothing but the recession speed of the particle horizon:
D RP H : (2.167)
From (2.135), one sees that D 0 for w D 1 (cosmological constant, de Sitter

expansion), > 0 for w > 1, and < 0 for w < 1. In the latter case the Hubble
parameter (horizon) increases (respectively, decreases) with time, a situation called
super-acceleration. Moreover, (2.127) and (2.166) yield the additional constraint
> 1 when w > 1=3. Summarizing,
<0 if w < 1 ; (2.168a)

D0 if w D 1 ; (2.168b)
0<<1 if 1 < w < 13 ; (2.168c)
>1 if w > 13 : (2.168d)
In particular, we have acceleration only if < 1 and power-law cosmologies (2.93)

with 0 0).
period of inflation (Ra > 0), while (2.168a) corresponds to super-inflation (H
2.10 Exact solutions: H D const. Consider a D-dimensional universe with

D 0 filled with a scalar field with potential V. Find the exact solutions of
the equations of motion for a constant Hubble parameter, listing a.t/, .t/,
.t/ and V. /. Recast the solutions in conformal time.
Solution The Friedmann and continuity equations are

!
D 2 2 P2 K
1 H D CV 2 ; (2.169)
2 D1 2 a
2
P D
H2 C H
2
V P2 ; (2.170)
D1 D2
0 D R C .D 1/H P C V; : (2.171)
For a constant Hubble parameter H.t/ D H,
a.t/ D eHt ; .t/ D 0 : (2.172)

In a flat universe, the exact solution of the Friedmann and scalar equations is just a
cosmological constant (de Sitter spacetime),
.D 1/.D 2/H 2
.t/ D 0 ; V. / D ; K D 0: (2.173)
2 2
The Friedmann equations show that there is no solution if K D 1, while there is
one for a closed universe, but only in D D 4:
r
2
˙ .t/ D˙ eHt ; K D 1;
2H2
while the continuity equation fixes the potential:
3H 2
V. / D C H2 2
: (2.174)
2
This solution is not de Sitter because spatial sections are not flat. (We recall that the
H D const cosmology corresponds mathematically to de Sitter spacetime only if
spatial sections are flat. In that case, the de Sitter hyperboloid is only half covered by
FLRW coordinates.) The scalar field ˙ rolls down its potential from t D 1 and
climbs it again after passing the global minimum. The solution is actually unique,
since cosmological equations of motion are invariant under time reversal,
t ! t ; (2.175)
and the direction of the rolling in a symmetric potential does not matter.
We can recast proper-time solutions into expressions in conformal time by
inverting .t/. For an H D const background,
Z
eHt
D dt eHt D ; (2.176)
H
so that
ln.H /
tD : (2.177)
H
Notice that runs from 1 to 0, so that the above expression is well defined. The
geometric background in is
1 a0 1
a. / D ; H. / :D D aH D ; (2.178)
Hj j a j j
where a prime denotes differentiation with respect to . The solution in the closed
universe is linear in ,
. /D 0j j: (2.179)
2.11 Exact solutions: power-law expansion. Repeat the same calculations

of the previous exercise for a power-law expansion, a / tp , where p > 0.
Solution In this case,

t p p 1
a.t/ D ; H.t/ D ; .t/ D ; (2.180)
Nt t p
where Nt is some reference time. Trying the profile .t/ D . 0 =q/tq in the (sum of
the) Friedmann equations, one finds that it must be q D 0. This suggests to consider
the limit q ! 0, which is a logarithmic profile:
.t/ D 0 ln.t=Nt/ : (2.181)
From now on, Nt D 1. This gives the exponential potential [110–113]
.D 1/p 1 2 2 =
V. / D 0e
0
: (2.182)
2
If the universe is flat, then
r
.D 2/p
0 D˙ ; K D 0; (2.183)
2
while for a curved universe only the case p D 1 is a solution:
r
D 2 C 2K
0 D˙ ; p D 1: (2.184)
2
This solution is real only if D > 2.1 K/. Therefore, it is always valid for a closed
universe, while for an open universe it exists only in D > 4.
For a power-law expansion, conformal time is
Z
t1p
D dt tp D ; p ¤ 1; (2.185)
1p
which is positive if p < 1. Inverting this expression,

1
t D Œ.1 p/ 1p : (2.186)
The solution with K D 0 is

1p
p
p 1
a. / D ; H. / D ; . /D 0 ln ; (2.187)
0 1p
while for p D 1
Z
dt
D D ln t ; (2.188)
t
and the solution is
a. / D e ; H. / D 1 ; . /D 0 : (2.189)
References
1. S. Cole et al. [The 2dFGRS Collaboration], The 2dF Galaxy Redshift Survey: power-spectrum
analysis of the final dataset and cosmological implications. Mon. Not. R. Astron. Soc. 362,
505 (2005). [arXiv:astro-ph/0501174]
2. http://www.aao.gov.au/local/www/6df
3. D. Heath Jones et al. [The 6dFGS Collaboration], The 6dF Galaxy Survey: final redshift
release (DR3) and southern large-scale structures. Mon. Not. R. Astron. Soc. 399, 683 (2009).
[arXiv:0903.5451]
4. http://www.sdss3.org, http://classic.sdss.org, http://www.youtube.com/watch?v=
08LBltePDZw
5. C.P. Ahn et al. [SDSS Collaboration], The tenth data release of the Sloan Digital Sky Survey:
first spectroscopic data from the SDSS-III Apache Point Observatory Galactic Evolution
Experiment. Astrophy. J. Suppl. 211, 17 (2014). [arXiv:1307.7735]
6. M. Betoule et al., Improved photometric calibration of the SNLS and the SDSS supernova
surveys. Astron. Astrophys. 552, A124 (2013). [arXiv:1212.4864]
7. M. Betoule et al. [SDSS Collaboration], Improved cosmological constraints from a joint
analysis of the SDSS-II and SNLS supernova samples. Astron. Astrophys. 568, A22 (2014).
[arXiv:1401.4064]
8. “2dfdtfe” by Willem Schaap – http://en.wikipedia.org/wiki/Sloan_Great_Wall. Licensed
under Creative Commons Attribution-Share Alike 3.0 via Wikimedia Commons – http://
commons.wikimedia.org/wiki/File:2dfdtfe.gif#mediaviewer/File:2dfdtfe.gif
9. http://www2.aao.gov.au/~TDFgg/Public/Pics/2dFzcone.jpg
10. http://map.gsfc.nasa.gov
11. C.L. Bennett et al. [WMAP Collaboration], Nine-year Wilkinson Microwave Anisotropy
Probe (WMAP) observations: final maps and results. Astrophys. J. Suppl. 208, 20 (2013).
[arXiv:1212.5225]
References 59
12. G. Hinshaw et al. [WMAP Collaboration], Nine-year Wilkinson Microwave Anisotropy Probe
(WMAP) observations: cosmological parameter results. Astrophys. J. Suppl. 208, 19 (2013).
[arXiv:1212.5226]
13. http://www.esa.int/Our_Activities/Space_Science/Planck
14. P.A.R. Ade et al. [Planck Collaboration], Planck 2015 results. XIII. Cosmological parameters.
Astron. Astrophys. 594, A13 (2016). [arXiv:1502.01589]
15. J.C. Mather, D.J. Fixsen, R.A. Shafer, C. Mosier, D.T. Wilkinson, Calibrator design for the
COBE Far Infrared Absolute Spectrophotometer (FIRAS). Astrophys. J. 512, 511 (1999).
16. D.J. Fixsen, The temperature of the cosmic microwave background. Astrophys. J. 707, 916
(2009). [arXiv:0911.1955]
17. R.P. Kirshner, Hubble’s diagram and cosmic expansion. Proc. Natl. Acad. Sci. 101, 8 (2004)
18. A.G. Riess et al. [Supernova Search Team Collaboration], Observational evidence from
supernovae for an accelerating universe and a cosmological constant. Astron. J. 116, 1009
(1998). [arXiv:astro-ph/9805201]
19. J.L. Tonry et al. [Supernova Search Team Collaboration], Cosmological results from high-z
supernovæ. Astrophys. J. 594, 1 (2003). [arXiv:astro-ph/0305008]
20. R.A. Knop et al. [Supernova Cosmology Project Collaboration], New constraints on ˝m ,
˝ , and w from an independent set of eleven high-redshift supernovae observed with HST.
Astrophys. J. 598, 102 (2003). [arXiv:astro-ph/0309368]
21. A.G. Riess et al. [Supernova Search Team Collaboration], Type Ia supernova discoveries at
z > 1 from the Hubble Space Telescope: evidence for past deceleration and constraints on
dark energy evolution. Astrophys. J. 607, 665 (2004). [arXiv:astro-ph/0402512]
22. W.M. Wood-Vasey et al. [ESSENCE Collaboration], Observational constraints on the nature
of the dark energy: first cosmological results from the ESSENCE supernova survey. Astro-
phys. J. 666, 694 (2007). [arXiv:astro-ph/0701041]
23. T.M. Davis et al., Scrutinizing exotic cosmological models using ESSENCE super-
nova data combined with other cosmological probes. Astrophys. J. 666, 716 (2007).
24. W. Rindler, Visual horizons in world-models. Mon. Not. R. Astron. Soc. 116, 662 (1956)
25. A. Einstein, Hamiltonsches Prinzip und allgemeine Relativitätstheorie. Sitz.-ber. Kgl. Preuss.
Akad. Wiss. 1916, 1111 (1916)
26. J.W. York, Role of conformal three-geometry in the dynamics of gravitation. Phys. Rev. Lett.
28, 1082 (1972)
27. G.W. Gibbons, S.W. Hawking, Action integrals and partition functions in quantum gravity.
Phys. Rev. D 15, 2752 (1977)
29. S.W. Hawking, Perturbations of an expanding universe. Astrophys. J. 145, 544 (1966)
30. G.F.R. Ellis, Relativistic cosmology, in General Relativity and Cosmology, Proceedings of the
XLVII Enrico Fermi Summer School, ed. by R.K. Sachs (Academic Press, New York, 1971)
31. G.F.R. Ellis, M. Bruni, Covariant and gauge-invariant approach to cosmological density
fluctuations. Phys. Rev. D 40, 1804 (1989)
32. M.S. Madsen, Scalar fields in curved spacetimes. Class. Quantum Grav. 5, 627 (1988)
33. S. Weinberg, Gravitation and Cosmology (Wiley, New York, 1972)
34. S. Colafrancesco, Dark matter in modern cosmology. AIP Conf. Proc. 1206, 5 (2010).
[arXiv:1004.3869]
35. Z. Ahmed et al. [CDMS-II Collaboration], Results from a low-energy analysis of the CDMS
II Germanium data. Phys. Rev. Lett. 106, 131302 (2011). [arXiv:1011.2482]
36. S. Galli, F. Iocco, G. Bertone, A. Melchiorri, CMB constraints on dark matter models with
large annihilation cross-section. Phys. Rev. D 80, 023505 (2009). [arXiv:0905.0003]
37. W. de Sitter, Einstein’s theory of gravitation and its astronomical consequences. Third paper.
Mon. Not. R. Astron. Soc. 78, 3 (1917)
38. R. Utiyama, B.S. DeWitt, Renormalization of a classical gravitational field interacting with
quantized matter fields. J. Math. Phys. 3, 608 (1962)
39. T.P. Singh, T. Padmanabhan, Notes on semiclassical gravity. Ann. Phys. (N.Y.) 196, 296
(1989)
40. Ya.B. Zel’dovich, The cosmological constant and the theory of elementary particles. Sov.
Phys. Usp. 11, 381 (1968)
41. Ya.B. Zel’dovitch, I.D. Novikov, Relativistic Astrophysics, vol. 1 (University of Chicago
Press, Chicago, 1971)
42. S. Perlmutter et al. [The Supernova Cosmology Project], Discovery of a supernova explosion
at half the age of the universe and its cosmological implications. Nature 391, 51 (1998).
43. S. Perlmutter et al. [The Supernova Cosmology Project], Measurements of ˝ and from 42
high-redshift supernovæ. Astrophys. J. 517, 565 (1999). [arXiv:astro-ph/9812133]
44. D.J. Eisenstein et al. [SDSS Collaboration], Detection of the baryon acoustic peak in the large-
scale correlation function of SDSS luminous red galaxies. Astrophys. J. 633, 560 (2005).
45. W.J. Percival et al. [SDSS Collaboration], Baryon acoustic oscillations in the Sloan Digital
Sky Survey Data Release 7 galaxy sample. Mon. Not. R. Astron. Soc. 401, 2148 (2010).
[arXiv:0907.1660]
46. N.G. Busca et al., Baryon acoustic oscillations in the Ly-˛ forest of BOSS quasars. Astron.
Astrophys. 552, A96 (2013). [arXiv:1211.2616]
47. L. Anderson et al. [BOSS Collaboration], The clustering of galaxies in the SDSS-III Baryon
Oscillation Spectroscopic Survey: baryon acoustic oscillations in the Data Release 10 and 11
galaxy samples. Mon. Not. R. Astron. Soc. 441, 24 (2014). [arXiv:1312.4877]
48. A.J. Ross, L. Samushia, C. Howlett, W.J. Percival, A. Burden M. Manera, The clustering of
the SDSS DR7 main galaxy sample I: a 4 per cent distance measure at z D 0:15. Mon. Not.
R. Astron. Soc. 449, 835 (2015). [arXiv:1409.3242]
49. M. Tegmark et al. [SDSS Collaboration], Cosmological constraints from the SDSS luminous
red galaxies. Phys. Rev. D 74, 123507 (2006). [arXiv:astro-ph/0608632]
50. L.-M. Wang, P.J. Steinhardt, Cluster abundance constraints on quintessence models. Astro-
51. S. Tsujikawa, A. De Felice, J. Alcaniz, Testing for dynamical dark energy models with
redshift-space distortions. JCAP 1301, 030 (2013). [arXiv:1210.4239]
52. M. Chevallier, D. Polarski, Accelerating universes with scaling dark matter. Int. J. Mod. Phys.
D 10, 213 (2001). [arXiv:gr-qc/0009008]
53. E.V. Linder, Exploring the expansion history of the universe. Phys. Rev. Lett. 90, 091301
(2003). [arXiv:astro-ph/0208512]
54. T. Padmanabhan, H. Padmanabhan, Cosmological constant from the emergent gravity
perspective. Int. J. Mod. Phys. D 23, 1430011 (2014). [arXiv:1404.2284]
55. A. Mazumdar, The origin of dark matter, matter-anti-matter asymmetry, and inflation.
arXiv:1106.5408
56. A. Melchiorri, L. Pagano, S. Pandolfi, When did cosmic acceleration start? Phys. Rev. D 76,
041301 (2007). [arXiv:0706.1314]
57. G.F.R. Ellis, Topology and cosmology. Gen. Relat. Grav. 2, 7 (1971)
58. M. Lachièze-Rey, J.-P. Luminet, Cosmic topology. Phys. Rep. 254, 135 (1995).
[arXiv:gr-qc/9605010]
59. G.D. Starkman, Topology and cosmology. Class. Quantum Grav. 15, 2529 (1998)
60. J.-P. Luminet, B.F. Roukema, Topology of the universe: theory and observation, in Proceed-
ings of the NATO Advanced Study Institute on Theoretical and Observational Cosmology,
ed. by M. Lachièze-Rey (Kluwer, Dordrecht, 1999); NATO Sci. Ser. C 541, 117 (1999).
61. J.J. Levin, Topology and the cosmic microwave background. Phys. Rep. 365, 251 (2002).
References 61
62. J.-P. Luminet, Cosmic topology: twenty years after. Grav. Cosmol. 20, 15 (2014).
[arXiv:1310.1245]
63. Ya.B. Zeldovich, A.A. Starobinsky, Quantum creation of a universe in a nontrivial topology.
Sov. Astron. Lett. 10, 135 (1984)
64. B.S. DeWitt, C.F. Hart, C.J. Isham, Topology and quantum field theory. Physica A 96, 197
(1979)
65. J.-P. Luminet, J. Weeks, A. Riazuelo, R. Lehoucq, J.-P. Uzan, Dodecahedral space topology
as an explanation for weak wide-angle temperature correlations in the cosmic microwave
background. Nature 425, 593 (2003). [arXiv:astro-ph/0310253]
66. E. Gausmann, R. Lehoucq, J.-P. Luminet, J.-P. Uzan, J. Weeks, Topological lensing in
spherical spaces. Class. Quantum Grav. 18, 5155 (2001). [arXiv:gr-qc/0106033]
67. R. Lehoucq, J. Weeks, J.-P. Uzan, E. Gausmann, J.-P. Luminet, Eigenmodes of three-
dimensional spherical spaces and their application to cosmology. Class. Quantum Grav. 19,
4683 (2002). [arXiv:gr-qc/0205009]
68. D.D. Sokolov, V.F. Shvartsman, An estimate of the size of the universe from a topological
point of view. Zh. Eksp. Teor. Fiz. 66, 412 (1974) [Sov. Phys. JETP 39, 196 (1975)]
69. D.D. Sokolov, A.A. Starobinsky, Globally inhomogeneous “spliced” universes. Sov. Astron.
19, 629 (1976)
70. N.J. Cornish, D.N. Spergel, G.D. Starkman, Does chaotic mixing facilitate ˝ < 1 inflation?
Phys. Rev. Lett. 77, 215 (1996). [arXiv:astro-ph/9601034]
71. J.J. Levin, E. Scannapieco, J. Silk, The topology of the universe: the biggest manifold of them
all. Class. Quantum Grav. 15, 2689 (1998). [arXiv:gr-qc/9803026]
72. G.I. Gomero, A.F.F. Teixeira, M.J. Rebouças, A. Bernui, Spikes in cosmic crystallography.
Int. J. Mod. Phys. D 11, 869 (2002). [arXiv:gr-qc/9811038]
73. J.R. Bond, D. Pogosian, T. Souradeep, CMB anisotropy in compact hyperbolic universes. 1.
Computing correlation functions. Phys. Rev. D 62, 043005 (2000). [arXiv:astro-ph/9912124]
74. J.R. Bond, D. Pogosian, T. Souradeep, CMB anisotropy in compact hyperbolic universes. 2.
COBE maps and limits. Phys. Rev. D 62, 043006 (2000). [arXiv:astro-ph/9912144]
75. J. Barrow, H. Kodama, The isotropy of compact universes. Class. Quantum Grav. 18, 1753
(2001). [arXiv:gr-qc/0012075]
76. G.I. Gomero, M.J. Rebouças, R.K. Tavakol, Detectability of cosmic topology in almost flat
universes. Class. Quantum Grav. 18, 4461 (2001). [arXiv:gr-qc/0105002]
77. J.D. Barrow, H. Kodama, All universes great and small. Int. J. Mod. Phys. D 10, 785 (2001).
78. G.I. Gomero, M.J. Rebouças, R.K. Tavakol, Are small hyperbolic universes observationally
detectable? Class. Quantum Grav. 18, L145 (2001). [arXiv:gr-qc/0106044]
79. G.I. Gomero, M.J. Rebouças, Detectability of cosmic topology in flat universes. Phys. Lett.
A 311, 319 (2003). [arXiv:gr-qc/0202094]
80. J. Weeks, R. Lehoucq, J.-P. Uzan, Detecting topology in a nearly flat spherical universe. Class.
Quantum Grav. 20, 1529 (2003). [arXiv:astro-ph/0209389]
81. G.I. Gomero, M.J. Rebouças, R. Tavakol, Limits on the detectability of cosmic topology in
hyperbolic universes. Int. J. Mod. Phys. A 17, 4261 (2002). [arXiv:gr-qc/0210016]
82. J.R. Weeks, Detecting topology in a nearly flat hyperbolic universe. Mod. Phys. Lett. A 18,
2099 (2003). [arXiv:astro-ph/0212006]
83. A. Riazuelo, J.-P. Uzan, R. Lehoucq, J. Weeks, Simulating cosmic microwave background
maps in multi-connected spaces. Phys. Rev. D 69, 103514 (2004). [arXiv:astro-ph/0212223]
84. A. de Oliveira-Costa, M. Tegmark, M. Zaldarriaga, A. Hamilton, The significance
of the largest scale CMB fluctuations in WMAP. Phys. Rev. D 69, 063516 (2004).
85. B. Mota, M.J. Rebouças, R. Tavakol, Constraints on the detectability of cosmic topology from
observational uncertainties. Class. Quantum Grav. 20, 4837 (2003). [arXiv:gr-qc/0308063]
86. B. Mota, G.I. Gomero, M.J. Rebouças, R. Tavakol, What do very nearly flat detectable cosmic
topologies look like? Class. Quantum Grav. 21, 3361 (2004). [arXiv:astro-ph/0309371]
87. N.J. Cornish, D.N. Spergel, G.D. Starkman, E. Komatsu, Constraining the topology of the
universe. Phys. Rev. Lett. 92, 201302 (2004). [arXiv:astro-ph/0310233]
88. A. Riazuelo, J. Weeks, J.-P. Uzan, R. Lehoucq, J.-P. Luminet, Cosmic microwave back-
ground anisotropies in multi-connected flat spaces. Phys. Rev. D 69, 103518 (2004).
89. B.F. Roukema, B. Lew, M. Cechowska, A. Marecki, S. Bajtlik, A hint of Poincaré dodec-
ahedral topology in the WMAP first year sky map. Astron. Astrophys. 423, 821 (2004).
90. A.D. Linde, Creation of a compact topologically nontrivial inflationary universe. JCAP 0410,
004 (2004). [arXiv:hep-th/0408164]
91. R. Aurich, S. Lustig, F. Steiner, CMB anisotropy of the Poincaré dodecahedron. Class.
Quantum Grav. 22, 2061 (2005). [arXiv:astro-ph/0412569]
92. B. Mota, M.J. Rebouças, R. Tavakol, The local shape of the universe in the inflationary limit.
Int. J. Mod. Phys. A 20, 2415 (2005). [arXiv:astro-ph/0503683]
93. R. Aurich, S. Lustig, F. Steiner, CMB anisotropy of spherical spaces. Class. Quantum Grav.
22, 3443 (2005). [arXiv:astro-ph/0504656]
94. R. Aurich, S. Lustig, F. Steiner, The circles-in-the-sky signature for three spherical universes.
Mon. Not. R. Astron. Soc. 369, 240 (2006). [arXiv:astro-ph/0510847]
95. M. Kunz, N. Aghanim, L. Cayon, O. Forni, A. Riazuelo, J.-P. Uzan, Constraining topology in
harmonic space. Phys. Rev. D 73, 023511 (2006). [arXiv:astro-ph/0510164]
96. R. Aurich, H.S. Janzer, S. Lustig, F. Steiner, Do we live in a small universe? Class. Quantum
Grav. 25, 125006 (2008). [arXiv:0708.1420]
97. B. Mota, M.J. Rebouças, R. Tavakol, Circles-in-the-sky searches and observable cosmic
topology in the inflationary limit. Phys. Rev. D 78, 083521 (2008). [arXiv:0808.1572]
98. B. Mota, M.J. Rebouças, R. Tavakol, Circles-in-the-sky searches and observable cosmic
topology in a flat universe. Phys. Rev. D 81, 103516 (2010). [arXiv:1002.0834]
99. G. Aslanyan, A.V. Manohar, The topology and size of the universe from the cosmic
microwave background. JCAP 1206, 003 (2012). [arXiv:1104.0015]
100. P.A.R. Ade et al. [Planck Collaboration], Planck 2013 results. XXVI. Background geometry
and topology of the Universe. Astron. Astrophys. 571, A26 (2014). [arXiv:1303.5086]
101. P.A.R. Ade et al. [Planck Collaboration], Planck 2015 results. XVIII. Background geometry
and topology of the Universe. Astron. Astrophys. 594, A18 (2016). [arXiv:1502.01593]
102. Yu.P. Goncharov, A.A. Bytsenko, The supersymmetric Casimir effect and quantum creation
of the universe with nontrivial topology. Phys. Lett. B 160, 385 (1985)
103. Yu.P. Goncharov, A.A. Bytsenko, The supersymmetric Casimir effect and quantum creation
of the universe with nontrivial topology (II). Phys. Lett. B 169, 171 (1986)
104. Yu.P. Goncharov, A.A. Bytsenko, Casimir effect in supergravity theories and the quantum
birth of the Universe with non-trivial topology. Class. Quantum Grav. 4, 555 (1987)
105. R.H. Brandenberger, C. Vafa, Superstrings in the early universe. Nucl. Phys. B 316, 391
(1989)
106. B. McInnes, Inflation, large branes, and the shape of space. Nucl. Phys. B 709, 213 (2005).
107. P.A.R. Ade et al. [Planck Collaboration], Planck 2013 results. XVI. Cosmological parameters.
Cambridge, 2009)
109. http://lhc.web.cern.ch/lhc
110. F. Lucchin, S. Matarrese, Power-law inflation. Phys. Rev. D 32, 1316 (1985)
111. J.J. Halliwell, Scalar fields in cosmology with an exponential potential. Phys. Lett. B 185,
341 (1987)
112. J. Yokoyama, K.-i. Maeda, On the dynamics of the power law inflation due to an exponential
potential. Phys. Lett. B 207, 31 (1988)
113. Y. Kitada, K.-i. Maeda, Cosmic no-hair theorem in power-law inflation. Phys. Rev. D 45, 1416
(1992)
Chapter 3
Cosmological Perturbations
Yet it is possible to see peril in the finding of ultimate

perfection. It is clear that the ultimate pattern contains its own
fixity. In such perfection, all things move toward death.
— Frank Herbert, Dune
Contents
3.1 Metric Perturbations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
3.1.1 Linearized Einstein Equations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
3.1.2 Gauge Invariance and Gauge Fixing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
3.1.3 Cosmological Horizons. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
3.1.4 Separate Universe Approach. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
3.2 Linear Tensor Perturbations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
3.2.1 Transverse-Traceless Gauge. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
3.2.2 Equation of Motion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
3.2.3 Mukhanov–Sasaki Equation and Solution. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
3.2.4 Discovery of Gravitational Waves. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
3.3 Scalar Perturbations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
3.3.1 Non-linear Perturbations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
3.3.2 Non-linear Perturbations at Large Scales. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
3.3.3 Linear Perturbations at Large Scales. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
3.4 Gaussian Random Fields. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
3.4.1 Power Spectrum. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
3.4.2 Bispectrum and Trispectrum. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88

DOI 10.1007/978-3-319-41127-9_3
64 3 Cosmological Perturbations
The cosmological principle is an idealization of what we observe in nature. At late

times, matter does not occupy the whole space uniformly due to gravitational
clustering. Near the big bang, a homogeneous and isotropic universe is not the most
general initial condition. Even if one chose FLRW as the classical initial state at
very early times, quantum fluctuations of particle fields would be of the same order
of the particle horizon, and they would give rise to strong inhomogeneities in the
metric and matter energy density. On the other hand, CMB anisotropies are rather
small and a perfect FLRW background is still a good lowest-order approximation.
A reasonable hypothesis, verified by experiments, is that CMB anisotropies come
from these metric and density perturbations. Because the effect is so small, the
formalism of linear perturbations is well suited for our purpose [1–4]. One should
also explain why non-linear inhomogeneities [5] do not play a major role in the
early universe. We shall postpone the issue to Chap. 5. In parallel, late-time non-
linear large-scale structures and the possible detection of tiny primordial non-linear
effects require to go beyond the first perturbative order. In preparation of this, we
review some results on linear and non-linear perturbation theory.
3.1 Metric Perturbations
According to the background field method, we decompose the metric in two parts,
an unperturbed background and a perturbation:
g D gQ C h : (3.1)
In this chapter, the background gQ is assumed to be a flat FLRW metric. The

assumption K D 0 is made both for simplicity and because the universe can
be argued to have been effectively almost flat during most part of its life.1 The
perturbation h should not be confused with the spatial projection tensor (2.41).
3.1.1 Linearized Einstein Equations
We now calculate the variation of the Riemann tensor, the Ricci tensor and the
Ricci scalar with respect to the generic symmetric linear perturbation (3.1) of the D-
dimensional background metric gQ .2 Preservation of the trace equation gQ gQ D
1
This actually constitutes a major problem in the hot big bang model, which we shall meet soon.
The problem will be solved eventually, hence the flatness assumption is justified.
2
The calculus of metric variations runs along exactly the same steps when h is interpreted as
an infinitesimal variation of the metric rather than a physical metric perturbation. In that case,
ıg ! h , ıg ! h and (2.22) stems from (3.87).
3.1 Metric Perturbations 65
D D g g implies that the inverse metric is
g D gQ h ; (3.2)
where h D gQ gQ h . With this definition of the perturbed inverse metric, h is
a genuine tensor and indices can be lowered and raised without changing the overall
sign,
A h D A h D A h :
In the following, a tilde will denote quantities in the background metric, while
indices in brackets represent symmetrized and anti-symmetrized combinations:
1 1
A. B/ :D A B C A B ; AŒ B :D A B A B :
2 2
In Problem 3.1, we derive step by step the linearized Einstein equations from the
Einstein–Hilbert action (2.18) plus matter:
ıG C h D 2 ıT ; (3.3)
where
ıG D ıR 12 h RQ 12 gQ ıR : (3.4)
The space-space off-diagonal component ˘˛ˇ D ıT˛ˇ , ˛ ¤ ˇ, is the anisotropic

stress.
The perturbation h can be split in three sectors:
• Tensor perturbations, described by the symmetric, transverse, traceless spatial
tensor h˛ˇ .
• Scalar perturbations, described by four scalar fields ˚; ; B and E.
• Vector perturbations, consisting in two divergence-free spatial vectors S˛ and F˛ .
In the linear approximation, the three types of perturbations evolve independently
and they can be studied separately. Vector perturbations will play no role in the
following, since they are typically damped away in the early universe. All these
functions represent corrections to the FLRW line element, and the identification
of h˛ˇ ; ˚; ; B and E with the metric components will be shown later, in (3.19)
and (3.52).
3.1.2 Gauge Invariance and Gauge Fixing
Following [2, Sect. I.3], let M be the spacetime manifold in which a coordinate
system fx g has been fixed, and let A be a tensor. At any point of M, we define a
Q
background function A.x/ whose form does not depend on the coordinate choice:
NQ x/ D A.N
given another system fNx g, one has A.N Q x/. We perturb A at linear order in
both coordinate frames,
Q
A.x/ D A.x/ C ıA.x/ ; N x/ D A.N
A.N Q x/ C ı A.N
N x/ :
Diffeomorphisms x 7! xN , which change coordinate systems but leave the

background invariant, are also called gauge transformations. For an infinitesimal
transformation x 7! xN D x C v .x/,
ı AN ıA D Lv AQ ; (3.5)
where Lv is the Lie derivative along the vector v (see below). If ı AN ıA D 0,

the perturbation is the same in both frames and the linear perturbation theory is
gauge invariant; in this case, one avoids the so-called gauge modes, unphysical
perturbative terms due to a bad choice of coordinates.
Alternatively, one can consider two manifolds, one physical and the other the
background with respect to which we perturb the former. A gauge choice is then
a one-to-one correspondence between points of the background and points in the
physical spacetime. A gauge transformation is a change in the mapping which keeps
background points fixed. This is different from a coordinate transformation, which
would change the point. A perturbation of A is the difference between the value
Q
A.x/ at the point x in spacetime and the value A.x/ in the gauge-equivalent point in
the background.
3.1.3 Cosmological Horizons
Perturbations are quite often studied in momentum space. In conformal time,

comoving coordinates and flat spatial slices, expand a D D 4 random scalar field
'. ; x/ in a Fourier integral,
Z C1
d3 k
'. ; x/ D 'k . / eikx : (3.6)
1 .2/3
If the coefficients 'k depend only on the comoving wave-number
2
k :D jkj D ; (3.7)
com
they describe an isotropic perturbation of comoving wave-length com .

One can compare the wave-number k of a given perturbation mode with the
cosmological horizon in any epoch. We want to define the cosmological horizon
distinguishing between the causal region of the universe and the “outside.” For
power-law scenarios with 0 1, conformal
time is negative ((2.176) and (2.185)) and the particle horizon, meant as the radius of
a causally connected region, is ill defined. In these cases, it is convenient to choose
the comoving Hubble horizon
1 1
rH D D (3.8)
aH H
as the boundary of the causal patch. Here H :D a0 =a. From the definition of particle
horizon, we have
Z rH
drH0
D ; (3.9)
rHi 1
where D 1 H0 =H2 is defined in (2.166). Assuming that rHi 0 and be

constant, one has
rH
D :
1
From (2.172), the comoving Hubble horizon of the de Sitter universe is
rH D j j ; (3.10)
while the power-law horizon (2.137) is
1p
rH D : (3.11)
p
We have already seen that for p in the radiation- or matter-domination eras, the
coefficient .1 p/=p is O.1/. But this is true also for p 1 ( 1) and, without
loss of generality, we can set (3.10) as the definition of our horizon both for de Sitter
and power-law expansion for any p. In particular, the power-law Hubble horizon is
indeed (3.10) in the limit p ! C1.
Therefore, we identify small scales (modes well inside the Hubble horizon) as
those modes k obeying the inequality
kj j 1; (3.12)
and large scales (modes outside the horizon) as
kj j 1 : (3.13)
The moment when a perturbation enters or exits the horizon at a given time is
called horizon crossing and is characterized by
kj j D 1 : (3.14)
Sometimes, one modifies the definition of horizon crossing with O.1/ coefficients
(see Sect. 3.2.3) but the physics is unchanged.
Once an horizon has been specified, we can split any perturbation field '. ; x/
into a coarse-grained and a fine-grained part, integrating over wave-lengths outside
and inside the horizon, respectively:
Z Z
d3 k d3 k
'. ; x/ D 'k . /eikx C 'k . /eikx
k<k .2/3 k>k .2/3
D: 'c . ; x/ C 'q . ; x/ : (3.15)
The subscripts c and q stem for “classical” and “quantum.” We shall explain these
names in Chap. 5.
3.1.4 Separate Universe Approach
An intuitive way of dealing with inhomogeneities at large scales is the separate

universe approach [6] (Fig. 3.1). Consider a perturbation of proper wave-length
greater than the Hubble horizon RH at time D 1 centered at the (proper) point x
(a or b in the figure). Define a patch Ux of the Universe as a region of proper size
s slightly larger than RH . Inside the patch, the physics can be described by a set of
homogeneous fields fx . /, where the subscript denotes the center of the patch, not
a spatial dependence. Thus, for each patch Ux we have a “local” scale factor ax . /,
a Hubble parameter Hx . /, and so on. The scale factor defines a local number of
e-foldings Nx . / D ln ax . / (in this chapter, we shall omit the subscript “a” in Na ).
Consider now a collection of patches of overall size 0 , much larger than and
within which spatial gradients are not negligible but still mild. At ultra-large scales
inhom > 0 , there may be strong inhomogeneities (according to the stochastic
model of eternal inflation described later). Then, we have the hierarchy of scales
RH . s < 0 . inhom : (3.16)

Fig. 3.1 The separate λ0

universe approach.
A perturbation described by
two locally FLRW patches is λ
let evolve in time (Reprinted
figure with permission from
[6]. ©2000 by the American
Physical Society)
t2
λs
cH
-1
t1
a b
The region within 0 is coarse-grained by patches of size s which are locally FLRW
and characterized by perturbations of wave-length . In this context, the definition
of a perturbation is clear and, as a great added bonus with respect to the traditional
definition (3.1), it goes beyond linear order. Let x . / be some locally measured
potential which enters the metric. A perturbation at time t along the spatial direction
˛ with wave-length ˛ D jx˛1 x˛2 j is then
ı .˛/ . / :D x˛1 . / x˛2 . /: (3.17)
If the perturbation is non-linear, the coarse graining in terms of local patches is

refined and the spatial index in local parameters is promoted to an actual spatial
dependence,
x. /D . ; x/ :
Perturbations can then be identified with gradients. One has
. ; x1 / . ; x2 /
' @˛ . ; x/ ;
x˛1 x˛2
so that, up to a numerical factor (no summation over ˛),
ı .˛/ ˛ @˛ . ; x/ :
The time evolution of . ; x/ between two slices D a and D b is determined

by the cosmic expansion via the local number of e-foldings, as shown later. Thus,
one has an intuitive setting where to study the dynamics of long-wave-length
inhomogeneities.
3.2 Linear Tensor Perturbations
In this section, we give a simple example of calculation in the linear theory: tensor
perturbations. The separate universe picture is not necessarily invoked at the linear
level.
3.2.1 Transverse-Traceless Gauge
Since the background metric defines a global Lorentz frame, one can fix the
gauge for the tensor perturbation to the transverse-traceless (or traceless-harmonic)
gauge. The proof of this statement can be sketched as follows on a flat FLRW
background. In conformal time the metric is conformally flat, so that we can do a
conformal transformation (Sect. 7.4.2) and consider Minkowski spacetime, where
the problem is simplified. There, one can always choose the traceless-harmonic
gauge [7]
h0 D 0 ; @ h D 0 ; h D 0; (3.18)
which is discussed in Problem 3.2. Then, the perturbed line element in conformal
time is

ds2 D a2 . / d 2
C ı˛ˇ C h˛ˇ dx˛ dxˇ : (3.19)
In four dimensions, a gravitational wave h˛ˇ propagating along, e.g., the x3 direction
can be decomposed into two polarization scalar modes,
h˛ˇ .x/ D hC .x/eC

˛ˇ C h .x/e˛ˇ ; (3.20)
where

C 1 0 01
e D ; e D : (3.21)
0 1 10
Interpreted as a particle, the gravitational wave h˛ˇ is called the graviton.

3.2 Linear Tensor Perturbations 71
3.2.2 Equation of Motion
On a conformally flat background, the traceless-harmonic gauge greatly simpli-

fies (3.86) and (3.87):
Q ;
ıR D 12 h ıR D 0 : (3.22)
Therefore, (3.4) becomes [8]
Q ˛ˇ :
ıG˛ˇ D 12 h (3.23)
The same result could have been obtained by perturbing the action at second
order in linear perturbations (Problem 3.3). In four dimensions, we have two degrees
of freedom:
Z
.2/ 1 X p
Q 2h2 :
ı Sh D 2
d4 x Qg h h (3.24)
4
DC;
The effective action of the independent polarization modes is that of two massive
scalars.
3.2.3 Mukhanov–Sasaki Equation and Solution
Ignoring the anisotropic stress ˘˛ˇ , we can solve the perturbed Einstein equations in
this approximation for a de Sitter and a power-law flat background. We can consider
each polarization mode separately. Call ' D h and m2 D 2. The equation
of motion for each mode is that for a massive scalar field. Since this equation
appears also in the scalar sector, from now on we forget we have derived it for the
tensor modes in four dimensions and leave the dimensionality D arbitrary. In linear
perturbation theory on a flat FLRW background, we have to solve the generic
equation
Q C m2 ' D 'R 1 2
0 D ' r ' C .D 1/H 'P C m2 ' ;
a2
P
where r 2 D ˛ @2˛ is the spatial Laplacian. It is convenient to go to momentum
space, where r 2 ! jkj2 D k2 . Switching to conformal time, we need the useful
formulæ
1 1 2
@t D @ ; @2t D .@ H@ / ; (3.25)
a a2
so that
'k00 C .D 2/H'k0 C .k2 C m2 a2 /'k D 0 : (3.26)
Before looking for analytic solutions, we can determine the qualitative asymp-
totic behaviour of the perturbation 'k by appealing to the results of Sect. 3.1.3. Let
us consider a massless scalar field, m2 D 0. At very small scales (kj j 1), the
curvature of the manifold is negligible and one can ignore the Hubble friction term.
The perturbation 'k is well inside the horizon and obeys the harmonic oscillator
equation
'k00 C k2 'k ' 0 ; kj j 1; (3.27)
whose solutions are incoming and outgoing plane waves,
'k ' Ak e˙ik ; kj j 1; (3.28)
where Ak is some normalization constant. On the other hand, outside the horizon the
effective mass term can be ignored and the perturbation is approximately constant,
'k ' Ck ; kj j 1 : (3.29)
The normalization Ck can be determined by a junction condition at horizon crossing:

ˇ ˇ ˇ ˇ
ˇ'.kj j 1/ˇ ' ˇ'.kj j 1/ˇ ) jCk j D jAk j : (3.30)
Classically, the plane-wave normalization is Ak D .2k/1=2 .

In order to solve (3.26) analytically, one remembers that the engineering
dimension of a scalar field in D dimensions is ˛ :D .D 2/=2. Since in natural
units Œa D 1, the scaling dimension of the field
wk :D a˛ 'k (3.31)
is zero (in the left-hand side, we anticipated that w will depend only on k by
isotropy). Scalar fields have zero scaling dimension in an effective two-dimensional
background, where there is no Hubble friction term (see (3.26)). Indeed, using

'k0 D a˛ w0k ˛Hwk ; 'k00 D a˛ w00k 2˛Hw0k C ˛.˛ 1 C /H2 wk ;
we obtain an equation without friction:
w00k C .k2 M 2 /wk D 0 ; (3.32)

where

D2 D
M 2 :D H 2 m2 a 2 : (3.33)
2 2
The field w is often called Mukhanov–Sasaki variable and (3.32) Mukhanov–Sasaki

equation.
On a de Sitter background (2.178), the effective mass reads

2 m2 1 4 2 1
M D D.D 2/ 4 2 D ; (3.34)
H 4 2 4 2
where
r
D.D 2/ C 1 m2
D 2: (3.35)
4 H
Thus, (3.32) has been rewritten as a Bessel-type equation. The general solution is a
superposition of Bessel functions [9, formula 8.491.5]:
p p
wk D C1 kj j J .kj j/ C C2 kj j Y .kj j/ : (3.36)
Asymptotically, at large scales
y1 1
J .y/ ' y ;
2 . C 1/
and
82 y
y1
< ln 2 C EM if D0
Y .y/ ' ;
: ./
2
y if >0
where EM 0:577 is the Euler–Mascheroni constant. At small scales,

s
y1 2
J .y/ ' cos y ;
y 2 4
s
y1 2
Y .y/ ' sin y :
y 2 4
On a power-law background (2.187), the Mukhanov–Sasaki equation is no longer

a Bessel equation and cannot be solved exactly, unless m2 D 0. One has
s
2
2 4 2 1 D2 D 1 p 1
M D ; D C : (3.37)
4 2 2 2 p 1p 4
In the limit p ! C1, one obtains (3.35). In this sense, the large-p limit of power-
law cosmology is de Sitter.
3.2.4 Discovery of Gravitational Waves
One hundred years after Einstein’s prediction that weak gravitational signals
propagate as transverse waves at the speed of light [10, 11], gravitational waves have
been detected for the first time in September 2015 by Advanced LIGO, the Laser
Interferometer Gravitational-Wave Observatory [12–15]. The instrument measured,
with a signal-to-noise ratio of about 24, the O.1021 / spatial strain created by the
waves coming from a binary black-hole system distant 1.3 billion light years from
us. Two black holes of about 30–40 solar masses each coalesced into a larger black
hole radiating a total energy in gravitational waves equivalent to 3 solar masses. The
strain pattern recorded by Advanced LIGO was in complete agreement with general
relativity (Fig. 3.2). Possible violations of Einstein’s theory in the event GW150914
cannot exceed 4 % in the noise-weighted signal correlation [16]. In particular, the
upper bound on a mass for the graviton results to be mg < 1:2 1022 eV at the
90 % confidence level. The first announcement of the discovery was soon followed
by the study of another significant event, detected in December 2015 and generated
by a merger with smaller masses [17, 18].
The discovery of gravitational waves marked one of the most momentous
successes of general relativity as well as the first direct confirmation of the existence
of binary black-hole systems. Yet, this is only the beginning of modern gravitational-
wave astronomy. Primordial gravitational waves, i.e., those produced in the early
universe during inflation, have not been detected yet and there are only (but quite
important) upper bounds on their amplitude (Sect. 4.4.2).
Fig. 3.2 The gravitational-wave event GW150914 observed by the LIGO Hanford (H1, left
column panels) and Livingston (L1, right column panels) detectors. Times are shown relative to
September 14, 2015 at 09:50:45 UTC. For visualization, all time series are filtered with a 35–
350 Hz band-pass filter to suppress other signals. Top row, left: H1 strain. Top row, right: L1 strain.
GW150914 arrived first at L1 and 6:9C0:50:4 ms later at H1; for a visual comparison the H1 data
are also shown, shifted in time by this amount and inverted (to account for the detectors’ relative
orientations). Second row: gravitational-wave strain projected onto each detector in the 35–350 Hz
band. Solid lines show a numerical relativity wave-form for a system with parameters consistent
with those recovered from GW150914 confirmed by an independent calculation. Shaded areas
show 90 % credible regions for two wave-form reconstructions: one that models the signal as a
set of sine-Gaussian wavelets and one that models the signal using binary-black-hole template
wave-forms. These reconstructions have a 95 % overlap. Third row: residuals after subtracting the
filtered numerical relativity wave-form from the filtered detector time series. Bottom row: a time-
frequency decomposition of the signal power associated with GW150914. Both plots show a signal
with frequency increasing with time (Source: [12, 15])
3.3 Scalar Perturbations
Scalar perturbations generate matter density fluctuations and it is important to

identify the primordial fluctuation fields seeding cosmic structures. To this and
other aims, we move to fully non-linear perturbations, later specializing to linear
variables.
3.3.1 Non-linear Perturbations
The most general treatment of non-linear perturbations is the covariant formalism

[19–28]. At large scales, it reduces to other approaches named gradient expansion
[29–33] and ıN formalism [34–37]. The separate universe approach in its original
formulation is the lowest-order approximation of the gradient expansion [31], but in
practice it encompasses all these methods.
Consider a non-FLRW universe filled with a perfect fluid defining the metric
decomposition (2.41). For two scalars f1;2 and from (2.47) and (2.51),
fP1 fP1
D f1 D f D @ f1 @ f2 : (3.38)
Pf2 2 fP2
We define the Lie derivative along the world-line u (we recall that u u D 1) on
a scalar f and on a covector v as
Lu f :D fP ; Lu v :D vP C v @ u : (3.39)
Assuming only the continuity equation (2.31), one can show that the non-linear
curvature perturbation on uniform density hypersurfaces
NP NP 1 1
:D D N D D @ N @ D @ N C @ (3.40)
P P D1CP
obeys a conservation equation. In the second and third steps we have used,
respectively, (3.38) and the projected continuity equation, which is (2.76) with H
replaced by the local Hubble parameter H.t; x/ D NP :
P D .D 1/NP . C P/ : (3.41)
Notice that is purely geometric and depends neither on the background nor on
the form of the total action. On a FLRW background, it vanishes.
3.3 Scalar Perturbations 77
Taking the Lie derivative of , we find (Problem 3.4)
NP
Lu . / D ; (3.42)
CP
where in the right-hand side we defined the non-linear non-adiabatic pressure
PP PP
:D D p D D @ p @ : (3.43)
P P
Another important quantity is the non-linear curvature perturbation on comoving

hypersurfaces, or comoving curvature perturbation
R :D D N : (3.44)
It obeys another continuity equation, which can be written in terms of matter

variables after using the Einstein equations [25]. By definition,
D
C R D : (3.45)
.D 1/. C P/
From this expression, the name of is clarified: this quantity coincides with
(minus) the curvature perturbation on hypersurfaces where the energy density is
unperturbed. These hypersurfaces are called uniform-density slices.
The comoving curvature perturbation is fundamental in the study of cosmologies
dominated by a scalar field. From (2.68), one notices that the projected covariant
derivative vanishes identically on a scalar, D D 0. By virtue of (3.38), the
definition (3.44) is equivalent to
NP
R D @ N C @ : (3.46)
P
3.3.2 Non-linear Perturbations at Large Scales
In the separate universe picture, we can find a rather useful formula for the curvature
perturbation at large scales, where gradient terms can be neglected. By “large” we
mean slightly larger than the Hubble horizon of the local patch, where long-wave-
length inhomogeneities are still smoothed out.
Define the perturbed spatial metric as
g˛ˇ D aQ 2 e2NL ˛ˇ ; (3.47)

where NL D NL .t; x/, aQ is the background scale factor and ˛ˇ is the 3-metric
with tensor perturbations included. The “local” scale factor
a.t; x/ D aQ .t/ eNL .t;x/ (3.48)
encodes also the perturbation ı K of the intrinsic curvature, which allows one to
identify NL with the non-linear curvature perturbation RNL .
At super-horizon scales, we can ignore gradient terms. The expansion rate is
(D D 4 from now on)
aP Q P NL :
' DH (3.49)
3 a
Integrating this in time, one obtains the non-linear, large-scale perturbation of the
number of e-foldings,
ıN :D N .t; x/ NQ .t/ D NL .ti ; x/ NL .t; x/ : (3.50)
This is called the ıN formula. Notice that we can always choose the initial slicing
to be flat, NL .ti ; x/ D 0. On uniform density slices, NL D RNL D NL is the
Salopek–Bond non-linear curvature perturbation [5]. Thus, the ıN formula can be
recast as
ıN D NL : (3.51)
The non-linear curvature perturbation on a uniform-density slice at time t is given

by the perturbation of the number of e-foldings of the expansion of a local patch
centered in x, starting from any flat slice. Below, we shall see that ˛ ' @˛ NL at
linear order.
3.3.3 Linear Perturbations at Large Scales
We now consider linear fluctuations. The line element in conformal time for linear
scalar perturbations on a flat background is
˚
ds2 D a2 . / .1 C 2˚/d 2
C 2.@˛ B/ dx˛ d
C .1 2 /ı˛ˇ C 2@˛ @ˇ E dx˛ dxˇ ; (3.52)
while the fluid perturbations are
D Q C ı ; P D PQ C ıP : (3.53)
3.3 Scalar Perturbations 79
Using suitable combinations of ˚ (called gravitational potential), (called

Bardeen potential), B and E, for a given perturbative order one can construct gauge-
invariant quantities under infinitesimal coordinate transformations. These quantities
determine which gauge choice we can (or cannot) make. Some of the most common
gauge choices are:
• Synchronous gauge, where g00 D 1 and g0˛ D 0. Coordinate time coincides
with proper time and the foliation is orthogonal (space-like hypersurfaces are
orthogonal to time-like world-lines).
• Newtonian or longitudinal gauge, where B D 0 D E. It corresponds to a
time-like hypersurface where expansion is isotropic. The fields ˚ and are
then automatically gauge invariant and, from the form of the line element, one
identifies ˚ with the gravitational potential associated with the perturbation.
If the anisotropic stress of the matter energy-momentum tensor vanishes, then
˚ D .
• Flat gauge, where D 0 D E and space-like hypersurfaces remain flat.
• Total-matter gauge, where E D 0 and space-like hypersurfaces (rest frame of
free-falling particles) are orthogonal to matter velocity v D B.
• Uniform density gauge, where the perturbation of the total energy density
vanishes, ı D 0.
The set of linearized dynamical equations is rather complicated and we shall not
present it here. However, we are in a position to relate the non-linear variables of
Sects. 3.3.1 and 3.3.2 with the scalar potentials ˚ and in longitudinal gauge. The
zero component 0 vanishes at linear order, since
!
NPQ NP
0 ' ı NP ı P ı PQ D 0 :
QP P
On the other hand, the spatial components are just gradients of a scalar,
NPQ
˛ ' @˛ ; :D ıN ı : (3.54)
PQ
To show that this is the perhaps more familiar linear curvature perturbation on
uniform density slices, ignore tensor and vector perturbations (which decouple from
the equations anyway) and linearize (3.47) to get
a D aQ C ıa ' aQ .1 / : (3.55)
Thus, in the line element (3.52) is the same NL in (3.47) at linear order.
From now, on we omit tildes since all expressions will be at linear order and there
is no risk of confusion. The linear version of (3.50) is now ıN D . Plugging
that into (3.54), we finally obtain
ı
D H ; (3.56)
P
which corresponds, as promised, to the definition of the linear [4]. Simply going to
the uniform density gauge, one immediately sees that is the linear version of NL .
Therefore, NL is the large-scale approximation of . To recast the conservation
P
equation (3.42) in linear form, consider (3.43). The left-hand side is Lu .˛ / ' @˛ .
Again, the 0 component vanishes at linear order, while

ıP ı
˛ ' @˛ ıPnad ; ıPnad :D PP : (3.57)
PP P
Then,
H
P ' ıPnad at large scales : (3.58)
CP
In the case of adiabatic (or isentropic) perturbations, different particle species

(baryons, photons, neutrinos, dark matter) have the same spatial distribution and
a density perturbation ı depending on the total background density and on the
equation of state. As a consequence,
ıPnad D 0 (3.59)
and the linear curvature perturbation is constant at sufficiently large scales.

Furthermore, the curvature perturbation is the same for all species (in particular,
for dust matter and radiation, m D r ). In flat gauge and for barotropic fluids,
ı 1 ı ıa
D H D D : (3.60)
P 3.1 C w/ a
Also, one can show that, at large scales where one can neglect the anisotropic stress
and in any era dominated by a barotropic fluid, the Bardeen potential is related to
as
3 C 3w
˚ ' ' : (3.61)
5 C 3w
3.4 Gaussian Random Fields 81
During matter domination, on large scales
3
D˚ D : (3.62)
5
In the presence of a scalar field, the scalar part of the linearized R˛ (3.46) is a
gradient, Rs˛ ' @˛ R, where
H
R :D C ı (3.63)
P
is the linear comoving curvature perturbation. Since the gradient of the scalar field
vanishes for a comoving observer, (3.63) relates two quantities defined on different
hypersurfaces, the curvature perturbation on comoving slices R and the fluctuation
ı on other suitably defined slices, in particular flat.
Since at large scales the density perturbation ı is negligible, from (3.45) we get
R ' : (3.64)
Comoving and uniform density curvature perturbations coincide at the linear level
and at large scales. The intuitive reason is that, for a canonical Klein–Gordon field,
ı ıP ' 2V; ı on large scales. The left-hand side vanishes on uniform density
slices, implying that these slices are also comoving.
3.4 Gaussian Random Fields
Before applying perturbation theory to cosmological observations, in this section

we describe some statistical properties of D D 4 isotropic Gaussian distributions
on space-like manifolds [38, 39]. The time dependence of the generic random scalar
field '.t; x/ will be omitted.
1. If the field '.x/ can be expressed as the Fourier superposition (3.6) of coefficients
'k and if the real and imaginary part of the latter are statistically independent
and with the same distribution for all k (isotropy condition), then the probability
distribution of '.x/ is Gaussian. In particular, if Re.'k / and Im.'k / obey an
isotropic Gaussian distribution, so will '.x/.
2. The statistical properties of isotropic causal Gaussian fields are completely
described by the two-point correlation function.
3. A random field is ergodic (that is, its spatial averages in a given realization
are equal to the expectation values on the entire ensemble) if, and only if, its
spectrum is continuous.
The last property will be discussed in the next chapter. As far as the first is
concerned, the Gaussian isotropic distribution with variance 2 is defined as
1 ' 2 .x/
f Œ'.x/ D p e 2 2 ; (3.65)
2 2
where 2 does not depend on x. The distribution f is normalized to 1,

Z C1
d' f Œ' D 1 : (3.66)
1
We denote with angular brackets the average of an operator O.'/

Z C1
hO.'/i :D d' f Œ'O.'/ : (3.67)
1
In particular, if ' is Gaussian, then 'c and 'q in (3.15) are separately Gaussian in
large- and small-wave-length ensemble averages,
hi D hic C hiq : (3.68)
Since f is even in ', correlation functions of odd order vanish,
.'/ ˝ ˛
2nC1 :D ' 2nC1 D 0 ; (3.69)
while
1
.'/ ˝ 2n
˛ 2n 2 Cn
2n D ' D p 2n : (3.70)

.'/
In particular, 2 D 2.
3.4.1 Power Spectrum
Random fields and their correlation functions can also be described in momentum
space. If '.x/ obeys a Gaussian isotropic statistics, then
h'k 'k0 i D: .2/3 ı.k C k0 /P' .k/ ; (3.71)
where P' is, by definition, the Fourier transform of the two-point correlation
function (Problem 3.5). It is customary to rescale P' and define a quantity P' , the
3.4 Gaussian Random Fields 83
power spectrum, which will play a fundamental role:
k3
P' .k/ :D P' .k/ : (3.72)
2 2
In Problem 3.5 we prove the Wiener–Khintchine theorem

Z C1
.'/ dk sin.k%/
2 .%/ :D h'.x1 /'.x2 /i D P' .k/ ; (3.73)
0 k k%
where % :D jx1 x2 j. For a power-law spectrum
P' .k/ D A' kn' ; (3.74)
with n' constant, an evaluation of (3.71) yields [9, formula 3.761.4]

.'/ .n' 1/ 1
2 .%/ D A' .n' 1/ sin : (3.75)
2 %n'
When the spectrum (3.74) is scale invariant,
n' D 0 ; (3.76)
an expansion of (3.75) around n' 0 gives, up to an irrelevant additive constant,
.'/
2 .%/ ln % : (3.77)
If the spectral index is positive, n' > 0, the spectrum (3.74) will have an ultraviolet
(UV) divergence and there will be more power at small scales (large k). Then the
spectrum is said to be blue-tilted. On the other hand, if n' < 0 the spectrum will
diverge in the infrared (IR) and will be said to be red-tilted.
3.4.2 Bispectrum and Trispectrum
In the next chapter, we shall be interested also in higher-order correlators. The

bispectrum in momentum space is
h'k1 'k2 'k3 i D: .2/3 ı.k1 C k2 C k3 /B' .k1 ; k2 ; k3 / ; (3.78)

where ki :D jki j. One can show that B' is the Fourier transform of the three-point
correlation function (Problem 3.5). For a Gaussian distribution, the bispectrum is
identically zero. Experimentally, its estimate can help to constrain the statistical
distribution of an almost Gaussian observable. Another estimator is the four-point
correlation function. For a Gaussian field,
h'.x1 /'.x2 /'.x3 /'.x4 /iG D 2 .%12 /2 .%34 / C 2 .%13 /2 .%24 /
(3.73)
C2 .%14 /2 .%23 / : (3.79)
In momentum space,
h'k1 'k2 'k3 'k4 iG D .2/6 ı.k1 C k2 /ı.k3 C k4 /P' .k1 /P' .k3 /
C.k3 $ k2 / C .k3 $ k1 / : (3.80)
In the case of coincident directions xi D x, (3.79) becomes

˝ ˛ ˝ ˛2
' 4 .x/ D 3 ' 2 .x/ (Gaussian) : (3.81)
The four-point correlation function (3.79) is always non-zero irrespectively of the

mutual angular separations. This is called the disconnected trispectrum. If the
statistics deviates from Gaussianity, there will also be a connected contribution
which vanishes at large angular separation. In momentum space, this non-Gaussian
trispectrum is defined as the difference between the total trispectrum and the four-
point auto-correlation function (3.81):
h'k1 'k2 'k3 'k4 iNG D: .2/3 ı.k1 C k2 C k3 C k4 /T' .k1 ; k2 ; k3 ; k4 / : (3.82)
3.1 Linearized Einstein equations. Expanding the Einstein–Hilbert

action (2.18) with matter around a generic background as in (3.1), find the
linearized Einstein equations (3.3).
Solution From the definitions of the Levi-Civita connection (2.15) and the Riemann
tensor (2.16), one has
p p p
ı g D 1
2 gQ Qg h D 12 gQ Qg h ; (3.83)

ı D 12 gQ rQ h C rQ h rQ h ; (3.84)
and
ıR D rQ ı rQ ı

D 1 gQ rQ rQ h C rQ rQ h
2 rQ rQ h

12 gQ rQ rQ h C rQ rQ h rQ rQ h

D gQ rQ Œ rQ h C rQ rQ Œ h C rQ rQ Œ h ; (3.85)
ıR D rQ ı rQ ı

D gQ rQ Œ rQ h C rQ rQ Œ h C rQ rQ Œ h ; (3.86)
ıR D RQ h C gQ ıR

D RQ h C gQ gQ rQ Œ rQ h C 2rQ rQ Œ h
D .RQ C rQ rQ gQ /
Q h : (3.87)
In the last line, we used the fact that the commutator of two vector operators
contracted with a symmetric rank-2 tensor vanishes. This can be seen also by
remembering the commutator of covariant derivatives on a tensor:
X
n
1 p1 pC1 n
2rŒ r A11
m D
n
Rp A1 m
pD1
X
m
1 n
R l A1 l1 lC1 m : (3.88)
lD1
In particular, if A is symmetric, then
rŒ r A D A . R / : (3.89)
From (3.83), (3.86) and (3.87), we obtain (3.3).
3.2 Transverse-traceless gauge. Show that it is always possible to choose

the transverse-traceless gauge 3.18 in D dimensions.
Solution Suppose to select a global D-dimensional Lorentz frame with D-velocity

u D ı0 . If the (linearized) gravity theory is covariant, it is by definition invariant
under a general coordinate transformation. In particular, a gauge transformation of
the gravitational wave h is
hQ D h C @ C @ : (3.90)
Let h :D h . The choice of such that @ @ D @ h C.1=2/@ h is called
harmonic:
@ hQ D 1
2
Q
@ h: (3.91)
We can always add an arbitrary harmonic function to , which corresponds to a

residual gauge freedom: hN D hQ C @ C @ , where @ @ D 0. We want to
fix the gauge so that hN u D hN 0 D 0. Hence,
hQ 0 C @ 0 C 0 D 0 : (3.92)
Also, hN D hQ C 2@ , so that the trace of hN vanishes if
@ D 12 hQ : (3.93)
Both conditions (3.92) and (3.93) are satisfied when @ hQ 0 D .1=2/hQ 0, but this
relation always holds because hQ satisfies the harmonic gauge (3.91). Therefore, in
a global Lorentz frame it is always possible to find a gauge in which h satisfies
the constraints (3.18).
3.3 Graviton action. Derive (3.24) by perturbing the Einstein–Hilbert

action (2.18) in four dimensions at second order.
Solution We get
Z h p i
1 p
ı .2/ Sh D d4 x RQ 2 ı .2/ g C Qgı .2/ R
2 2
1 p
Z
1

D 2 Q
d x Qg R 2 h ıR h
4
2 2
Z p
1 1 Q ˛ˇ h˛ˇ h˛ˇ ;
D 2 d4 x Qg h˛ˇ h (3.94)
2 2
where in the first line we dropped a vanishing term proportional to the trace of h .
Using (3.20), we obtain (3.24).
3.4 Covariant formalism. Prove (3.42).
Solution From definition (3.39) of the Lie derivative, and given a scalar f , one finds
the useful relation
Lu .@ f / D u @ @ f C .@ u /@ f D u @ @ f C .@ u /@ f

D @ fP : (3.95)
Then, from (3.40) we get

1 1 1 1
Lu . / D Lu .@ N / C Lu .@ / C Lu @
D1CP D1 CP
1 1 P C PP
D @ NP C @ P @
(3.95)
D1CP .D 1/. C P/2

!
NP PP
D @ P @ :
(2.76)
(3.96)
CP P
The bracket in the right-hand side coincides with (3.43).
3.5 Wiener–Khintchine theorem. Show that P' and B' , defined in (3.71)
and (3.78), are the Fourier transform of the two-point and three-point cor-
relation function, respectively. Prove the Wiener–Khintchine theorem (3.73).
Solution From definition (3.71),

Z
d3 k1 d3 k2
h'.x1 /'.x2 /i D h'k1 'k2 iei.k1 x1 Ck2 x2 /
.2/3 .2/3
Z
d3 k
D P' .k/eik.x1 x2 / ;
(3.71)
(3.97)
.2/3
while from definition (3.78)

Z
d3 k1 d3 k2 d3 k3 i.k1 x1 Ck2 x2 Ck3 x3 /
h'.x1 /'.x2 /'.x3 /i D e
.2/3 .2/3 .2/3
h'k1 'k2 'k3 i
Z
d3 k1 d3 k2 iŒk1 .x1 x3 /Ck2 .x2 x3 /
D
(3.78)
e
.2/3 .2/3
B' .k1 ; k2 ; k3 / : (3.98)
In particular,
Z
˝ 3
˛ d3 k1 d3 k2
' .x/ D B' .k1 ; k2 ; k3 / : (3.99)
.2/3 .2/3
Let ˇ be the angle between k and x1 x2 and let % D jx1 x2 j. We recall that the
infinitesimal “surface” (.D 2/-dimensional volume) element of the .D 2/-sphere
(boundary of the unit .D 1/-ball) is
d˝D2 D d sin '1 d'1 sin2 '2 d'2 sinD3 'D3 d'D3 : (3.100)
The surface of the .D 2/-sphere is

Z
2 .D1/=2
˝D2 D d˝D2 D : (3.101)
Œ.D 1/=2
In particular, ˝2 D 4. Since d3 k D d˝2 dkk2 , (3.97) gives
.'/
2 .%/ :D h'.x1 /'.x2 /i
Z C1 Z 1
2
D dkk2 P ' .k/ d.cos ˇ/eik% cos ˇ
0 .2/3 1
Z C1
dk k3 sin.k%/
D 2
P' .k/ ;
0 k 2 k%
as anticipated.
References
(1984)
Phys. Rep. 215, 203 (1992)
3. J.M. Bardeen, Gauge-invariant cosmological perturbations. Phys. Rev. D 22, 1882 (1980)
4. J.M. Bardeen, P.J. Steinhardt, M.S. Turner, Spontaneous creation of almost scale-free density
perturbations in an inflationary universe. Phys. Rev. D 28, 679 (1983)
5. D.S. Salopek, J.R. Bond, Nonlinear evolution of long wavelength metric fluctuations in
inflationary models. Phys. Rev. D 42, 3936 (1990)
References 89
6. D. Wands, K.A. Malik, D.H. Lyth, A.R. Liddle, A new approach to the evolution of cosmolog-
ical perturbations on large scales. Phys. Rev. D 62, 043527 (2000). [arXiv:astro-ph/0003278]
7. C.W. Misner, K.S. Thorne, J.A. Wheeler, Gravitation (Freeman, New York, 1973)
8. E. Lifshitz, On the gravitational stability of the expanding universe. Zh. Eksp. Teor. Fiz. 16,
587 (1946) [J. Phys. JETP 10, 116 (1946)]
9. I.S. Gradshteyn, I.M. Ryzhik, Table of Integrals, Series, and Products (Academic Press,
London, 2007)
10. A. Einstein, Näherungsweise Integration der Feldgleichungen der Gravitation. Sitz.-ber. Kgl.
Preuss. Akad. Wiss. 1916, 688 (1916)
11. A. Einstein, Über Gravitationswellen. Sitz.-ber. Kgl. Preuss. Akad. Wiss. 1918, 154 (1918)
12. B.P. Abbott et al. [LIGO Scientific and Virgo Collaborations], Observation of gravita-
tional waves from a binary black hole merger. Phys. Rev. Lett. 116, 061102 (2016).
[arXiv:1602.03837]
13. B.P. Abbott et al. [LIGO Scientific and Virgo Collaborations], Properties of the binary black
hole merger GW150914. Phys. Rev. Lett. 116, 241102 (2016). [arXiv:1602.03840]
14. B.P. Abbott et al. [LIGO Scientific and Virgo Collaborations], Observing gravitational-
wave transient GW150914 with minimal assumptions. Phys. Rev. D 93, 122004 (2016).
[arXiv:1602.03843]
15. https://losc.ligo.org/events/GW150914
16. B.P. Abbott et al. [LIGO Scientific and Virgo Collaborations], Tests of general relativity with
GW150914. Phys. Rev. Lett. 116, 221101 (2016). [arXiv:1602.03841]
17. B.P. Abbott et al. [LIGO Scientific and Virgo Collaborations], GW151226: observation of
gravitational waves from a 22-solar-mass binary black hole coalescence. Phys. Rev. Lett. 116,
241103 (2016). [arXiv:1606.04855]
18. B.P. Abbott et al. [LIGO Scientific and Virgo Collaborations], Binary black hole mergers in
the first Advanced LIGO observing run. Phys. Rev. X 6, 041015 (2016). [arXiv:1606.04856]
19. G.F.R. Ellis, M. Bruni, Covariant and gauge-invariant approach to cosmological density
fluctuations. Phys. Rev. D 40, 1804 (1989)
20. G.F.R. Ellis, J. Hwang, M. Bruni, Covariant and gauge-independent perfect-fluid Robertson–
Walker perturbations. Phys. Rev. D 40, 1819 (1989)
21. M. Bruni, G.F.R. Ellis, P.K.S. Dunsby, Gauge invariant perturbations in a scalar field dominated
universe. Class. Quantum Grav. 9, 921 (1992)
22. M. Bruni, P.K.S. Dunsby, G.F.R. Ellis, Cosmological perturbations and the physical meaning
of gauge-invariant variables. Astrophys. J. 395, 34 (1992)
23. P.K.S. Dunsby, M. Bruni, G.F.R. Ellis, Covariant perturbations in a multifluid cosmological
medium. Astrophys. J. 395, 54 (1992)
24. D. Langlois, F. Vernizzi, Evolution of non-linear cosmological perturbations. Phys. Rev. Lett.
95, 091303 (2005). [arXiv:astro-ph/0503416]
25. D. Langlois, F. Vernizzi, Conserved non-linear quantities in cosmology. Phys. Rev. D 72,
103501 (2005). [arXiv:astro-ph/0509078]
26. D. Langlois, F. Vernizzi, Nonlinear perturbations for dissipative and interacting relativistic
fluids. JCAP 0602, 014 (2006). [arXiv:astro-ph/0601271]
27. D. Langlois, F. Vernizzi, Nonlinear perturbations of cosmological scalar fields. JCAP 0702,
017 (2007). [arXiv:astro-ph/0610064]
28. A. Naruko, A general proof of the equivalence between the ıN and covariant formalisms.
Europhys. Lett. 98, 69001 (2012). [arXiv:1202.1516]
29. G.L. Comer, N. Deruelle, D. Langlois, J. Parry, Growth or decay of cosmological inhomo-
geneities as a function of their equation of state. Phys. Rev. D 49, 2759 (1994)
30. J. Parry, D.S. Salopek, J.M. Stewart, Solving the Hamilton–Jacobi equation for general
relativity. Phys. Rev. D 49, 2872 (1994). [arXiv:gr-qc/9310020]
31. G.I. Rigopoulos, E.P.S. Shellard, The separate universe approach and the evolution of
nonlinear superhorizon cosmological perturbations. Phys. Rev. D 68, 123518 (2003).
32. G.I. Rigopoulos, E.P.S. Shellard, Non-linear inflationary perturbations. JCAP 0510, 006
(2005). [arXiv:astro-ph/0405185]
33. G.I. Rigopoulos, E.P.S. Shellard, B.J.W. van Tent, Non-linear perturbations in multiple-field
inflation. Phys. Rev. D 73, 083521 (2006). [arXiv:astro-ph/0504508]
34. A.A. Starobinsky, Dynamics of phase transition in the new inflationary universe scenario and
generation of perturbations. Phys. Lett. B 117, 175 (1982)
35. M. Sasaki, E.D. Stewart, A general analytic formula for the spectral index of the density pertur-
bations produced during inflation. Prog. Theor. Phys. 95, 71 (1996). [arXiv:astro-ph/9507001]
36. M. Sasaki, T. Tanaka, Super-horizon scale dynamics of multi-scalar inflation. Prog. Theor.
Phys. 99, 763 (1998). [arXiv:gr-qc/9801017]
37. D.H. Lyth, K.A. Malik, M. Sasaki, A general proof of the conservation of the curvature
perturbation. JCAP 0505, 004 (2005). [arXiv:astro-ph/0411220]
38. R.J. Adler, The Geometry of Random Fields (Wiley, London, 1981)
39. R.J. Adler, J. Taylor, Random Fields and Geometry (Springer, New York, 2007)
Chapter 4
Cosmic Microwave Background
Philolaus puts fire in the middle, around the center, which he

calls furnace of everything and abode of Zeus and mother of the
gods and altar and junction and measure of nature. And then
another fire at the top, surrounding the whole.
— Aëtius (ed. H. Diels), Doxographi Graeci, II 7,7
Contents
4.1 Cosmic Background Radiation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
4.1.1 Boltzmann Equation and Spectral Distortions. . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
4.1.2 Last-Scattering Surface. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
4.2 Temperature Anisotropies: Formalism. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
4.2.1 Spherical Harmonics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
4.2.2 Gaussian Spectrum. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
4.2.3 Ergodic Hypothesis and Cosmic Variance. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
4.3 Temperature Power Spectrum. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
4.3.1 What we Observe. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
4.3.2 Angular Scales. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
4.3.3 Sachs–Wolfe Plateau (` . 60). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
4.3.4 Acoustic Peaks (60 . ` . 1000). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
4.3.5 Damping Tail (` & 1000). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
4.3.6 Secondary Anisotropies. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
4.4 Cosmological Parameters and Observational Constraints. . . . . . . . . . . . . . . . . . . . . . . . . 119
4.4.1 Shape of the Angular Spectrum and Parameters. . . . . . . . . . . . . . . . . . . . . . . . . . 119
4.4.2 Primordial Spectra. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
4.5 Polarization. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
4.5.1 Formalism. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
4.5.2 Spectra. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
4.5.3 What we Observe. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
4.6 Non-Gaussianity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
4.6.1 Bispectrum. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
4.6.2 Trispectrum. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
4.6.3 Physical Origin. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
4.6.4 Current Estimates. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146

DOI 10.1007/978-3-319-41127-9_4
92 4 Cosmic Microwave Background
The standard hot big bang model predicts that today the universe has a temperature
of a few Kelvin [1]. In 1964, a background signal was discovered and found
consistent with a black-body spectrum at the temperature of about 3 K [2], which
was soon recognized as radiation from the primordial universe [3]. Later observa-
tions confirmed the black-body spectrum and defined the main characteristics of
this radiation, such as the presence of tiny anisotropies in an otherwise extremely
isotropic background (Fig. 4.1).
Until two decades ago, theoretical predictions of cosmological models were so
precise with respect to the data then available, and the number of well-motivated
scenarios so large, that estimates of cosmological parameters did not allow to make
a decisive choice among the models. With the last generations of experiments
inaugurated by COBE, observational cosmology has radically changed face by
keeping up with, and in some cases going beyond, the theory. By now, the cosmic
microwave background is recognized as one of the best bridges between models of
structure formations, high-energy scenarios within and beyond the Standard Model,
and their experimental verification. Its physics, therefore, is important not only for
cosmology and astrophysics. In this chapter, we will give a qualitative description
of CMB processes.
Fig. 4.1 The microwave sky from its discovery to WMAP. The red band in the COBE map is the
microwave emissions from the Milky Way (Credit: NASA/WMAP Science Team [4])
4.1 Cosmic Background Radiation 93
4.1 Cosmic Background Radiation
After less than a second since the big bang, the universe presents itself as a hot
plasma of matter and radiation in thermal equilibrium. Known matter is made of
free baryons and electrons, not bound in atomic structures and which interact with
themselves, dark matter and radiation. Nucleosynthesis of light elements (D, He,
Li, . . . ) happens a few minutes after the birth of the universe. Baryons remain
in the form of plasma until about 360;000 years, when the universe has cooled
down to a temperature T 3000 K 0:3 eV. At this point, protons can capture
electrons and form neutral hydrogen atoms; this process is called recombination
and happens during decoupling at z D zdec , (2.131). The composition of the
universe at decoupling is shown in Fig. 4.2. Recombination does not begin at the
temperature corresponding to the ionization energy of hydrogen, 13:6 eV, mainly
due to the strong compensation between recombination rate and photo-dissociation.
Each electron captured in the fundamental state emits a photon which reionizes
other atoms already formed. Up to recombination, photons and free electrons scatter
and the universe is opaque to electromagnetic radiation. After that, the density
of diffusion centers decreases, matter and radiation decouple, and the universe
becomes transparent to light. Therefore, it is not possible to observe by light the
cosmos beyond the last-scattering surface (Fig. 4.3).
Fig. 4.2 Composition of the

universe today and at
decoupling. The percentages
in the top pie chart are based
on WMAP data and have
been slightly changed by
PLANCK , as reported in the
text (Credit: NASA/WMAP
Science Team [4])
Fig. 4.3 Last-scattering surface (Credit: NASA/WMAP Science Team [4])
Cosmic radiation emitted at decoupling reaches us almost unperturbed from the

last-scattering surface and its intensity peak lies in the microwave spectral region.
The almost perfect CMB black-body spectrum [5, 6], which we have described in
Sect. 2.5.2, is shown in Fig. 4.4.
4.1.1 Boltzmann Equation and Spectral Distortions
The physics of the interactions between matter and radiation during the history of
the universe, in particular during recombination, is based upon the radiative transfer
or Boltzmann equation,
df
D CŒ f ; (4.1)
dt
4.1 Cosmic Background Radiation 95
Fig. 4.4 Black-body CMB spectrum measured by the COBE FIRAS instrument [6]. The
horizontal-axis variable is the frequency in cm1 . The vertical axis is the power per unit area
per unit frequency per unit solid angle in megajanskies per steradian (1 Jy D 1026 watts per
square meter per hertz). The error bars are magnified 400 times! (Credit: E.L. Wright [11])
which governs the dynamics of the photon distribution f .x ; p /. The left-hand side
takes gravity into account:
df dx @f dp @f @f @f
D
C D p p p ;
dt dt @x dt @p @x @p
where we used the geodesics equation. The right-hand side describes interactions
with other particles. If the collision operator C vanishes, then the photon number
density is conserved along geodesics in the state space.
The simple form of (4.1) is deceiving, since the mechanisms involved are
extremely heterogeneous. For instance, in the collision term one must include the
contributions of Compton scattering and thermal Bremsstrahlung. Each of these
processes then branches into a number of secondary effects; for example, Compton
scattering gives rise to Doppler shifts, distortion effects, and so on.
The construction of realistic models and the solution of radiative transfer
equations, propagating through matter-radiation decoupling on an expanding cos-
mological background, demands a remarkable computing effort. In fact, one has to
solve from a dozen coupled differential equations in analytic approaches [7–10] to
several thousands of equations in numerical systems involving baryon distributions
and dynamics, dark matter, photons and neutrinos. In this sense, the anthology of
phenomena proposed below is only qualitative.
The thermal equilibrium established by all three processes discussed in
Sect. 2.5.2 is preserved until z 107 (T 1 keV), after which any mechanism
modifying radiation energy will distort the spectrum. The fact that it is difficult to
thermalize the spectrum after this epoch, and that we observe an almost perfect
black-body emission, constitutes solid evidence in favour of the primordial origin
of the CMB. Among the distortion mechanisms, we mention three.
4.1.1.1 Sunyaev–Zel’dovich Effect
If double Compton scattering and Bremsstrahlung are dominated by inverse Comp-

ton scattering with high-energy electrons, low-frequency photons are blue-shifted
towards Wien’s region. When photons pass through a hot relativistic plasma (e.g.,
in dense clusters of galaxies) they are diffused at higher energies. Observing the
plasma in the Rayleigh–Jeans region of the CMB spectrum, one sees less photons
and a spot colder than the average [12, 13]. The effect is independent of the redshift
of the plasma. Distortions of this kind measured in the CMB black-body spectrum
allow to detect galaxy clusters.
4.1.1.2 Non-thermal Sources
Consider photons emitted by non-thermal processes such as the decay of massive

particles [14] or electromagnetic phenomena. If these occur at z & 107 , they
reach thermal equilibrium with CMB photons thanks to Compton scattering and
Bremsstrahlung, and they leave no imprint. However, if this happens at 105 .
z . 107 (t 107 – 109 s), non-thermal photons can reach kinetic but not chemical
equilibrium, since interactions changing the number of photons are slower than
the cosmic expansion rate. One will then obtain a Bose–Einstein spectrum (2.100)
with non-zero chemical potential, deviating from the Planck distribution at low
frequencies. Data from the COBE spectrophotometer FIRAS give an upper limit
for the absolute value of the chemical potential, jj < 9 105 at 95 % CL [6].
4.1.1.3 Compton Distortion (Comptonization)
Photons emitted before recombination at 103 . z . 105 (t 109 – 1012 s) are

unable to reach either kinetic or thermal equilibrium. They produce a spectral
distortion due to the energy transfer from the baryonic fluid to radiation [15]:
1 !3 n x h x io
I.!; x; y/ D 2
1Cy x
x coth 4 ;
2 e 1
x 1e 2
where x :D !=T and y is a distortion parameter. If y 1, the photon energy after

last scattering is about twice the initial one. The same effect takes place when CMB
photons are rediffused by intergalactic plasma after recombination. The FIRAS
constraint is jyj < 15 106 at 95 % CL [6].
4.2 Temperature Anisotropies: Formalism 97
4.1.2 Last-Scattering Surface
Since recombination does not happen instantaneously, the last-scattering surface

has a finite thickness z. To determine it, one defines the total ionization fraction
xe .t/ :D ne .t/=np .t/ (ne and np are the number density of free electrons and protons,
respectively) and the optical depth
Z t0
.t/ :D T dt0 ne .t0 / ; (4.2)
t
where T is Thomson scattering cross-section. Since T ne is the probability per unit

time that a photon scatters, exp. / is the photon flux attenuation factor, i.e., the
probability that a photon last scattered at time t. Eventually, one finds [16–18] that
the visibility function
g.t/ :D T ne .t/ e .t/
can be approximated, in redshift units, by a Gaussian centered at zdec and of width

z 80. Therefore, the last scattering surface can be regarded as a very thin shell
with radius (2.155) of about 30 billion light years.
4.2 Temperature Anisotropies: Formalism
The temperature anisotropies detected by COBE, WMAP, PLANCK and other

experiments are the result of inhomogeneities of matter distribution at the time of
last scattering. Compton scattering is an isotropic process in the electron rest frame,
so that earlier anisotropies were cancelled before decoupling of matter and radiation.
What remains are only inhomogeneities, perceived by the observer as temperature
anisotropies. Later on, gravitational collapse of super-dense regions has given rise
to the formation of large-scale structures such as galaxies and clusters of galaxies.
4.2.1 Spherical Harmonics
The microwave sky is conveniently described using a chart in polar coordinates

.#; '/ of the last-scattering sphere S2 , with metric gab

1 0
gD : (4.3)
0 .sin #/2
The spherical harmonics are defined as

s
mCjmj 2` C 1 .` jmj/Š
Y`m .#; '/ D .1/ 2 P`jmj .cos #/ eim' ; (4.4)
4 .` C jmj/Š
where 0 6 # 6 , 0 6 ' 6 2, ` D 0; 1; 2; : : : are called multipoles, m D

`; : : : ; ` and P`jmj .x/ are the associated Legendre functions:
.1 y2 /m=2 d`Cm 2
P`m .y/ :D .y 1/` ;
2` `Š dy`Cm
.` m/Š
P`;m .y/ D .1/m P`m .y/ :
.` C m/Š
These obey the orthogonality relation

Z 1
2 .` C m/Š
dy P`0 m .y/P`m .y/ D ı``0 : (4.5)
1 2` C 1 .` m/Š
The spherical harmonics have parity .1/m under complex conjugation,

Y`m .#; '/ D .1/m Y`;m .#; '/ ;
and constitute a complete orthonormal set,

Z

d˝2 Y`m .#; '/Y`0 m0 .#; '/ D ımm0 ı``0 ;
S2
where * denotes complex conjugation. The first spherical harmonics are

r r
1 3 3
Y00 D p ; Y10 D cos # ; Y11 D sin # ei' :
4 4 8
If #12 is the angle between the unit vectors e1 and e2 corresponding to directions
with coordinates .#1 ; '1 / and .#2 ; '2 /, then cos #12 D e1 e2 and one can show that
X̀ 1

Y`m .#1 ; '1 /Y`m .#2 ; '2 / D .2` C 1/P` .cos #12 / ; (4.6)
4
mD`
where P` is the Legendre polynomial of degree ` and parity .1/` :
1 d` 2
P` .y/ :D P`0 .y/ D .y 1/` :
2` `Š dy`
They are orthogonal according to (4.5),

Z 1
2
dy P`0 .y/P` .y/ D ı``0 :
1 2` C 1
The first Legendre polynomials are
P0 .y/ D 1 ; P1 .y/ D y ; P2 .y/ D 12 .3y2 1/ ; P3 .y/ D 12 .5y3 3y/ : (4.7)
A formula we shall use once is the Legendre expansion
X C1
sin.%je1 e2 j/
D .2` C 1/j2` .%/P` .e1 e2 / ; (4.8)
%je1 e2 j
`D0
where % is a variable and

`
1 d sin %
j` .%/ :D .%/` (4.9)
% d% %
are the spherical Bessel functions j of integer order `.
4.2.2 Gaussian Spectrum
Let us denote with
ıT.e/
.e/ :D (4.10)
T0
the temperature fluctuation (with respect to the mean value T0 ) observed in the
direction of the unit vector e whose tip lies on the sphere S2 of comoving radius
centered at the observer. The CMB temperature fluctuation has been produced at
last scattering, so that ' 0 and the field
.e/ :D .e 0 / (4.11)
can be parametrized by the inclination and azimuth angles .#; '/. In terms of
spherical harmonics,
C1
X X̀
.#; '/ D a`m Y`m .#; '/ ; (4.12)
`D0 mD`
where
Z

a`m D d˝2 .e/Y`m .e/ (4.13)
S2
are complex coefficients whose statistics is not known a priori. For (4.12) to be real,
it must be
a`m D .1/m a`;m : (4.14)
In particular, a`0 is real.

At this point, we make an assumption which is justified experimentally, and that
will be explained later by a theoretical model. Namely, we assume that temperature
anisotropies come from a causal process with isotropic Gaussian distribution.
From (4.13), this means that the real and imaginary parts of the coefficients a`m
are statistically independent (for m > 0, by virtue of (4.14)) and have isotropic
Gaussian distributions. For each `, defining the .2` C 1/-dimensional set
˚ ˇ n p p ˇ o
N D 1; 2; : : : ; 2` C 1 :D a`0 ; 2Re.a`m /; 2Im.a`m / ˇ 1 6 m 6 ` ;
˛`;mN ˇ m
(4.15)
one has [19–21]

2
˛`; N
1
2`2
m
f` Œ˛`;mN D q e ; ` 2 N; (4.16)
2`2
where `2 is the variance of the distributions f` . Written as
C`TT D `2 ; (4.17)
it is called temperature angular spectrum.

The isotropy assumption (statistical rotation invariance, Fig. 4.5) ensures that the
variance does not depend on m, so that it is the same for all ˛`;mN for a given multipole
`. As in Sect. 3.4, the distribution f` is normalized to 1 and correlation functions of
odd order vanish. On the other hand, even-order correlation functions are completely
determined by C`TT .
We denote with angular brackets the average over an “ensemble of skies,” i.e., a
collection of skies observed at different points in the Universe for a given `. For an
observable O.˛/ measured in the realization f˛g f˛`;mN j ` 2 Ng, the average is
Z 2`C1
!
C1 Y
hO.˛/i :D d˛`;mN f` Œ˛`;mN O.˛/ : (4.18)
1 N
mD1
Fig. 4.5 Statistical isotropy

of n-point angular correlation
functions. As long as its
configuration is preserved, we
can average .e1 / : : : .en /
over all possible orientations
and positions on the sky
(Credit: caption from [22])
The results of Sect. 3.4 hold with the random field ' replaced by a`m : the average of
the coefficients a`m over an ensemble is zero,
ha`m i D 0 ; (4.19)
while (Problem 4.1)
ha`m a`0 m0 i D C`TT ı`0 ` ım0 m : (4.20)
From these formulæ, there descend the properties of the anisotropies: all the
correlation functions of odd order vanish,
h.e1 /.e2 / .e2nC1 /i D 0 ;
while the correlation functions of even order are completely described by the angular
spectrum C`TT . In particular, the temperature-temperature spectrum and the four-
point correlation function are (Problem 4.1)
C1
1 X
K.#12 / :D h.e1 /.e2 /i D .2` C 1/C`TT P` .cos #12 / ; (4.21)
4
`D0
h.e1 /.e2 /.e3 /.e4 /i D K.#12 /K.#34 / C K.#13 /K.#24 /

C K.#14 /K.#23 / : (4.22)
Orthogonality of the P` ensures that the modes ` do not couple with each other and
can be treated separately.
Since all non-vanishing correlation functions depend only on ` for a perfectly
isotropic Gaussian distribution, it is convenient to define the quantity
X̀
a2` :D ja`m j2 ; (4.23)
mD`
whose ensemble average is proportional to the angular spectrum:
ha2` i D .2` C 1/C`TT : (4.24)

q
From (4.16), it can be shown that a` :D a2` obeys the normalized distribution
qP
2
F` Œa` D ı a` N ˛`;m
m N
r a2
2 1 a2`
`
`2
D e 2`
; ` 2 N: (4.25)
.2` 1/ŠŠ `2`C1
This can be used instead of (4.18) to calculate ensemble averages of observables

which depend only on a` , such as temperature correlation functions:
Z C1
hO.a` /i D da` F` Œa` O.a` / : (4.26)
0
In comoving coordinates and on flat spatial slices, the Fourier expansion of is

Z C1
d3 k
.e/ D k eike ; (4.27)
1 .2/3
where the coefficients k obey a statistics inherited from the temperature field
in comoving space. For the CMB fluctuation field at last scattering ( D 0 ) and
from (4.13),
Z
`
C1
d3 k O
a`m D 4.i/ k Y`m .k/ ; (4.28)
1 .2/3
where kO D k=jkj. The analogue of (4.20) is (3.71) with the replacements 'k ! k
and x ! 0 e. Let ˇ be the angle between k and e1 e2 . From the Wiener–Khintchine
theorem (3.73),
Z C1
dk sin.k 0 je1 e2 j/
h.e1 /.e2 /i D P .k/ : (4.29)
0 k k 0 je1 e2 j
Using (4.8) and comparing (4.29) with (4.21), we obtain

Z C1
dk
C`TT D 4 P .k/j2` .k 0 / : (4.30)
0 k
4.2.3 Ergodic Hypothesis and Cosmic Variance
The averages considered above are to be taken over the ensemble of all possible
skies. Since the available data belong to just one position, we have to extract
all the information from just one element of the ensemble and to keep track
of an unavoidable theoretical error. At first, one might hope to achieve a great
simplification by making the following assumption:
Ergodic hypothesis. Averages over the ensemble of all possible skies are
equivalent to spatial averages over one sky:
Z
‹ 1
hOi D hOisky :D d˝2 O : (4.31)
4 S2
For a causal process to be spatially ergodic, it is necessary (but not sufficient) that
correlation functions decay in the large-distance limit. In this case, averages over the
ensemble can be replaced by spatial integrals over one realization. However, a causal
process on a two-dimensional sphere (as is the case for CMB anisotropies) cannot
be ergodic because the angular spectrum C`TT is labelled on a discrete set, while a
random field is ergodic if, and only if, its spectrum is continuous (Sect. 3.4). We can
also sketch a direct proof of the statement. A necessary (but not sufficient) condition
for the fluctuation field ıT=T on the sphere to be ergodic is that it be homogeneous
and isotropic. This is true by virtue of (4.21). Furthermore, the integral over S2 of
the two-point correlation function yields
hK.#12 /isky D 02 ; (4.32)
which vanishes if, and only if, the monopole variance is identically zero, i.e., when
a00 D 0. In this case, the average of the fluctuation field is equal to the average over
the sphere, since
a00
h.e/isky D p D 0 D h.e/i : (4.33)
4
However, because of (4.32) it is not true that h.e1 /.e2 /isky D K.#12 /: integrating
.e1 /.e2 / by parts and using (4.33), one finds that the left-hand side of this
equation vanishes, h.e1 /.e2 /isky D 0.
Despite this result, we are forced to perform spatial averages over the only
available sky. In doing so, we make an error which can be estimated. This error
will be smaller for small angular scales, where it is possible to average over a
large number of pairs of independent directions separated by the same angle. This
corresponds to have many modes m for a given ` 1. At large scales, there are
fewer samples and estimates of averages are more difficult. This effect is called
cosmic variance and stems from the present impossibility to perform complete
measurements, no matter how precise and accurate, of theoretical quantities.
Let us quantify cosmic variance for the two-point correlation function.
From (4.21) and (4.26), the observed temperature-temperature spectrum is
determined by a single realization of coefficients a`m ,
C1
1 X
Kobs .#12 / D .2` C 1/Còbs P` .cos #12 / ;
4
`D0
where Còbs :D a2` =.2` C 1/. Then,

˝ ˛
VarŒK.#12 / :D ŒKobs .#12 / K.#12 /2
2 C1 X
1
D .2` C 1/2 C2` P2` .cos #12 / ;
4
`D0
where (superscript TT omitted) [23]
˝ ˛ ˝ ˛ 2
C2` :D .Còbs C` /2 D Còbs 2 C`2 D C2 : (4.34)
2` C 1 `
The cosmic variance C2` is smaller at large ` and lowest-order multipoles are
afflicted by a greater theoretical uncertainty.
4.3 Temperature Power Spectrum
4.3.1 What we Observe
The comparison between experimental data and theoretical predictions is rather

complex and can be divided into several steps illustrated in Fig. 4.6 for the
temperature-temperature spectrum [24, 25]:
1. Compression of the set of time-ordered data into sky maps at different frequen-
cies in order to minimize the effect of various noise signals. The PLANCK satellite
[26–29] is sensitive to O.103 / band-powers, with 9 main band channels, from
which about O.105 / maps with O.107 / pixels each is generated.
2. Compression of multi-frequency maps into a single map to minimize the
contribution of instrumental noise and foreground contamination.
3. Compression of the CMB map into the angular spectrum. In the PLANCK case,
the spectrum has about O.103 / points.
4. Conversion of spectrum measurements into constraints of O.10/ cosmological
parameters.
The quantities of Sects. 4.2.2 and 4.2.3 do not take into account instrumental
effects and experimental conditions. Temperature fluctuations (and, as we shall see
later, polarization) cannot be determined at an exact point in the sky. A receiver has
a finite angular resolution limiting the determination of the spectrum at momenta
4.3 Temperature Power Spectrum 105
Fig. 4.6 The analysis of a large CMB data set is conveniently broken down into four steps: map-
making, foreground removal, power spectrum extraction and parameter estimation (see also [24])
(Credit: [25])
smaller than a certain threshold. Conventionally, the sensitivity of the instrument

is encoded in a window function W` . For instance, the temperature two-point
correlation function would be smeared as
C1
1 X
K.#12 / D .2` C 1/C`TT W` .#12 /P` .cos #12 / ;
4
`D0
where the symmetry of the apparatus allows one to consider W` as a function of the
separation angle only. In general, the window function depends on several factors,
including the position of the sensor as a function of time and the observational
strategy. If the sensor has a Gaussian response of width , the window function is
W` D expŒ`.` C 1/ 2 [30], which cuts high multipoles. A low-multipole cut for
an experiment taking pairwise measurements separated by an angle is controlled
by the window function [23]
W` D 2Œ1 P` .cos / expŒ`.` C 1/ 2 :
There are several other sources of uncertainty which depend on the experiment. For
instance, thermal maps covering only a limited portion of the sky carry an error
2
sample , called sample variance, due to the fact that one is taking measurements for a
multipole ` using only some of the momenta m. The error is determined by cosmic
variance and the observed solid angle ˝obs [31]:
2 4 2
sample ' :
˝obs C`
Just to give the reader an idea of the progress of full-sky CMB experiments as far
as temperature anisotropies are concerned, we can compare the angular resolution
of COBE, WMAP and PLANCK. The DMR experiment (Differential Microwave
Radiometer) on board of COBE, which produced the first thermal map of the sky,
had D 7ı , corresponding to measurements of the spectrum at ` . 15. The angular
resolution of WMAP was about 33 times better, & 0:2ı , ` . 1000. The level of
detail of PLANCK maps has been improved by a factor of 2, 0:07ı (` < 2500).
We already have a satisfactory knowledge of the temperature spectrum at large and
intermediate scales. At small scales, further observations will be necessary to place
stronger constraints on some cosmological parameters [32].
To date, the most recent determination of the angular spectrum is based upon the
2015 PLANCK data release. The PLANCK experimental (binned) points are shown
in Fig. 4.7 for the conventionally rescaled power spectrum
`.` C 1/ TT
C` ; (4.35)
2
together with the effect of cosmic variance. The solid line in the plots is the best-fit
curve of the CDM concordance model, which contains O.10/ free parameters. To
understand why the spectrum has this particular shape, we have to give a qualitative
explanation of the CMB physical mechanisms.
4.3.2 Angular Scales
Suppose to observe a small patch in the sky of angular scale # (in radians; the angle
in degrees will be denoted as D 360ı#=.2/). One can show that, in the patch,
Fig. 4.7 The temperature (TT) power spectrum (4.35) detected by PLANCK . The curve is the
CDM model best fit. The colored band represents cosmic variance. The error bars show ˙1
uncertainties and include cosmic variance (©ESA and the Planck Collaboration [33])
the approximate relation # ' 2=` holds [16],
360ı
' : (4.36)
`
The greater the multipole `, the smaller is the angular scale.

We have seen in (2.155) that the comoving distance of the last-scattering surface
is approximately 0 . Then, the comoving distance subtending an angle in the
microwave sky is
' 0# ' 250 Mpc : (4.37)

1ı
For large `, there is a simple relation between comoving wave-number and multipole
order,
`
k' : (4.38)
0
The first term of the harmonic expansion (4.12) is the monopole, ` D 0, a

correction of the mean temperature measured in a single sky with respect to the
average over the ensemble. This contribution is not measurable.
The dipole term, ` D 1, is dominated by an anisotropy of order 103 due
to the relative motion of our galaxy with respect to the CMB frame. This anisotropy
varies as 1Ccos , where is the angle between the line of sight and the direction of
motion. The velocity of the Local Group (the cluster of galaxies to which the Milky
Way belongs) with respect to the CMB frame turns out to be .627 ˙ 22/ km s1
towards the Hydra–Centaurus super-cluster [34]. The velocity of the Sun, on the
other hand, is v 371 km s1 1:2 103 c. This term is the largest source of
anisotropies, ıT=T D O.mK/; it can be removed from CMB maps and will be
ignored from now on.
The first contribution of intrinsic anisotropies is the quadrupole, ` D 2. Below,
we will see that the concordance-model best fit of the CMB spectrum predicts a
quadrupole amplitude rather different from the observed value. Sometimes, this
fact is invoked to justify, or constrain, theoretical models beyond the concordance
scenario. However, we should stress that the cosmic variance C2` is still too high at
these scales. This does not mean that we have to throw away data at low multipoles,
because best-fit curves also interpolate points at intermediate scales, where error
bars are small. A deviation of the concordance model from low-` data can be
ascribed to cosmic variance and other contaminations [35].
In Table 4.1 and Fig. 4.8, we make a rough distinction between large, interme-
diate and small scales, according to (4.36) and (4.37). Very-large-scale anisotropies
are greater than the particle horizon at last scattering (Problem 4.2), while those
on small scales are typically non-linear and associated with galactic formations.
Consequently, fluctuations smaller than a few degrees strongly depend on the details
of the cosmological model (matter content, reionization, and so on), while large-
scale predictions are founded upon more general assumptions.
Due to the finite width of the last-scattering surface, anisotropies smaller than
about 2:40 are washed out (Problem 4.3).
4.3.2.1 Geometry
If the universe is not flat globally, projection effects modify temperature

anisotropies. A comoving distance on the last-scattering surface is approximately
given by (2.154) for D 0. Equation (4.37) is therefore replaced by
0
#' ' #flat : (4.39)
.0/ .0/
Figure 4.9 shows that 0 =.0/ > 1 (< 1) for a closed (respectively, open) universe.
Intuitively, light rays are focussed (respectively, diverge) in closed (open) geometry.
Therefore, in a closed (open) universe a given comoving scale subtends a larger
(smaller) angle with respect to the flat case. The aspect of actual CMB maps is very
close to that simulated for zero curvature (Fig. 4.10).
Table 4.1 Multipoles versus

` (ı ) Scales (Mpc)
angular scales versus
comoving scales Large scales 2 – 60 >6 >1500
Intermediate scales 60 – 1000 0:4 – 6 100 – 1500
Small scales >1000 <0:4 <100
Fig. 4.8 Angular scales of the CMB temperature power spectrum (Credit: NASA/WMAP Science
Team [4])
1.2
1.0
0.8
c 0.6
0.4
0.2
0.0
0.0 0.2 0.4 0.6 0.8 1.0
tt
Fig. 4.9 Comoving distance . / as a function of conformal time for an open, flat and closed
universe (increasing thickness). At last scattering 0
Fig. 4.10 Anisotropy distortion from geometry. Simulated CMB features are characterized
by larger angular scales in a closed universe. Observations favour the flat universe (Credit:
NASA/WMAP Science Team [4])
4.3.3 Sachs–Wolfe Plateau (` . 60)
Having described the statistics of temperature fluctuations, we review some of the

most important physical mechanisms giving rise to anisotropies. The latter depend
on the observed angular scale, both because they stem from different mechanisms
and because the details of the same phenomenon can change with the scale. CMB
anisotropies can be divided in two groups: primary anisotropies, originated by
processes taking place before and during decoupling, and later-time secondary
anisotropies.
4.3.3.1 Scalar Sachs–Wolfe Effect
At large scales, the dominant contribution to anisotropies is the Sachs–Wolfe effect

[36, 37]: photons travelling from the last-scattering surface to the observer pass
through metric perturbations of the background which lower their frequency. Over-
dense regions of the last-scattering surface correspond to deeper potential wells,
so that the gravitational redshift of the photons leaving these regions is partly
compensated by the intrinsic temperature excess of the gas compressed in the wells.
This gives rise to cold spots in CMB maps.
Consider a perturbed metric g D gQ C h , where gQ is the background

metric and the perturbation h contains both scalar and tensor modes. Ignore for the
moment the tensor contribution and consider the gravitational potential ˚ D h00 =2.
A calculation of the photon geodesics in the perturbed background yields, up to a
dipole term,
Z
ˇe 1 e
dQx dQx
SW D je C ˚ ˇo C d h0 ; (4.40)
2 0
d d
where e and o indicate values at the emission and observation points ( e D dec ) and
xQ . / is the photon unperturbed geodesic path parametrized by conformal time.
The first term is the intrinsic fluctuation of the last-scattering surface due to
the local photon density perturbations. For adiabatic perturbations, the ratio of the
number densities of two species is everywhere the same and the specific entropy s
is constant for every species i:

n ıi 3 ı
ıs :D ı / D 0: (4.41)
ni i 4
Then, one can show that the first term in (4.40) is
1 ı 1 ım 2
je D D ' ˚e : (4.42)
4 3 m 3
The second term in (4.40) represents the gravitational redshift due to the net change
of potential from the emission point to the observer. Since ˚o would only give
an isotropic temperature shift, we can set it to zero. The last term in (4.40) is
responsible for secondary anisotropies and will be discussed later. Thus, one obtains
(subscript e omitted)
1
SW D ˚: (4.43)
3
In the case of isocurvature perturbations, the total density remains unperturbed
while pressure (or, equivalently, entropy) fluctuates. One can show that isocurvature
fluctuations produce an anisotropy six times larger, ıT=T ' 2˚.
Equation (4.43) means that the temperature fluctuation at large scales is deter-
mined by the scalar primordial perturbation via the Sachs–Wolfe effect. The
observed angular spectrum in Fig. 4.7 tells us that `.` C 1/C`TT const in this
region, which is called Sachs–Wolfe plateau. How do we explain this result?
Recall that C`TT can be written as an integral over wave-number modes at last
scattering, (4.30). As we have just seen, the primordial spectrum P .k/ is actually
proportional to the spectrum of primordial scalar perturbations, P .k/ ' P˚ .k/=9.
At asymptotically large scales, we can approximate the spectrum to a power law,
P .k/ ' A . 0 k/n ; (4.44)

where A D const is the dimensionless amplitude of the perturbation and n is the

spectral index. Equation (4.30) can be integrated exactly to give [21, 23]
Z "p #
C1 . 2n
2 / .` C

2 /
n
C`TT ' 4A d% %n 1 j2` .%/ D A :
0 . 3n
2

/ .` C 4n
2
/
If the primordial spectrum is exactly scale invariant, n D 0, then the quantity in

square brackets is equal to 2Œ`.` C 1/1 and we obtain
`.` C 1/ TT
C` ' A ; ` 100 : (4.45)
2
Here is the reason why the left-hand side of this equation is preferred in spectral
plots such as Fig. 4.7: at large scales, a scale-invariant (Harrison–Zel’dovich
[38, 39]) scalar primordial spectrum generates a flat angular spectrum. This is only
an approximation, since n ¤ 0 and the primordial spectrum is not perfectly power-
law in general (the spectral index depends on the momentum) but it is a good
qualitative description of what we observe at large scales.
Of course, we have not explained much in terms of physics. We have simply
shifted the original question “Why is the Sachs–Wolfe region a plateau?” to a dif-
ferent level: Why is the primordial spectrum of scalar perturbations approximately
scale invariant? Or, in other words and to yet another level: What are the primordial
seeds of metric perturbations? We will answer that in the next chapter.
4.3.3.2 Gravitational Spectrum
Tensor modes hC; do not give rise to perturbations of scalar densities and their
only contribution to the temperature spectrum is the last term of (4.40). Solving
the perturbed Einstein equations shows that tensor modes decay with the comoving
wave-number scale (Problem 4.4), so that the main effect is to increase the amplitude
of the total spectrum in the region ` < 100. Perturbations with wave-numbers
corresponding to low ` were outside the horizon at last scattering. At small scales,
the tensor spectrum rapidly decays by cosmic expansion.
Measurements of the CMB temperature-temperature spectrum do not distinguish
between the contributions of the scalar (C`s , described above) and the tensor (C`t )
primordial seeds. The observed angular spectrum is actually
C`TT D C`TT;s C C`TT;t : (4.46)
As a consequence, in (4.45) one has
A ' ˇs As C ˇt At ; ` 100 ; (4.47)

Fig. 4.11 Scalar and tensor temperature-temperature spectra for a realistic cosmological model.
The main features of the spectrum at large, intermediate and small angular scales are, respectively,
the Sachs–Wolfe plateau, acoustic oscillations and the damping tail (Credit: [40])
where As and At are the amplitudes of the primordial scalar and tensor perturbations,
while ˇs and ˇt are approximately constant O.1/ coefficients. Therefore, scale
invariance at low multipoles is guaranteed if both the scalar and the tensor pri-
mordial spectrum are scale independent. Actually, this requirement is too stringent
because it turns out that realistic models predict a very low tensor-to-scalar
amplitude ratio At =As (Fig. 4.11). To date, we have only an upper bound on this
ratio (At =As 1) which we will discuss in Sect. 4.4.2.
4.3.3.3 Late Integrated Sachs–Wolfe Effect
The last term in (4.40), called integrated Sachs–Wolfe effect (ISW), is due to the time
variation of the metric. The tensor contribution to the angular spectrum is indeed
part of the ISW effect. In the scalar ISW effect, the gravitational potential between
source and observer decays along the signal path.
The late ISW effect modifies large angular scales and takes place in models where
today’s matter density is smaller than the critical density. In particular, ˝m0 < 1
is a characteristic feature of open universes or flat universes with a cosmological
constant. In these cases, matter-curvature (respectively, matter-) equality is at
13
˝K 0 ˝0
1 C zK D or 1 C z D : (4.48)
˝m0 ˝m0
Due to opposite contributions of under- and over-densities, the late ISW effect is
cancelled for fluctuations of wave-length smaller than the horizon at this epoch.
Since zK > z for a fixed density contribution, matter domination ends (and the
potential decays to zero) at earlier times in open models. Thus, K < , `K > `
and, for a fixed ˝m0 , the late ISW effect in flat models suffers cancellation at
lower multipoles ` with respect to open models. In other words, at large scales the
late ISW effect is greater for open universes than for flat universes with . Also,
in the first case there is a more pronounced shift in the position of the acoustic
peaks [24].
In Fig. 4.7, one can see a feature in the Sachs–Wolfe plateau at very low
multipoles called ISW rise. Observations ascribe the ISW rise to a cosmological
constant.
4.3.4 Acoustic Peaks (60 . ` . 1000)
4.3.4.1 Oscillations
Take a simplified model of photons interacting with baryonic matter. Before

recombination, both components feel a gravitational force F in regions where the
total density is perturbed. Density perturbations are directly associated with scalar
perturbations of the metric, in such a way that in conformal time and momentum
space
Fk k2
' ˚k C k00 ; (4.49)
1CR 3
where
3b
R :D ' 3 104 .1 C z/1 ˝b0 h2 : (4.50)
4
In the longitudinal gauge, ignoring matter pressure one has ˚k ' k . In the ensuing
collapse, the cosmic fluid is compressed until photon pressure starts dominating.
Then, one can show that the temperature fluctuation obeys the dynamics of a driven
harmonic oscillator,
Fk
k00 C k2 c2s k ' ; (4.51)
1CR
where
pP 1
cs :D Dp (4.52)
P 3.1 C R/
is the speed of sound of the total fluid. Equation (4.51) can be obtained from the
Boltzmann and Euler equations of the fluid, in the approximation R const on
small and intermediate scales and during recombination.
For adiabatic perturbations, the initial condition is (4.42), k .0/ D 2˚k =3 and
k0 .0/ D 0. In a static potential (˚k0 D 0) and if R 1, the constant force shifts the
zero point of the oscillation to ˚k : (4.51) becomes
.k C ˚k /00 C k2 c2s .k C ˚k / ' 0 : (4.53)
Therefore, the initial amplitude is k .0/ C ˚k D ˚k =3 and evolves as
1 1
k . / C ˚k D ˚k cos.kcs / ' ˚k cos.krs / ; (4.54)
3 3
where
Z
rs . / :D d 0 cs . 0 / ' cs D O.1/ (4.55)
0
is the comoving sound horizon, which is of the same order of magnitude than the
comoving particle horizon.
At last scattering, photons decouple from baryons, climb the potential wells, and
suffer a redshift equal to ˚ cancelling the zero-point offset. The behaviour of a
perturbation of wave-number k D =.cs / is determined according its size with
respect to rs .
Fluctuations of comoving wave-length longer than the horizon at last scattering,
krs . 0 / 1, do not oscillate and evolve from the initial condition only by
cosmological redshift: the amplitude is then fixed at k . / ' ˚k =3, which is the
Sachs–Wolfe effect (4.43).
Modes within the horizon leave an oscillatory pattern in the cosmic radiation
[41, 42]. From (4.54), the fluctuation peaks are located at krs D n. Odd peaks
n D 1; 3; 5; : : : correspond to compression phases (temperature maxima), while
even peaks n D 2; 4; 6; : : : represent rarefaction phases (temperature minima).
When two causally disconnected regions merge, the relative pressure gradient
forms matter structures. An isocurvature Fourier mode entering the horizon would
begin to oscillate from its minimum rather than its maximum. For isocurvature
initial conditions (.0/ D 0, 0 .0/ D const), one has k . /C˚k D 0 .0/ sin.krs /.
The peaks of the fluctuation correspond to krs D .n1=2/ and compression occurs
at even n. With respect to the adiabatic scenario, acoustic peaks are out of phase by
=2.
Summarizing, the scale determining the behaviour of a mode is the comoving
wave-number at horizon crossing (Sect. 3.1.3),

k D :
cs
The amplitude of each mode is constant while outside the sound horizon. As soon
as a mode enters the causal region, its amplitude starts oscillating like a stationary
acoustic wave. Large-wave-length modes (small k, small `) are the last to enter the
horizon and they oscillate more slowly. Therefore, the first acoustic peak in the
spectrum C`TT corresponds to the last mode entering the horizon before decoupling.
4.3.4.2 Baryon Drag
The previous calculation did not take baryons into account, R 1. The effective
mass meff / 1 C R of the oscillator (4.51) increases with the baryonic matter
density. When baryons collapse in the gravitational potential wells, they drag
photons along, causing a displacement from the zero point, C ˚ ! C
.1 C R/˚, and thus a greater compression. The effect survives after last scattering
and heightens compression peaks with respect to under-density peaks (Fig. 4.12).
For what said above, the order of the peaks allows us to discriminate between
adiabatic and isocurvature perturbations. A visual inspection of Fig. 4.7 is not
sufficient, but a cross-check of different features of the CMB spectrum shows that it
is the adiabatic scenario to be realized in Nature.
Fig. 4.12 (a) Acoustic oscillations ( in the figure is the perturbation ˚ in the text). Photon
pressure resists gravitational compression of the fluid setting up acoustic oscillations (left panel,
real space =2 . kx . =2). Springs and balls schematically represent the fluid pressure and the
effective mass, respectively. Gravity displaces the zero point such that cos.kx/ D cos.kx/
at equilibrium with oscillations in time of amplitude =3 (right panel). The displacement is
cancelled by the redshift cos.kx/ a photon experiences when climbing out of the well. Velocity
oscillations lead to a Doppler effect Vk shifted by =2 with respect to the phase of the temperature
perturbation. (b) Baryon drag increases the gravitating mass, causing more infall and a net
zero-point displacement, even after redshift. Temperature crests (compression) are enhanced
over troughs (rarefaction) and Doppler contributions (Reprinted by permission from Macmillan
Publishers Ltd: Nature [43], ©1997)
4.3.4.3 Doppler Effect
Each Fourier mode of the density field induces a mode in the field of peculiar
velocities, whose oscillations are out of phase by =2 with respect to the oscillations
of the baryonic fluid. This Doppler effect fills the valleys of the angular spectrum,
which would go to zero otherwise. While the velocity of the observer induces a
dipole anisotropy, the velocity of the fluid is associated with smaller scales and with
anisotropies dependent on the line-of-sight.
4.3.4.4 Early Integrated Sachs–Wolfe Effect
The early ISW effect happens just after recombination if matter and radiation
densities are still comparable. In the adiabatic case, the potential ˚ decays between
tdec and complete matter domination. For the particular imprint of this secondary
anisotropy on C`TT , the first acoustic peak is widened at scales larger than the particle
horizon at recombination (Problem 4.2).
4.3.5 Damping Tail (` & 1000)
4.3.5.1 Photon Diffusion and Reionization
At small scales, the details of the interaction between matter and radiation affect
the CMB temperature anisotropies [8–10]. In Fig. 4.7, we see that small-scale
fluctuations are progressively damped for ` & `dec , where `dec is the multipole
scale characterizing the horizon at recombination (Problem 4.2). This is because
photons diffuse within the horizon at tdec , when their mean free path is of order of the
particle horizon. The greater the thickness of the last-scattering surface, the greater
the optical depth (4.2) and the damping of the spectrum. At scales smaller than dec ,
is suppressed by a factor e . The details of the Silk damping [44, 45] strongly
depend on the cosmological parameters and their study can break the degeneracy
between models of structure formation.
At some point during its history, the neutral hydrogen of the Universe was ionized
by highly energetic objects associated with stellar evolution. Photons interacted
with free electrons and diffused via Thomson scattering. In the case of an early
reionization, fluctuations generated at recombination would be deleted, while others
would be created due to the motion of the new diffusion centers. In the simple model
of “instantaneous reionization” at zrei from the neutral state to the fully ionized state,
one has [16]
' 2:5 103 .1 C zrei /3=2 :

Again, is suppressed by a factor e at scales smaller than the horizon at this

epoch (large multipoles `). The farther reionization is in the past, the greater the
damping of the spectrum. Also, a non-negligible ionization fraction during the
history of the universe would lengthen the period of Compton drag [46–49] after
recombination, thus slowing down the growth of density perturbations. Actually
this does not happen and reionization has a negligible effect on the CMB, since the
optical depth is very small [50] (PLANCK TTClowPClensing):
D 0:066 ˙ 0:016 ; zrei D 8:8C1:7

1:4 (68 % CL) : (4.56)
4.3.6 Secondary Anisotropies
Secondary anisotropies are due to processes happening after decoupling and their
origin is typically rather different from that of intrinsic CMB fluctuations. We
mention only a few:
• Sunyaev–Zel’dovich effect (Sect. 4.1.1) and kinematic (or velocity) Sunyaev–
Zel’dovich effect [51]. See [52, 53] for first detections.
• Late and early ISW effects (Sects. 4.3.3 and 4.3.4).
• The Rees–Sciama effect [54], i.e., the late ISW effect for potentials associated
with the gravitational collapse of non-linear structures such as clusters of
galaxies. In this case, one has to go beyond linear perturbation theory. The
blueshift experienced by photons when falling into the potential is not cancelled
by a redshift at the exit. In general, the effect is smaller than intrinsic anisotropies
and is localized in correspondence with galaxy structures or voids [55, 56]. The
CDM model does not produce large voids, so that the size of hot and cold spots
in the CMB, together with the projected galaxy distribution therein, can constrain
the dark-energy paradigm. At the time of writing, no evidence of super-voids has
been found [35, 50].
• The Ostriker–Vishniac effect [57], a Doppler shift due to the peculiar velocity of
the diffusing intergalactic medium.
• Weak gravitational lensing [58–60]. CMB photons can be deflected by potential
gradients (non-linear structures, galaxy clusters) along our line of sight. Their
frequency does not change and the spectrum remains black-body. However,
a non-linear distortion (mode coupling) is produced which redistributes the
temperature-temperature spectrum at small angular scales (100 ). Gravitational
lenses can produce and distort non-Gaussian and O.K/ polarization signals.
• Contributions from several extra-galactic sources, detectable by multi-frequency
observations.
4.4 Cosmological Parameters and Observational Constraints 119
4.4 Cosmological Parameters and Observational Constraints
It is time to define the standard cosmological parameters tabulated in experimental

results and entering analytic and numerical manipulations. We begin with density
and equation-of-state parameters.
4.4.1 Shape of the Angular Spectrum and Parameters
So far, we got acquainted with some parameters such as
h; ˝m0 ; ˝b0 ; ˝0 ; ˝0 ; (optical depth); w ; : : : :
1=2
The position `1 ' 200˝0 [61] of the first acoustic peak gives information on the
total density parameter and, in particular, on the geometry of the universe and on
the cosmological constant (Fig. 4.13). The BOOMERanG [62, 63] and MAXIMA
experiments [64–66] took the first precision measurement of `1 [67]. In the estimate
of [68], `1 D 212 ˙ 17. This alone is not sufficient to conclude that ˝0 1. In
Fig. 4.13 Dependence of the CMB temperature spectrum on the geometry (Credit:
Fig. 4.14 Dependence of the CMB temperature spectrum on the cosmological constant (Credit:
fact, only the joint determination of the amplitude and the position of the peaks can
constrain the allowed region in the parameter space [24].1
Also a cosmological constant induces a shift in the position of the acoustic
peaks because, for a given normalization, the anisotropies of models with ¤ 0
correspond to smaller angular scales. However, the effect is less pronounced than
for curved models [24]. On the other hand, changes the large-scale normalization
via the late ISW effect (Fig. 4.14).
From (4.50) and the discussion on baryon drag, it follows that measurements of
the relative height of the peaks determine ˝b0 h2 . Changing the baryon density and
h modifies the CMB spectrum accordingly, in particular the height of the first peak
(Fig. 4.15).
4.4.2 Primordial Spectra
The set of standard cosmological parameters must be completed with information on

the primordial spectra. For scalar perturbations, we elect the curvature perturbation
1
For instance, a non-scale-invariant primordial spectrum shifts (but only mildly) the position of the
peaks, ì =ì ' ni .ns 1/, where ni 1 [67] and ns will be defined in (4.58).
Fig. 4.15 Dependence of the peaks of the CMB temperature spectrum on baryon density (Credit:
, equation (3.54), as the primordial scalar degree of freedom, and its spectrum
Ps .k/ :D P .k/ (4.57)
as the primordial spectrum of scalar perturbations. Its scale dependence is governed

by the scalar spectral index
d ln Ps
ns 1 :D : (4.58)
d ln k
The 1 in the left-hand side is by convention. For a power-law spectrum, ns is

constant but, in general, it will be k-dependent and the running of the scalar index
will be non-vanishing:
dns
˛s :D : (4.59)
d ln k
The power spectrum of generates P through a series of evolution processes

encoded in a scale-dependent transfer function,
k D T` .k/k : (4.60)
T` .k/ depends only on the modulus of the momentum because the dynamical
equations are rotation invariant. From (3.62) and (4.43), ' ˚=3 ' =5 and
one can see that the transfer function is constant at large scales,
1
T` .k/ : (4.61)
5
To make contact with observations, we assume (4.57) to be the power law
Ps .k/ D As .k 0 /ns 1 ; (4.62)
where As is the large-scale scalar amplitude which appeared in (4.47). Comparing

this expression with (4.44) and n D ns 1, and noting that P .k/ ' P .k/=25,
we find that the coefficient ˇs in (4.47) is
1
ˇs D :
25
The primordial spectrum of tensor perturbations is
X
Pt .k/ :D Ph .k/ D 2Ph .k/ ; (4.63)
DC;
while the tensor spectral index is
d ln Pt
nt :D : (4.64)
d ln k
The large-scale tensor amplitude At in (4.47) is defined via
Pt .k/ D At .k 0 /nt ; (4.65)
while the coefficient ˇt in (4.47) has a mild `-dependence which was calculated in
[69]:

48 2
ˇt D 1 C ˇ` ;
385 18
where ˇ2 1:118, ˇ3 0:878, ˇ4 0:819 and ˇ` 1 at large `.

Another physical observable is the tensor-to-scalar ratio
Pt
r :D ; (4.66)
Ps
which replaces At as an observable upon using (4.62) and (4.65) in the scale-
invariant limit. To summarize, there are five parameters associated with early-time
physics:
As ; ns ; ˛s ; r; nt :
One could define other parameters such as the running of the tensor index ˛t and
higher-order momentum derivatives of the spectrum, but presently one can place
significant constraints only on the above set of observables. In practice, observations
can estimate the parameters in a finite range of momentum scales and, for a given
set of data, one can choose a characteristic pivot scale k0 within this range. Note
that k0 is not fixed observationally, except from the fact that we can choose any
value among the scales relevant to the experiment. In general, the constraints on the
parameter space, and in particular the likelihood contours, depend (even strongly)
on the choice of the pivot scale [70].
The scalar spectrum with non-trivial running is parametrized as [71]

ns .k0 /1C 1 ln k
k 2 k0 ˛s .k0 /
Ps .k/ D Ps .k0 / ; (4.67)
k0
while the tensor spectrum is written as

nt .k0 /
k
Pt .k/ D Pt .k0 / ; (4.68)
k0
where we have already set ˛t D 0.

Together with the uncertainty due to cosmic variance, we should also consider
that observations yield a finite discrete set of quantities C`TT , ` D 2; 3; 4; : : :, which
can be fitted by an infinite number of possible spectra. The ensuing degeneration is
sometimes called “cosmic confusion” and only the combined experimental efforts
on CMB, large-scale structure, matter distribution and astronomical explorations are
able to reduce it.
In computer simulations, it is customary to restrict the parameter space either by
discretizing the prior interval of some of the parameters or by fixing them altogether
(marginalization). We can give an example of how parameter constraints change
with the enlargement of the parameter space. The experimental values of some of
the parameters have been quoted in Chap. 2. Here we report on the others.
The PLANCK mean for the scalar amplitude (r D 0) is [50] (PLANCK

TTClowPClensing)
Ps .k0 / D .2:139 ˙ 0:063/ 109 at k0 D 0:05 Mpc1 (68 % CL) :

(4.69)
The pivot scale k0 corresponds to ` 700 via (4.38). Including tensor modes but
excluding scalar running (˛s D 0) and assuming a flat CDM model, the PLANCK
TTClowPClensing likelihood yields a scalar spectral index and a tensor-to-scalar
ratio at the pivot scale k0 D 0:002 Mpc1 (` 30) [33]
ns D 0:9688 ˙ 0:0061 (68 % CL) ; (4.70)

r0:002 < 0:114 (95 % CL) : (4.71)
This result excludes the Harrison–Zel’dovich spectrum above 5.

These bounds can be slightly modified in at least two ways. One is to consider
generalized reionization scenarios [72]. The other is to allow for a scalar running,
as one can see in Fig. 4.16. For PLANCK TTClowPClensing, one gets [33]
ns .k0 / D 0:9690 ˙ 0:0063 (68 % CL) ;

˛s D 0:0076C0:0092
0:0080 (68 % CL) ; (4.72)
r < 0:176 (95 % CL) ;
ΛCDM+running+tensors
0.008
0.4
ΛCDM+tensors
0.3
−0.008
dns /d ln k
r0.002
0.2
−0.024
0.1
−0.040
0.0
0.90 0.95 1.00 1.05 1.10
ns,0.002
Fig. 4.16 Constraints on the tensor-to-scalar ratio r at k0 D 0:002 Mpc1 in the CDM model
with running, using PLANCK TT (samples, colored by the running parameter) and PLANCK
TTClensingCBAO (black contours). Dashed contours show the corresponding constraints also
including the BKP B-mode likelihood. These are compared to the constraints when ˛s D 0 (blue
contours) (Credit: [50], reproduced with permission ©ESO)
at k0 D 0:05 Mpc1 , while r0:002 < 0:186 at 95 % CL. Constraints on the running
depend on both prior assumptions and the data set. For instance, for r D 0 the
PLANCK TTClowPClensing likelihood gives ˛s D 0:0033 ˙ 0:0074 at 68 % CL,
while for PLANCK TTClowPClensingCextCBKP ˛s .k0 / D 0:0065 ˙ 0:0076
[50]. This and other estimates all agree on the fact that a power-law spectrum with
constant index is consistent with observations.
Information about the spectrum of gravitational waves may be gained both from
polarization spectra (as we shall see in the next section) and from independent
sources other than the CMB. Observations by the telescopes BICEP2 and Keck Array
at the South Pole agree with PLANCK. A joint analysis (BKP) in early 2015 placed
the upper bound
r < 0:12 (95 % CL) (4.73)
at k0 D 0:05 Mpc1 [73], while a PLANCK 2015 TTClowPClensingCextCBKP

likelihood analysis yields r0:002 < 0:09 at k0 D 0:002 Mpc1 [50], the same upper
bound found for k0 D 0:05 Mpc1 by the more recent BICEP2+Keck Array data
analysis [74]. Combining the latter data with PLANCK constraints gives the most
stringent upper bound on r to date, at k0 D 0:05 Mpc1 [74]:
r < 0:07 (95 % CL) : (4.74)
The tensor-to-scalar ratio r can be bound indirectly also through observations

of large-scale structures (galaxy correlation, cluster abundance, Lyman-˛ forest
statistics, and so on) [75]. Also, upper bounds on the tensor spectral index are
obtained by combining the CMB bound on r with other experiments. The spectrum
of primordial gravitational waves is not directly observed because, after inflation,
it evolves into a stochastic background ˝gw (gravitational-wave energy spectrum)
through radiation and matter eras. This process can be encoded by a transfer function
T .k/ [76], so that [77, 78]
1 dgw
˝gw :D D T 2 .k/Pt .k/ ; (4.75)
crit;0 d ln k
where crit;0 D 3H02 = 2 is the critical energy density today and gw is the energy
density of the gravitational waves. Then [79],
2
1 14 h ˝gw
nt ' ln 2:29 10 ; (4.76)
ln f ln f0 r
where f D k=.2/ is the frequency of the signal and f0 D k0 =.2/ D 3:10 1018
Hz corresponds to the pivot scale k0 D 0:002 Mpc1 . Observations of pulsar
timing (slightly changed by gravitational waves passing between the pulsar and
the observer), interferometer experiments (upgrades of Advanced LIGO [80, 81],
eLISA [82]) and the theory of big-bang nucleosynthesis can place constraints on
the tensor index [79]. For instance, taking an upper bound r < 0:30, from pulsar
timing nt . 0:79, while from BBN nt . 0:15. These bounds constrain models of
primordial perturbations predicting blue-tilted spectra. Standard inflation typically
does not fall into this category.
4.5 Polarization
The processes responsible for temperature anisotropies can also give rise to
polarization of CMB photons [23, 83–97]. In particular, Thomson scattering always
produces polarized radiation even if the incoming photons are not polarized.
In general, the Thomson differential cross-section depends on the polarization
directions "0 and " (directions of the electric field E) of the incident and scattered
light, dT =d˝2 / j"0 "j2 . Consider the simple case where the photon scatters at =2
with respect to the incident direction. The outgoing polarization must be orthogonal
to the outgoing direction, so that the polarization component of incoming radiation
parallel to the outgoing direction does not scatter. Thus, only one polarization state
is left.
Thomson scattering produces a net linear polarization only if the incident
radiation has a quadrupole temperature anisotropy. In fact, if incoming light were
isotropic, contributions from orthogonal directions would cancel each other and the
scattered radiation would be unpolarized (Fig. 4.17). Therefore, one expects to find
a polarization pattern in the cosmic microwave background (Fig. 4.18).
Fig. 4.17 Thomson

scattering of radiation with a
quadrupole anisotropy
generates linear polarization.
The bars orthogonal to
scattering directions are the
intensity of the electric field
in the two polarization
directions. Blue colors (thick
lines) and red colors (thin
lines) represent, respectively,
hot and cold radiation
(Credit: [97])
4.5 Polarization 127
Fig. 4.18 The polarization of the CMB as detected by PLANCK , on a patch of the sky measuring
20 degrees across and a zoom. The colors represent the temperature variations above (red) and
below (blue) the average temperature of the CMB, while the textures that cut across the colors
show the direction and intensity of the polarized light. The curly textures are characteristic of E-
mode polarization, which is the dominant type for the CMB (©ESA and the Planck Collaboration
[26])
4.5.1 Formalism
Following a classical treatment, let
E1 D E01 .t/ cosŒ!t 1 .t/ ; E2 D E02 .t/ cosŒ!t 2 .t/
be the components of the electric field at a given point of space for a monochromatic
electromagnetic wave of angular frequency !, propagating along the x3 direction.
This wave is imagined to come from a series of Thomson scattering events.
Polarized radiation is described by the Stokes parameters I; Q; U; V. These are
the temporal averages
˝ 2˛ ˝ 2˛ ˝ 2˛ ˝ 2˛
I :D E01 t
C E02 t
; Q :D E01 t
E02 t ;
U :D h2E01 E02 cos. 1 2 /it ; V :D h2E01 E02 sin. 1 2 /it ;
where I is the radiation intensity and V characterizes circular polarization. The

Stokes parameters are additive in plane-wave superpositions. In unpolarized radi-
ation, phases and amplitudes are not correlated and Q D U D V D 0.
For CMB radiation, the time averages do not depend on the choice of the time
interval; also, there is no circular polarization and we set V D 0. While I and V are
invariant under coordinate transformations, Q and U transform as the independent
components of a rank-2 tensor Pab :2

1 1 Q.e/ U.e/ sin #
P.e/ D p Q.e/gC U.e/g D p ; (4.77)
2 2 U.e/ sin # Q.e/ sin2 #
where
p
C C 1 0 0 sin #
g D ge D ; g D det g e D ;
0 .sin #/2 sin # 0
(4.78)
are the polarization tensors eC;
ab (equation (3.21)) on the sphere. Pab is traceless
(with respect to the metric g (4.3)) and symmetric, and as such it can be decomposed
into a curl and a gradient part. These are conventionally denoted, respectively, with
a superscript “B” and “E” (in some early literature, “C” and “G”) in analogy with
electric and magnetic fields (not to be confused with the radiation field itself).
Therefore, the polarization tensor P can be expanded in spherical tensor harmonics,
C1
X X̀ h i
Pab .e/ D T0 aE`m Y`m.ab/
E
.e/ C aB`m Y`m.ab/
B
.e/ ; (4.79)
`D2 mD`
p
2
This definition differs by a factor 1= 2 from that given in [95].
where we have omitted the monopole and dipole terms ` D 0; 1 because they do not
E B
contribute [98]. Y`m.ab/ and Y`m.ab/ are constructed from (4.4),
s
2.` 2/Š 1
E
Y`m.ab/ :D Y`mIaIb gab Y`mIc Ic ;
.` C 2/Š 2
s
.` 2/Š
B
Y`m.ab/ :D Y`mIaIc c b C Y`mIbIc c a ;
2.` C 2/Š
where a semicolon denotes the covariant derivative with respect to gab and D
p
det g eC e is the anti-symmetric tensor with components 12 D 21 D sin #.
Notice that under a parity transformation
.#; '/ ! . #; '/ ; (4.80)
the Y’s transform as
Y`m . #; '/ D .1/` Y`;m .#; '/ ;

E
Y`m.ab/ . #; '/ D .1/` Y`;m.ab/ .#; '/ ; (4.81)
B
Y`m.ab/ . #; '/ D .1/` Y`;m.ab/ .#; '/ :
The tensor harmonics form a complete orthonormal basis for, respectively, the
irrotational and curl components of the polarization field P:
Z
E .ab/
E
d˝2 Y`m.ab/ .e/Y`0 m0 .e/ D ımm0 ı``0 ;
S2
Z
B .ab/
B
d˝2 Y`m.ab/ .e/Y`0 m0 .e/ D ımm0 ı``0 ; (4.82)
S2
Z
B .ab/
E
d˝2 Y`m.ab/ .e/Y`0 m0 .e/ D 0 :
S2
Explicit expressions of the tensor harmonics in terms of the associated Legendre

functions have been obtained in [95, 98, 99]:
s
.` 2/Š
E
Y`m.ab/ .e/ D W`m .e/gC C iX`m .e/g ;
2.` C 2/Š
s
.` 2/Š
B
Y`m.ab/ .e/ D W`m .e/g iX`m .e/gC ;
2.` C 2/Š
where

m2
W`m .e/ D @2# cot #@# C Y`m .#; '/ D 2 @2# `.` C 1/ Y`m .#; '/
.sin #/2
s
2` C 1 .` m/Š im' cos #
D 2.1/ m
e .` C m/ P`1;m .cos #/
4 .` C m/Š .sin #/2

` m2 `.` 1/
C P `m .cos #/
.sin #/2 2
and
m
X`m .e/ D 2 .@# cot #/ Y`m .#; '/
sin #
s
2` C 1 .` m/Š im'
D 2.1/m e
4 .` C m/Š
m
Œ.` 1/ cos #P`m .cos #/ .` C m/P`1;m .cos #/ :
.sin #/2
Another formalism, based on spin-weighted spherical harmonics [100, 101], has

also become a standard [93].
4.5.2 Spectra
The coefficients in (4.79) can be obtained by the orthonormality properties (4.82):

Z
1 E .ab/
aE`m D d˝2 Pab .e/ Y`m .e/ ; (4.83)
T0 S2
Z
1 B .ab/
aB`m D d˝2 Pab .e/ Y`m .e/ : (4.84)
T0 S2
Given a polarization field Pab .e/, these two equations permit to separate the E and
B components in the polarization signal and find the corresponding spectra.
Since temperature anisotropies and polarization are both generated by the same
primordial density fluctuations, the statistical properties of the coefficients (4.83)
and (4.84) will be the same as for the temperature coefficients (4.13), which we
denote with a superscript T. In particular, for an isotropic Gaussian statistics one
can define the mixed angular spectra (cross-correlated two-point functions) as
˝ ˛ ˝ ˛
aT`0 m0 aT`m D C`TT ı``0 ımm0 ; aE`0 m0 aE`m D CÈE ı``0 ımm0 ; (4.85)
˝ ˛ ˝ ˛
aT`0 m0 aE`m D C`TE ı``0 ımm0 ; aB`0m0 aB`m D C`BB ı``0 ımm0 ; (4.86)
˝ T B ˛ ˝ E B ˛
a`0 m0 a`m D C`TB ı``0 ımm0 ; a`0 m0 a`m D CÈB ı``0 ımm0 : (4.87)
From the parity properties (4.81), it follows that if parity is preserved in the physics
responsible for anisotropies and polarization, then
C`TB D 0 D CÈB : (4.88)
If this is the case, as we assume, then the statistics is completely specified by the
spectra (4.85) and (4.86). In general, the polarization amplitude is smaller than the
temperature amplitude because polarization has been formed after recombination,
when the number of diffusors was already reduced. The thicker the last-scattering
surface, the greater the signal.
Studying the Boltzmann equation for scalar, vector and tensor perturbations,
one finds that also polarization spectra display a sequence of oscillations and
a damping tail. Intuitively, this is because each perturbation mode still obeys a
cosmological harmonic oscillator equation, but with a non-zero source coming
from the anisotropic stress of radiation. Due to the velocity field at last scattering,
the peaks of the polarization spectra are out of phase with respect to the thermal
spectrum [91].
As in (4.30), polarization spectra are related to the primordial power spectrum
P .k/ by transfer functions Ti` .k/,
Z C1
ij dk
C` D 4 Ti` .k/Tj` .k/P .k/ ; i; j D T; E; B : (4.89)
0 k
The polarization pattern changes according to the type of primordial perturbation

seeding it. For instance, the curl component aB`m is zero for scalar perturbations,
which produce only TE and EE spectra [95]. Tensor modes generate a BB spectrum,
which is somewhat smaller than the tensor-produced EE spectrum (C`BB =CÈE
8=13 at large ` [16]).3
4.5.3 What we Observe
As in the case of the temperature spectrum, observations of polarization angular

spectra suffer from a number of experimental uncertainties due to cosmic variance,
instrumental noise, limited sky coverage and angular resolution [95]. From what
3
Also vector modes can potentially give rise to a curl-curl signal (for them, C`BB =CÈE 6 at large
` [16]).
Fig. 4.19 WMAP 3-year CMB map. The white bars show the polarization direction of the oldest
light. Polarization direction is usually depicted as a discontinuous vector field, as explained in
[102] or [95] (Credit: NASA/WMAP Science Team [4])
said above the observed spectra are, ignoring vector perturbations, (4.46) and
CÈE D CÈE;s C CÈE;t ; (4.90)

C`TE D C`TE;s C C`TE;t ; (4.91)
C`BB D C`BB;t : (4.92)
These spectra can be obtained from the two-point correlation functions of Q, U and
, as described in [95] (see also [52]).
The first experiment to positively detect a polarization pattern in the CMB was
WMAP, first at large scales (Fig. 4.19) and then in correspondence with hot and cold
spots of angular size 1ı (Fig. 4.20), as predicted by the theory [89, 90].
In models with a standard reionization scenario, the (re-normalized) EE spectrum
is flat at low multipoles. Evaluating the posterior distribution of a band power with
constant spectrum over the range 2 6 ` 6 7, the 7-year WMAP data yielded [103]
`.` C 1/ EE
C` D 0:074C0:034 2
0:025 K ; 2 6 ` 6 7;
2
at 68 % CL and with cosmic variance taken into account. Comparing with the
temperature quadrupole, one sees that
CÈE
103
C`TT
at large scales. Indeed, polarization is more difficult to detect.

The TE cross-spectrum observed by PLANCK is shown in Fig. 4.21 together with
the CDM best fit. The acoustic oscillations are clearly visible. The cosmological
Fig. 4.20 With the 7-year results, WMAP has produced a visual demonstration that the
polarization pattern around hot and cold spots (angular size 1ı ) follows the pattern expected
in the cosmic concordance model [89, 90] (Credit: NASA/WMAP Science Team [4])
Fig. 4.21 The TE (top figure) and EE (bottom figure) polarization spectra `.` C 1/C`TE;EE =.2/
measured by PLANCK . Solid lines are the theoretical spectra computed from the CDM best fit
to the PLANCK data. The error bars show ˙1 errors. A detailed discussion on the polarization
signal in the PLANCK 2015 release can be found in [112] (©ESA and the Planck Collaboration
[33])
primordial B-mode polarization has not been detected by WMAP or PLANCK 2015.
In contrast, non-primordial B-modes sourced by gravitational lensing have been
observed at high multipoles (intermediate and small scales) [104], also by SPTPOL
(the polarization-sensitive receiver mounted on the South Pole Telescope) [105] and
the POLARBEAR experiment in Chile [106].
The TB cross-spectrum in Fig. 4.22 is consistent with the null result (4.88). By
late 2016, no TB nor EB spectra have been detected by the Antarctic telescopes
BICEP2 and Keck Array either [73]. These results are in agreement both with parity
conservation [52] and the notion that the primordial tensor amplitude At (the one
capable of seeding a BB spectrum) is much smaller than the scalar amplitude As .
Foreground signals and systematic effects can produce non-zero B signals, so that
observations provide also a test for residual polarization contamination.
In general, given a set of values for the spectral observables, one can obtain the
ij
form of the C` ’s via dedicated numerical codes such as CAMB [107] (also available
as an online tool [108]) implemented in the Fortran code COSMOMC [109], or the
4.6 Non-Gaussianity 135
1.0
0.8 TB
(l+1)Cl /2π (μK2)

0.6
0.4
0.2
0.0
–0.2
–0.4
10 100 500 1000
Multipole moment l
Fig. 4.22 The temperature-polarization (TB) cross-power spectrum measured by WMAP. This
spectrum is predicted to be zero in the basic CDM model and the measured spectrum is consistent
with zero. Note that the plotted spectrum is .` C 1/C`TB =.2/, not `.` C 1/C`TB =.2/ (Credit:
analogous CLASS package [110] for the Python code MONTE PYTHON [111]. The
input values may either come as predictions of a model, in which case the output
curve can be compared with data, or as best-fit values from data analysis.
4.6 Non-Gaussianity
So far, the results of this chapter have been obtained by assuming that the
statistical distribution of the multipole coefficients ai`m , i D T; E; B, is isotropic
and Gaussian. In this case, all odd-point correlation functions vanish, while all
even-point correlation functions can be found through the two-point correlation
function (Sect. 3.4).
The Gaussian hypothesis has been verified by observations, but experimental
uncertainties do not exclude small deviations from it. These deviations in the
temperature fluctuation (and the multipole coefficients a`m , via (4.28)) could
be generated by exotic physics in the primordial perturbation spectrum. A non-
Gaussian statistical distribution of the primordial field or fields can be quantified by
the three-point and higher-order correlation functions. In parallel, an observational
campaign on CMB, large-scale structure, high-z galaxies, Lyman-˛ forest, and
gravitational lensing can detect non-Gaussian signals and clarify their physical
origin [114, 115].
4.6.1 Bispectrum
The counterpart in momentum space of the three-point temperature correlation

function (the bispectrum introduced in Sect. 3.4.2) is [116–118]
Bm 1 m2 m3
`1 `2 `3 :D ha`1 m1 a`2 m2 a`3 m3 i : (4.93)
We can construct an estimator of the bispectrum observed in a single sky. Maintain-

ing the isotropy hypothesis, one can write [118]
Bm 1 m2 m3 m1 m2 m3
`1 `2 `3 D G`1 `2 `3 b`1 `2 `3 ; (4.94)
where b`1 `2 `3 is a real and symmetric arbitrary function of the momenta ì called
reduced bispectrum, while G`m11`m2 `23m3 is the Gaunt integral, real and symmetric under
permutations of momenta ì and mi :
Z
G`m11`m2 `23m3 :D d˝2 Y`1 m1 .e1 /Y`2 m2 .e2 /Y`3 m3 .e3 /
S2
r
.2`1 C 1/.2`2 C 1/.2`3 C 1/ `1 `2 `3 `1 `2 `3
D ;
4 0 0 0 m1 m2 m3
where the two tables are the 3j Wigner symbol. Equation (4.94) is the generic
rotation-invariant form of the bispectrum. From the properties of the Wigner
symbol, it follows that the bispectrum Bm 1 m2 m3
`1 `2 `3 must obey three conditions:
• m1 C m2 C m3 D 0 (triangular closure condition);

• `ˇ 1 C `2ˇC `3 D 2n, where n 2 N;
• ˇì `j ˇ 6 `k 6 ì C `j for every cyclic permutation.
If one of these properties is not satisfied, the Gaunt integral (and the bispectrum)
vanishes identically. Using the identity
r
X `1 `2 `3 m m m
.2`1 C 1/.2`2 C 1/.2`3 C 1/ `1 `2 `3

G`1 `2 `3 D
1 2 3
;
m ;m ;m
m1 m2 m3 4 0 0 0
1 2 3
one can see that the estimator (bispectrum averaged over momenta m) is
X `1 `2 `3 m m m
B`1 `2 `3 :D B`11`2 `23 3
m1 ;m2 ;m3
m 1 m 2 m 3
r
.2`1 C 1/.2`2 C 1/.2`3 C 1/ `1 `2 `3
D b`1 `2 `3 :
4 0 0 0
It is affected by cosmic variance, as discussed in [16, 22].

A quantity which is sometimes used is the limit of zero angular separation of
the bispectrum, called skewness: S3 :D h 3 .e/i=h 2 .e/i3=2 . In alternative, one can
take the equilateral three-point correlation function, where e1 e2 D e3 e1 D e2 e3 .
In momentum space, we parametrize the level of non-Gaussianity at leading

order via the momentum-dependent non-linear parameter fNL :
6 X
B .k1 ; k2 ; k3 / D fNL .k1 ; k2 ; k3 / P .k˛ /P .kˇ / ; (4.95)
5
˛<ˇ
where ˛; ˇ D 1; 2; 3 and we hint, through the subscripts , that the origin of

primordial non-Gaussianity can be traced back to the statistical properties of the
curvature perturbation on uniform density slices. In spherical momentum space, the
corresponding reduced bispectrum is [116, 118]
Z C1
b`1 `2 `3 D 2 d%%2 b`1 b`2 bN
`3 C .`3 $ `2 / C .`3 $ `1 / ; (4.96)
0
where
Z C1
2
b` .%/ :D dkk2 P .k/T` .k/j` .k%/ ;
0
Z C1
2
` .%/
bN :D dkk2 fNL .k1 ; k2 ; k3 /T` .k/j` .k%/ :
0
4.6.2 Trispectrum
The connected trispectrum in momentum space has been introduced in Sect. 3.4.2.
In spherical momenta space, a rotationally invariant estimator T``13``24 of the connected
trispectrum is [119, 120]
C1
X X
L
`1 `2 L
ha`1 m1 a`2 m2 a`3 m3 a`4 m4 iconnected D .1/ M
m1 m2 M
LD0 MDL

`3 `4 L
T``13``24 .L/ ; (4.97)
m3 m4 M
where L is the third edge of the triangles formed with `1 and `2 and `3 and `4 . Parity
invariance requires that `1 C `2 C L D 2n, `1 C `2 C L D 2n0 , m1 C m2 M D 0
and m3 C m4 C M D 0.
In momentum space, we parametrize the next-to-leading order of non-

Gaussianity via the non-linear parameters NL and gNL :
X
T .k1 ; k2 ; k3 ; k4 / D NL .k1 ; k2 ; k3 / P .ki /P .kj /ŒP .kil / C P .kjl /
i<j<l
54 X
C gNL .k1 ; k2 ; k3 / P .ki /P .kj /P .kl / : (4.98)
25 i<j<l
4.6.3 Physical Origin
Once instrumental noise and foreground contamination are under control, it remains
to explain the physical origin of a non-Gaussian signal. The plausible contributions
are many but here we mention only a few.
First of all, due to cosmic variance even primordial perturbations with a perfectly
Gaussian distribution would have a non-zero probability of producing a non-
Gaussian statistics in the sky [121]. Apart from this statistical source, there do exist
physical mechanisms of non-Gaussianity. Secondary mechanisms are the Sunyaev–
Zel’dovich effect, gravitational lensing, and radio and infrared extragalactic sources.
After matter domination, perturbations enter a phase of gravitational ampli-
fication which deforms the primordial Gaussian distribution to a non-Gaussian
one. This is the typical non-linear regime of the advanced evolutionary phase of
cosmic structure formation, where the density contrast is ı= 1. Such type of
non-Gaussianity has a specific form that can be obtained via analytic approxima-
tions and numerical simulations, is related with the problem of galaxy bias, and can
be constrained by observations of the galaxy distribution [122–124].
Non-linearity is also the source of primary non-Gaussianity. The seed of
matter cosmic structures and temperature fluctuations is, via (3.62), the primordial
curvature perturbation , which we have defined at linear order in (3.56). When
higher-order perturbations are taken into account, they naturally generate a non-
Gaussian signal.
4.6.3.1 Local Form
In momentum space, the bispectrum and trispectrum are given by (4.95) and (4.98),
respectively. The form of the non-linear parameters depend on the model of
primordial perturbations. In the simplest case [118, 120, 125, 126], one decomposes
the Salopek–Bond non-linear curvature perturbation NL .x/ (3.51) into a Gaussian
linear part and a non-linear part:
3 local 2 9
NL D C N D C fNL h 2 i C glocal 3 ; (4.99)
5 25 NL
NL are constant. By definition, hNL i D

local
where the non-linear parameters fNL and glocal
hi D 0. In the following, we set gNL D 0. Then, a direct calculation of the
bispectrum shows that (Problem 4.5)
fNL .k1 ; k2 ; k3 / D fNL

local
: (4.100)
For the trispectrum, similar steps lead to

2
6fNL
local
NL .k1 ; k2 ; k3 / D : (4.101)
5
The decomposition (4.99) is point-wise in position space and, for this reason, it is
called the local form of non-Gaussianity. For a power-law scalar spectrum (4.62),
the local bispectrum reads
6 local 2 X 1
Blocal
.k1 ; k2 ; k3 / D fNL A ; (4.102)
5 .k˛ kˇ /4ns
˛<ˇ
where A D 2 2 As 0ns 1 . Translating (4.99) in terms of the spherical harmonics

decomposition (4.13), one can split the coefficients a`m into a linear and a non-
linear part, aNL`m D a`m C a`m . Substituting (4.13) into (4.93) and using (4.122)
N
and (4.94), we obtain (4.96) with constant fNL local

. Using the full transfer function T` ,
one finds positive and negative acoustic peaks with twice the period of the angular
spectrum [118]. Numerical and semi-analytic details of the shape of the bispectrum
can be found in [118, 125, 127].
Expression (4.102) peaks at the squeezed limit where one of the edges of the
triangle .k1 ; k2 ; k3 / collapses [128, 129]:
k1 ' k2 k3 ; k3 0 : (4.103)
Sending, e.g., k3 ! 0, by conservation of momenta one has k1 k2 and
12 local
Blocal
.k1 ; k1 ; k3 0/ ' f P .k1 /P .k3 / : (4.104)
5 NL
local
Measuring the bispectrum in this configuration, one can obtain an estimate of fNL .
In the local bispectrum, small- and large-scale modes are coupled together.
The squeezed limit can be understood in a fairly intuitive way in all models
where the curvature perturbation is constant at large scales [128, 130]. Consider
the natural splitting (3.15) into a coarse-grained and a fine-grained perturbation. In
the limit (4.103), k3 is larger than the Hubble horizon and can be treated as constant
in time. From (3.48), it is clear that
.x3 / c .x3 /
(4.103)
(4.105)
defines a new coordinate background x0 ' Œ1 C c .x3 /x inside the horizon. In the
new coordinates and up to linear order,
d d
q .x0 / ' q .x/ C .x0 x/ q .x/ ' q .x/ C c .x3 / x q .x/ : (4.106)
dx dx
If the linear perturbation q .x/ is Gaussian, in the squeezed limit we have (Prob-
lem 4.6)
./ ./
h.x1 /.x2 /.x3 /i ' .ns 1/2 .0/2 .%/ : (4.107)
Comparing this expression with (4.104), we finally obtain
5
local
fNL ' .1 ns / : (4.108)
12
For spectra which are almost scale-invariant at large scales, the level of non-
Gaussianity is very low, fNL 1. Tensor modes produce an even lower signal.
4.6.3.2 Equilateral Form
Other models of primordial seeds predict a bispectrum which can be approximated

by the equilateral form [131] (i; j; l D 1; 2; 3)
8
18f
equil < X 1 2

equil 2
B .k1 ; k2 ; k3 / D NL
A C
5 : .ki kj /4ns .ki kj kl /2.4ns /=3
i<j<l
" #9
X 1 1 =
C C ; (4.109)
.ki kj2 kl3 /.4ns /=3 .ki kj kl4 /.4ns /=3 ;
i¤j¤l
equil
where fNL is constant. In the limit k1 k2 k3 , it coincides with (4.102). In the
equilateral form, small- and very-large-scale modes are not coupled. The local and
equilateral forms can be measured almost independently.
4.6.3.3 Orthogonal Form
Another Ansatz which is nearly orthogonal to the other two is the orthogonal
form [132]
8
18f ortho < X 3 8

2
Bortho
.k 1 ; k 2 ; k 3 / D NL
A C
5 : .ki kj /4ns .ki kj kl /2.4ns /=3
i<j<l
" #9
X 3 3 =
C C ; (4.110)
.ki kj2 kl3 /.4ns /=3 .ki kj kl4 /.4ns /=3 ;
i¤j¤l
ortho
where fNL is constant.
In the next chapter, we will discuss theoretical models predicting these three
types of bispectrum.
4.6.4 Current Estimates
To date, we do not have significant constraints on the trispectrum but we have obser-
vational bounds on the non-linear parameter fNL for the three types of primordial
non-Gaussianity described above. At 68 % CL and combining temperature and E-
mode polarization data, these estimates are [133]
local
fNL D 0:8 ˙ 5:0 ;
equil
fNL D 4 ˙ 43 ; (4.111)
ortho
fNL D 26 ˙ 21 :
Bounds combining CMB data and large-scale structure surveys can change consid-
erably depending on the data set, the estimator and the chosen model of galaxy
power spectrum [134]. The progress in the determination of the level of non-
Gaussianity has been remarkable since the beginning of precision cosmology. One
can compare the 1 errors in (4.111) by PLANCK with the final WMAP results,
equil
local
fNL 20 and fNL 140, back to the first measurements by COBE where the
error on local non-Gaussianity (the only one determined by that experiment) was
local
100 times larger, fNL 600.
Local non-Gaussianity can also be tested against a residual k-dependence in the
non-linear parameter. This is done by estimating the index [135]
d ln jfNL j
nfNL :D : (4.112)
d ln k
Observations do not disagree with the local model, since 2:5 < nfNL < 2:3 [136].
Constraints on the trispectrum parameters can be found in [133, 136]. For fNL local
D
0, the PLANCK 2015 bound on gNL in (4.99) is gNL D .9:0 ˙ 7:7/ 104 at
local local
68 % CL.
Secondary and “post-primordial” effects easily generate non-Gaussianity but
they can be subtracted from the maps. Primordial CMB temperature fluctuations are
Gaussian within the present experimental confidence level. A detection of a non-
Gaussian signal could constrain a number of high-energy models predicting large
deviations from a Gaussian distribution. Several of these models will be analyzed in
Chaps. 5 and 13.
4.1 Two-point function. Derive (4.20) and (4.21).
Solution A direct calculation yields

Z 2`C1
!
C1 Y
ha`m a`0 m0 i D ı`0 ` ım0 m d˛`;mN f` Œ˛`;mN ja`m j2
(4.19)
1 N
mD1
Z C1
2
D ı`0 ` ım0 m d˛`;mN f` Œ˛`;mN ˛`;
(4.15)
mN
1
D C`TT ı`0 ` ım0 m :

(3.70)
(4.113)
In the second
p N is silent and can pick any value in its range. The
equality, the index m
factor 2 in the definition (4.15) of the ˛`;mN ’s was introduced to get this simple
result. The temperature spectrum is
C1 ` 0
X X̀ X
h.e1 /.e2 /i D ha`m a`0 m0 i Y`m .#1 ; '1 /Y`0 m0 .#2 ; '2 /
(4.12)
`;`0 D0 mD` m0 D`0
C1 ` 0
X X̀ X ˝ ˛
D a`m a`0 m0 Y`m .#1 ; '1 /Y`0 m0 .#2 ; '2 /
`;`0 D0 mD` m0 D`0
C1
X X̀

D .#1 ; '1 /Y`m .#2 ; '2 /
(4.113)
C`TT Y`m
`D0 mD`
C1
1 X
D .2` C 1/C`TT P` .cos #12 / :
(4.6)
(4.114)
4
`D0
4.2 Angular scales 1. Determine the angular, the multipole and the wave-
number scale of anisotropies the size of the particle horizon at equality and
decoupling for a flat universe.
Solution From (4.36), (4.37), (4.38) and Table 2.1, one gets
eq 0:6ı ; èq 640 ; k 0:04 Mpc1 ; (4.115)
and
dec 1:3ı ; `dec 280 ; k 0:02 Mpc1 : (4.116)
Anisotropies with ` `dec depend more strongly on the causal astroparticle model
describing photon-matter interaction. On the other hand, outside the last-scattering
horizon the features of anisotropies are rather general and well described by linear
perturbation theory.
4.3 Angular scales 2. Determine the comoving, the angular and the multi-
pole scale at which anisotropies are blurred by effect of the finite thickness of
the last-scattering surface. Assume K D 0.
Solution According to Sect. 4.1.2, the visibility function can be approximated by a

Gaussian centered at zdec and of width z 80. During the period marked by the
initial and final redshift
z
z˙ D zdec ˙ 1090 ˙ 40 ;
2
photon-electron scattering washes out anisotropies smaller than the comoving
particle horizon. Therefore, on the CMB sphere we cannot see fluctuations with
wave-length smaller than the comoving scale
. z/ :D .z / .zC / : (4.117)
The calculation of comoving particle horizons at different epochs was done in

Problem 2.6, where we saw that, at decoupling, we can ignore the cosmological
constant but that radiation brings a correction factor (2.146) to the matter-dominated
result. Then,

2:36 1 1
. z/ ' 1:18Œ CDM .z / CDM .zC / D p p
H0 1 C z 1 C zC
12 Mpc : (4.118)
In terms of angular and multipole scales ((4.37) and (4.36)), we lose information on
primordial fluctuations for
. 0:047ı 2:80 ; ` & 7700 : (4.119)
Experimental limitations practically increase (respectively, reduce) the lower limit

of angular resolution (the upper limit of usable multipoles).
4.4 Linear tensor perturbations. Briefly discuss linear tensor adiabatic

perturbations at decoupling. After matter domination, the anisotropic stress
˘˛ˇ can be neglected in the linearized Einstein equations. Ignore .
Solution Neglecting and for a matter-dominated universe, tensor perturbations

are given by (3.36) and (3.37) with D D 4 and p D 2=3 ( > 0, D 3=2). In the
small wave-length limit,
r r
p 2 p 2
k J .k / ' cos.k / ; k Y .k / ' sin.k / : (4.120)

Since primordial fluctuations are adiabatic, we choose the initial conditions at dec
0 as hk .0/ D const ¤ 0, h0k .0/ D 0, which set C2 D 0. From (3.31) and (2.187),
each tensor mode is
r

h .k / D 3C .k /3=2 J3=2 .k / ; (4.121)
2
where we have chosen the normalization so that h .k / ! C outside the horizon.

This solution is plotted in Fig. 4.23 together with the discarded unphysical solution
/ .k /3=2 Y3=2 .k /, which diverges inside the horizon.
4.5 Local form of non-Gaussianity. Derive (4.100).
Solution The Fourier transform of the non-linear part of the Salopek–Bond pertur-
bation (4.99) is
" Z #
3 local 3 2 d3 p
kN D fNL .2/ ı.k/h i C p pk :
5 .2/3
Fig. 4.23 Solutions of the tensor modes equation in terms of the Bessel function J (continuous
curve, (4.121) with C D 1) and Y (dashed curve). Adiabatic perturbations are described by the
first, which goes to a non-zero constant outside the horizon, k ! 0
The first term stems from the fact that the auto-correlation function is independent
of x. Since not all momenta can vanish at the same time, this piece can be
thrown away. The second term enters into the three-point function, which at lowest
order is
hkNL NL NL i ' hk1 k2 kN3 i C .k3 $ k2 / C .k3 $ k1 /

1 k2 k3
Z
3 local d3 p
D fNL hk k p k3 p i
5 .2/3 1 2
3 local
D .2/3 fNL P .k1 /P .k2 /
(3.80)
5
Z
d3 p Œı.k1 C p/ı.k2 C k3 p/
Cı.k2 C p/ı.k1 C k3 p/ C .k3 $ k2 / C .k3 $ k1 /

3 local
D .2/3 ı.k1 C k2 k3 / fNL 2P .k1 /P .k2 /
5
C.k3 $ k2 / C .k3 $ k1 / : (4.122)
This yields (4.100) after comparing (3.78) and (4.95).

4.6 Squeezed limit. Derive (4.107) using (3.75).
Solution In the squeezed limit,

(4.103) ˝ ˛
h.x1 /.x2 /.x3 /i q .x01 /q .x01 /c .x3 /

d
c2 .x3 /x1
(4.106)
' q .x1 /q .x2 /
dx1
˝ ˛ d ˝ ˛
' c2 .x3 / c x1
(3.68)
q .x1 /q .x2 / q
dx1
./ d ./
D 2 .0/ .%/ ;
d ln % 2
where in the second line we exploited translation invariance and in the last line we
used % D jx1 x2 j and @%=@x1 D x1 =%. From (3.75),
./
d2 .%/ ./
D .ns 1/2 .%/ ; (4.123)
d ln %
so that (4.107) holds.
References
1. R.A. Alpher, R.C. Hermann, Remarks on the evolution of the expanding universe. Phys. Rev.
75, 1089 (1949)
2. A.A. Penzias, R.W. Wilson, A measurement of excess antenna temperature at 4080 Mc/s.
Astrophys. J. 142, 419 (1965)
3. R.H. Dicke, P.J.E. Peebles, P.G. Roll, D.T. Wilkinson, Cosmic black-body radiation. Astro-
phys. J. 142, 414 (1965)
5. J.C. Mather et al., Measurement of the cosmic microwave background spectrum by the COBE
FIRAS instrument. Astrophys. J. 420, 439 (1994)
6. D.J. Fixsen, E.S. Cheng, J.M. Gales, J.C. Mather, R.A. Shafer, E.L. Wright, The cosmic
microwave background spectrum from the full COBE/FIRAS data set. Astrophys. J. 473,
576 (1996). [arXiv:astro-ph/9605054]
7. W. Hu, N. Sugiyama, Anisotropies in the cosmic microwave background: an analytic
approach. Astrophys. J. 444, 489 (1995). [arXiv:astro-ph/9407093]
8. W. Hu, N. Sugiyama, Toward understanding CMB anisotropies and their implication. Phys.
Rev. D 51, 2599 (1995). [arXiv:astro-ph/9411008]
9. W. Hu, N. Sugiyama, Small scale cosmological perturbations: an analytic approach. Astro-
10. W. Hu, M.J. White, The damping tail of CMB anisotropies. Astrophys. J. 479, 568 (1997).
11. http://www.astro.ucla.edu/~wright/spectrum.gif
References 147
12. R.A. Sunyaev, Ya.B. Zel’dovich, Microwave background radiation as a probe of the contem-
porary structure and history of the universe. Ann. Rev. Astron. Astrophys. 18, 537 (1980)
13. M. Birkinshaw, The Sunyaev–Zel’dovich effect. Phys. Rep. 310, 97 (1999).
14. W. Hu, J. Silk, Thermalization constraints and spectral distortions for massive unstable relic
particles. Phys. Rev. Lett. 70, 2661 (1993)
15. Ya.B. Zel’dovich, R.A. Sunyaev, The interaction of matter and radiation in a hot-model
universe. Astrophys. Space Sci. 4, 301 (1969)
Cambridge, 2009)
2005)
19. L.F. Abbott, M.B. Wise, Anisotropy of the microwave background in the inflationary
cosmology. Phys. Lett. B 135, 279 (1984)
20. L.F. Abbott, M.B. Wise, Large-scale anisotropy of the microwave background and the
amplitude of energy density fluctuations in the early universe. Astrophys. J. 282, L47 (1984)
21. R. Fabbri, F. Lucchin, S. Matarrese, Multipole anisotropies of the cosmic background
radiation and inflationary models. Astrophys. J. 315, 1 (1987)
23. J.R. Bond, G. Efstathiou, The statistics of cosmic background radiation fluctuations. Mon.
Not. R. Astron. Soc. 226, 655 (1987)
24. W. Hu, S. Dodelson, Cosmic microwave background anisotropies. Ann. Rev. Astron.
25. M. Tegmark, M. Zaldarriaga, Current cosmological constraints from a 10 parameter CMB
analysis. Astrophys. J. 544, 30 (2000). [arXiv:astro-ph/0002091]
26. http://www.esa.int/Our_Activities/Space_Science/Planck
27. http://pla.esac.esa.int/pla/aio/index.html
28. http://planck.caltech.edu/index.html
29. R. Adam et al. [Planck Collaboration], Planck 2015 results. I. Overview of products and
scientific results. Astron. Astrophys. 594, A1 (2016). [arXiv:1502.01582]
30. J. Silk, M.L. Wilson, Residual fluctuations in the matter and radiation distribution after the
decoupling epoch. Physica Scripta 21, 708 (1980)
31. D. Scott, M. Srednicki, M.J. White, ‘Sample variance’ in small scale CMB anisotropy
experiments. Astrophys. J. 421, L5 (1994). [arXiv:astro-ph/9305030]
32. J.B. Peterson et al., Cosmic microwave background observations in the post-Planck era.
arXiv:astro-ph/9907276
33. P.A.R. Ade et al. [Planck Collaboration], Planck 2015. XX. Constraints on inflation. Astron.
34. A. Kogut et al., Dipole anisotropy in the COBE DMR first year sky maps. Astrophys. J. 419,
1 (1993). [arXiv:astro-ph/9312056]
35. P.A.R. Ade et al. [Planck Collaboration], Planck 2013 results. XVI. Cosmological parameters.
36. R.K. Sachs, A.M. Wolfe, Perturbations of a cosmological model and angular variations of the
microwave background. Astrophys. J. 147, 73 (1967) [Gen. Relat. Grav. 39, 1929 (2007)]
37. M.J. White, W. Hu, The Sachs–Wolfe effect. Astron. Astrophys. 321, 8 (1997).
38. E.R. Harrison, Fluctuations at the threshold of classical cosmology. Phys. Rev. D 1, 2726
(1970)
39. Ya.B. Zel’dovich, A hypothesis, unifying the structure and the entropy of the universe. Mon.
Not. R. Astron. Soc. 160, 1P (1972)
40. S. Eidelman et al. [Particle Data Group], Review of particle physics. Phys. Lett. B 592, 1
(2004). [arXiv:astro-ph/0406567]
41. A.D. Sakharov, The initial stage of an expanding universe and the appearance of a nonuniform
distribution of matter. Zh. Eksp. Teor. Fiz. 49, 345 (1965) [Sov. Phys. JETP 22, 241 (1966)]
42. P.J.E. Peebles, J.T. Yu, Primeval adiabatic perturbation in an expanding universe. Astrophys.
J. 162, 815 (1970)
43. W. Hu, N. Sugiyama, J. Silk, The physics of microwave background anisotropies. Nature 386,
37 (1997). [arXiv:astro-ph/9604166]
44. J. Silk, Fluctuations in the primordial fireball. Nature 215, 1155 (1967)
45. J. Silk, Cosmic black-body radiation and galaxy formation. Astrophys. J. 151, 459 (1968)
46. W.T. Hu, Wandering in the Background: A CMB Explorer. Ph.D. thesis, UC Berkeley,
Berkeley (1995). [arXiv:astro-ph/9508126]
47. P.J.E. Peebles, The black-body radiation content of the universe and the formation of galaxies.
Astrophys. J. 142, 1317 (1965)
48. M.J. Rees, Cosmology and galaxy formation, in The Evolution of Galaxies and Stellar
Populations, ed. by B.M. Tinsley, R.B. Larson (Yale University Observatory Publications,
New Haven, 1977)
49. C.J. Hogan, A model of pregalactic evolution. Mon. Not. R. Astron. Soc. 188, 781 (1979)
51. R.A. Sunyaev, Ya.B. Zel’dovich, The velocity of clusters of galaxies relative to the microwave
background. The possibility of its measurement. Mon. Not. R. Astron. Soc. 190, 413 (1980)
52. E. Komatsu et al., Seven-year Wilkinson Microwave Anisotropy Probe (WMAP) observa-
tions: cosmological interpretation. Astrophys. J. Suppl. 192, 18 (2011). [arXiv:1001.4538]
53. M. Zemcov et al., First detection of the Sunyaev–Zel’dovich effect increment at < 650m.
Astron. Astrophys. 518, L16 (2010). [arXiv:1005.3824]
54. M.J. Rees, D.W. Sciama, Large-scale density inhomogeneities in the universe. Nature 217,
511 (1968)
55. N. Padmanabhan, C.M. Hirata, U. Seljak, D. Schlegel, J. Brinkmann, D.P. Schneider,
Correlating the CMB with luminous red galaxies: the integrated Sachs–Wolfe effect. Phys.
56. B.R. Granett, M.C. Neyrinck, I. Szapudi, An imprint of superstructures on the microwave
background due to the integrated Sachs–Wolfe effect. Astrophys. J. 683, L99 (2008).
[arXiv:0805.3695]
57. J.P. Ostriker, E.T. Vishniac, Generation of microwave background fluctuations from nonlinear
perturbations at the era of galaxy formation. Astrophys. J. 306, L51 (1986)
58. A. Blanchard, J. Schneider, Gravitational lensing effect on the fluctuations of the cosmic
background radiation. Astron. Astrophys. 184, 1 (1987)
59. S. Cole, G. Efstathiou, Gravitational lensing of fluctuations in the microwave background
radiation. Mon. Not. R. Astron. Soc. 239, 195 (1989)
60. A. Lewis, A. Challinor, Weak gravitational lensing of the CMB. Phys. Rep. 429, 1 (2006).
61. M. Kamionkowski, D.N. Spergel, N. Sugiyama, Small scale cosmic microwave background
anisotropies as a probe of the geometry of the universe. Astrophys. J. 426, L57 (1994).
62. P. de Bernardis et al. [Boomerang Collaboration], A flat universe from high-resolution
maps of the cosmic microwave background radiation. Nature 404, 955 (2000).
63. P. de Bernardis et al. [Boomerang Collaboration], Multiple peaks in the angular power spec-
trum of the cosmic microwave background: significance and consequences for cosmology.
64. A. Balbi et al., Constraints on cosmological parameters from MAXIMA-1. Astrophys. J. 545,
L1 (2000); Erratum-ibid. 558, L145 (2001). [arXiv:astro-ph/0005124]
65. A.T. Lee et al., A high spatial resolution analysis of the MAXIMA-1 cosmic microwave
background anisotropy data. Astrophys. J. 561, L1 (2001). [arXiv:astro-ph/0104459]
References 149
66. R. Stompor et al., Cosmological implications of the MAXIMA-1 high resolution cos-
mic microwave background anisotropy measurement. Astrophys. J. 561, L7 (2001).
67. W. Hu, M. Fukugita, M. Zaldarriaga, M. Tegmark, CMB observables and their cosmological
implications. Astrophys. J. 549, 669 (2001). [arXiv:astro-ph/0006436]
68. R. Durrer, B. Novosyadlyj, S. Apunevych, Acoustic peaks and dips in the CMB power
spectrum: observational data and cosmological constraints. Astrophys. J. 583, 33 (2003).
69. A.A. Starobinsky, Cosmic background anisotropy induced by isotropic flat-spectrum
gravitational-wave perturbations. Sov. Astron. Lett. 11, 133 (1985)
70. M. Cortês, A.R. Liddle, P. Mukherjee, On what scale should inflationary observables be
constrained? Phys. Rev. D 75, 083520 (2007). [arXiv:astro-ph/0702170]
71. A. Kosowsky, M.S. Turner, CBR anisotropy and the running of the scalar spectral index. Phys.
72. S. Pandolfi, A. Cooray, E. Giusarma, E.W. Kolb, A. Melchiorri, O. Mena, P. Serra, Harrison–
Zel’dovich primordial spectrum is consistent with observations. Phys. Rev. D 81, 123509
(2010). [arXiv:1003.4763]
73. P.A.R. Ade et al. [BICEP2 and Planck Collaborations], Joint analysis of BICEP2/Keck Array
and Planck data. Phys. Rev. Lett. 114, 101301 (2015). [arXiv:1502.00612]
74. P.A.R. Ade et al. [BICEP2 and Keck Array Collaborations], Improved constraints on
cosmology and foregrounds from BICEP2 and Keck Array cosmic microwave background
data with inclusion of 95 GHz band. Phys. Rev. Lett. 116, 031302 (2016). [arXiv:1510.09217]
75. J.P. Zibin, D. Scott, M.J. White, Limits on the gravity wave contribution to microwave
anisotropies. Phys. Rev. D 60, 123513 (1999). [arXiv:astro-ph/9901028]
76. M.S. Turner, M. White, J.E. Lidsey, Tensor perturbations in inflationary models as a probe of
cosmology. Phys. Rev. D 48, 4613 (1993). [arXiv:astro-ph/9306029]
77. B. Allen, The stochastic gravity-wave background: sources and detection, in Relativistic
Gravitation and Gravitational Radiation, ed. by J.-A. Marck, J.-P. Lasota (Cambridge
University Press, Cambridge, 1997)
78. S. Chongchitnan, G. Efstathiou, Prospects for direct detection of primordial gravitational
waves. Phys. Rev. D 73, 083511 (2006). [arXiv:astro-ph/0602594]
79. A. Stewart, R. Brandenberger, Observational constraints on theories with a blue spectrum of
tensor modes. JCAP 0808, 012 (2008). [arXiv:0711.4602]
80. J. Aasi et al. [The LIGO Scientific Collaboration], Advanced LIGO. Class. Quantum Grav.
32, 074001 (2015). [arXiv:1411.4547]
81. B.P. Abbott et al. [LIGO Scientific and Virgo Collaborations], GW150914: the Advanced
LIGO detectors in the era of first discoveries. Phys. Rev. Lett. 116, 131103 (2016).
[arXiv:1602.03838]
82. https://www.elisascience.org
83. M.J. Rees, Polarization and spectrum of the primeval radiation in an anisotropic universe.
Astrophys. J. 153, L1 (1968)
84. M.M. Basko, A.G. Polnarev, Polarization and anisotropy of the relict radiation in an
anisotropic universe. Mon. Not. R. Astron. Soc. 191, 207 (1980)
85. M. Kaiser, Small-angle anisotropy of the microwave background radiation in the adiabatic
theory. Mon. Not. R. Astron. Soc. 202, 1169 (1983)
86. J.R. Bond, G. Efstathiou, Cosmic background radiation anisotropies in universes dominated
by nonbaryonic dark matter. Astrophys. J. 285, L45 (1984)
87. A.G. Polnarev, Polarization and anisotropy induced in the microwave background by
cosmological gravitational waves. Astron. Zh. 62, 1041 (1985)
88. D.D. Harari, M. Zaldarriaga, Polarization of the microwave background in inflationary
cosmology. Phys. Lett. B 319, 96 (1993). [arXiv:astro-ph/9311024]
89. D. Coulson, R.G. Crittenden, N.G. Turok, Polarization and anisotropy of the microwave sky.
90. R.G. Crittenden, D. Coulson, N.G. Turok, Temperature-polarization correlations from tensor
fluctuations. Phys. Rev. D 52, 5402 (1995). [arXiv:astro-ph/9411107]
91. M. Zaldarriaga, D.D. Harari, Analytic approach to the polarization of the cosmic
microwave background in flat and open universes. Phys. Rev. D 52, 3276 (1995).
92. M. Kamionkowski, A. Kosowsky, A. Stebbins, A probe of primordial gravity waves and
vorticity. Phys. Rev. Lett. 78, 2058 (1997). [arXiv:astro-ph/9609132]
93. U. Seljak, M. Zaldarriaga, Signature of gravity waves in polarization of the microwave
background. Phys. Rev. Lett. 78, 2054 (1997). [arXiv:astro-ph/9609169]
94. M. Zaldarriaga, U. Seljak, An all-sky analysis of polarization in the microwave background.
Phys. Rev. D 55, 1830 (1997). [arXiv:astro-ph/9609170]
95. M. Kamionkowski, A. Kosowsky, A. Stebbins, Statistics of cosmic microwave background
polarization. Phys. Rev. D 55, 7368 (1997). [arXiv:astro-ph/9611125]
96. D.N. Spergel, M. Zaldarriaga, CMB polarization as a direct test of inflation. Phys. Rev. Lett.
79, 2180 (1997). [arXiv:astro-ph/9705182]
97. W. Hu, M.J. White, A CMB polarization primer. New Astron. 2, 323 (1997).
98. A. Stebbins, Weak lensing on the celestial sphere. arXiv:astro-ph/9609149
99. F.J. Zerilli, Tensor harmonics in canonical form for gravitational radiation and other applica-
tions. J. Math. Phys. 11, 2203 (1970)
100. E.T. Newman, R. Penrose, Note on the Bondi–Metzner–Sachs group. J. Math. Phys. 7, 863
(1966)
101. J.N. Goldberg, A.J. MacFarlane, E.T. Newman, F. Rohrlich, E.C.G. Sudarshan, Spin-s
spherical harmonics and dH . J. Math. Phys. 8, 2155 (1967)
102. S. Dodelson, Modern Cosmology (Academic Press, San Diego, 2003)
103. D. Larson et al., Seven-year Wilkinson Microwave Anisotropy Probe (WMAP) observa-
tions: power spectra and WMAP-derived parameters. Astrophys. J. Suppl. 192, 16 (2011).
[arXiv:1001.4635]
104. P.A.R. Ade et al. [Planck Collaboration], Planck 2015 results. XV. Gravitational lensing.
105. D. Hanson et al. [SPTpol Collaboration], Detection of B-mode polarization in the cosmic
microwave background with data from the South Pole Telescope. Phys. Rev. Lett. 111, 141301
(2013). [arXiv:1307.5830]
106. P.A.R. Ade et al. [POLARBEAR Collaboration], A measurement of the cosmic microwave
background B-mode polarization power spectrum at sub-degree scales with POLARBEAR.
Astrophys. J. 794, 171 (2014). [arXiv:1403.2369]
107. http://camb.info
108. http://lambda.gsfc.nasa.gov/toolbox/tb_camb_form.cfm
109. http://cosmologist.info/cosmomc
110. http://www.class-code.net
111. http://montepython.net
112. N. Aghanim et al. [Planck Collaboration], Planck 2015 results. XI. CMB power spec-
tra, likelihoods, and robustness of parameters. Astron. Astrophys. 594, A11 (2016).
[arXiv:1507.02704]
113. C.L. Bennett et al. [WMAP Collaboration], Nine-year Wilkinson Microwave Anisotropy
Probe (WMAP) observations: final maps and results. Astrophys. J. Suppl. 208, 20 (2013).
[arXiv:1212.5225]
114. E. Komatsu et al., Non-Gaussianity as a probe of the physics of the primordial universe and
the astrophysics of the low redshift universe. arXiv:0902.4759
115. L. Verde, R. Jimenez, M. Kamionkowski, S. Matarrese, Tests for primordial non-Gaussianity.
116. L. Wang, M. Kamionkowski, Cosmic microwave background bispectrum and inflation. Phys.
117. A. Gangui, J. Martin, Cosmic microwave background bispectrum and slow roll inflation. Mon.
Not. R. Astron. Soc. 313, 323 (2000). [arXiv:astro-ph/9908009]
References 151
118. E. Komatsu, D.N. Spergel, Acoustic signatures in the primary microwave background
bispectrum. Phys. Rev. D 63, 063002 (2001). [arXiv:astro-ph/0005036]
119. W. Hu, Angular trispectrum of the cosmic microwave background. Phys. Rev. D 64, 083005
(2001). [arXiv:astro-ph/0105117]
120. N. Kogo, E. Komatsu, Angular trispectrum of CMB temperature anisotropy from primordial
non-Gaussianity with the full radiation transfer function. Phys. Rev. D 73, 083007 (2006).
121. R. Scaramella, N. Vittorio, Non-Gaussian temperature fluctuations in the cosmic microwave
background sky from a random Gaussian density field. Astrophys. J. 375, 439 (1991)
122. J.N. Fry, Gravity, bias, the galaxy three-point correlation function. Phys. Rev. Lett. 73, 215
(1994)
123. S. Matarrese, L. Verde, A.F. Heavens, Large-scale bias in the Universe: bispectrum method.
124. E. Sefusatti, E. Komatsu, The bispectrum of galaxies from high-redshift galaxy surveys:
primordial non-Gaussianity and non-linear galaxy bias. Phys. Rev. D 76, 083004 (2007).
[arXiv:0705.0343]
125. A. Gangui, F. Lucchin, S. Matarrese, S. Mollerach, The three-point correlation function of
the cosmic microwave background in inflationary models. Astrophys. J. 430, 447 (1994).
126. L. Verde, L. Wang, A. Heavens, M. Kamionkowski, Large-scale structure, the cosmic
microwave background, and primordial non-Gaussianity. Mon. Not. R. Astron. Soc. 313,
L141 (2000). [arXiv:astro-ph/9906301]
127. E. Komatsu, D.N. Spergel, B.D. Wandelt, Measuring primordial non-Gaussianity in the
cosmic microwave background. Astrophys. J. 634, 14 (2005). [arXiv:astro-ph/0305189]
128. J.M. Maldacena, Non-Gaussian features of primordial fluctuations in single field inflationary
models. JHEP 0305, 013 (2003). [arXiv:astro-ph/0210603]
129. D. Babich, P. Creminelli, M. Zaldarriaga, The shape of non-Gaussianities. JCAP 0408, 009
(2004). [arXiv:astro-ph/0405356]
130. P. Creminelli, M. Zaldarriaga, Single field consistency relation for the 3-point function. JCAP
0410, 006 (2004). [arXiv:astro-ph/0407059]
131. P. Creminelli, A. Nicolis, L. Senatore, M. Tegmark, M. Zaldarriaga, Limits on non-
Gaussianities from WMAP data. JCAP 0605, 004 (2006). [arXiv:astro-ph/0509029]
132. L. Senatore, K.M. Smith, M. Zaldarriaga, Non-Gaussianities in single field inflation and their
optimal limits from the WMAP 5-year data. JCAP 1001, 028 (2010). [arXiv:0905.3746]
133. P.A.R. Ade et al. [Planck Collaboration], Planck 2015 results. XVII. Constraints on primordial
non-Gaussianity. Astron. Astrophys. 594, A17 (2016). [arXiv:1502.01592]
134. F. De Bernardis, P. Serra, A. Cooray, A. Melchiorri, Constraints on primordial non-
Gaussianity from WMAP7 and luminous red galaxies power spectrum and forecast for future
surveys. Phys. Rev. D 82, 083511 (2010). [arXiv:1004.5467]
135. C.T. Byrnes, S. Nurmi, G. Tasinato, D. Wands, Scale dependence of local fNL . JCAP 1002,
034 (2010). [arXiv:0911.2780]
136. J. Smidt, A. Amblard, C.T. Byrnes, A. Cooray, D. Munshi, CMB constraints on primordial
non-Gaussianity from the bispectrum and trispectrum and a consistency test of single-field
inflation. Phys. Rev. D 81, 123007 (2010). [arXiv:1004.1409]
Chapter 5
Inflation
Le Réel, c’est l’impossible.

— Jacques Lacan, Le Séminaire. Livre XI, XIII.3
The Real is the impossible.
Contents
5.1 Problems of the Hot Big Bang Model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
5.1.1 Planck and GUT Scale. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
5.1.2 Flatness Problem. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
5.1.3 Horizon Problem. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
5.1.4 Monopole Problem. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
5.1.5 Primordial Seeds Problem. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
5.2 Inflationary Mechanism. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
5.2.1 Solution of the Flatness Problem. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
5.2.2 Solution of the Horizon Problem. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
5.2.3 Solution of the Monopole Problem. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
5.2.4 Solution of the Primordial Seeds Problem. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
5.3 Cold Big Bang. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
5.3.1 Equation of State. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
5.3.2 Chaotic Inflation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
5.3.3 Reheating. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
5.3.4 Observable Inflation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
5.3.5 Timeline of the Early Universe. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
5.4 Scalar Field: Background Dynamics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
5.4.1 Hamilton–Jacobi Formalism. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
5.4.2 Slow-Roll Parameters. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
5.4.3 Inflationary Attractor. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
5.5 Models of Inflation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
5.5.1 Large-Field Models. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
5.5.2 Small-Field Models. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
5.5.3 Multi-field Inflation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
5.6 First Glimpse of the Quantum Universe. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
5.6.1 Decoherence. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
5.6.2 From Quantum Fields to Classical Spectra. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
5.6.3 Choice of Vacuum. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
(continued)

DOI 10.1007/978-3-319-41127-9_5
154 5 Inflation
5.6.4 Mukhanov–Sasaki Equation Revisited. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193

5.6.5 Eternal Inflation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
5.7 Cosmological Spectra. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198
5.7.1 Gaussianity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198
5.7.2 Linear Tensor Perturbations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
5.7.3 Linear Scalar Perturbations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200
5.7.4 Consistency Relations and Lyth Bound. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
5.8 Non-Gaussianity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205
5.8.1 Stochastic Inflation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
5.8.2 Multi-field Non-Gaussianity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
5.9 Observational Constraints on Inflation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
5.9.1 Temperature Spectra. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
5.9.2 Polarization. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
5.9.3 Non-Gaussianity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
5.10 Unsolved and New Problems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
5.10.1 Graceful Entry Problem. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
5.10.2 Graceful Exit Problem. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
5.10.3 Trans-Planckian Problem. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
5.10.4 Naturalness or Model-Building Problem. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
5.11 The Inflaton and Particle Physics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
5.11.1 Not Only Scalars. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
5.11.2 Higgs Inflation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
5.12 Supersymmetry and Supergravity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
5.12.1 Global Supersymmetry. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
5.12.2 Supergravity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228
5.12.3 -problem. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
5.12.4 Inflation in Supergravity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
5.13 Problems and Solutions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239
5.1 Problems of the Hot Big Bang Model
The hot big bang model is based upon some fundamental assumptions:
(i) The laws of physics verified today were valid also in primordial epochs.
(ii) The cosmological principle holds.
(iii) The initial conditions of the universe at the big bang tbb are such that
˝.tbb / 1 and the universe is in thermal equilibrium at some temperature
Tbb > 100 MeV.
(iv) Large-scale cosmic structures (CMB anisotropies and galaxy distributions)
formed from a primordial spectrum of almost Gaussian density fluctuations.
This standard cosmology has achieved remarkable results including the correct
prediction of light elements abundances (big-bang nucleosynthesis) and the natural
5.1 Problems of the Hot Big Bang Model 155
explanation of the CMB as a fossil imprint of the initial hot phase. We have already
seen that the model contains at least two conceptual problems (the presence of a
big-bang singularity and a very small cosmological constant), but there are actually
other issues that deserve our attention.
5.1.1 Planck and GUT Scale
So far, we have followed a beaten path of knowledge that combined the cosmology
of classical general relativity with ingredients of the Standard Model of particles.
A problem in physics is almost invariably a point where the path stops and the
explorer must resort to their own inventiveness and curiosity. Inevitably, in carving
a new track one will follow false threads and meet dead ends, and often a certain
amount of speculation will be involved. Such is the case while approaching the first
instants of life of the Universe and energies we have not yet probed in accelerators.
For instance, in grand unification theories (GUT) the symmetry group of funda-
mental interactions is larger than the Standard Model group SU.3/ ˝ SU.2/ ˝ U.1/
and at high energies the strong and electromagnetic forces are unified. The minimal
umbrella group is SU.5/ (Georgi–Glashow model [1, 2]), but there exist other
proposals with larger groups such as SO.10/ [3, 4]. Typically, GUTs predict a finite
lifetime for the proton [5–7]. The present experimental limit on the proton decay
[8] almost rules out the SU.5/ model [9, 10], while SO.10/ is safe [6, 11, 12].
Regardless the chosen symmetry group, the characteristic energy scale at which
symmetry breaking occurs is [13–18]
TGUT 1015 – 1016 GeV ; zGUT 1028 – 1029 : (5.1)
At higher energies, not only should gravity be as important as the other forces,
but the very concepts of gravity and matter might no longer be separable and a
description in terms of particle fields might even fail. In Chap. 1 we introduced the
Planck time tPl , equation (1.3), which is interpreted as the time interval within which
quantum fluctuations exist at Planck-length scale, equation (1.2). The other Planck
quantities are obtained from these by dimensional arguments. From the point of
view of cosmology, Planck time represents the time at which quantum effects are
the size of the horizon, H 1 lPl . From what said above, it is conventional to
identify the “beginning of time” or big bang at tbb tPl . One also expects that field
theory breaks down at densities above m4Pl . Beyond the Planck scale, only a theory of
quantum gravity might be able to describe physical phenomena correctly. None of
these features fit rigorously in a classical framework such as the hot big bang model
of Chap. 2.
156 5 Inflation
5.1.2 Flatness Problem
From (2.84), the deviation from the critical density parameter is proportional to the
square of the comoving Hubble horizon:
j˝K j D jKjrH2 D jKj . 1/2 2

:
(3.11)
(5.2)
If we assume, as suggested by observations, that matter and radiation dominated the

universe throughout most of its history, then the comoving Hubble horizon (2.137)
increases with time. Consequently, any deviation from flat geometry are amplified.
However, we know from (2.128) that today ˝K0 is small, so that the curvature
parameter must be fine tuned to a high degree near the big bang. More precisely,
consider the ratio of the Hubble horizon at two different times tA and tB for a power-
law cosmology. From (2.142),
1p 1 1p
rH;A A 1 C zB B 2
D D D : (5.3)
rH;B B 1 C zA A
For a radiation-dominated universe (p D 1=2),

14
A 1 C zB B TB
D D D ; (5.4)
B 1 C zA A TA
while for a matter-dominated scenario (p D 2=3)

12 16
A 1 C zB B
D D : (5.5)
B 1 C zA A
Then,
2 2
rH .t/ 2 eq
j˝K j D j˝K0 j D j˝K0 j (5.6)
rH .t0 / 0 eq
2 2
eq 1 C zeq p 2 (5.4) 1
D j˝K0 j ' 2; z > zeq ;
(5.3)
(5.7)
0 1Cz z
where in the last step we used (2.150) and (2.128) and radiation domination at early
times. Already at radiation-matter equality, the curvature density should have been
quite small, j˝K;eq j 106 . Going backwards in the thermal history of the universe,
this fine tuning is worsened. At energy scales of TeV order (the highest we can probe
at the LHC),
j˝K;LHC j 1032 :
5.1 Problems of the Hot Big Bang Model 157
At the GUT scale (5.1), corresponding to t 1038 – 1036 s, one has
1026 – 1025 ; j˝K;GUT j 1058 – 1056 :

GUT (5.4)
(5.8)
eq
At the Planck scale,
j˝K;Pl j 1064 :
We face a problem of initial conditions. The hot big bang model does not explain
why the initial density of the universe was so close (up to one part over 1060 or
less) to the critical density. The probability that Nature realized such a special initial
condition is almost zero and it does not seem reasonable to accept that without a
fight. An obvious way out is to assume that the universe has flat geometry,
K D 0:
In this case, however, we would like to have a theory of particles and geometry
explaining why the flat case is elected above the others.
5.1.3 Horizon Problem
One of the main assumptions of the big bang model is that the universe is
homogeneous. The cosmological principle justifies the assertion that the Universe
was born at the initial time t D tbb in a state of infinite energy (temperature).
From CMB and large-scale observations, we know that the primordial universe
was homogeneous and isotropic up to one part over 105 . This indicates a strong
correlation of physical conditions in every region of the sky. The horizon at last
scattering (4.116) subtends angular scales 1:3ı . Zones smaller than about one
degree were in causal contact at the time when matter and radiation decoupled, while
larger regions were outside the horizon dec . These regions later entered the causal
patch as soon as the growing horizon encompassed them. However, we observe
temperature isotropy at scales far larger than dec and no causal process could have
thermalized these ultra-horizon regions, since dec 0 . In power-law models with
0 0 ;
distinguishing “in” and “out” regions. The big bang is also responsible for the
fine tuning of the curvature energy density, but we have seen that the tuning is
158 5 Inflation
removed by a brute-force choice of geometry. In this respect, the horizon problem

is truly endemic in big-bang expanding scenarios and more serious than the flatness
problem.
5.1.4 Monopole Problem
Spontaneous symmetry breaking plays a major role in modern theoretical physics,

from the particle Standard Model to condensed matter systems, from string theory to
quantum gravity scenarios. Certain systems invariant under a discrete or continuous
symmetry, and depending on one or more time-varying parameters, may undergo
a phase transition when these parameters acquire a critical value. In this case, the
symmetry is broken. An example is ferromagnetism. Above the Curie temperature,
the spins of atoms in a ferromagnet are randomly oriented; a microscopic inspection
of the material shows that its properties are invariant under spatial rotations. When
the temperature drops below the critical value, the spins get the same orientation
and the material acquires magnetic properties. Microscopically, there is a privileged
direction and rotation invariance has been lost. This symmetry breaking is the
transition from a high-entropy state to an ordered state.
Another example is the electroweak Higgs mechanism giving mass to the W and
Z gauge bosons of the Standard Model. At even higher energies of order of (5.1),
the mother group (e.g., SU.5/) is reduced to SU.3/ ˝ SU.2/ ˝ U.1/ and magnetic
monopoles form. Monopoles are a particular type of structure generated via the
Kibble mechanism [19]. During a cosmological phase transition, correlation lengths
are limited by the particle horizon. Uncorrelated transitions take place at super-
horizon scales, but at the border between causal regions false vacuum structures,
called topological defects, are formed. Their geometry depends on the topology of
the boundary. One can then have defects which are point-wise (monopoles), one-
dimensional (cosmic strings), two-dimensional (domain walls), three-dimensional
(textures), and so on, together with other formations such as Kibble gradients and
fractal defects [20–22].1
The monopole problem is the excessive density of defects produced during the
phase transition at a mass scale M [25–27]. A heuristic calculation yields (e.g., [28])
˝M0 > 1015 ;
incompatible with the observed quasi-critical density. This large monopole density
can be avoided in certain models [29], and whenever one excludes grand unification.
Therefore, the monopole and the flatness problems are less severe than the horizon
problem.
1
Cosmic strings can give rise to characteristic signatures in the CMB. See, e.g., [23, 24].
5.2 Inflationary Mechanism 159
5.1.5 Primordial Seeds Problem
We have not yet identified what originated the primordial fluctuations. In doing so,
we would like also to explain why the level of anisotropy is as low as in (4.69),
ıT
105 ;
T0
and why the statistical distribution is Gaussian.
5.2 Inflationary Mechanism
If the exposition of the cosmological equations has been clear so far, it should
be easy for the reader to devise a solution of the flatness and horizon problems.
However, at the time of its proposal this solution was not so obvious and triggered
a revolution in our way of thinking the universe [30–33].
We define as inflation any period of accelerating expansion,
aR > 0 , < 1: (5.9)
From (2.168) and the discussion in Problem 2.9, this is equivalent to impose an
equation of state with w < 1=3:
1
P < : (5.10)
3
In terms of exact solutions, the power-law solution (2.180) is inflationary when
p > 1. The de Sitter background (2.172) realizes the prototype of extreme infla-
tionary expansion:
a.t/ D eHt ; H.t/ D H ; .t/ D 0 : (5.11)
Many approximate solutions will share almost all the qualitative features of de Sitter,
including an exponential expansion of the scale factor, a eHt , and a small first
slow-roll parameter, 1. For this reason, de Sitter cosmology has become almost
a synonym of inflation. However, we must always keep in mind that de Sitter is an
ideal setting and many realizations of inflation can, or actually must, deviate from it.
We assume that inflation lasted from some initial time ti > tbb until some time te ,
after which the universe was radiation-dominated. Inflation is characterized by an
“improved” definition of the number of e-foldings (compare with (2.48)), which is
160 5 Inflation
the logarithm of the ratio of comoving Hubble horizons [34],

rH;i
N :D ln : (5.12)
rH
During inflation, the Hubble parameter is almost constant and N . / ' Na .
5.2.1 Solution of the Flatness Problem
Inflation solves the flatness problem almost by definition, since the comoving
Hubble length decreases in time during an accelerated era (equation (3.9)):
1
rPH D Ra rH2 D .1 / < 0 : (5.13)
a
At primordial times, ˝ naturally approaches the critical density even if ˝ ¤ 1

as initial condition (Fig. 5.1). Because of the accelerated expansion, the universe
inflates and its characteristic curvature scale is stretched in a short amount of time.
After inflation, a local observer no longer feels any intrinsic curvature effect. This
is the reason why we often ignored curvature so far.
We can estimate the amount of inflation sufficient to remove the fine-tuning
problem of flatness. Suppose inflation starts soon after the Planck-scale big bang.
As initial time, we set
ti & tbb tPl 1044 s ; (5.14)
Fig. 5.1 Inflation and the flatness problem. Starting from a general initial condition j˝K i j 1,
the curvature density parameter j˝K j is strongly damped during inflation, ti < t < te (time axis
arbitrary). Until today (t0 ), the density parameter remains close to the critical value ˝K 0.
Depending on the future evolution of the universe, the dark-energy era might end and curvature
effects might eventually dominate
5.2 Inflationary Mechanism 161
while in most models it is reasonable to assume that inflation ended at the GUT
scale. Taking the upper limit of (5.1),
te tGUT 1038 s : (5.15)
After the end of inflation, there may be a matter-dominated period called reheating
that precedes the radiation era. For the time being, we assume it is so short
we can ignore it, an approximation called instantaneous reheating. In this case,
from (5.2), (5.12) and the lower energy bound of (5.8),
j˝K;i j D j˝K;e je2Ne ' 100:87Ne 58 :
To have a natural initial condition j˝K;i j 1, we need at least
Ne & 64 : (5.16)
This means that the expansion factor of the universe during inflation is at least
e64 1028 .
5.2.2 Solution of the Horizon Problem
During inflation, the comoving Hubble length decreases as (5.13) and regions of the
observable universe which were in causal contact disappear beyond the shrinking
horizon. After the end of inflation, rH increases and the external regions re-enter the
causal patch. However, a post-inflationary observer would see these regions for the
first time. This phenomenon is depicted in the upper panel of Fig. 5.2 for a comoving
scale com D =a, which is constant by definition.
In the synchronous coordinate system (lower panel of Fig. 5.2), consider the
physical scale inside the Hubble horizon at t D ti . This scale can represent
an inhomogeneity of wave-length , generated by some causal process within the
horizon. Before inflation, both the scale .t/ a.t/ and the Hubble distance H 1
increase with time. During inflation, the scale .t/ eHt increases exponentially
but the Hubble distance almost remains constant. After the end of the accelerated
expansion, the Hubble radius increases and begins to encompass nearby regions
again, eventually englobing .
If the comoving horizon at the beginning of inflation is larger than the present
one,
rH .ti / > rH .t0 / ;
then the horizon problem is solved because the regions we observe projected in the
sky entered in causal contact before inflation, and they had time to reach thermal
162 5 Inflation
Fig. 5.2 Inflation and the horizon problem. The upper panel shows the time evolution of a
physical scale or perturbation (thin curve) which starts inside the Hubble horizon RH (thick
curve). The curves are not smooth because we assume for simplicity a sudden begin and end of
inflation at ti and te , respectively. Before and after inflation, we assume a 0 < p < 1 power-law
evolution, ln ln a p ln t and ln RH ln t C j ln pj, while for ti < t < te the scale factor
expands quasi exponentially, ln t, ln RH const. The perturbation leaves the horizon at
t D t and re-enters after the end of inflation, at tk . The same process in comoving coordinates is
presented in the bottom panel
equilibrium before horizon exit. In instantaneous reheating scenarios, this places a

lower bound on the number of e-foldings. For inflation ending at the GUT scale,
rH .ti / rH .ti / rH .te /

D eNe ' 100:43Ne 28 ;
eq
1< D
GUT
(5.17)
rH .t0 / rH .te / rH .t0 / eq 0
thus yielding again (5.16). To solve both the horizon and the flatness problem,
inflation must have lasted more than 60 e-foldings.
In Sect. 3.1.3, we mentioned the fact that the particle horizon radius rp is negative
on inflationary backgrounds, so that it must be replaced by the Hubble horizon rH
5.3 Cold Big Bang 163
as an estimator of the size of the causal patch. Nevertheless, it still gives an intuitive
physical insight into the causal properties of cosmological spacetimes. The power-
law FLRW results (2.185) and (2.188) are sufficient to this aim. For 0 < p < 1,
the range of conformal time is the same as for proper time, 2 .0; C1/ , so that
the distance a signal can cover in the time interval D 0 D since the
singularity at D 0 is no greater than rp D c . For p > 1, conformal-time range
is 2 .1; 0/ , meaning that for a given instant any finite distance c D
c. N / > 0 can be covered by a signal travelling since the big bang at D 1.
5.2.3 Solution of the Monopole Problem
During the exponential expansion of the inflationary period, monopoles and other
relics are diluted and expelled from the horizon. Their observable density contri-
bution is lowered to acceptable levels. This mechanism can work if the reheating
temperature (see below) is not so high as to favour the thermal creation of new
defects. In another scenario dubbed eternal inflation, which we shall discuss later,
monopoles can actually inflate and constitute the seeds of other universes.
5.2.4 Solution of the Primordial Seeds Problem
Historically, inflation as a kinematical paradigm was invoked to solve the flatness,

horizon and monopole problems [31–33, 35, 36]. Although CMB anisotropies were
discovered later, it was soon recognized that the dynamics of the new mechanism
could account for the observed density contrast ı= 105 – 104 [37–42].
While (5.9) is a kinematical definition, (5.10) is a model-dependent dynamical
statement which implicitly requires a knowledge of the macro- and microscopic
properties of the cosmic fluid. The cosmic fluid should be constituted by one or more
fields which are approximately homogeneous at the classical level but that naturally
fluctuate upon quantization. The small-scale fluctuations of these fields are the
primordial cosmic seeds we are looking for. In the next sections, we will illustrate
this point by studying the classical and quantum dynamics of a real scalar field. The
scalar field is perhaps the simplest and most intuitive realization of inflation, and by
far the most popular and best studied.
5.3 Cold Big Bang
Consider the Friedmann equations (2.81) and (2.82) for a universe with no
cosmological constant and no curvature term. What type of fluid dominated the
first instants after the big bang? For simplicity, we can imagine that all forms of
164 5 Inflation
matter and radiation were originally enclosed in a single real scalar field , which
later decayed into known particles and dark matter. The scalar is a “matter field”
in this particle-field sense and the neutral spin-0 boson which it represents is called
inflaton. The choice of a scalar field may be partially justified by supersymmetry,
where scalars are abundant and play an important role. However, the inflaton does
not have a precise place within field theory, like the other unknowns in cosmology
(dark matter and dark energy).
5.3.1 Equation of State
We assume the following working hypotheses:

• The energy density of the inflaton dominates at the beginning, .ti / ' .
• The inflaton is minimally coupled with gravity, so that the total action is
effectively given by (2.20) and (2.59), and the matter equation of motion is (2.66).
In many models (scalar-tensor theories) a non-minimal coupling is allowed, but
we will not consider this case until Sect. 7.4.
• The universe is flat, K D 0.
• The inflaton is spatially homogeneous, .x/ D .t/. From (2.79), this implies
that the effective inflaton equation of state is
1 P2
P 2 V. /
w D D 1 P2
: (5.18)
2 C V. /
The assumption of flat geometry is motivated both by simplicity (the dynamical

equations being easier to solve) and by experiments, which place stringent bounds
on the curvature of the universe [43]. Spectrum predictions different from the flat
case are more difficult to accommodate in a natural way, since the location of the first
CMB peak and of the Silk-damping drop-off are strongly affected by the parameters
of curved models, both for a closed [44, 45] and an open universe [46–59]. Open
models have recently experienced a raise of interest, due to their possible embedding
in string theory [60, 61]; see Sect. 13.2.1.
Depending on the equation of state of , we can identify various regimes.
• In the extreme slow-roll (ESR) limit
P2 V ; (5.19)
the inflaton behaves as an effective cosmological constant, w 1. For this

reason, the ESR limit is actually (flat) de Sitter. In this regime, one can neglect
the scalar kinetic term in the first Friedmann equation and the second-derivative
term in the scalar equation of motion:

s
8
H' V; (5.20)
3m2Pl
V;
P' : (5.21)
3H
• The non-extreme slow-roll regime corresponds to (2.168c), 1 < w < 1=3.

This ensures inflation, equation (5.9).
• In the kinetic fast-roll regime, or kination [62], the equation of state is w 1
(stiff matter).
Then, the range of the barotropic index is
1 < w < 1: (5.22)
Notice, however, that relaxing the homogeneity condition extends this range. In fact,
the energy density and pressure with gradients included are (equation (2.65))
.x/ D 1 P 2 C V C 1 @˛ @˛ ; (5.23)
2 2
P .x/ D 1 P2 V 1 ˛
2 6 @˛ @ : (5.24)
In the gradient-dominated regime, w 1=3, a value already included in (5.22);

but in the static regime
P 2 minŒV; .@˛ /2 ; (5.25)
the barotropic index can be w < 1. Such values of w are often associated with a
phantom scenario where the kinetic term of the inflaton has the wrong sign. We have
just seen that the static regime offers a more natural mechanism (see Sect. 7.4.4 for
an alternative).
5.3.2 Chaotic Inflation
The old inflationary scenario, originally and independently proposed by Sato,

Kazanas and Guth [30–33], was based on three fundamental assumptions. (i)
Initially, the Universe expands in a high-temperature state of thermal equilibrium,
during which the inflaton sits in a false vacuum (local maximum of the potential
at D 0). This is a homogeneous universe with the same value of temperature
and energy density everywhere. (ii) Inflation ends with the decay of the false-
vacuum state and the breaking of the action symmetry through a second-order phase
transition. (iii) The decay of the unstable vacuum gives rises to the formation of
166 5 Inflation
bubbles where acquires the expectation values of the true degenerate vacuum.
These bubbles expand at the speed of light, collide and eat up the false-vacuum
regions. Linde [42] proposed an alternative and more natural model, called chaotic
inflation, which has almost become a synonym of inflation since then.
As a quantum field operator, the inflaton will be denoted as O . Initially, the
expectation value h O i is assumed to be strongly position-dependent, so that it
spans a wide range of values (hence the name chaotic).2 The most natural initial
conditions for the inflaton should be defined as close as possible to the Planck time
ti D tPl , since this is the earliest time when we can still hope to make sense of
initial conditions for classical fields on classical spacetimes. The Universe emerges
from the Planck era with the scalar field shifted arbitrarily from the minima of its
potential, so that the typical energy is
.ti / V. i / m4Pl (5.26)
or V. i / MPl4 . In regions where the field has appropriate initial conditions, inflation
begins. These regions have a typical size of order of the Planck length lPl . One can
show that spatial gradients of the field do not prevent the onset of inflation at least
in some regions [63].
The pre-inflationary Universe is highly inhomogeneous and anisotropic. The
Planck scale acts as a coarse-graining or cut-off scale for inhomogeneities, so that
the scalar field is approximately homogeneous in regions of volume
Vi a3i l3Pl ;
where, in the second member, we normalized the comoving volume to 1. These are
the primordial analogues of the local patches of the separate universe approach. In
some of the patches, the initial conditions are such that the slow-roll approximation
is valid.
Assume a quadratic potential
V. / D 1
2
m2 2
: (5.27)
Obviously, the chaotic inflationary scenario can be applied to any well-motivated

potential, but (5.27) is the simplest toy model of inflation and it well describes
the dynamics near a local minimum. In the ESR regime, the dynamical equations
are (5.20) and (5.21),
r
2
4 m P ' m mPl m
H' ; ' p ;
3 mPl 3 H 12
2
Here the term “chaotic” loosely refers to the statistical distribution of the initial conditions, not
to precise stochastic properties of chaos theory. In Chap. 6, we will see an example of chaotic
evolution in the latter mathematical sense.
leading to a mild time dependence of ,

mPl m
' p t;
12
and a scale factor

4 2 2
. i /
a ' ai eH.tti / ' ai e mPl
2
: (5.28)
Observational bounds will later constrain the inflaton mass to be
m 106 mPl : (5.29)
The initial condition (5.26) in the inflating regions requires i 106 mPl . The end
of inflation, as we have seen, can be placed at about the GUT scale, implying e
mPl i . From (5.28), the total number of e-foldings is
ae
Ne ' ln 1013 ; (5.30)
ai
which is much larger than the minimum 60 e-folds (5.16). What is the physical size
of a closed Universe after inflation? If the Universe at the big bang was a Planck-size
sphere of radius ai lPl 1035 m, at te it is
12 12
ae ai 1010 1010 m : (5.31)
In comparison, the linear size of the observable universe today is a tiny 0 1026 m.
Our causal patch is only a negligible part of a region which was very small
in the primordial chaotic Universe. In this minute region, statistical fluctuations
determined an appropriate initial mean value of the scalar field to trigger inflation.
Thus, the cosmic principle and the slow-roll approximation combine naturally.
5.3.3 Reheating
The temperature of the scalar fluid during inflation does not need to be high. Then,
Ti Te 0 : (5.32)
At the end of inflation, the universe is dominated by a non-relativistic scalar particle

and we can assume a dust equation of state of the cosmic fluid, w D 0.
In cosmological and astrophysical surveys as well as in accelerators we have
never observed effects which can be ascribed to a fundamental scalar particle
168 5 Inflation
of the typical mass of the inflaton. It is possible to explain this by our present
technological limitations. However, a more plausible notion is that we do not see
the boson because it decayed soon after inflation, thus generating all the matter
(in cosmological jargon, matter and radiation) of the universe. This process is called
reheating because it entails a rapid increase of temperature [64–79] (see [80] for a
review). Reheating is typically divided into three phases.
1. In the first, called preheating [78, 81–83], the scalar field, after rolling slowly
down its potential, oscillates around a local minimum and decays into massive
bosons through a resonance. Gradually, oscillations are damped due to cosmo-
logical friction and the energy transfers to the new particles.
2. In the second phase, the new particles interact and decay into the actual matter
constituents. This is the end of the cold big bang.
3. In the third stage, the final products are thermalized reaching the reheating
temperature which, in the presence of supersymmetry, has an upper bound Treh .
106 – 1010 GeV [84, 85], corresponding to a cosmic age t 1026 – 1018 s. In
this model, the lower limit for the redshift at the end of reheating is
zreh > 1023 : (5.33)
From there on, the universe is described by the radiation-dominated hot big bang
model of Chap. 2. Assuming instead the non-supersymmetric Standard-Model
particle content and comparing the energy density at reheating with the one at
horizon crossing (see equations (5.195) and (5.196) below), one obtains the more
conservative upper bound
14 14
30 reh 30 3H2
Treh ' < < 7:6 1015 GeV : (5.34)
2 106:75 2 106:75 2
In most of the models, the reheating process is explosive: this justifies, a posteriori,
our initial assumption that the universe was radiation-dominated just after inflation.
To summarize, inflation might have taken place during the interval 1043 s < t <
10 s, followed by a reheating phase which ended at t 1018 s. As anticipated,
30
these numbers are somewhat speculative and depend significantly on assumptions

involving semi-classical conjectures on quantum gravity and supersymmetric exten-
sions of the Standard Model. Actually, the most conservative bound for exotic
physics is dictated by the big-bang nucleosynthesis. The hot big bang model must
start at T 10 MeV at the latest, in order to validate the standard calculations of
light elements formation. The upper bound for the end of inflation is then
te < 102 s: (5.35)

5.3.4 Observable Inflation
To calculate the observable amount of inflation [86], consider a comoving scale

k crossing and re-entering the horizon at t and tk , respectively during and after
inflation. We can use j j instead of rH in power-law cosmologies. In particular,
j k j D 1=k. At horizon exit, N e-foldings have passed since the beginning of
inflation. If we observe the scale k today, then k < 0 , and the portion of inflation
which affected our local patch corresponds to the last e-foldings since horizon exit,

Nk D Ne N D ln : (5.36)
e
Since the horizon at exit and re-entry is the same,
rH .tk / k 0 eq reh e
1D D :
rH .t / 0 eq reh e
Using (5.4) and (5.5),

2 0 1 23 3
1
@ e A 7
4
0 eq reh 6 0 Treh
Nk D ln.k 0 / C ln D ln.k 0 / C ln 4 5
eq reh e eq Teq Treh
2 1 23 1 1
3
0
max 3
Treh TGUT 3
Treh e6
D ln.k 0 / C ln 4 max 13 23
5;
eq Teq T Treh GUT
max
where Treh D 1010 GeV is the upper bound of the reheating temperature and TGUT D
15
10 GeV is the lower limit of the GUT scale. Therefore, from (2.150),
2 1015 GeV 1 1010 GeV

Nk ' 52 ln.k 0 / ln 1=4
ln : (5.37)
3 e 3 Treh
This is the number of e-folds we can observe today, which is in general smaller (but
see [87]) than the minimum amount of inflation (5.16) required to solve the horizon
problem. In turn, Nk can be much smaller than the actual duration of inflation Ne
(e.g., (5.30)).
Fix the perturbation wave-length to the horizon size today (largest observable
scale), khor D 01 . Shorter and shorter scales took more e-folds to cross the horizon,
so that N0 :D NkDkhor is the upper bound of “observable inflation.” The minimum
1=4
N0 occurs for the BBN bound e Treh 10 MeV, N0 17. For instantaneous
1=4
reheating at the GUT scale, e Treh TGUT , N0 56 (see Problem 5.1).
170 5 Inflation
5.3.5 Timeline of the Early Universe
We have now to refine the timeline of the universe before BBN by adding the early
cold big bang phase to the table in Problem 2.8 (Fig. 5.3). One can split the post-
inflationary evolution into a matter-dominated era between the end of inflation and
the end of reheating, and a radiation-dominated era between the end of reheating
and matter-radiation equality. Table 5.1 summarizes the results of this section. The
expansion ratio of the inflationary era is estimated via the lower bound (5.16),
rH rH rH
D eNe > 1028 ;
rHi rHe rHe
while to calculate the ratio of comoving Hubble horizons after reheating we use
rH0 0 T
' 0:1 z ' 109
eq
D :
rH eq 1 MeV
Looking at the second column from the right in the table, one sees that the
universe expanded more in the first 1036 seconds than in the rest of its life.
How much more, it greatly depends on the details of inflation: compare, for
instance, (5.16) with (5.30).
Fig. 5.3 Timeline of the universe (Credit: NASA/WMAP Science Team [88])
5.4 Scalar Field: Background Dynamics 171
Table 5.1 Simplified history of the early universe from the cold big bang. The values of this table
should be taken only as indicative
1=4 (GeV) T (GeV) t (s) rH .t0 /=rH .t/ Event

19 44 52
10 0 10 >10 Cold big bang? Inflation
begins at Planck scale?
15 36 24
10 0 10 10 Inflation ends at GUT
scale? Reheating begins?
10 7 22
10 10 10 Reheating ends?
2 2 10
10 10 10 Latest end of cold
big bang model
5 8
10 200 10 BBN
109 1012 100 Radiation-matter equality
1013 1017 1 Today
5.4 Scalar Field: Background Dynamics
In this section, we will describe the classical dynamics of a homogeneous, canonical

scalar field. Let D D 4 and K D 0. The Friedmann and scalar field equations (2.169)
and (2.171) are
!
2 2 P 2
H D CV ; (5.38)
3 2
0 D R C 3H P C V; : (5.39)
5.4.1 Hamilton–Jacobi Formalism
In Sect. 2.1.2, we briefly mentioned the possibility of using matter as an internal

clock. For a scalar-field dominated universe, is the natural choice. Assume without
loss of generality that P > 0. From (2.135),
2
P D P2 :
H (5.40)
2
Thus, a canonical homogeneous scalar field in standard general relativity can never
P < 0. If we regard H as a function of , and if
lead to super-inflation, since H
varies monotonically with time, then
2 P
H; D : (5.41)
2
172 5 Inflation
Equation (5.38) then gives
3 2 2 4
H;2 H C V D 0: (5.42)
2 2
The first slow-roll parameter can be recast in several different ways:
2 P 2
D (5.43a)
2 H2

2 H; 2
D 2 (5.43b)
H

2 a 2
D ; (5.43c)
2 a;
so that (5.42) becomes
1
VD .3 /H 2 : (5.44)
2
Equation (5.43c) stems from (5.41) via
2
a ; H; D aH : (5.45)
2
Equations (5.41) and (5.42) (or (5.44) and (5.45)) are the Hamilton–Jacobi equa-
tions [89, 90]. They hold only if P does not change sign and they break down, for
instance, when the field oscillates around one of its minima. While the Hamilton–
Jacobi formalism is not suited for reheating, it is useful during inflation, when the
field is rolling far from the minima. In that case, it allows one to find the dynamical
profiles of the model starting from the Hubble parameter H. /: (i) replace H. /
in (5.42) to obtain V. /; (ii) integrate (5.41) to get .t/ and, hence, H.t/ and a.t/. In
general, it is not convenient to choose first the potential V. / because (5.44) is non-
linear; an exception is the exponential potential associated with power-law inflation
(Problem 2.11), where is constant. Then, also H is exponential, H. / / e = 0 .
5.4.2 Slow-Roll Parameters
The use of the slow-roll formalism [34, 91–93] simplifies the study of inflation;
however, it can also be considered as an effective notation for some recurrent
dimensionless combinations of cosmological quantities, without imposing any
condition on their magnitude. We will keep calling these parameters “slow-roll”
in this case, too. The most commonly used SR towers rely upon two different
quantities, the geometric Hubble parameter H and the dynamical inflaton potential
V. We will name these towers H-SR and V-SR, respectively, and explore some of
their properties. Other SR towers can be constructed for particular cosmological
scenarios or analyses [94–97].
5.4.2.1 H-SR Tower
The H-SR tower is defined as
Yn 1
d ln H .i/ n
0 :D ; n :D ; n > 1; (5.46)
iD1
d ln a
where .i/ is the i-th derivative. The first three parameters, which are those
appearing in all the main expressions for cosmological observables, are
3 P2
D 0 D ; (5.47)
P 2 C 2V
d ln P R
:D 1 D D ; (5.48)
d ln a HP
!:
2 2 1 R «
:D 2 D 2 D 2 : (5.49)
H P H2 P
The conditions
1; jj 1 (5.50)
define the ESR regime. The second inequality is equivalent to assume solu-
tion (5.21). Formulæ truncated at the n-th power of these parameters will be referred
to as “n-th order SR.” The first-order SR is precisely the ESR approximation. If
jj > 1 or jj > 1, inflation could still take place but it would soon end due to the
R D 2H H,
rapid variation of . Noting that H P we have
P D 2H. / ; (5.51)
2
P D H. / : (5.52)
Differentiation with respect to the scalar field yields n; D Pn = P . By (5.43a), the
resulting prefactor H= P can be expressed as
r
H 2
DC ; (5.53)
P 2
174 5 Inflation
where the plus sign has been chosen in order to have a slow rolling down the
potential with P > 0. This is always possible by a redefinition ! .
A comment is in order. A slow-roll tower is dynamical (i.e., it does say
something about the dynamics of the Hamilton–Jacobi equations) either when
external constraints on the form and magnitude of the SR parameters are applied,
or when distinct SR definitions are related through the Hamilton–Jacobi equations
themselves. For example, the fundamental building block of the H-SR tower
is the parameter ; as long as the latter is not linked with the matter content
(equation (2.166)), it is clear there will be no knowledge about the evolution of
the system. However, when rewriting the H-parameters in terms of P through the
first Hamilton–Jacobi equation, these parameters become dynamically informative.
An interesting discussion on related issues can be found in [96].
5.4.2.2 V-SR Tower
The H-SR hierarchy is an elegant instrument of analysis coming from the Hamilton–
Jacobi formulation of the equations of motion. However, often the natural starting
point is not H but a reasonable inflaton potential V, for instance suggested by field-
theory high-energy models. Given V, the Hubble parameter must be determined by
the Hamilton–Jacobi equations, which are not always readily solvable. Therefore,
it is convenient to define another SR tower and try to relate it to the original one,
namely,
2 " # 1n
1 V; 1 V .nC1/ V;n1
V 0 :D 2 ; V n :D 2 ; n > 1; (5.54)
2 V Vn
where, again, we have introduced the first parameter separately. Therefore,
V :D V 0 ; (5.55a)
1 V;
V :D V 1 D ; (5.55b)
2 V
1 V; V;
V2 :D V22 D 4 ; (5.55c)
V2
and their derivatives with respect to the scalar field are
V;
V ; D .2V V / ; (5.56)
V
1 V;
V ; D 2V V V2 ; (5.57)
2V V
where
V; p
D 2 2 V : (5.58)
V
The conditions V 1 and jV j 1 are necessary to drop the kinetic term in (5.38)
and the acceleration term in (5.39) but they are not sufficient. In general, the V-SR
formalism requires a further assumption, namely (5.21), which is easy enough to be
satisfied. This determines the minus sign in (5.58), provided P > 0.
For a potential with a mass term V; D m2 , the condition jV j 1 is roughly
equivalent to
m2 H 2 : (5.59)
Thus, the mass of the inflaton is expected to be much lighter than the Hubble
energy during the accelerated expansion. Many models in supergravity and string
cosmology face the challenge of respecting (5.59) without fine tuning. This “-
problem” will be discussed in Sect. 5.12.3.
5.4.2.3 Horizon-Flow Tower
The horizon-flow parameters [94, 95] are defined by
Hinf d ln jH i1 j

H0 :D ; Hi :D ; i 1; (5.60)
H dNa
where Hinf is the Hubble rate at some chosen time and Na D ln.a=ai/ is the number
of e-folds.3 As shown in [96], these parameters (and others similarly defined) do
not properly encode inflationary dynamics even if they provide a good algorithm
for reconstructing the inflationary potentials. In fact, because of the absence of the
1=n power, definition (5.60) does not permit a power truncation similar to that of
the traditional SR towers (5.46) and (5.54), unless one imposes a constraint such as
@i H D 0 for i > imax .
The evolution equation for the horizon-flow parameters is given by
PHi D HHi H iC1 : (5.61)
3
Our definition counts Na forward in time, in accordance with [94] where Na .ti / D 0 and goes up
to Na .t/ > 0. In [95], the “backward” definition is used, where Nk D ln.ae =a/ is the number of
remaining e-folds at the time t before the end of inflation.
176 5 Inflation
5.4.2.4 Relation Between SR Parameters
It is possible to map the three SR towers one to the other by some simple relations.
Here we will restrict ourselves to the first three parameters. From (5.44), we get the
exact relation
2
3
V D : (5.62)
3
Then, noting that
V; D P H. 3/ ; V; D H 2 Œ3. C / 2 2 ; (5.63)
one has
3. C / 2 2
V D : (5.64)
3
Finally, since V; D 3.3 C 2 /H 3 C O. 3 / we obtain, to first H-SR order,
' V ; ' V V ; 2 ' V2 3V V C 3V2 : (5.65)
These equations allow us to shift from one hierarchy to the other, according to the
most convenient approach.
The horizon flow parameters are related to the first H-SR parameters by
H1 D ; H2 D 2. / ; H2 H3 D 2.2 2 3 C 2 / : (5.66)
5.4.3 Inflationary Attractor
The predictive power of inflation depends on the behaviour of cosmological

solutions with different initial conditions. If there exists an attractor such that
the differences of these solutions rapidly vanish, then the inflationary and post-
inflationary physics will generate observables which are independent of the initial
conditions. Let Ho . / > 0 be a generic expanding solution (denoted with the
subscript o) of the Hamilton–Jacobi equations and consider a linear perturbation
ıH. / which does not reverse the sign of P > 0. The linearized equation of
motion (5.42) (V is fixed) is
2Ho; ıH; D 3 2 Ho ıH ;
so that the perturbation is [89]

Z
3 2 Ho
ıH. / D ıH. o / exp d ; (5.67)
2 o
Ho;
where we have integrated from some initial value o . All linear perturbations are
exponentially suppressed when the integrand is negative definite, which is the case
during inflation as one can see from (5.41) and (5.53).
The number of e-foldings (5.12) can be written as an integral in the scalar field,
Z rH Z t
N D d ln rH D dt H.1 /
rHi ti
Z s
2
D d Œ1 . / :
(5.53)
(5.68)
i
2. /
In the SR approximation ' const, (5.43b) and (5.68) yield
ıH. / ' ıH. o / e3.1/.N No / : (5.69)
During the accelerated expansion, inflationary solutions all approach one another at
exponential rate. This is known as “the inflationary attractor.” The non-linear case
is more complicated but the above result holds as soon as perturbations enter the
linear regime.
In general, the classical stability of a solution .a.t/; aP .t/; i .t// with various
matter components i is checked by linearizing the background equations against
a homogeneous perturbation
0 1
ıa.t/
ıX.t/ :D @ ı aP .t/ A ; (5.70)
ıi .t/
and solving with respect to the latter. The set of linearized equations can be written
in matrix form,
P D M ıX ;
ıX (5.71)
where the entries Mij of the matrix M are calculated on the background solution.
The characteristic equation
det.M 1/ D 0 (5.72)

178 5 Inflation
determines the eigenvalues and a solution is stable provided
Re./ 6 0 : (5.73)
When the eigenvalues are time dependent, they are interpreted as evaluated at a
given time t [98].
To make sense physically, linear inhomogeneous perturbations must be defined
on backgrounds which are classically stable in the above sense. In the case of
inflationary solutions, we have just seen that they are all attractors. This means
that we can use any of these solutions as a background whereon to calculate the
primordial spectra.
5.5 Models of Inflation
Given a potential V. /, we have seen how to obtain a complete cosmological profile

and a set of SR parameters for the homogeneous scalar field .t/. In preparation
of calculating the primordial spectra, we summarize the recipe to obtain all the
quantities appearing in cosmological observables.
1. Fix the inflaton potential V. / by phenomenological arguments or via a high-
energy theory, for instance an effective field theory stemming from a suitable
model of microscopic physics.
2. Determine the first V-SR parameters via (5.55) (or (5.65), if the model is an exact
solution).
3. According to the chaotic scenario, the value of the inflaton i at the onset of
inflation is defined by
V. i / D m4Pl : (5.74)
On the other hand, we can estimate the field value at the end of inflation by setting
. e / D 1 : (5.75)
One may adopt the alternative criterion V . e / D 1, but the difference is small as
long as the number of e-folds is large enough.
4. To first SR order, the total number of e-foldings (2.48) is
s Z
4 e
d
Ne ' sgn P p ; (5.76)
m2Pl i . /
5.5 Models of Inflation 179
Fig. 5.4 Large-field inflationary models: potentials n =n with n D 2; 4; 6 (increasing thickness)

and exponential potential (dashed curve). Slow rolling begins at i mPl and terminates
somewhere near the minimum at e 0
while at horizon crossing

s Z
4 e
d
Nk ' sgn P p : (5.77)
m2Pl . /
The value of the field at horizon crossing is obtained by inverting the last
expression and fixing Nk . Typically, Nk 50 – 70.
5. From , one obtains the value of the slow-roll parameters D . /, D
. /, . . . , when the cosmological scale k left the causal region. We observe the
imprint of inflation at these scales in the CMB.
A standard classification [99] identifies three types of models when the potential
has at least one global minimum4: large-field, small-field and multi-field. A more
detailed charting of the inflationary models can be found in [100].
5.5.1 Large-Field Models
In large-field models, inflation starts with the scalar field displaced away from a
minimum at the origin, i D .ti / ¤ 0. From this configuration, the inflaton rolls
down towards the minimum, e i . Polynomial potentials of the form (2.61)
with n > 0 fall in this category. We consider two extreme examples: (2.182) and a
monomial potential (Fig. 5.4).
4
Classically, it is sufficient to have a local minimum, but when considering a quantum theory
tunneling effects should be taken into account.
180 5 Inflation
5.5.1.1 Exponential Potential
The exponential potential (2.182) in four dimensions,
3p 1 2 2 =
V. / D 0e
0
; (5.78)
2
corresponds to power-law inflation (2.180), which is an exact solution of the
equations of motion discussed in Problem 2.11. In the flat case,
r
2p
0 D˙ ; K D 0: (5.79)
2
Inflation does not have a natural end, since all the SR parameters are constant:
1
DD D : (5.80)
p
To terminate inflation, we have to force reheating by hand in the model. Despite

this issue, called graceful entry problem, power-law inflation is important because
it is not only the background around which inhomogeneous perturbations are often
defined, but also the starting point for the construction of approximate solutions.
The potential (2.182) may arise in various settings beyond the cosmological
standard model. In Kaluza–Klein scenarios, the Universe is allowed to have D1 >
3 spatial dimensions, all but three of which are compactifiedR to an unobservable
p
size today. An example is an Einstein–Maxwell theory S6 D d4 x d2 y OgŒc6 RO
.1=4/FABF AB 26 defined on a six-dimensional spacetime with structure M4 S2 ,
where M4 3 x is a four-dimensional Lorentzian manifold and S2 3 y is the
2-sphere with a spacetime-dependent radius L.x/, frozen at some constant scale
L0 at late times [101]. Call l.x/ D L.x/=L0 . The matrix gO AB is decomposed
into a four-dimensional metric gO .x/ D l2 .x/g .x/ and an internal S2 matrix
gO ab .x; y/ D l2 .x/gab .y/. After integrating over the S2 coordinates and by suitably
tuning the constants, the effective four-dimensional action of the model is S4 D
p
d4 x gŒR=.2 2 / C Ll , where the Lagrangian for the scalar l is of the form
Ll D .@l=l/2 V0 l2 .1 l2 /2 . Defining :D .2=/ ln l, one obtains the effective
potential V. / / e .1 e /2 , which reduces to (5.78) with p D 2 in the large-
limit. The procedure can be extended to M4 Sd and d compact dimensions,
where the .4 C d/-dimensional Lagrangian density is higher-order in curvature
invariants, L6 D O.R/ C O.R2 / [102, 103]. The resulting effective potential for the
scalar is V. / / ed .1 e2 /2 .1 C ce2 /2 ed . Still in the context of
Kaluza–Klein compactification, the six-dimensional Nishino–Salam–Sezgin N D 2
supergravity model [104–107] produces, when compactified to four dimensions,
two mutually coupled scalar fields with exponential potentials [108–110]. None of
these scenarios, however, can give sustainable inflation, since p D O.1/ typically.
Other models leading to exponential potentials and acceleration will be discussed in
Sect. 7.5.
5.5.1.2 Monomial Potential
The monomial potential (n > 0)

n
V. / D n
; n > 0 ; (5.81)
n
does not correspond to any exact background solution, except for the H D const,
K D 1 case (2.174), where n D 2. In the ESR approximation (5.20) and (5.21), the
K D 0 approximate solution features a Hubble parameter
s
8n n
H. / ' 2 ; (5.82)
3n m2Pl
while integrating (5.21) with H const , one has

r
2 n2 nn m2Pl
' t: (5.83)
24
Then,
8 2 2
2 Œ i .t/
a.t/ ' ai e n mPl : (5.84)
The slow-roll parameters and field values at ti and te read (n > 1)
n2 m2Pl
D ; (5.85a)
16 2
n.n 2/ m2Pl
D 2
; (5.85b)
16
4 2n
2 nmPl
D ; (5.85c)
i
n
2 n2 2 2
D m : (5.85d)
e
16 Pl i
The total number of e-foldings is ( P < 0)
2n
4 2 2
4 nm4Pl n
Ne D D ; (5.85e)
nm2Pl i e
nm2Pl n 4
182 5 Inflation
while
2 n.4Nk C n/ 2
D mPl ; (5.85f)
16
n
D ; (5.85g)
4Nk C n
n2
D : (5.85h)
4Nk C n
The time at which inflation ends is, from (5.83) and (5.85c),
s 2n 12
24 nm4Pl
te ' : (5.86)
nn m2Pl n
Inflation induced by a linear potential suffers from the same graceful exit problem
as power-law inflation, since D D const.
The linear term of an arbitrary potential can be reabsorbed by a field translation
and it can be ignored without loss of generality. The n D 1 case will be resuscitated
in Sect. 13.4.5 in the context of a fundamental theory of Nature (string theory),
together with other potentials with non-integer n D 2=5; 2=3; 4=5; 4=3.
In the case of the quadratic potential (n D 2, 2 D m2 ),
p mPl
i D 2 mPl ; (5.87a)
m
mPl
e D p ; (5.87b)
4
m2Pl
Ne ' 4 ; (5.87c)
m2
2 Nk 2
m 10 m2Pl ;
' (5.87d)
2 Pl
1
' 102 ; (5.87e)
2Nk
D 0 : (5.87f)
The SR parameters are small. For m D 106 mPl , one recovers the total number of
e-foldings (5.30), Ne 1013 , while inflation ends at te 1013 tPl 1031 s. The
minimum amount of e-folds Ne 60 is obtained, from (5.87c), with a larger mass,
m 0:5mPl . The lighter the scalar field, the longer inflation.
The quadratic potential is the prototype of large-field models and one among
the first proposals for chaotic inflation. Even if the potential energy V keeps below
the Planck energy density, having field excursions mPl may pose an issue
about the physical viability of these models (trans-Planckian problem). In fact, at

scales comparable with or larger than the Planck mass, effective field theory can
break down to give way to new degrees of freedom. To assess this possibility, it is
necessary to explore these degrees of freedom in a fundamental theory (in jargon,
to make a UV completion of the effective models of this chapter). That will be done
in Chap. 13 (for the quadratic case, in Sect. 13.4.5).
For a quartic potential (n D 4) the numbers are similar. To get enough inflation
1=2
in order to solve the horizon problem, 4 . 102 . Since te ' 104 tPl , in that case
inflation lasts only until te 10 s. For 4 D 10 , one has te 1035 s. Quartic
42 16
monomial inflation, however, is ruled out by data, while the quadratic potential is
under strong pressure [43, 111].
5.5.2 Small-Field Models
In small-field models, the scalar field is initially very near a local maximum at
the origin, i 0, and rolls towards the minimum at 0 ¤ e i . This
case is sometimes called hilltop inflation because slow rolling takes place in a
neighborhood of the local maximum. This type of potential is typically originated
in spontaneous-symmetry-breaking scenarios (see Sect. 7.1.1).
A double-well (Mexican hat) example is shown in Fig. 5.5. Near the maximum,
the typical potential can be approximated as
n
4
V. / D 1 ; (5.88)
2f
Fig. 5.5 Small-field inflationary model: double-well potential V D 0 2 =2 C 4 4 =4 with

0 ; 4 > 0. Slow rolling begins near the local maximum at i 0 and terminates somewhere near
the minimum at e MPl
184 5 Inflation
where =.2f / 1 and and f are mass scales. For n ¤ 2, one has
2.n1/
n2 MPl2
' ; (5.89a)
8 f2 2f
n2
n.n 1/ MPl2
' ; (5.89b)
4 f2 2f
4 n
1
2 MPl
i ' 1 64 4 2f ; (5.89c)

1
8 f 2 2.n1/
e ' 2f ; (5.89d)
n2 MPl2
and
2n
4 f2 i
Ne ' ; (5.89e)
n.n 2/ MPl2 2f
2n n.n 2/ MPl2

' .2f /2n Nk ; (5.89f)
4 f2
n1 1
' ; (5.89g)
n 2 Nk
where we used the reduced Planck mass for later convenience. The case n D 2
approximates, at small =f , the cosine potential

4
V. / D 1 C cos : (5.90)
2 f
This model is dubbed natural inflation, where the inflaton is a Nambu–Goldstone

pseudo-scalar such as the axion and f is a spontaneous-symmetry-breaking mass
scale called decay constant [112–114]. In the =f 1 limit,
2
1 MPl2
' ; (5.91a)
8 f2 f
1 MPl2
' ; jj ; (5.91b)
2 f2
4 2
1
2 MPl
i ' 1 64 4 2f ; (5.91c)

p f2
e ' 8 ; (5.91d)
MPl
and
f2 e
Ne ' 2 ln ; (5.91e)
MPl2 i

p f2 MPl2 Nk
' 8 exp 2 : (5.91f)
MPl f 2
Supergravity and superstring realizations of (5.90) will be presented in Sects. 13.4.2,

13.4.3, and 13.4.4.
5.5.3 Multi-field Inflation
The dynamics of multi-field inflationary scenarios is considerably richer than for

a single field. Simple models of multiple inflation feature two or several non-
interacting scalar fields, which drive a series of inflationary periods [115–118] or
just one period of accelerated expansion as in assisted inflation [119, 120]. In the
latter case, even if each field has a potential too steep to roll slowly and cannot lead
to inflation individually, their combined effect can drive acceleration. For instance,
the scalar spectral index generated by one field with exponential potential (5.78) of
slope p (equation (5.79)) is ns 1 D 2=p (see (5.153) below), and it is close to scale
invariance only if p 1. On the other hand, a collection of n non-interacting
P scalar
fields with exponential potentials of slope pi is ns 1 D 2=. i pi / and one can
achieve scale invariance even if some of the slopes are too steep, pi < 1 [119].5 In
the case of decoupled fields (no cross-terms in the total potential), they interact only
through minimal gravitational coupling. More generally, the background equations
of motion with an Einstein–Hilbert gravitational action and n homogeneous scalars
5
Extensions of these models include the effect of curvature and a barotropic perfect fluid [121],
Bianchi backgrounds [122] and polynomial potentials [123]. The stability of solutions as critical
points in phase space was studied for decoupled [120, 124, 125] and coupled [126–128] exponential
potentials, for decoupled inverse power-law potentials [125] and for general potentials with or
without cross-interactions [129]. Late-time dark-energy scenarios can be found in [125, 130].
Multi-field inhomogeneous perturbations have also been considered at the linear [131] and non-
linear level [132–134].
186 5 Inflation
read
!
1 X P2
n
2 2 Ri C 3H P i C W; i D 0 ;
H D CW ; (5.92)
3 2 iD1 i
where W :D W. 1 ; 2 ; : : : ; n / is the potential with both self-interaction P and

interaction terms. These two equations can be combined to give H P D .1=2/ i P i2 .
Multi-field assisted inflationary scenarios can be physically motivated within high-
energy multi-dimensional theories. In fact, as we shall see in Chap. 13, they naturally
emerge in Kaluza–Klein compactifications [123], M-theory and string theory.
In other settings, inflation is driven by one scalar field coupled with light or
massless scalar spectators (e.g., an axion) [135–138]. These extra fields can play a
variety of roles. For instance, curvature perturbations may be generated by a scalar
companion of the inflaton, in which case the former is called curvaton [138–141].
Or, if the partner’s mass is larger than that of the inflaton, it can effectively stop
inflation in a natural way (Sect. 5.10.2). The potential of this hybrid inflation
[142–146] is
V. ; '/ D 14 .' 2 M 2 /2 C 12 m2 2
C 12 2 2
' ; (5.93)
where and are the coupling constants of the auxiliary scalar field '. A priori,
the masses m and M range between 1 TeV (electroweak scale) and the Planck mass.
When the vacuum expectation value of the inflaton is greater than the crit-
ical value N 2 D M 2 =, the potential has a degenerate local minimum at
h'i D '.t/ D 0 (Fig. 5.6). Then, inflation is driven by the effective potential
Veff . / D 14 M 4 C 12 m2 2
: (5.94)
The dynamics is similar to a large-field model. The inflaton slowly rolls at the
bottom of the ' D 0 trough, then at j j . j N j it quickly falls into one of the
Fig. 5.6 Hybrid inflation

potential: is the inflaton and
' an auxiliary scalar field
5.6 First Glimpse of the Quantum Universe 187
wells corresponding to the true minimum of the auxiliary field,

s
2
'˙ . / D ˙M 1 :
N2
After rolling into either well, the scalar ' oscillates until the cosmic expansion
damps its motion. If the vacuum energy associated with ' dominates the poten-
tial (5.94), then the phase transition stops inflation almost instantaneously.
The SR parameters in the false-vacuum (inflationary) regime (5.94) are
2
1 mmPl 2
' 42 ; ' : (5.95)
m2Pl 2 M 2
The vacuum-domination regime is equivalent to the condition
: (5.96)
Also, in this scenario inflation ends at N , which is different, in general, from the
value e obtained from the . e / D 1 condition (with some choices of the couplings
magnitude, the end of inflation rather occurs at e > N ). Also,

2 1
' 2m4Pl M 4 ; (5.97a)
i
m2 2
2
1
Ne ' ln i2 ; (5.97b)
2 N
' N eNk ; (5.97c)

2
M 2Nk
' 42 e : (5.97d)
m2Pl
The two-field potential is shown in Fig. 5.6. Typical values of the coupling constants
are 101 , m 102 GeV, M 1011 GeV [146]. Because of (5.96)
and (5.153), hybrid models of this type predict blue-tilted scalar spectra and a very
low tensor-to-scalar ratio. They are therefore ruled out experimentally [147].
5.6 First Glimpse of the Quantum Universe
So far in this book, we have treated fields and perturbations as classical objects, but
quantum mechanics is expected to dominate the history of the universe at and near
the big bang. What is the relation between the inflaton quantum fluctuations and
classical perturbations spectra, and how did the transition between the quantum and
the classical world happen?
188 5 Inflation
5.6.1 Decoherence
Imagine we are able to describe the early universe as a quantum system where matter
and gravitational degrees of freedom are encoded in field operators on a Hilbert
space of physical states. This scenario is part of a broad area of investigation known
under the name of quantum cosmology. In this section, we discuss a generic scalar
field operator O , eventually identified with the inflaton or a tensor polarization mode.
In Chap. 10, we will quantize the Universe as a whole, which is the cosmological
equivalent of the problem of how to quantize gravity. Here we anticipate that the
Copenhagen interpretation of quantum mechanics (where macroscopic apparatus
are treated as fundamentally classical in contrast with the quantum phenomena they
observe) is replaced by the many-worlds or Everett interpretation [148–152], where
the Universe itself is subject to quantum laws. Hence, the Universe is regarded as
a superposition of quantum states and the space of states is spanned by all possible
Universes.6 In the Everett interpretation, it is assumed that the linear structure of
a quantum theory is exact, so that equations such as the Schrödinger equation or
the quantum constraints of canonical quantum gravity have universal validity and
non-linear terms in the wave-functions are excluded.
Regardless of the interpretation, the canonical quantization of a bosonic field O
on a curved manifold proceeds as in Minkowski spacetime. Namely, one defines
a Fock space and a no-particle vacuum state j0i for the Heisenberg operator O .
The field is then expanded in creation and annihilation operators like the quantum
harmonic oscillator. In the multiverse or “third” quantization, however, the vacuum
state j˝i is not an eigenstate of O . We can expand j˝i into a combination of
eigenstates of O ,
X
j˝i D ci j0ii ; O j0ii D i j0ii ;
i
where the sum is over the ensemble of all possible universes. The probability to
find a certain field distribution is Pi Œ i jci j2 . In usual quantum experiments,
wave-functions lose phase correlations due to interaction with the environment and
performing a measurement means to choose an eigenstate. This process is called
decoherence, that is, a delocalization of interference terms or, in other words, a
local suppression of interference. In collapse models, decoherence amounts to a
global damping or destruction of interference (off-diagonal) terms in the density
matrix associated with the eigenstates (see [152, 154] for reviews).
Within standard quantum theory, decoherence is a well-established and physi-
cally tested concept [155]. Quantum cosmology, however, lies on rather different
grounds. Contrary to the laboratory set-up, choosing an eigenstate of O does not
correspond to performing a measurement in a traditional sense, because there is
no observer external to the quantum Universe. Here we face the same impasse we
6
A third type of interpretation in quantum cosmology is Bohm interpretation [153].
met in relation with CMB correlators and multi-sky averages. To continue, we must
assume that the universe we live in corresponds to a particular choice i D I in
the ensemble. The problem is to understand what mechanism forced the vacuum
state into one of its eigenstate components. The issue of cosmic decoherence is still
unclear in many respects, but some authors have considered it in great detail [156–
164].
Here, we refrain from entertaining ourselves with this difficult subject but
notice that the problem of classical-to-quantum transitions affects cosmology at
different levels, not just the Universe as a whole. Let us either assume a second-
quantized universe or admit that, in a multiverse scenario, some mechanism makes
the scalar quantum field O lose its coherence and selects a given spectrum of
cosmological perturbations. (In the rest of this chapter, the choice of interpretation
is not important.) These perturbations are born as quantum fluctuations but the
observer records them as a classical spectrum. At some point, a decoherence process
must have taken place. In fact, we already have a hint of an answer. We have seen
that “ultraviolet” and “infrared” scales are defined by the Hubble or particle horizon.
The horizon signals the transition between small and large scales and a change of
behaviour of classical perturbations, oscillating inside the horizon and frozen out at
ultra-large scales. We can naturally regard the horizon as the watershed governing
the transition of fluctuations between the quantum and the classical regime. Loosely
speaking, the Hubble and particle horizons are the cosmological analogue of the
laboratory measurement apparatus.
5.6.2 From Quantum Fields to Classical Spectra
An important assumption of inflationary dynamical models is that the inflaton is in

its vacuum state j0i (corresponding to some specific realization j0iI in the multiverse
interpretation). We can imagine that all non-inflatonic fields adjust themselves in the
state of minimum energy while O evolves. These fields, spatially averaged over a
comoving region of the size of the Hubble horizon, constitute a classical background
which defines the state j0i.
Then, we reinterpret the homogeneous classical field . /, solution to the
classical equations, as the expectation value of the quantum field O calculated on
the vacuum state,
˝ ˛
O . ; x/ :D h0j O . ; x/j0i D: . / ; (5.98)
where we use conformal time to eventually match with the notation for classical
perturbations. In Heisenberg picture, the quantum fluctuation ı O of O is defined as
O . ; x/ D . / C ı O . ; x/ ; (5.99)
and has zero vacuum expectation value, hı O i D 0. At this point, we can quantize
the fluctuation on a FLRW background. In general, the canonical field will be
190 5 Inflation
proportional to ı O by some time-dependent factor:
uO . ; x/ D f . /ı O . ; x/ : (5.100)
We can in fact be more specific and consider a field of the form
uO . ; x/ D a. /'.
O ; x/ ; (5.101)
where 'O will represent either the inflaton fluctuation ı O or a tensor polarization
mode hO . One imposes equal-time commutation relations to uO ,
uO . ; x1 /; uO . ; x2 / D 0 D ˘O u . ; x1 /; ˘O u . ; x2 / ; (5.102a)
uO . ; x1 /; ˘O u . ; x2 / D iı.x1 ; x2 / ; (5.102b)
where ˘O u D uO 0 is the conjugate momentum density. The fluctuation field can be

expanded in spatial comoving Fourier modes,
Z
d3 k
uO . ; x/ D uO k . / eikx ; (5.103)
.2/3
where the operator uO k can be written as a combination of harmonic oscillators,

"
uO k . / D uk . /ak C uk . /ak : (5.104)
Here, uk . / is a complex function of k D jkj and . The time-independent creation

"
and annihilation operators ak and ak are defined on the Fock space of normalized
n-particle states jni as
"
ak j0i D 0 ; ak j0i D j1i ; (5.105)
and satisfy canonical bosonic commutation relations,

"
Œak1 ; ak2 D ı.k1 ; k2 / ; (5.106)
" "
Œak1 ; ak2 D 0 D Œak1 ; ak2 : (5.107)
"
Since uO k D uO k , the quantum operator uO .t; x/ is Hermitian. Plugging (5.104)
into (5.102b) and using (5.106), one gets the Wronskian
0
uk uk uk u0k D i : (5.108)
The same quantization scheme is applied to gravitational waves [165, 166], in which
case they are called gravitons.
The classical fluctuation uk . / (D uk;I . / in the multiverse scenario) is the eigen-

value of uO k associated with the realization I. This is the primordial cosmological
perturbation whose spectrum we observe and measure. Therefore,
"
hjOuk . /j2 i :D h0j uO kuO k j0i D juk . /j2 ; (5.109)
and the primordial spectrum is
k3 2
Pu .k/ :D juk 1 . /j ; (5.110)
2 2
where in the right-hand side we have used the asymptotic form of the solution uk
at large scales and evaluated everything at horizon crossing (3.14): j j D 1=k.
The spectral index and its running, introduced in Sect. 4.4.2, can be obtained
from (5.110).
5.6.3 Choice of Vacuum
In the absence of gravity, there exists a global inertial frame, the vacuum state of
the free field uO is uniquely determined by (5.105) and there is a clear distinction
between positive- and negative-frequency modes. On curved spacetimes, however,
the Poincaré group is no longer a symmetry group and there is neither a global Fermi
frame nor a unique vacuum state [167]. An observer can build particle detectors and
determine the local vacuum state j0i accordingly but for other, distant observers the
N Their detectors will measure
state j0i will not correspond to their local vacuum j0i.
the presence of particles:
N D 0;
aN k j0i aN k j0i ¤ 0 : (5.111)
This scenario is often described impressionistically by saying that particles are

created from spacetime geometry. More precisely, the Fock vacuum (5.105) is
regarded as the vacuum at some fixed time i , annihilated by ak D ak . i /, while the
vacuum at generic time is defined by the time-dependent creation and annihilation
operators
" " "
ak . / D v1k . / ak C v2k . / ak ; ak . / D v1k . / ak C v2k . / ak : (5.112)
These relations are the Bogoliubov transformations relating the operators of (5.105)
with those of (5.111), aN k D ak . /. The commutation relations (5.106) are preserved
192 5 Inflation
in time if, and only if,
jv1k j2 jv2k j2 D 1 :
Writing the field operator uO k as the canonical harmonic oscillator in Heisenberg

picture,
"
uO k . / D ak . / C ak . / ; (5.113)
one recovers (5.104) provided

uk D v1k C v2k :
The choice of vacuum j0i is an Ansatz (5.105) for uO k at some fixed time i.
Physical considerations can guide us into its selection.
• Adiabatic or Bunch–Davies vacuum [167, 168]. This state is asymptotically
Minkowski in the remote past/future, where all inertial observers would see no
particles. The infinite-past time-like surface I at ! 1 (defined rigorously
in Sect. 6.1.1) corresponds to the “in” region bounded by the Hubble horizon,
where the mode uk is a positive-frequency plane wave. In the infinite-future time-
like surface I C at ! C1, the mode uk is a negative-frequency plane wave:
˙ik
! ˙1 e
uk ! p : (5.114)
2k
In the absence of a cut-off scale at high energies, there is no reason to impose

initial conditions at a finite scale k D k. i /. Then, the adiabatic vacuum at the infi-
nite past is the unique choice compatible with the Wronskian condition (5.108).
• Local or instantaneous Minkowski vacuum [169]. It is a generalization of the
adiabatic vacuum but imposed as a Cauchy problem at a finite initial time:
1
uk . i / D p ; u0k . i / D ˙ik uk . i / ; (5.115)
2k
where i is momentum-dependent and is determined by some new physics at high

energies.
• Minimal-energy vacuum [170], so called because it minimizes the energy density
[171]:
1
uk . i / D p ; u0k . i / D ˙ik0 uk . i / : (5.116)
2k0
• de Sitter vacua [172–175]. This family of vacua was studied in [176–180] and,
with particular reference to cosmology, in [181–185]. The family is defined by
the Cauchy problem
iH
u k . i / D p ei ; u0k . i / D ik uk . i / ; (5.117)
2k
where i D ˇ=k is momentum dependent, H is constant and and ˇ are real

parameters. Often these vacua are labelled by one complex parameter ˛, for
which reason they are also called ˛-vacua. Since they correspond to states of
minimum uncertainty [181, 182], they can be named minimal uncertainty vacua.
In the limit ! ˇ ! C1, (5.117) reduces to the infinite-past (positive
frequency) adiabatic vacuum, while the local Minkowski vacuum is recovered
for ! 0 and i H D ˇ=a ! 1 (here we used the horizon-crossing relation
k D aH). In general, the ˛-vacua are all inequivalent and non-thermal, meaning
that a detector in a de Sitter vacuum will not end up in thermal equilibrium.
The assumption that the Fourier modes uO k have zero occupation number is necessary
to guarantee a sufficiently long period of inflation, since a non-vacuum state with
too many particles would drag and brake the accelerated expansion [186, 187].
5.6.4 Mukhanov–Sasaki Equation Revisited
A very general result of linear perturbation theory is that the equation of motion for
uk takes the form of the Mukhanov–Sasaki equation (3.32),
u00k C .k2 M 2 /uk D 0 ; (5.118)
where
M 2 D .2 /H2 m2 . / (5.119)
and m. / is an effective mass whose details depend on the nature of the perturbation
(scalar, tensor, choice of f in (5.100)) and on the dynamics (definition of the action,
background Friedmann equations). During inflation or a power-law expansion, it is
possible to write M 2 as
4 2 1
M2 D ; (5.120)
4 2
where is a constant. In this case, the solution of (5.118) was given by (3.36) in
Sect. 3.2.3. Instead of the Bessel functions, one can express the solution also in
194 5 Inflation
terms of the Hankel functions of the first and second kind,
H.1/ .y/ D J .y/ C iY .y/ ; H.2/ .y/ D J .y/ iY .y/ ; (5.121)
which describe outgoing (advancing) and incoming cylindrical waves, respectively.

Then,
p p
uk . / D C1 kj j H.1/ .kj j/ C C2 kj j H.2/ .kj j/ : (5.122)
Asymptotically (we assume > 0),
y1 2 ./
H.1/ .y/ ' iY .y/ ' i y ;

y1 2 ./
H.2/ .y/ ' iY .y/ ' i y ;

s
.1/ y1 2 i .C 1 / iy
H .y/ ' e 2 2 e ;
y
s
y1 2 i .C 1 / iy
H.2/ .y/ ' e2 2 e :
y
At sub-horizon scales (kj j 1), the solution is a superposition of plane waves,

r r
kj j 1 2 i .C 1 / ikj j 2 i .C 1 / ikj j
uk . / ' C1 e 2 2 e C C2 e2 2 e

D CC eikj j C C eikj j : (5.123)
As we know, perturbations with small wave-lengths oscillate, in accordance with

the fact that the quantum scalar field uO locally behaves as a Minkowski harmonic
oscillator. In turn, the quantum harmonic oscillator is defined through a choice of
the vacuum state, which fixes the coefficients CC and C . With the Bunch–Davies
infinite-past vacuum (during inflation, lower signs in (5.114)), for (5.123) one has
1
CC D p ; C D 0 : (5.124)
2k
The local Minkowski and minimal-energy vacua actually agree for the standard
Mukhanov–Sasaki equation (5.118), where the dispersion relation of the field is
.k0 /2 C k2 D 0 : (5.125)
In fact, the two branches of the local Minkowski vacuum (5.115) are
eik i eik i
CC D 0 ; C D p ; and CC D p ; C D 0 ;
2k 2k
while the minimal-energy state coincides with the previous one: initially and locally,
the mode does not feel curvature effects. In models with modified dispersion
relations [171], which can encode some quantum-gravity effects via a phenomeno-
logical cut-off (Sect. 5.10.3), the small-scale solution may differ from (5.123) and
the local Minkowski and minimal-energy states are physically inequivalent. Notice
also that, in general relativity and in de Sitter spacetime, a large class of initial states
approach the adiabatic vacuum after a few e-foldings [188]. Therefore, at least in
general-relativistic inflation the Bunch–Davies choice is very reasonable.
In the long wave-length limit, the appropriately normalized solution is
22 ./ 1
juk 1 j D p jk j 2 : (5.126)
k .3=2/
In all cases of interest the constant is
3
D C ; (5.127)
2
where D 0 in de Sitter and 1 in the slow-roll regime. Therefore, (5.110)
becomes
" #2
k2
32 C
Pu .k/ D 2 : (5.128)
4 2 32
Recalling (5.101), one immediately gets P' D Pu =a2 . For power-law inflation
(constant ), the relation (3.11) between k and the Hubble parameter at horizon
crossing is
k D aH.1 / ; (5.129)
so that
" 3 #2 2
2 2 C H
P' D .1 / 2 3 : (5.130)
2
2
In pure de Sitter, (5.130) gives
2
H
P' D : (5.131)
2
196 5 Inflation
The right-hand side is the square of the gravitational temperature associated with a
FLRW spacetime:
H
TH :D : (5.132)
2
In particular, the de Sitter temperature is constant. We shall come back to this
concept in Sect. 7.7.
Equation (5.131) is the leading-order term for the spectrum on a general
background in slow-roll regime. Using the expansion
3
2 C
2 3 ' 1 C C ; 1 ;
2
where C D 2ln 2 EM 0:73 (recall that EM is the Euler–Mascheroni constant),

we get
2
H
P' ' .1 2 C 2C / : (5.133)
2
Below we will not consider next-to-leading slow-roll expressions such as this

[189, 190], although they can be of interest in the future when high-precision
cosmological tests become available.
We can rewrite (5.130) in another useful form. For power-law inflation, a /
j j1=.1/ D k1=.1/ at horizon crossing (equation (2.187)), so that
2
P' .k/ D A' .k 0 / 1 ; (5.134)
where A' is a dimensionless constant. The spectra in power-law backgrounds a D tp

admit exact expressions in the SR parameters, which are all equal to 1=p [191].
Typically, however, one has inflation in mind and truncates these expressions to
lowest or next-to-lowest SR order.
In all the above formulæ, we have omitted the subscript in slow-roll parameters,
which are all evaluated at horizon crossing. Note that this computation did involve
the Friedmann equations but only indirectly, through the effective mass term in the
Mukhanov–Sasaki equation.
We can now revisit the issue of the quantum-to-classical transition of linear
cosmological fluctuations. When omitting the negative-frequency part in (5.123),
one is actually throwing away a large-scale decaying mode of the field operator
in Heisenberg picture. The correlation between the decaying and non-decaying
mode determines coherence, so that in this approximation one obtains “decoherence
without decoherence,” i.e., without invoking any interaction of the modes with
an “environment” [181]. However, the presence of the cosmological horizon is
crucial to determine the decaying, constant or growing nature of the modes. Beyond
linear level, however, where different modes interact, the problem of decoherence
forces itself again upon our attention. What one usually does, in this case, is to
assume the Everett interpretation, start with the standard inflaton vacuum and turn
it dynamically into a squeezed vacuum. This squeezed vacuum state can be written
as a superposition of excited states and can decohere [192].
To summarize, once outside the horizon quantum fluctuations are frozen and
become classical. One can reach the same conclusion by noting that the energy per
quantum ! D k=a decreases by redshift, and by energy conservation the number of
quanta n! E=! for a given frequency (E is the Hamiltonian eigenvalue coming
from the effective action of ') must increase accordingly outside the horizon [193].
5.6.5 Eternal Inflation
The super-horizon modes of the scalar-field quantum fluctuations acquire the almost
constant classical spectrum (5.131) with ' D ı , when the slow-roll approximation
of the dynamics holds. Thus, during the time of a Hubble expansion ıt H 1 ,
the classical field evolution .t/ is modified by jumps of size ı ˙TH . If
these random jumps dominate the dynamics, there will always be a non-vanishing
probability that the field acquires values suitable to trigger a new inflationary era.
Then, in an expanding region of Hubble size some sub-regions will end their
accelerating expansion and reheating will begin, but in other sub-regions suitable
initial conditions will be generated by the quantum fluctuations and new seedbeds of
inflation will form. This scenario goes under the name of eternal inflation [194–203]
(for updated references on the subject, see [204, 205]).
The broad picture is one of regions in accelerated expansion which, at some
point, stop inflating and subsequently thermalize. Thermalized patches are separated
by still inflating domains, which are continuously created by quantum fluctuations.
The Universe is composed by thermalized and inflating regions. The comoving
volume of inflating regions vanishes at t ! C1, while their physical volume grows
exponentially: the Universe never thermalizes completely. Inflation, once started,
reproduces itself ad libitum in the future. Specifying that the eternal process of
reproduction is future-directed is important. If this scenario extended also to the
infinite past, it would avoid the big-bang problem. We will see in Chap. 6 that this is
not the case [206, 207] and eternal inflation does not extend indefinitely in the past.
On the plus side, however, the graceful entry problem (Sect. 5.10.1) is alleviated in
this framework [196, 197, 204, 208].
Many models of inflation admit fluctuation-dominated regimes. In the highly
homogeneous universe of chaotic inflation, for instance, an infinite number of
causally disconnected inflationary regions are expected to form. As we will see in
Sect. 5.8.1, the dynamics of quantum fluctuations can be interpreted as a stochastic
process, in particular a Brownian motion [37, 195, 209]. The distribution of
198 5 Inflation
reheating and inflating regions is characterized by a fractal geometry [142, 196, 197,
200, 202], which can be described by different choices of measure [202, 204, 210–
217].
5.7 Cosmological Spectra
Once solved the horizon, flatness and monopole problems, inflation does not
exhaust its task and plays a fundamental role in cosmic structure formation. Any
pre-inflationary inhomogeneity and anisotropy is washed away by the primordial
accelerated expansion, so that the origin of irregularities in the energy distribution
of the universe must be traced back to events during or after inflation. It turns
out that the observed anisotropies can be explained as being the Planck-size
quantum fluctuations of the inflaton stretched at horizon scales. Tensor perturbations
(gravitational waves) are generated via the same mechanism. By coincidentia
oppositorum, the study of the large-scale structures is also the study of the smallest
quantum scales and high-energy processes. Several orders of magnitude beyond the
probing capability of accelerators is, at least potentially, cosmology.
The SR formalism gives good control over the theoretical shape and amplitude of
cosmological perturbations. Here we shall restrict ourselves to the linear first-order
approach [218–220], although it is possible to extend the discussion to second-order
perturbations [221–223] and, as we have seen in Chap. 3, even to a non-perturbative,
non-linear set-up.
The standard procedure to compute the perturbation spectra is: (i) Write the
linearly perturbed metric in terms of gauge-invariant scalar and tensor quantities
separately; (ii) Compute the effective action and the associated linearized equations
of motion for a given background solution with constant or small SR parameters;
solve the linearized equations with respect to the perturbations; (iii) Write the scalar
and tensor perturbation amplitudes in terms of this solution; (iv) Since the observed
fluctuations are originated at horizon crossing, the perturbation spectra are evaluated
at this point (equation (3.14), kj j D 1).
5.7.1 Gaussianity
We are in a position to understand why linear primordial fluctuations are Gaussian.

Once chosen a realization i D I for the universe, we can write the complex
coefficient uk as
uk D juk jei#k D Re.uk / C i Im.uk / : (5.135)
By the cosmological principle, the distribution of each mode in (5.135) is isotropic,

implying that the modes uk as well as the spectrum Pu depend only upon the absolute
value k D jkj.
5.7 Cosmological Spectra 199
We use now the properties of random fields enunciated in Sect. 3.4. The
Mukhanov–Sasaki equation (5.118) is linear in uk . Then, the real and imaginary
part of each mode behave like two independent harmonic oscillators for each k. In
" "
the vacuum state, the operators Re.uk /ak and Im.uk /ak have the same probability
distribution, given by the ground-state wave eigenfunction of an harmonic oscillator
(in this case, a Gaussian). The set of modes in (5.135) are then statistically
independent and with the same distribution, and the probability distribution of the
field u is Gaussian. Another way to see this is to note that the phases #k are mutually
independent and randomly distributed in the interval 0 6 #k < 2. If the phase of
each mode is random, then the central limit theorem guarantees that the (classical
version of the) superposition (5.103) is Gaussian if the number of modes is large.
Since the power spectrum Pu .k/ is continuous, the field uk is ergodic, in
agreement with the previous cosmic-choice assumption for any continuous transfer
function describing the time evolution of the perturbation. We can summarize all
these results as:
Gaussianity. In the linear approximation, inflationary fluctuations have a
Gaussian probability distribution and are completely described by the power
spectrum in momentum space. The statistical properties of the perturbations
are evaluated in the ensemble of spatial points in the sky vault.
The Gaussianity of the statistical distribution for the perturbations is a direct
consequence of (i) neglecting second-order terms in the equations of motion and (ii)
taking the cosmological principle for granted. Both are only approximations of the
real world, although very good ones according to experiments. Going beyond the
linear theory and accepting some deviations from perfect isotropy, as CMB probes
indicate, small departures from the Gaussian distribution appear and generate new
interesting features we shall explore later.
5.7.2 Linear Tensor Perturbations
The scalar field perturbation induces fluctuations on the metric and tensor per-
turbations are produced by back-reaction. In Sect. 3.2.3, we have seen that the
Mukhanov–Sasaki equation for each tensor mode is (3.32) with
wk D ah;k (5.136)
and (see equation (3.37)) [224]

D ; (5.137)
1
200 5 Inflation
for power-law inflation with D 0. To tie in with the quantum picture, one notices
that (3.24) suggests to define the canonical field
wk
uk D ; (5.138)
2
so that in pure de Sitter
2
2 H
Ph D 4 ; (5.139)
2
and our final result for the spectrum of tensor perturbations (4.63) is, to lowest SR
order,
2
H
Pt D 8 2 : (5.140)
2
In pure de Sitter, this is a constant. During inflation, quantum fluctuations of the

gravitational field are pushed outside the horizon, where they remain frozen.
The tensor spectral index (4.64) can be calculated in two equivalent ways. For the
lowest-order de Sitter expression (5.140), it is convenient to recast the momentum
derivative into a time derivative,
d 1 d
' ; (5.141)
d ln k H dt
and then act on (5.140) regarding H as non-constant, so that
nt ' 2 : (5.142)
Alternatively, one uses (5.134) and obtains
2 2
nt D D ; (5.143)
1 p1
which is an exact expression in agreement with (5.142) for small . The inflationary
spectrum of tensor perturbations in general relativity is almost scale invariant,
always with a slight red tilt (nt . 0).
5.7.3 Linear Scalar Perturbations
Scalar perturbations are more involved than the tensor sector. The perturbed Einstein
equations give a set of expressions for the inflaton fluctuation ı and the gauge-
invariant metric scalar perturbations ˚ and . At large scales, one can ignore
the anisotropic stress and set ˚ D , so that there are only two independent
dynamical equations. Combining them together, one finds an equation for the
canonical variable
0
u D zR D aı C ; (5.144)
H
where z :D a 0 =H and R is the curvature perturbation (3.63). The exact (in the
SR parameters) Mukhanov–Sasaki equation for the linear perturbation uk reads
[225–227]

z00
u00k 2
C k uk D 0 ; (5.145)
z
where
z00
D H2 2 C 2 3 4 C 2 2 C 2 C 2 : (5.146)
z
To understand where this equation comes from, we can either sketch its derivation
from the linearized Einstein equations or employ a very efficient trick. Let us first
briefly review the traditional approach.
The general scalar equation of motion is (2.66), V; D 0. Splitting
the conformal metric into a background and an inhomogeneous perturbation as
in (3.52), the d’Alembertian is divided into a pure FLRW operator a2 Q D @2
2H@ and a contribution containing the metric back-reaction:
Q C r2
D C ı : (5.147)
a2
Linearizing the equation of motion, we get (tilde omitted in background quantities)
2
@ C 2H@ r 2 C a2 V; ı a2 .ı/ D 0 : (5.148)
If rolls slowly down its potential, in a zero-order approximation we can ignore the
metric back-reaction and forget the last term ı. In this case, ı is said to be a test
field. In many circumstances, however, the metric back-reaction is not negligible.
This happens, for instance, in eternal inflation or in certain high-energy models, and
one should check the consistency of this assumption explicitly. First, one computes
.ı/ , which turns out to be a function of the scalar perturbations of the metric. In
longitudinal gauge and from (3.52),
2˚ p
ıg00 D ; ı g D a4 .˚ 3 / I
a2
202 5 Inflation
plugging that into (2.67) and linearizing,

1 p 4H p
a2 .ı/ D a2 . 00
C 4H 0 /ıg00 C 0
a2 .ıg00 /0 4
.ı g/0 C 4 ı g
a a
0
D .˚ 0 C 3 0 / 2a2 ˚V; : (5.149)
In momentum space, (5.148) is identical to (3.26), with u D aı . The equation of

motion for yields (e.g., [193])
H
' 0
ı ;
so that the mass contribution in (5.119) is
H
m2 D a 2 V ; C 2a2 0
V; C O. 2 / ' H2 .3 3/ ;
0
where we recalled (5.63) and neglected the term in (5.149). The effective mass
term is of the form (5.120), with (5.127) and
' 2 (5.150)
at linear SR order, in agreement with (5.146).

The two-line trick mentioned above is the following. One obtains a friction-
free Mukhanov–Sasaki equation from (5.148) if uk aı , which determines z
in (5.144). At large scales, one can ignore Laplacian terms / k2 in the linearized
equations. Also, the comoving curvature perturbation is approximately constant
(see (3.58) and (3.64)), so that at super-horizon scales u00k ' z00 Rk and
z00
u00k uk ' 0 :
z
The only missing term, which can be readily added, is the Laplacian k2 (with
coefficient 1, as an inspection of (5.148) immediately shows). We have thus
recovered (5.145) from the conservation equation of a gauge-invariant perturbation.
The scalar spectrum can be found by solving (5.145) or, to lowest SR order,
directly from (5.148). For pedagogical purposes, the second option is clearer. If the
SR approximation holds and can be treated as a constant, then the only effect
of metric back-reaction is to change the normalization of the power spectrum and
we can ignore the details of the effective mass z00 =z. This is tantamount to going to
the flat gauge ˚ D D 0 and neglecting the contribution (5.149). Therefore, the
spectrum of scalar field fluctuations in de Sitter is simply7

2
H
Pı D : (5.151)
2
The primordial scalar spectrum (4.57) is the spectrum of the curvature perturbation
. At large scales, coincides with the comoving curvature perturbation (3.64). In
flat gauge, R D .H= 0 /ı ; taking (5.43a) into account, we get
2
2 1 H
Ps D : (5.152)
2 2
Using (5.51) and (5.141), one obtains the scalar spectral index
ns 1 ' 2 4 : (5.153)
For an exact power-law expansion, the scalar index is the same as (5.143),
2 2
ns 1 D D : (5.154)
1 p1
The SR approximation guarantees that the scalar spectrum be almost scale invariant.
To lowest order, the running (4.59) is quadratic in the SR parameters,
˛s ' 2.5 4 2 2 / ; (5.155)
where we used (5.51) and (5.52). Note that the SR parameters in all the observ-
ables (5.142), (5.152) and (5.153) are evaluated at horizon crossing. We omitted,
and will do so again in similar expressions, the subscript * in and everywhere.
In this section, we have used three types of approximation: linearity of perturba-
tions, large-wave-length limit and extreme SR conditions. The latter allowed us to
neglect the inflaton mass and metric back-reaction. This is no longer possible in the
case of fast-roll inflation, where the SR approximation is not enough.
7
For power-law inflation, the exact expression for Pı is (5.130) with given again by (5.137)
[191].
204 5 Inflation
5.7.4 Consistency Relations and Lyth Bound
The tensor-to-scalar ratio (4.66) is very small in inflationary models. This is because
the scalar amplitude (5.152) is much larger than the tensor amplitude (5.140):
r ' 16 : (5.156)
We can collect (5.142), (5.153) and (5.156) in the set of first-order consistency
equations
r
nt ' ; (5.157)
8

r 3
˛s ' r C 5.ns 1/ 2 2 : (5.158)
16 4
If
jj min.; jj/ ; (5.159)
then the set closes and the scalar running depends only on first-order observables.
This does not happen in power-law inflation, where jj D O.; jj/. The consistency
equations relate cosmological observables in a way typical of inflationary scenarios,
where the scalar and gravitational spectra have a common physical origin. They
are a typical result of inflation which other models of structure formation cannot
reproduce. Often, (5.157) is used in data analysis to lower the number of parameters
in the model (Sect. 4.4.2). In doing so, one is implicitly assuming that inflation was
realized by an ordinary scalar field.
Combining (5.76) and (5.156), one canpexpress the excursion of the inflaton
RN
during N e-foldings as =MPl ' 0 dN 0 r=8. Since the tensor-to-scalar ratio is
approximately constant during inflation,
r
r
'N : (5.160)
MPl 8
The scales at which r is observed are the 2 < ` . 100 multipoles of the CMB
spectrum, corresponding to modes which left the horizon during the last Nlow-` 4
e-folds. The observed anisotropies in the sky have been produced in the last Nk
60 e-foldings, so that the excursion Nlow-` sets both a lower bound on the total
variation , 8. =MPl /2 > Nlow-` 2
r, and the upper limit r < . =MPl /2 =2 for
the tensor-to-scalar ratio, the Lyth bound [228]. A more refined estimate, which
takes into account the experimental constraints on the scalar spectral index and the
variation of r.N / across all scales, sets Nmin 30 as an effective lower limit on the
number of e-folds replacing Nlow-` [229]:

2 2
8 2
r< 2 ' 10 : (5.161)
30 MPl MPl
On the other hand, the maximum observable tensor signal is set by Nk ,

2 2
8 3
r> 2 ' 10 : (5.162)
60 MPl MPl
Assuming a total excursion of order of the reduced Planck mass ( ' MPl ), we
obtain the allowed range for the tensor-to-scalar ratio in single-field inflation:
103 < r < 102 : (5.163)
A detection at the threshold (5.161) would signal an O.MPl / field excursion.

Conversely, models where the field variation is smaller than the Planck scale MPl
are usually characterized by a low inflationary scale V 1=4 and produce a negligible
gravitational spectrum, well below the range (5.163). We will see examples with
r 0 in Chap. 13
5.8 Non-Gaussianity
In Sect. 4.6.3, we have reviewed some of the mechanisms which can be responsible
for a non-Gaussian statistical distribution of perturbations. Now that we have a
model of the high-energy early universe, we can complete the discussion on non-
linear perturbations. Before doing that, we mention just one more possible source
of non-Gaussianity. It is the production of topological defects, which occurs during
the whole history of the universe. Their evolution is highly non-linear and naturally
generates non-Gaussian statistical distributions. Numerical simulations show that
the combined effect of large populations of defects results in almost Gaussian
perturbation spectra, in agreement with present observations. However, inflation is
the main, if not the only, responsible of primordial fluctuations, so that the role of
non-inflationary primordial sources is expected to be relatively marginal.
Within the inflationary paradigm, the standard way to generate non-Gaussian
spectra is to consider self-interaction.8 In general, even if one assumes a classical
quadratic potential, quantum corrections to the effective action of the inflaton will
8
Inflationary non-Gaussianity from self-interaction has been studied extensively in [230–248].
Non-Gaussianities are also generated, for instance, by the inclusion of higher-dimension operators
in the inflaton Lagrangian [249], in warm inflation [250, 251], ghost inflation [252, 253] and when
assuming that the inflaton does not sit in a vacuum state [186, 187, 254].
206 5 Inflation
give rise to higher-order terms, for instance of the form (5.81). If the coupling
constant is small, quantum fluctuations of the field D h i C ı are approximately
Gaussian with variance 2 :D hjı j2 i. However, if self-interaction is important
we can no longer neglect second-order perturbations. For instance, for a monomial
potential V / n , in the ESR regime given by (5.20) and (5.21) the comoving
curvature perturbation in flat gauge is
H 2
R 0
ı ı D h iı Cı ;
and one can immediately see that the statistics is not Gaussian: hRi ¤ 0, hR3 i ¤
0, and so on. This is precisely the non-linear mechanism discussed in Sect. 4.6.3,
realized concretely by the inflaton. We already computed the non-linear parameter
fNL (4.108) in the local form of non-Gaussianity. That can be obtained also in the
ıN formalism, via the ıN formula (3.51) [87]. Thinking N as a function of the
scalar field and expanding ıN in a Taylor series around the background trajectory
.t/,
N;
NL .t; x/ D ıN .t; x/ ' N; ı .t; x/ C 12 N; Œı .t; x/2 D N; ı C .N; ı /2 ;
2N;2
where ı .t; x/ :D .t; x/ .t/. In the local-form case, comparing this expression
with (4.99) we get
5 N;
local
fNL D : (5.164)
6 N;2
Noting that N is the backward-oriented number of e-folds,

r
H (5.53) 2 ; (5.51) 2
N; D D ; N; D N; D N; . / ; (5.165)
P 2 2
local
one has fNL D .5=12/.2 2/ D .5=12/.1 ns 2/ from (5.153), which
misses an extra term 2 in brackets to get (4.108). This contribution comes from the
back-reaction of the metric, which we ignored in (5.164). An improved calculation
eventually yields the correct result in the squeezed limit.
5.8.1 Stochastic Inflation
Another framework wherein to study scalar-field non-linearities is stochastic infla-

tion [198, 199, 230, 233, 255–262]. Stochastic inflation is an approximated method
according to which the solutions of the equation of motion for the scalar perturbation
in the long-wave-length limit k aH are connected to those in the k aH limit at

the Hubble horizon. This approach allows one to model the coupling between large-
and small-wave-length modes.
The scalar field (or any other derived scalar quantity ', such as ' D ı )
is separated into a “classical” or coarse-grained contribution 'c , encoding all the
modes outside the Hubble horizon, and a quantum or fine-grained part 'q taking into
account the in-horizon modes (equation (3.15)). Therefore, the classical part is the
average of the scalar field on a comoving volume with radius R D rH =", 0 < " < 1,
larger than the shrinking Hubble radius rH D .aH/1 , while the oscillations of 'q
cancel one another in the average. With this decomposition, the equation of motion
for ' becomes a Langevin equation with a stochastic noise source generated by
the fine-grained contribution of the quantum fluctuations. If self-interaction is not
negligible, the evolution of 'c is non-linear even if the quantum fluctuations 'q are
exactly Gaussian.
Low frequencies are cut by a window function W.jx x0 j=R/ falling off at
distances larger than R. Denoting with W the Fourier transform of W, we can choose
3
W.k/ D 1 j1 .kR/ :
kR
Its asymptotic limits are
lim W D 0 ; lim W D 1 :
kR!0 kR!C1
We can approximate the window function to a Heaviside distribution,
W ' .k "aH/ ; (5.166)
where 0 < " < 1. The average of 'c is performed over a volume R3 D .rH ="/3 .
Let ' be a test field and let us ignore metric perturbations. Within the horizon,
the scalar-field modes are approximated by
Z h i
d3 k "
'q .x/ D 3
W.k/ 'k .t/ak eikx C 'k .t/ak eikx ; (5.167)
.2/
while the long-wave-length modes 'q have the same definition with .k "aH/
replaced by ."aH k/. Substituting (3.15) in the equation of motion

r2
@2t C 3H@t 2 ' C V;' D 0 ; (5.168)
a
one gets

2 r2
@t C 3H@t 2 'c D V;'c C 3H ; (5.169)
a
208 5 Inflation
where V;'c D @V=@'j'D'c is the average of V;' over a comoving volume and

1 2 r2
.x/ :D @t C 3H@t 2 'q : (5.170)
3H a
The noise term .x/ stems from quantum fluctuations, which become classical at
horizon exit and contribute to the background wherein 'c lives. In this sense, one
often refers to as the field back-reaction.
If the conditions of extreme slow-roll regime are satisfied, one can neglect the
term 'Rc in (5.169) and obtain the Langevin equation
1 V;'c
'Pc .x/ ' r 2 'c .x/ C .x/ : (5.171)
3Ha2 3H
We briefly mention that the Langevin equation is associated with the probability
distribution PŒ'c ; t of the value of the scalar field at time t at a given point x in the
comoving volume. One can show that P obeys the Fokker–Planck equation [256,
263, 264]

@ @ 1 r2 @2 H 3
PŒ'c ; t D ' c V ;' PŒ' c ; t C PŒ'c ; t ; (5.172)
@t @'c 3H a2 c
@'c2 8 2
presented here in Itô’s version. Solutions to this equation determine the coupling
among modes at horizon scale.
For a de Sitter universe (H D const) the noise term, calculated from the equation
of motion of the massless scalar (5.167), is [233]
Z
d3 k 1 "
.x/ D i"H 3 a 3
p ı.k "aH/ ak eikx ak eikx ; (5.173)
.2/ 2k 3
where a D exp.Ht/. Its vacuum expectation value and two-point correlation

function are
H3 sin."aH%/
h.x/i D 0 ; h.x/.x0 /i D 2
ı.t t0 / ;
4 "aH%
where % D jx x0 j. In particular,
H3
h.x/.x0 /ixDx0 D ı.t t0 / : (5.174)
4 2
The last equation characterizes as a white noise (infinitesimally short correlation
time) but stochastic inflation can be extended to colored-noise models [265].
From (5.171), one can recognize two sources of non-linearity. The first is the self-
interaction of the inflaton (e.g., if its potential contains terms such as n , n > 2),
while the second is the back-reaction of the field fluctuations on the background,
encoded in the noise term. The statistical distribution of the field 'c is non-Gaussian,
even if quantum fluctuations are completely random. While fluctuations relevant for
the visible universe are Gaussian, at scales much larger than the present Hubble
horizon the field distribution can be more complicated and, in particular, non-
Gaussian [137, 257]. Therefore, we expect a small non-Gaussian effect in the
observable region, at least in the case of a single scalar field.
A calculation at first SR order recovers the non-linear parameter in the squeezed
limit (4.108) via (5.153) [244]. A powerful approach obtaining this result in a
reasonably easy and robust way is the space-gradient formalism [235, 266–268],
a development of the separate universe method [269] (Sects. 3.1.4 and 3.3.1). It is
convenient to leave the gauge unspecified until towards the end, and work with the
metric
ds2 D N 2 .t; x/dt2 C a2 .t; x/dx˛ dx˛ ; (5.175)
where N.t; x/ and a.t; x/ are a locally defined lapse function (see Sect. 9.1.2) and
scale factor, respectively. In synchronous gauge, N D 1. Physical quantities such
as H.t; x/ D aP =.Na/ and the scalar field .t; x/ are defined on an inhomogeneous
background and evolve, separately in each homogeneous patch (“at each point”),
through the dynamical equations once the initial conditions have been specified.
Then, one can convert time derivatives into spatial gradients. For instance, at lowest
order in the gradient expansion and at large scales,
H2
@˛ H ' @˛ ; @˛ ˘ ' H@˛ ; (5.176)
˘
where
P P
H ˘P
˘ :D ; D ; D : (5.177)
N NH 2 NH˘
The non-linear extension of the Mukhanov–Sasaki variable (5.144) is
U˛ :D zR˛ ; (5.178)
where
a.t; x/˘.t; x/
z :D (5.179)
H.t; x/
210 5 Inflation
and R˛ is the non-linear comoving curvature perturbation (3.44) (here we consider

only spatial indices since R0 0 at lowest order). From (3.46),

˘
U˛ D a @˛ @˛ ln a : (5.180)
H
At large scales, R˛ is conserved in time, so that differentiating twice with respect to

t one gets
zR
UR˛ U˛ ' 0 : (5.181)
z
However, it is more convenient to keep the decaying mode implicitly dropped in the
recursion of UP˛ D .Pz=z/U˛ . Then, the equation of motion can be recast as
UR˛ C F UP˛ M 2 U˛ ' 0 at large scales ; (5.182)
where
NP zP zR
F :D NH ; M 2 :D F C D 2 C O./ : (5.183)
N z z
In the linear approximation and in momentum space, (5.182) holds for U˛ '
@˛ u ! ik˛ uk . As in the linear case, the equation of motion for U˛ can be written
as an equation for the coarse-grained part U˛c sourced by a stochastic noise term.
In momentum space, the coarse-grained part of the Mukhanov–Sasaki variable is
U˛;k
c
D U˛;k W.kR/. With " sufficiently smaller than 1 (R rH ), we can safely
discard the k2 term (second-order gradient) in the Mukhanov–Sasaki equation.
Therefore, at large scales
UR˛c C F UP˛c M 2 U˛c D ˛ ; (5.184a)

Z
d3 k R C W.2@
P
˛ .t; x/ D .ik˛ /eikx ŒW t C F/uk ˇk C c.c. ;
.2/3
(5.184b)
where c.c. stands for complex conjugate and ˇk is a complex stochastic variable
with ensemble average
hˇk ˇk0 i D .2/3 ı.k k0 / I (5.185)
it simulates the continuous crossing of modes outside the horizon and their fueling
of the coarse-grained part. As it stands, (5.184) is the non-linear extension of the
Langevin equation (5.171) and it properly encodes the full stochastic contribution.
This would not be the case if we started from (5.181), where the velocity degree of
freedom associated with the decaying mode has been absorbed [267].
Equation (5.184) can be expressed as a Langevin differential equation in the
curvature perturbation (superscript c omitted):

zP P ˛ D ˛ :
RR ˛ C 2 C F R (5.186)
z z
R ˛ term, one gets

To lowest SR order, zP =z ' NH and, neglecting the R
˛
RP ˛ ' : (5.187)
.2NH C F/z
At first order in a perturbative gradient expansion,
.1/
˛
RP .1/
˛ ' ) RP ' ; (5.188)
Œ.2NH C F/z.0/ .2NH C F/z
.1/ .1/
where we used the gradient expansion, R˛ ' @˛ R and ˛ ' @˛ (Sect. 3.3.3)
and in the second expression the superscript .0/ has been omitted, so that z, N and
H are defined on the homogeneous background. The power spectrum (5.152) is
recovered from the solution of (5.188) (Problem 5.2).
At second order in the perturbation,
.2/
˛
RP .2/ .1/
z1 g.1/ .0/ :
˛ D ˛ fŒ.2NH C F/N (5.189)
z
.2/
The second term is higher order, ˛ D O. 2 /, and can be dropped. To calculate
the first term, we fix the gauge so that the time coordinate coincides with the natural
time variable during inflation, the logarithm of (the local value of the inverse of) the
comoving Hubble radius [257, 267]:
t D ln.aH/ :
Consistently, the right-hand side increases with time. Differentiating this expression
with respect to t, we obtain
1
NH D ;
1
212 5 Inflation
P
so that, to lowest SR order, N=N P
' H=H and 2NH C F ' 3. We need also the
relation
˘ a a˘
@˛ z D @˛ a C @˛ ˘ 2 @˛ H
H H H
a a˘
D U˛ C a@˛ C @˛ ˘ 2 @˛ H
(5.180)
H H
(5.176)
' U˛ C a.1 C /@˛ : (5.190)
Notice that in synchronous gauge one would have had to expand Hz D a P rather
than z, thus loosing the term in crucial for the the final result. On surfaces of
constant time, by definition @˛ D 0, so that @˛ a=a D @˛ H=H [267]. Therefore,
U˛ D .1 /a@˛ and, combining this with (5.190), we get
@˛ z D .2 /U˛ D 12 .1 ns /U˛ :
Then,
.1/
" # .1/
1 .1/ ˛ @2 @ˇ @ˇ z.1/ ˛
RP .2/
˛ ' .z / D
3 z.0/ 3z.0/
' 12 .1 ns /R.1/ RP .1/

˛ ;
from which
R.2/ ' 12 .1 ns /R2 : (5.191)
Now we are ready to write down the curvature perturbation at second order,9
RNL ' R C 12 R.2/ D R 14 .1 ns /R2 : (5.192)
This expression should be compared with the local form of the non-linear comoving
curvature perturbation,
3 local 2
RNL ' R fNL R hR2 i ; (5.193)
5
derived from (4.99) (the sign in front of fNL

local
is because R ' at large
scales). Thus, upon computing the bispectrum, we have reobtained the non-linear
parameter (4.108).
9
In [270], various definitions of the second-order curvature perturbation are reviewed.
5.9 Observational Constraints on Inflation 213
5.8.2 Multi-field Non-Gaussianity
In scenarios with dynamical scalar spectators during inflation, a non-Gaussian

statistics arises when the light fields reach the bottom of the potential while
their quantum fluctuations and the interaction with the inflaton generate non-
linear effects [135–138, 271–276]. Also the curvaton scenario is characterized by
appreciable non-Gaussianity [140, 141, 277–280]. In the presence of many light
fields, the curvature perturbation can evolve after horizon exit due to the presence
of isocurvature modes. Thus, one could expect a more pronounced non-Gaussianity
statistics [237, 270, 281]. In general, however, non-Gaussianity is small in multi-
field inflation, of which assisted inflation is a special case. Clear indications of this
are given by analytical treatments (mainly, the ıN formalism) of the multi-field
bispectrum [87, 237, 270, 281–288] and trispectrum [100, 239, 240, 289, 290].
5.9 Observational Constraints on Inflation
5.9.1 Temperature Spectra
Since standard inflationary models produce fluctuations of the total density, the
ensuing scenario of structure formations is adiabatic, in pleasant accordance with
CMB and large-scale structure observations. Also, in Sect. 4.4.2 we saw the
experimental bounds on the primordial spectral amplitudes and indices. These
translate into constraints on the inflaton potential.
First, rewrite (5.152) in the ESR regime as
8 V
Ps D : (5.194)
3 m4Pl
Assuming no tensor signal (r D 0), (4.69) fixes the normalization at the pivot scale
k0 D 0:05 Mpc1 . Considering that < 1, this corresponds to an upper bound on the
inflaton potential at horizon crossing: V 1=4 < 6:5 1016 GeV at k0 D 0:05 Mpc1 .
Including also a tensor signal and expressing in terms of r via the consistency
relation (5.156), the PLANCK constraint (4.69) gives
1=4 r
3 1=4
V 1=4 D rPs .k0 / mPl ' 1:93 1016 GeV (5.195)
128 0:12
at k0 D 0:05 Mpc1 . Using the upper bound (4.73), one gets a 95 % CL upper bound
for the Hubble parameter,
lPl H < 7:2 106 at k0 D 0:05 Mpc1 ; (5.196)

214 5 Inflation
while for (4.74) the coefficient is lowered to 5:5. The energy density at the end of
inflation is at least 12 orders of magnitude smaller than the Planck density.
For the scalar potential (5.81) and from (5.85g), (5.85h), (5.153) and (5.156), we
have
2n C 4 16n
ns 1 ' ; r' : (5.197)
4Nk C n 4Nk C n
The scalar spectrum is always red tilted. For a given number of e-folds Nk , these
equations identify a map between monomial potentials n and points in the .ns ; r/
plane.
The border of large-field models (0 6 < , V; > 0) and small-field models
( < 0 < , V; < 0) corresponds to n D 1 (V; D 0), giving [291]
8
r D .ns 1/ :
3
The region spanned by large-field, small-field and hybrid models in the .ns ; r/ plane
are depicted in Fig. 5.7. The border between large-field and hybrid models is at
D ,
r D 8.ns 1/ :
Figures 5.8 and 5.9 show the marginalized constraints at the 1 and 2 level for the
scalar spectral index versus the tensor-to-scalar ratio [43, 111, 147]. This likelihood
analysis, which can be performed via computer packages such as COSMOMC [292],
is one of the so-called “top-down” approaches: one asks what the probability is
that a theory predicting a given set of observables would realize the observed
experimental data [293]. The Harrison–Zel’dovich spectrum is excluded at more
Fig. 5.7 Single-field models in the .ns ; r/ plane

0.25 Planck TT+lowP

Planck TT+lowP+BKP
+lensing+ext
N=
N=
60
0.20
50
φ2
0.15
Co
r0.002
nv
Co ex
nc
ave
0.10
φ
0.05
0.00
0.95 0.96 0.97 0.98 0.99 1.00
ns
Fig. 5.8 Two-dimensional joint marginalized constraint (68 % and 95 % CL) on the primordial
scalar index ns and the tensor-to-scalar ratio r0:002 at k D 0:002 Mpc1 and for zero running,
derived from the data combination of PLANCK and other data sets (“ext” includes BAO, supernovæ
and H0 observations, while BKP is the BICEP2CKeck ArrayCPLANCK joint likelihood on
B modes). These contours change when the running is included; see Fig. 4.16 (Credit: [43],
reproduced with permission ©ESO)
Fig. 5.9 Joint marginalized constraints in the .ns ; r0:002 / plane from PLANCK in combination with
other data sets, compared with the theoretical predictions of selected inflationary models. Hilltop
quartic models have a potential V / 1 . = 0 /4 (Credit: [111], reproduced with permission
©ESO)
216 5 Inflation
than 99 % CL, and so are monomial potentials with large n. According to data,
the quartic potential 4 is ruled out and even the quadratic potential is under tight
pressure. Small-single-field models such as natural inflation and higher-derivative
gravity seem to be favoured. The PLANCK constraint f & 6:9 MPl [33] disfavours
the original interpretation of natural inflation as a pseudo-Nambu–Goldstone boson,
where f . MPl . However, a multi-axion generalization of the same scenario can
easily produce a effective super-Planckian f [114].
The formulæ (5.142) and (5.153) express the spectral indices nt and ns in terms
of the slow-roll parameters. If the accelerated expansion is rapid enough (; 1),
perturbations of cosmological scale exit the horizon soon one after another. Since,
with good approximation, the physical conditions are about the same when small
and large scales cross the horizon, the fluctuation spectra will be invariant. The fact
that inflation does not predict perfect scale invariance is a great success. The zero-
order Harrison–Zel’dovich scalar spectrum ns D 1 is ruled out by observations and
it is necessary to know at least the first-SR-order expressions of the spectral index.
Inflationary models with different Lagrangians often produce similar observables
ns and r, so that a single type of observation cannot distinguish among inequivalent
theories. The combination of various experiments and their improving precision can
break this degeneracy [294].
5.9.2 Polarization
Cosmic variance forbids an accurate determination of the primordial gravitational-

wave spectrum at large scales, where the latter has the largest amplitude,10 while
small scales are more affected by the uncertainty on cosmological parameters.
Polarization maps can help to decompose the tensor contribution from the scalar
one. Since the inflaton is a scalar field, no B-modes arise from density perturbations
to linear order. Inflationary gravitational waves are not directly detectable but they
leave a BB imprint. Magnetic-type polarization is produced by vector and tensor
modes but it is negligible in both cases. In particular, polarization from gravitational
waves is quite weak due to the small tensor-to-scalar ratio predicted in standard
inflationary models.
As for the temperature spectra, the polarization signal of topological defects is
quite distinct from that of inflation. Due to vector modes, the signal of the B-type
polarization is enhanced to the order of 1 K at small angular scales [296]. The
position and amplitude of the observed temperature and TE-polarization peaks,
together with the upper bounds on BB, already rule out defect models. The role
of B polarization will be relevant to constrain other scenarios alternative to standard
inflation and non-inflationary seeds generating a strong gravitational-wave signal or
vorticity.
10
A preheating phase can enhance the tensor signal after inflation, thus enhancing the expectation
for detection of a stochastic gravitational-wave background [295].
An even more interesting situation would be determined by a non-zero cosmo-

logical contribution to the cross-correlation spectral C`TB and CÈB , showing that the
primordial perturbation spectrum has definite handedness.
Parity violations may arise from several physical mechanisms. In scalar-tensor
theories and other non-minimally coupled models, a scalar field ˚ can inter-
act with the electromagnetic field strength F via terms of the form L˚;F /
f .˚/ F F . If f .˚/ D ˚, the field ˚ is a pseudo-scalar [297–300]. A
non-minimal axion inflaton in the same class of theories can couple with gravity

through interactions such as the Pontryagin term L˚;R / f .˚/ R R .
TB EB
Non-trivial C` and C` spectra are generated by both types of terms [301–303]
as well as by others with gravity-electromagnetic axial coupling [304]. Another
intriguing possibility, arising in models of quantum gravity, is that gravity itself
may be chiral. In metric formalism, this means
p that one can identify left- and right-
handed graviton modes hL;R D .hC ˙ ih /= 2 coupling differently with matter, via
some effective Newton constants GL;R . The physical origin of such a situation [305–
309] can be understood in first-order formulation (see Chap. 9), where GL;R depend
on the Barbero–Immirzi parameter. We postpone a discussion of axion inflation in
string theory to Sect. 13.4.
5.9.3 Non-Gaussianity
Current experimental bounds on the non-linear parameter fNL (Sect. 4.6.4) are
compatible with the small level of non-Gaussianity predicted by inflation. In the
local model, this is (4.108),
5
local
fNL ' .1 ns / : (5.198)
12
We might regard this equation as a lowest-SR-order consistency relation joining the
set (5.157) and (5.158) [247, 310].11
One should note that the post-inflationary era greatly enhances non-Gaussianity,
post post
up to fNL D O.1/ [311–313]12 or even fNL D O.100/ from a suitable preheating
phase [314, 315]. In addition to the post-inflationary contribution, one must consider
also angular averaging. The total observed fNL is thus fNLobs
D O.1/ C fNL rather than
the bare inflationary result (4.108) [316]. Therefore, the non-linear effect of standard
11
A small non-Gaussian component, proportional to the tensor amplitude and spectral index nt ,
also comes from the three-point functions involving the graviton zero-mode [245]. Since the tensor
amplitude is much smaller than the scalar one, we can neglect this term with respect to the scalar
bispectrum.
12
This is due to the fact that, at second order in perturbation theory, the longitudinal gauge condition
˚ D 0 is modified as ˚ .2/ .2/ D 4 2 at large scales, thus providing a non-trivial second-
order correction to the Sachs–Wolfe effect [222].
218 5 Inflation
single-field inflation, if the SR approximation holds, is always sub-dominant.

Fortunately, we are able to discriminate between primordial and post-inflationary
contributions and separate the signals according to their features. The constraints
quoted in Sect. 4.6.4 are on the first type of contribution.
5.10 Unsolved and New Problems
The big bang and the cosmological constant still constitute unsolved issues. Inflation
neither resolves the big-bang singularity nor explains the fine tuning of . In
parallel, it solves four puzzles of the hot big bang model, but at a price. It is
almost the rule in theoretical physics that a good solution to a problem entails the
emerging of other problems. This happens because an answer to a question often is
just another, more instructive way to ask the same question.
5.10.1 Graceful Entry Problem
To develop the idea of inflation, we have first considered a kinematical and,

then, a dynamical model in a FLRW spacetime, Sects. 5.2 and 5.3.1. Then, we
set the initial conditions via the pre-inflationary chaotic scenario, which is far
from being FLRW. Linear inhomogeneous perturbations fall short of justifying the
homogeneous approximation in the face of the extremely inhomogeneous initial
conditions. The question remains whether inflation is a likely event in a spacetime
with general metric or not. Sometimes, this is called graceful entry problem. The
question of the naturalness of inflation in cosmological models with non-FLRW
initial conditions has been debated in early papers [317–322].
In [317, 318], it was conjectured that all expanding universes with a positive
cosmological constant tend asymptotically to de Sitter. There are obvious counter-
examples to this, for instance a closed universe collapsing before inflation. However,
under certain hypotheses this became a “no-hair” theorem for Bianchi universes
[320], later generalized to inhomogeneous cosmologies [321]. The name suggests
that a universe endowed with an arbitrarily complicated structure (“hair”) would
lose its memory (get “bald”) in an exponentially fast time scale. With due caution,
the no-hair theorem can be stated as follows.
No-hair theorem. Given a spacetime .M; g/ such that:
1. initially, it is expanding;
2. it is endowed with a a positive cosmological constant ;
3. it is described by an energy-momentum tensor obeying the strong and
dominant energy conditions;
4. the spatial Ricci curvature scalar .3/ R is non-positive in “sufficiently” large
regions.
Then, these regions evolve towards de Sitter spacetime.
5.10 Unsolved and New Problems 219
The statement requires some comments. The first assumption, which we would
translate as H.ti / > 0 in a cosmological setting, adapts the theorem to the
physical scenario. The presence of a cosmological constant is essential. For an
inflationary universe, mimics the slow-rolling inflaton with effective equation
of state w 1. The energy-momentum tensor described in point 3 represents the
contribution of all matter fields except the inflaton, the latter corresponding to the
cosmological constant. Point 4 is somewhat delicate. It is possible to require that
space has negative or zero curvature .3/ R at all points, corresponding to an open or
flat universe. In this case, one is excluding the existence of regions collapsing into
black holes, but these regions do not interfere with inflation if they are sufficiently
small in comparison with expanding volumes.
The theorem, unfortunately, does not really demonstrate the naturalness of
inflation. In fact, a cosmological constant is much less than a (set of) dynamical
slow-rolling field(s), even if the two notions blur in the extreme slow-roll regime. In
part, however, it justifies the use of the FLRW metric during inflation and strongly
suggests that an accelerated expansion is a rather general and model-independent
feature of the early universe.
The problem of initial conditions is relaxed in eternally inflating models
(Sect. 5.6.5).
5.10.2 Graceful Exit Problem
As already stressed by Guth, the old inflationary scenario suffered from a graceful
exit problem: if the formation rate of the true-vacuum bubbles is greater than the
expansion rate of the universe, then the phase transition is very fast and inflation
does not begin. Conversely, if the vacuum decay is too slow the post-inflationary
universe becomes highly inhomogeneous and, in some of its parts, the phase
transition is not completed.
Although the bubble scenario has been replaced by chaotic inflation, the exit
problem has been persisting in other forms. The simplest models of chaotic inflation
do not contain in themselves the physics necessary to limit the acceleration period.
For instance, all models with constant < 1 never cease to inflate (te D 1) and
they need to be supplemented by an ad hoc engineered reheating mechanism.
A simple solution preserving the advantages of chaotic inflation is to introduce
another scalar field ' whose decay determines the end of inflation. Such a field
configuration can be realized in hybrid inflation, already described in Sect. 5.5.3.
5.10.3 Trans-Planckian Problem
In most of the above models, the number of e-foldings of accelerated expansion

is such that all observable cosmological scales were sub-Planckian at the onset of
220 5 Inflation
inflation. Since we expect that the quantum-field-theory description breaks down at

these scales, one might question the validity of the models themselves. Inflation,
however, does work, so it is plausible that this trans-Planckian problem [323] is due
to the qualitative nature of the set of initial conditions. Another possibility is that
the latter will be determined in a satisfactory way only in a robust theory of high-
energy physics. At Planck scale, exotic physics should modify the effective action of
the inflation. For instance, this can happen in the form of a non-standard dispersion
relation
.k0 /2 C jkj2 D 0 ! .k0 /2 C f .k0 ; k/jkj2 D 0 : (5.199)
The resulting inflationary predictions can drastically change and observations can
place constraints on phenomenology at the Planck scale [171, 182–185, 324–337].
5.10.4 Naturalness or Model-Building Problem
We saw that the parameters of common inflationary potentials require a certain level
of tuning. For instance, according to (5.29) the mass in a quadratic potential should
be six orders of magnitude smaller than the Planck mass. This is a consequence
of having a small CMB temperature fluctuation ıT=T 105 . Sometimes, this is
perceived as a fine-tuning issue.
The naturalness of a theory can be assessed by different criteria. The first,
formulated by Dirac, establishes that a physical theory is “natural” when all its
dimensionless parameters are of order 1. This condition, however, is too restrictive
because it would label as unnatural very successful models such as the electroweak
theory (fine structure constant ˛ 102 , electron Yukawa coupling 105 , and
so on).
Another criterion was introduced by ’t Hooft [338] and defines a theory as natural
if a new symmetry is acquired when sending to zero all the dimensionless parame-
ters in the action. Consequently, all the parameters associated with this approximate
symmetry are stable against radiative corrections, so that they do not receive
significant (i.e., O.1/) contributions from higher-order quantum effects [339].
The second criterion or its adaptations13 can remove the fine-tuning issue related
to the inflationary potential but V. / depends on the chosen high-energy model. The
problem is that observations are fit by a number of early-universe “natural,” more
or less convincing theories, from supersymmetric particle physics to string theory,
13
In Polchinski’s formulation [340], a field theory is natural if all masses are forbidden by
symmetries. The Standard Model is not natural due to the Higgs mass but, interestingly, this
naturalness argument suggests that a new symmetry should appear at electroweak energies.
5.11 The Inflaton and Particle Physics 221
from loop quantum cosmology to modified gravity, and so on. A task for the present
generation of physicists is to remove the theoretical degeneracy. Much of this book
is devoted to this problem.
5.11 The Inflaton and Particle Physics
The construction of a physical picture of the early universe would be incomplete if

one could not embed the inflaton in a quantum field theory of particles. This entails
three types of questions: (i) whether inflation can be triggered by fields different
from a real scalar, (ii) whether the inflaton, in whatever incarnation, can be identified
with a specific particle field, and (iii) how to apply quantum-field-theory techniques
in order to improve our understanding of the inflationary era. In this section, we
comment on the first two questions, leaving the third aside.14
5.11.1 Not Only Scalars
We have seen that inflation, a generic mechanism producing a sufficiently long

period of accelerated expansion, can be realized dynamically by a real scalar field.
One can ask oneself if other fields can perform a similar job. For instance, a vector
field on cosmological backgrounds [349, 350] is an alternative candidate to play the
role of the inflaton [351–359].15 The simplest example is that of an Abelian massive
vector with Lagrangian

1 1 R
L D F F m2 A A ; (5.200)
4 2 6
where F D 2rŒ A is the vector field strength. One can envisage more
complicated self-interactions, so that A is non-gauge; these interactions can arise,
for instance, from a spontaneous breaking of the U.1/ symmetry. Rescaling the field
as B D A =a on an FLRW background, the field equations r F m2 A D 0
become
BR ˛ C 3H BP ˛ C m2 B˛ D 0 ; B0 D 0 : (5.201)
14
Examples of applications of quantum field theory are the effective field theory approach
[341] and the study of inflationary infrared divergences and loop contributions to cosmological
observables (see the review [342] and later papers [343–348]).
15
Vector fields have also been considered as curvaton or auxiliary fields [360–366] and in the
context of dark energy [367–373].
222 5 Inflation
The non-minimal term in (5.200) is required in order to make the dynamics

compatible with the slow-roll approximation.
Compared with the scalar case, some extra difficulties arise. First, the Einstein
equations are not satisfied by a homogeneous field because the spatial part of the
stress-energy tensor contains off-diagonal components which are of the same order
of magnitude as the diagonal components. Intuitively, this happens because the
vector field selects a preferred direction. The stress-energy tensor can be made
diagonal statistically by a large number of randomly oriented vector fields or,
.i/
exactly, by a triplet of orthogonal fields A˛ D ı˛i , where i D 1; 2; 3 and
is a scalar mode. Second, perturbation theory is more complicate because of the
coupling between scalar, vector and tensor sectors already at the linear level. In the
limit of a large number of random fields, this problem is relaxed. Third, perturbation
analyses possibly indicate the presence of an unstable mode from the longitudinal
component of the field, and hence an instability.
Vector inflation can be generalized to p-forms [374–380]. Contrary to the vector
case (1-form), 3-forms are consistent with an FLRW background (their deploy in
cosmology is not new [381]); also, a non-minimal coupling is not necessary to
support an inflating background. Unstable modes, apparently, persist also for p > 2
but in some regimes they can be avoided. Finally, both inflation and dark energy
can be realized by a fermionic field [382–391], even via a condensation mechanism
[392–398]. The level of fine tuning entailed in these models and required to match
with observations is still under inspection.
5.11.2 Higgs Inflation
p-forms are often invoked as alternative inflationary fields on the ground that,
since we have observed only one fundamental scalar in Nature, we should keep
an open mind to other types of particles. Yet, scalars maintain a certain theoretical
fascination. In supersymmetric scenarios and models with spontaneous symmetry
breaking, a pseudo-Goldstone boson can emerge (the axion, which made a brief
appearance in Sect. 5.5.2) as the driver of natural inflation [112, 113, 143, 399].
In Chap. 13, we shall see how the inflaton can be embedded in string theory as a
modulus field. Here we mention a more minimalistic possibility.
In the non-supersymmetric Standard Model, there is only one fundamental field,
the Higgs. Its origin from spontaneous symmetry breaking in the electroweak sector
will be very briefly sketched in Sect. 7.1.1. By itself, the Higgs cannot sustain
inflation. In fact, from (5.195), and considering an inflaton field scale mPl ,
the mass of the inflaton is close to the GUT scale, m 1013 GeV. Therefore,
one cannot identify with the much lighter minimally coupled Higgs field h. The
recent discovery at LHC of the Higgs neutral boson found the value mh 125 GeV
[400–402]. The coupling of the quartic potential term is also quite different, being
it 1013 for the inflaton and 101 for the Higgs [403].
5.11 The Inflaton and Particle Physics 223
Interestingly, by allowing a non-minimal coupling between the Standard Model

and gravity the Higgs can act as the inflaton. This type of coupling (in contrast
with the minimal one, where matter interacts with gravity only via the determinant
p
of the metric g) arises naturally in the renormalization of the theory on curved
backgrounds [167] and can be further justified [404, 405]. Advances in precision
cosmology eventually revived the interest in early attempts to identify the Higgs
and the inflaton [118, 406] within a realistic particle-physics scenario [407–431].
The total Lagrangian is
R 1
LD 2
C h" hR D h" D h .h" h v 2 /2 C LSM ; (5.202)
2 2 2 4
where are are coupling constants, h is the Higgs doublet, D is the SU.2/˝U.1/
covariant derivative, v is the symmetry-breaking scale of the Standard Model, and
LSM is the Lagrangian of all the other Standard-Model particles. For instance,p for
1 104 , at the tree level viable inflation is achieved by tuning 104
104 mh =v, where in the last step we defined the Higgs effective mass
m2h D 2v 2 : (5.203)
Radiative corrections and a one-loop renormalization-group improvement already

modify the constraints on the parameters [408, 412, 416, 418], while in a
renormalization-group analysis at 2-loop level the scalar spectral index ns is shown
to be correlated to the Higgs mass [411, 415]. Higgs inflation has been embedded
also in supersymmetric extensions of the Standard Model [432] and in supergravity
(Sect. 5.12).
A Higgs mass of about mh 125 GeV leads, in general, to a red-tilted scalar
spectrum (ns 0:97), a small negative running (˛s D O.104 /) and a tiny
tensor-to-scalar ratio (r 103 ), all compatible within 68 % CL with PLANCK
observations [147, 433]. p
The inflaton Higgs field takes values h mPl = E above the cut-off E D
mPl = H of the effective theory (which is non-renormalizable, due to gravity); this
highlights a unitarity problem [413, 414, 420–425, 427, 428]. The latter can also be
seen as a fine-tuning problem, akin to that affecting ad hoc models of inflation.
The actual cut-off, however, depends on the background, and the argument about
unitarity is true only for small field values. The model can be rescued [425, 430]
by introducing new physics beyond the Standard Model, which spoils the original
simplicity of the Higgs-inflation idea [427].
Another possible resolution is to take a different prescription for the Lagrangian
allowing for non-minimal couplings of derivative type (“new Higgs inflation”) [434,
435]. These couplings replace the second and third terms in (5.202) in the form
" 1 1
h hR D h" D h ! g ˇ 2 G D h" D h ; (5.204)
2 2 2
224 5 Inflation
where ˇ is a real constant and G D R g R=2 is Einstein’s tensor (2.24).
Viable inflation is obtained if ˇ ' 108 1=4 m1 Pl
, so that ˇ 1 109 – 1010 TeV for
the Higgs values of . Again, the spectral scalar index is within the observational
bounds, ns 0:97. Unitarity bounds seem to be respected throughout the evolution
of the universe (but see [427]).
A second proposal to address the unitarity issue is to couple the Higgs with
an extra scalar field [436] but this is not a genuine completion of Higgs inflation,
since it is the extra field to act as the inflaton [437]. A fourth option is to resort
to Palatini formalism, where the affine p connection is independent of the tetrad
and the cut-off is higher by a factor of [428]. Yet another model, built with
the same purpose of achieving unitarity, does not feature derivative non-minimal
couplings between the Higgs and gravity but it prescribes a non-canonical kinetic
term .1 C h2 /@ h@ h [438–440], which can easily arise in supergravity. The
ensuing cosmological consequences can be discriminated from those of the other
implementations [431].
Without introducing new matter degrees of freedom and complicating an orig-
inally parsimonious model, one could turn the attention to another medicine for
unitarity, namely, a UV improvement of the gravitational sector (for instance,
asymptotic safety [427] or loop quantum gravity [428]). A fully satisfactory
realization of the inflationary paradigm within particle physics and cosmology might
lie in the realm of quantum gravity. By no means, there is no conclusive indication
that the latter will be a necessary ingredient of a full theory of the early universe.
Yet, as the Higgs example shows, research is in progress in many directions.
5.12 Supersymmetry and Supergravity
The inflaton can be embedded also in supergravity (SUGRA), the supersymmetric

formulation of covariant general relativity. Models of SUGRA inflation are impor-
tant for several reasons. First and foremost, the quest for supersymmetry at LHC will
give, in the very near future, crucial information about the energy scale of the theory,
even in the least exciting case where no supersymmetric partners of the Standard-
Model particles were detected. Since supergravity scenarios are among the most
promising fundamental theories, it is interesting to see whether and how inflation
can be realized therein. Second, supersymmetry is one of the key ingredients of
string theory and SUGRA is the low-energy limit of that framework (Sect. 12.2).
However, not all potentials of SUGRA inflation can originate from strings and
we should differentiate between pure SUGRA models and string models. Here we
concentrate on the former, postponing a discussion of string cosmology to Chap. 13.
Vacua in supergravity will be discussed in Sect. 7.1.3.
Since the machinery of supersymmetry will play a key role in the cosmological
constant problem and will be essential to formulate string theory, we will delve upon
it in some detail.
5.12 Supersymmetry and Supergravity 225
5.12.1 Global Supersymmetry
Let us first recall the main idea of N D 1 supersymmetry in D D 4 Minkowski

spacetime. The RPoincaré algebra of translation generators (hat symbol of operators
omitted) P D dx T 0 and Lorentz generators J ,
ŒP ; P D 0 ; (5.205a)
ŒP ; J D i. P P / ; (5.205b)
ŒJ ; J D i. J J C J J / ; (5.205c)
R " R "
is augmented by a set of operators Qa D dx Ja0 and Qb D dx Jb0 called

supercharges, stemming from a conserved supercurrent vector Ja where the indices
a; b run from 1 to 2. Supercharges are two-component Weyl spinors obeying the set
of commutation relations
ŒQa ; P D 0 ; ŒQa ; J D . /ca Qc (5.206)

" "
fQa ; Qb g D 2. /ab P ; fQa ; Qb g D 0 D fQ"a ; Qb g ; (5.207)
where fA; Bg D AB C BA is the anti-commutator and . /ca D Œ. /ab .N /bc
. /ab .N /bc =4 are the generators of the special linear algebra sl.2; C/. In this
expression, are the 2 2 Pauli matrices

01 0 i 1 0
0 D 12 ; 1 D ; 2 D ; 3 D ; (5.208)
10 i 0 0 1
and N D . 0 ; ˛ /. The same supercharge algebra can be rewritten in terms of a

four-component Majorana spinor, in which case one uses the 4 4 Dirac matrices

in the Weyl basis,

0 0 0 ˛ 0 ˛
D ; D ; (5.209)
0 0 ˛ 0
obeying the Clifford algebra f ; g D 2 14 .

Taking the trace of (5.207) over spinor indices, only the D 0 component
survives to yield the Hamiltonian H D P0 :
1X
HD fQa ; Q"a g : (5.210)
4 a
Global supersymmetry is the requirement that the supercharges annihilate the Fock
"
vacuum, Qa j0i D 0 D Qa j0i, so that the vacuum expectation value of the
226 5 Inflation
Hamiltonian is
vac D h0j H j0i D 0 ; (5.211)
to all orders in perturbation theory.

A second way to recast this result is to consider the simplest analogue of (7.1),
the Wess–Zumino model [441–444]. Its on-shell matter content is a scalar multiplet
made of a real scalar, a real pseudo-scalar and a Majorana spinor. It is convenient to
work in superspace, tagged by the 4-vector x and four anti-commuting complex
"
coordinates a , a [443, 444]. The latter are Grassmann variables such that all
their anti-commutators vanish: f ; g D f " ; " g D f ; " g D 0. Integration over
Grassmann coordinates can be defined with a measure d2 D .1=4/d a d b ab
(and its conjugate; ab
R is the two-index
R Levi-Civita
R symbol) and obeys some simple
rules. For instance, d2 D 0 D d2 , d2 D 1, and so on (notice the
engineering dimensions Œ D 1=2, Œd D 1=2). On superspace, one can define
superfields such as the scalar ˚.x; ; " /. The Wess–Zumino particle fields can
be grouped into a chiral superfield ˚.y; /, which does not depend on " except
through the combination y :D x C i " . Using the so-called Fierz identities
for Grassmann variables (tabulated in, e.g., [445]), one can show that
p
˚.y; / D .y/ C 2 .y/ C F.y/
1 p
D .x/ C i "
@ .x/ C " "
.x/ C 2 .x/
4
i
p @ .x/ "
C F.x/ ; (5.212)
2
where omitted spinor indices are contracted via ab , (often called A in the
literature) is a complex field encoding both the proper and pseudo real scalar
particles, a is a complex left-handed Weyl spinor and F is a scalar auxiliary field.
The anti-chiral superfield ˚ " is obtained by Hermitian conjugation. By our choice
of units, Œ˚ D 1 D Œ , Œ D 1=2 and ŒF D 2. The particle fields and have
their usual scaling in four dimensions. R
The Wess–Zumino action SWZ D d4 x LWZ is invariant under supergauge
transformations:
Z Z
2 2 " " 2
LWZ D d d ˚ ˚ d W.˚/ C H:c: ; (5.213a)
1 1
W.˚/ D m˚ 2 C ˚ 3 ; (5.213b)
2 3
where m is the superfield mass, is the interaction coupling and “H.c.” stands
for Hermitian conjugate. The functional W.˚/ is called superpotential and has
dimension ŒW D 3. Then, Œm D 1, Œ D 0 and ŒLWZ D 4 (the Lagrangian is

a four-dimensional mass density).
Let us consider only the bosonic part of the model and ignore the fermion
in (5.212). Expanding the Lagrangian (5.213) and eliminating the auxiliary field F
by its equations of motion, one eventually obtains
LWZ D @ @ V. ; / C LŒ ; ; (5.214a)

ˇ ˇ
ˇ @W ˇ2
V. ; / D ˇˇ ˇ D jm C 2 j2 : (5.214b)
@˚ ˇ˚ D
The Wess–Zumino model can be generalized to an arbitrary number of chiral

superfields ˚ i with generic holomorphic superpotential W.˚ i /, independent of the
"
complex conjugate of the superfields, and a generic kinetic term K.˚ i ; ˚ j / called
Kähler potential. Indices in the argument of the functionals are omitted from now
on:
Z Z Z
S D d4 x L ; L D d2 d2 " K.˚; ˚ " / d2 W.˚/ C H:c: : (5.215)
In terms of particle fields, this action gives rise to what is known as a non-linear
sigma model [446, 447]:
X ˇˇ @W ˇˇ2
i j" " " ˇ ˇ
L D Gij @ @ V. ; /; V. ; /D ˇ @˚ i ˇ ; (5.216)
i ˚D
where we ignored Yang–Mills gauge fields, the dimensionless Hermitian matrix

ˇ
@2 K.˚; ˚ " / ˇˇ
Gij :D ˇ (5.217)
@˚ i @˚ j " ˇ ˚D
is called Kähler metric and the associated complex Riemannian manifold is a Kähler
manifold. The i act as coordinates on the Kähler manifold. To summarize, the units
we have used in this sub-section are
Œ˚ D Œ D 1 ; ŒGij D 0 ; ŒK D 2 ; ŒW D 3 ; ŒV D 4 :

(5.218)
The reader is warned that we will change convention from the end of the next sub-
section on.
228 5 Inflation
5.12.2 Supergravity
When gravity is included, supersymmetry becomes a local gauge symmetry [448–

450]. In D D 4, N D 1 supergravity, instead of (5.215) one has [446, 451–454]
Z 2 Z
3 2 2 " " 2 Q
LD 2 d d E exp K.˚; ˚ / d E W.˚/ C H:c: ;
3
(5.219)
where E and EQ are, respectively, the determinant of superspace and the chiral deter-
minant [448, 455–458]. Other choices for the kinetic term are possible [459], but
for the sake of the main argument we limit our attention to (5.219). Again, we drop
contributions (called D-terms) generated by Yang–Mills gauge fields [453, 460–
p
462]. In spacetime, the bosonic part of the Lagrangian is L D g LSUGRA , where
R j"
LSUGRA D Gij @ i @ V. ; "
/; (5.220a)
2 2
2K
V D e G ij Di W.Dj W/" 3 2 jWj2 ˚D
: (5.220b)
The first term in (5.220a) is the Einstein–Hilbert Lagrangian, G ij is the inverse of

the Kähler metric and
@W @K
Di W :D C 2 i W : (5.221)
@˚ i @˚
The contribution
2
F 2 :D e K G ij Di W.Dj W/" (5.222)
is called F-term. Supersymmetry is locally preserved as long as
Di W D 0 ; (5.223)
corresponding to field configurations such that V is stationary.

In Minkowski spacetime and in the limit 2 ! 0 (vanishing G), the expres-
sions (5.219), (5.220a) and (5.220b) reduce to (5.215) and (5.216), which do not
depend on Newton’s constant. At this point, we change units to make all the rest of
the formulæ throughout the book simpler. We rescale ˚ ! 1 ˚, K ! 2 K and
W ! 3 W, so that
Œ˚ D ŒGij D ŒW D ŒK D 0 ; ŒV D 4 ; (5.224)

while equations (5.220b) and (5.221) are
@W @K
4 V D eK G ij Di W.Dj W/" 3jWj2 ˚D
; Di W D C W:
@˚ i @˚ i
(5.225)
Notice that the Wess–Zumino potential becomes
1 1 m
WD M˚ 2 C ˚ 3 ; M :D : (5.226)
2 3
5.12.3 -problem
To realize inflation in supergravity, it is not sufficient to reproduce one of the

potentials discussed in this chapter. The inflaton mass must be light enough to drive
acceleration without fine tuning the parameters of the Kähler and superpotentials.
However, this is not always possible due to the -problem [146, 463].
Consider first the single-superfield case. The action (5.215) with (5.225) is
invariant under the transformations
K.˚; ˚ " / ! K.˚; ˚ " / C f .˚/ C f " .˚ " / ; W.˚/ ! ef .˚ / W.˚/ ; (5.227)
R R
provided d2 f " .˚ " / D 0 D d2 " f .˚/. In particular, the metric (5.217) is
determined up to shifts of the Kähler potential by a holomorphic function f . We
can make a holomorphic field redefinition ˚old ! ˚ D h.˚old / so that G ij D ı ij at
˚ D 0 (canonical kinetic term) and then cancel the purely holomorphic part of K
via (5.227), so that
K D j˚j2 C : : : ; eK D 1 C j˚j2 C : : : ; (5.228)
up to higher-order terms. Unfortunately, the exp K / exp j j2 term dominates the

potential with its steepness and (5.59) is violated. Expanding (5.225) around the
origin, one has V. / D V.0/.1 C j j2 / C : : : and the second slow-roll parameter is
(as is dimensionless, there is no 2 prefactor here)
V;
V D D 1 C ::: : (5.229)
V
In order to cancel the O.1/ term and obtain sufficient slow-roll, one must fine tune
the contributions “. . . ” stemming from jD Wj2 3jWj2 [464–469].
230 5 Inflation
In the absence of further input this is unlikely to happen, especially in the multi-
superfield case where there are many independent parameters involved. SUGRA
models typically feature a number of scalar fields called moduli that we will meet
quite often in string theory (Sects. 12.1.4 and 12.3.5). Field redefinitions and the
Kähler transformation (5.227) make the metric diagonal at the origin (G ij .˚ i D
0/ D ı ij ) and the above argument is repeated without changes for each individual
field i .
The origin of the -problem can be intuitively traced back to the fact that the
inflaton is just one among many moduli. If one assumes (or proves) that all the
other moduli are stabilized (i.e., sit at their minimum) when inflation occurs, then
their masses will be much larger than the Hubble scale during this epoch, m2i
H 2 . However, any such stabilization mechanism will, in general, stabilize also the
modulus we dub as the inflaton. Therefore, as long as we have a dynamical inflaton,
either there are also dynamical moduli with m2i H 2 (which may lead to the
moduli problem of Sect. 13.2.4 and spoil inflation) or, if the moduli are stabilized
beforehand, the inflaton has a large mass m2' D O.H 2 /.
5.12.4 Inflation in Supergravity
There are ways out of the -problem, but with caveats. One is to enforce some
symmetry in the SUGRA action that relaxes the fine tuning of the parameters
[463]. Another, which we will explore in this sub-section, is to tailor the super- and
Kähler potentials to obtain inflationary potentials with a plausible parameter range.
A disadvantage of this approach is its ad hoc nature and a lack of justification, from
first principles, in the choice of K and W. This issue will be partly addressed in
string cosmology (Chap. 13).
A very important class of supergravity scenarios with applications to inflation and
to the cosmological constant problem (Sect. 7.1.3) are the no-scale models [470–
477]. Let ˚ i D T; Ss ; ˚ c be a set of scalar fields constituting the bosonic sector of
a chiral multiplet. The superpotential and Kähler potential we consider are
W D W1 .˚/ C W2 .S/; Q
K D 3 ln jT C T h2 .˚; ˚ /j C K.S; S /;
(5.230)
where h2 and KQ are real functions of, respectively, ˚ and S (and their complex
conjugate). Notice that the superpotential does not depend on T, so that the TT
component of the Kähler metric G exactly cancels the second term in (5.225). Then,
one can show that the potential (5.225) is

eKQ
VD G s1 s2 Ds1 W.Ds2 W/
jT C T h2 j3

2 @W1 @W1
C jT C T h2 j f c1 c2 ; (5.231)
3 @˚ c1 @˚ c2
where
@2 h2
. f c1 c2 /1 D fc1 c2 :D : (5.232)
@.˚ c1 / @˚ c2
The independence of the exponential factor exp K from ImT is the key to solve the
-problem when identifying the inflaton with this field. In Sect. 12.3.5, we will see
how some (but not all) no-scale models can emerge in string theory.
No-scale models have flat directions and they can provide an embedding for
inflation [478–481]. Having seen a class of Kähler potentials for a generic W,
we present a non-exhaustive list of some four-dimensional supergravity models
with specific inflationary potentials. Some are no-scale models, others are not. We
consider the bosonic part of three chiral multiplets, a “matter” scalar ˚ and two
“moduli” S and T (all dimensionless, while the inflaton Œ' D 1), the first called
dilaton and the second volume modulus. This nomenclature will find a rationale in
Chap. 12. After a rescaling, the inflaton can be identified either with the matter field
or with the volume modulus T. The second case will be analyzed also in Sect. 13.3.2
in the context of string theory. The reason why one usually demands the dilaton
to sit at its minimum during inflation, as well as a discussion of scenarios where
acceleration is driven precisely by S, can be found in Sect. 13.7.5.
• Quadratic potential V.'/ D 12 M 2 ' 2 , stemming from the two-superfield model
W D mS˚ [482] and [482, 483]
1
K D jSj2 .˚ ˚ /2 jSj4 ; (5.233)
2 3
p
where is a constant and ' D 2Re˚. Both Im˚ and S are sub-dominant
with respect to ' mPl during inflation and their vacuum is stable thanks to the
term (see below).
• Polinomial potential V.'/ D b0 C b2 ' 2 C b3 ' 3 C b4 ' 4 , stemming (with b0 D 0)
from the single-superfield Wess–Zumino model (5.226)–(5.214) [484] (but this
case has global supersymmetry, since the potential is (5.216) here) or from its
two-superfield generalization with K D c1 .˚ ˚ / C c2 .˚ ˚ /2 C jSj2 [485].
Then, ' D Re˚ and Im˚ D 0 D S.pThe same potential (but with b0 ¤ 0 and
b3 D 0) arises with (5.233) and W D b0 S.1 const ˚ 2 / [486].
232 5 Inflation
• Non-minimally coupled inflation, which includes Higgs inflation and related

models with L D .=2/' 2 R .r'/2 =2 .=4/' 4 [487–494] as well as
conformally-coupled two-field models [495].
• Starobinsky inflation [496–504] is a minimally-coupled single-field model with
potential
3 4 p2 2
V.'/ D M 1 e 3 ' : (5.234)
4
Computing the slow-roll parameters, one finds that
2 12
ns ' 1 0:967 ; r' 3 103 ; (5.235)
Nk Nk2
where we calculated the observables at Nk D 60 e-foldings. After the release of

the first PLANCK data, Starobinsky inflation has been among the most favoured
models of early acceleration, due to the low tensor-to-scalar ratio well below the
upper bounds reported in Sect. 4.4.2.
A realization of the Starobinsky potential has the Wess–Zumino superpoten-
tial (5.226) with [505, 506]

j˚j2
K D 3 ln T C T : (5.236)
3
After a certain field redefinition D .˚/ and taking the vacuum expectations
values hImTi D 0, hReTi D m2 =.62 / ¤ 0 and hImi D 0 (at the minimum
of the potential corrected by non-perturbative effects), the inflaton is identified
with ' D Re. Stability of the minimum along the non-inflaton T-direction
requires to break the no-scale structure of (5.236) by adding terms which we will
see below.
Another two-superfield model corresponding to Starobinsky inflation has
[507–509]

1 2 4
W D 3mS.T 1/ ; K D 3 ln T C T jSj C jSj : (5.237)
3 3
The modulusp S sits at the minimum hSi D 0 and the resulting potential is (5.234),
where ' D 2ReT. If D 0 [507] the S direction has a tachyonic instability.
The Kähler potentials in (5.236) and (5.237) are no-scale SUGRA scenarios.
Other no-scale models with different Kähler potentials, and where both the
matter and volume-modulus field can play the role of the inflaton, are presented
in [506, 510–513], where supersymmetry breaking in no-scale inflationary
scenarios is also discussed. On the other hand, non–no-scale SUGRA models
of Starobinsky inflation and an implementation in the SUGRA version of f .R/
theories (Sect. 7.5.2) can be found, respectively, in [509, 514, 515] and [516, 517].
Notice, however, that higher-order corrections (in the field strengths) to the
SUGRA Lagrangian can reintroduce the -problem [515].
A different origin of Starobinsky inflation, from a quadratic Ricci-curvature
term, will be given in Sect. 7.5.4, while a string-theory realization can be found
in Sect. 13.4.5.
• Goncharov–Linde inflation [518, 519] is based on the potential
" r !# r !
M4 2 3 2 3
V.'/ D 4 tanh ' tanh ' (5.238)
9 2 2
and it generates the cosmological observables
2 4
ns ' 1 0:967 ; r' 4 104 (5.239)
Nk 3Nk2
for Nk D 60 e-foldings. The tensor-to-scalar is even smaller than Starobin-

sky’s. This scenario was originally formulated from the single-superfield model
[518, 519]
p
m2 sinh2 . 3˚/ 1
WD p ; K D .˚ ˚ /2 ; (5.240)
3 cosh. 3˚/ 2
where ' D ReŒ.˚/ for some suitable field redefinition of ˚. Equation (5.238)
can be recovered also from [520]

m 4 .Z Z /2
W D p Z 2 .1 Z 2 / ; 2
K D 3 ln 1 jZj ; (5.241)
3 3 9 1 jZj2
where
T 1
Z :D : (5.242)
T C1
Kähler potentials of the form K D 3 ln.1 jZj2 jhj2 /, where h D h.Z; S; ˚/

is a generic term, are of no-scale type if h D .T C 1/1 A.S; ˚/ for an arbitrary
function A. In fact, from
p(5.242) and after a Kähler transformation (5.227) with
f .T/ D 3 lnŒ.T C 1/= 2, one has
.T C 1/h
K D 3 ln.1 jZj2 jhj2 / ! 3 ln T C T jh0 j2 ; h0 :D p I
2
(5.243)
we recover (5.230) if @T h0 D 0.
234 5 Inflation
It is also possible to reproduce any potential V.'/ or certain parametric classes

V.˛/ .'/. The above potentials are thus recovered in a more general setting:
• Generic potentials V.'/ [479, 483, 521, 522]. For an arbitrary holomorphic
function f .˚/, one takes the superpotential
W D S f .˚/ (5.244)
and either the Kähler potential (5.233) or

1 2 1 2 4
K D 3 ln 1 jSj C .˚ ˚ / C jSj : (5.245)
3 6 3
In both cases, freezing both S and Im˚ at their minimum

p where they vanish, the
potential for the canonically normalized inflaton ' D 2Re˚ takes the form
ˇ ˇ
ˇ ' ˇ2
ˇ
V.'/ D ˇf p ˇˇ :
4
(5.246)
2
This is a dynamical attractor, since the trajectory S D 0 D Im˚ is stable thanks

to the higher-order term jSj4. In fact, the mass of S in the vacuum trajectory
can be shown to be m2S D V;SS jSD0DIm˚ D 4H 2 C f;'2 for (5.233) and m2S D
2.6 1/H 2 C f;'2 for (5.245). In order for S D 0 to be a stable minimum,
one needs m2S > 0, which is achieved for any f when > 0; 1=6, respectively.
Moreover, to avoid that S excitations spoil inflation, the mass must be greater
than the Hubble energy, which is guaranteed if > 1=4.
Generic non-minimally coupled models are obtained for a different choice
of K.
• ˛-attractors [520, 523–532]. The members of this class of models are called
˛-attractors because they are all comprised between a finite vertical line in the
.ns ; r/ plane parametrized by ˛ > 0 and because their properties for ˛ . O.1/
are not very sensitive topthe shape of the potential, which can be of the form
V.˛/ .'/ / F 2 Œtanh.'= 6˛/ for a wide range of choices of F . Here we take
F .Z/ D Z.
The simplest ˛-attractors are the so-called T-models [483]. The total
Lagrangian is L D R=.2 2/ C L˛ , where the Einstein–Hilbert Lagrangian is
augmented by the contribution of a non-canonical real-valued scalar field with a
quadratic potential (here and m are dimensionless):
2 @ @ m2
4 L˛ D 2
; (5.247)
2 1 2 =.6˛/ 2
p p
where ˛ > 0. With the field redefinition D 6˛ tanh.'= 6˛/, we can
transform (5.247) into the canonical Lagrangian

1 ' 2
L˛ D @ '@ ' 3˛M 4 tanh p ; (5.248)
2 6˛
where M 2 D m= 2 . The inflationary observables are [524]
2 12˛
ns ' 1 ; r' : (5.249)
Nk Nk .Nk C 3˛=2/
Just like the scalar spectral index, the scalar spectral amplitude does not depend
on ˛ and, for Nk D 60, it agrees with observations when m D O.105 /.
The T-models (5.247)–(5.248) can be derived from specific super- and
Kähler potentials. Up to stabilization terms we will ignore, two possibilities are
[524, 526]
p 3˛1
WD 3˛ m S Z.1 Z 2 / 2 ; (5.250)

2 1 2
K D 3˛ ln 1 jZj jSj ; (5.251)
3˛
and [495]
p
WD 3˛ m S Z.1 Z 2 / ; (5.252)

2 ˛ 1 .Z Z /2 1 2
K D 3 ln 1 jZj C jSj ; (5.253)
2 1 jZj2 3
where Z is given by (5.242). The potential V.S; Z/, calculated from (5.225), has
a minimum at S D 0 D Z. After stabilizing the field S at S D 0, or taking a
nilpotent
p field S2 D 0, the potential (5.225) is the one in (5.247) with ReZ D
= 6˛. The geometric interpretation of ˛ / jRK j1 is that it is the inverse of
the negative curvature RK of the Kähler manifold. The case ˛ ! C1 of a flat
Kähler manifold is also meaningful [528].
For special values of ˛, one recovers the predictions of several of the earliest
models of inflation. For ˛ & 103 (and exactly, in the limit ˛ ! C1), this is
the usual model of quadratic inflation ((5.197) with n D 2 and for large Nk ) in
tension with PLANCK data, at the border with the 2-level region in the .ns ; r/
plane. The 1 region is entered for ˛ . 40. For ˛ D 1 and ˛ D 1=9, one hits
the point in the .ns ; r/ plane corresponding, respectively, to (5.235) (Starobinsky
inflation; compare (5.251) and (5.253) for ˛ D 1 with (5.237) and (5.243))
and (5.239) (Goncharov–Linde inflation). Starobinsky’s potential is recovered
exactly with W D SF .Z/.1 Z 2 / for F .Z/ / Z=.1 C Z/.
Since V.S D 0; Z D 0/ D 0 at the minimum, these models preserve
supersymmetry. Supersymmetry can be broken and the vanishing minimum can
236 5 Inflation
be lifted to some small value by introducing new parameters in the superpotential

W [495]. This can be done in such ways that either the inflaton potential is
modified but the cosmological predictions are not (for ˛ 1 and ˛ 1), or
the form of the inflaton potential remains the same up to an additive constant,
V.'/ ! V.'/CV0 [495]. In either case, ˛-attractors with broken supersymmetry
on a de Sitter vacuum can realize models of quintessential inflation (Sect. 7.3.6).
Another class of ˛-attractors (called E-models) consists in exponential models
such that [525]
p2 2
V.'/ D 3˛M 4 1 e 3˛ ' : (5.254)
˛ D 1 is precisely Starobinsky’s model (5.234). The potential (5.254) stems from

the super- and Kähler potential
p p
W D ˛m2 6T .1 C 3 S/.T ln T T C 1/; (5.255)

˛ 1 .T T /2
K D 3 ln T C T C C jSj2 : (5.256)
2 T C T
Again, one can deform the superpotential to break supersymmetry on a de Sitter

vacuum. More two-field ˛-attractors coming from different combinations of
Kähler and superpotentials can be found in [520].
Two classes of single-field models can be devised for
K D 3˛ ln.T C T / (5.257)
and for a suitable ˛-dependent superpotential W.T/ which gives rise either to
attractors for ˛ > 1 [528] or to attractors for ˛ < 1 [529]. After the imaginary
part ofpthe volume modulus is settled in the minimum ImT D 0 and for T D
expŒ 2=.3˛/' 1, the potential at large ' reads
p2
V.'/ ' V0 V1 e 3˛ ' : (5.258)
For ˛ D 1=9, the ˛-attractors of [529] exactly reproduce the Goncharov–Linde

superpotential in (5.241).
For simplicity, we have presented single-field SUGRA potentials, but one can
generalize to multi-field situations where not all non-inflatonic degrees of freedom
have been stabilized in advance. In a no-scale model partially motivated by
string theory, it was checked that non-Gaussianity can be low enough to appease
observations [511, 512].
Finally, we mention the case where the main contribution to the inflaton energy
density is not the F-term (5.222) but a D-term, V D F 2 3eK jWj2 ! F 2 3eK jWj2 C
D2 [533–537]. This contribution, which we have ignored so far, is generated by a
gauge symmetry and is of the form D2 / Re. fab 1
/Da Db , where fab is the gauge
j
kinetic function, Da D ˚ i .ta /i @j K C a , ta are the generators of the gauge group
and a are constants which are non-zero only for U.1/. If the D-term dominates
over the F-term responsible for the -problem, it is possible to achieve the slow-roll
condition V 1 without fine tuning. We do not delve into D-term inflation here as
we will dedicate most of Chap. 13 to related string scenarios.
5.1 Observable inflation. Determine how many e-foldings before the end
of inflation the modes k0 D 0:002 Mpc1 , k00 D 0:05 Mpc1 (pivot scales
in observations) and kdec D 0:02 Mpc1 (size of the horizon at decoupling,
equation (4.116)), crossed the horizon. Assume instantaneous reheating at the
1=4
GUT scale, e Treh TGUT .
Solution With instantaneous reheating at the GUT scale,
Nk D N0 ln.k 0 / ' 56 ln.k 0 / : (5.259)
The pivot scale k0 D 0:002 Mpc1 crossed the horizon ln.k0 0 / 3:6 4 e-
foldings before khor , so that NkDk0 60, while the mode k00 D 0:05 Mpc1 crossed
the horizon less than half e-folding before khor . For the scale corresponding to the
Hubble radius at decoupling, ln.kdec 0 / 1:3 1 and NkDkdec 57.
5.2 Inflationary scalar spectrum. Recover the power spectrum (5.152)

from (5.188) in the quasi-de Sitter approximation. Use synchronous gauge,
N D 1. The linear comoving curvature perturbation R at large scales is
obtained by integrating =.3Hz/ from the initial time ti to t and then sending
both extrema to infinity. The asymptotic-past limit ti ! 1 sets the initial
conditions, while the asymptotic-future limit t ! C1 corresponds to the
large-scale regime k 1. Use a Gaussian window function
W.k/ D exp.k2 R2 =2/ : (5.260)
Solution The calculation is rather simple in the quasi-de Sitter (ESR) regime, pwhere
a ' exp.Ht/ and both H and are constant; then, by virtue of (5.43a), z D a 2= 2
and can be taken outside the time integral. Also, at large scales the scalar mode uk
238 5 Inflation
is given by (5.126), with D 0 in quasi-de Sitter:
aH
uk ' p ; (5.261)
2k3
so that
Z t
1
R ' lim dt0
3H .ti ;t/!.1;C1/ ti z
r Z Z C1
.5.261/ 1 2 d3 k eikx 1 R P
' p ˇk dt ŒW C W.2@t C H/a C c.c. :
3 2 .2/3 2k3 1 a
From the window function (5.260), W P ' .kR/2 HW and W

R ' Œ.kR/2
2 2
2.kR/ H W; since aP D Ha, the integral is
Z C1 ˇC1
H2 dt Œ.kR/2 C 1.kR/2 W D HWŒ.kR/2 C 3ˇ1 D 3H ;
1
where we used R.t/ D ."H/1 exp.Ht/. Therefore,

r Z
2H2 d3 k eikx
R.x/ D p ˇk C c.c. : (5.262)
2 .2/3 2k3
By definition, the power spectrum is Ps D k3 Ps =.2 2 /, where (Problem 3.5)

Z
d3 k
hjR.x/j2 i D: Ps .k/ :
.2/3
Using the stochastic relation (5.185), one obtains the scalar spectrum (5.152).
5.3 Patch cosmology. Consider a modified effective Friedmann equation of

the form
H 2 D ˇ ; (5.263)
where 1 6 6 1 is a constant and ˇ > 0 is a coupling with energy

dimension Œˇ D 2 . Assume a standard continuity equation and a nearly
homogeneous scalar field as matter content, with power-law potential (5.81).
Calculate the number of e-foldings, the scalar spectrum, its index and the
index running for a test field on de Sitter (lowest-order slow-roll expressions).
(continued)
References 239
Express the scalar index as a function of the number of e-foldings, as done in

Sect. 5.5.1.2. Do the inflationary predictions change much with respect to the
standard case D 0?
Sketch of solution We follow [538–540], where the reader can find more details.
As functions of the slow-roll parameter, the scalar observables read
6ˇ H 2C
Ps D ; (5.264a)
.2 /25 2 2
ns 1 D .2 4/ C O. 2 / ; (5.264b)
˛s D 2 2. 2/ 2 C 5 2 : (5.264c)
For the scalar potential (5.81), we have in particular
n1q n.6q 4/ C 4
ns 1 D 2C.q1/n
; (5.265)
6.nˇ /q
n.q1/C2
where q :D 2=.2 /. Setting D 1 at the end of inflation, one gets e '
qn3q =.6ˇ nq1 / and the number of e-foldings (5.85e) is modified as
q
q
3ˇ nq1 2Cn.q1/ qn
Ne D ; (5.266)
nq Œn.q 1/ C 2 2Œn.q 1/ C 2
which is valid for n ¤ 2=.1 q/. Then, (5.265) reads
2n.3q 2/ C 4 2.n C 2/ C 2 .n 1/
ns 1 D D : (5.267)
2Ne .nq n C 2/ C nq Ne .4 2 C n / C n
For a quartic potential (n D 4), we have ns 1 D 3=ŒNe C 2=.2 C /. Since the
correction 2=3 6 2=.2 C / 6 2 is much smaller than the number of e-foldings
Ne 60, the inflationary predictions do not change much.
The effective Friedmann equation (5.263) arise in high-energy regimes of certain
braneworld inflationary scenarios which we shall revisit in Sect. 13.7.1.
References
1. H. Georgi, S.L. Glashow, Unity of all elementary particle forces. Phys. Rev. Lett. 32, 438
(1974)
2. H. Georgi, H.R. Quinn, S. Weinberg, Hierarchy of interactions in unified gauge theories. Phys.
Rev. Lett. 33, 451 (1974)
3. H. Georgi, The state of the art—Gauge theories. AIP Conf. Proc. 23, 575 (1975)
240 5 Inflation
4. H. Fritzsch, P. Minkowski, Unified interactions of leptons and hadrons. Ann. Phys. (N.Y.) 93,
193 (1975)
5. A.J. Buras, J.R. Ellis, M.K. Gaillard, D.V. Nanopoulos, Aspects of the grand unification of
strong, weak and electromagnetic interactions. Nucl. Phys. B 135, 66 (1978)
6. G. Lazarides, Q. Shafi, C. Wetterich, Proton lifetime and fermion masses in an SO.10/ model.
Nucl. Phys. B 181, 287 (1981)
7. J.R. Ellis, D.V. Nanopoulos, S. Rudaz, A phenomenological comparison of conventional and
supersymmetric GUTs. Nucl. Phys. B 202, 43 (1982)
8. Y. Hayato et al. [Super-Kamiokande Collaboration], Search for proton decay through
p ! N K C in a large water Cherenkov detector. Phys. Rev. Lett. 83, 1529 (1999).
[arXiv:hep-ex/9904020]
9. H. Murayama, A. Pierce, Not even decoupling can save minimal supersymmetric SU.5/.
Phys. Rev. D 65, 055009 (2002). [arXiv:hep-ph/0108104]
10. B. Bajc, P. Fileviez Perez, G. Senjanović, Proton decay in minimal supersymmetric SU.5/.
11. K.S. Babu, J.C. Pati, F. Wilczek, Fermion masses, neutrino oscillations, and proton decay in
the light of SuperKamiokande. Nucl. Phys. B 566, 33 (2000). [arXiv:hep-ph/9812538]
12. B. Dutta, Y. Mimura, R.N. Mohapatra, Neutrino mixing predictions of a minimal
SO.10/ model with suppressed proton decay. Phys. Rev. D 72, 075009 (2005).
[arXiv:hep-ph/0507319]
13. S. Dimopoulos, H. Georgi, Softly broken supersymmetry and SU(5). Nucl. Phys. B 193, 150
(1981)
14. S. Dimopoulos, S. Raby, F. Wilczek, Supersymmetry and the scale of unification. Phys. Rev.
D 24, 1681 (1981)
15. N. Sakai, Naturalnes in supersymmetric GUTS. Z. Phys. C 11, 153 (1981)
16. W.J. Marciano, G. Senjanović, Predictions of supersymmetric grand unified theories. Phys.
Rev. D 25, 3092 (1982)
17. M.B. Einhorn, D.R.T. Jones, The weak mixing angle and unification mass in supersymmetric
SU(5). Nucl. Phys. B 196, 475 (1982)
18. L.E. Ibáñez, G.G. Ross, Low-energy predictions in supersymmetric grand unified theories.
Phys. Lett. B 105, 439 (1981)
19. T.W.B. Kibble, Some implications of a cosmological phase transition. Phys. Rep. 67, 183
(1980)
20. M.B. Hindmarsh, T.W.B. Kibble, Cosmic strings. Rep. Prog. Phys. 58, 477 (1995).
21. A. Vilenkin, E.P.S. Shellard, Cosmic Strings and Other Topological Defects (Cambridge
22. D. Tong, TASI lectures on solitons. arXiv:hep-th/0509216
23. R.J. Danos, R.H. Brandenberger, G. Holder, Signature of cosmic strings wakes in the CMB
polarization. Phys. Rev. D 82, 023513 (2010). [arXiv:1003.0905]
24. N. Bevis, M. Hindmarsh, M. Kunz, J. Urrestilla, CMB power spectra from cosmic
strings: predictions for the Planck satellite and beyond. Phys. Rev. D 82, 065004 (2010).
[arXiv:1005.2663]
25. Ya.B. Zel’dovich, M.Y. Khlopov, On the concentration of relic magnetic monopoles in the
universe. Phys. Lett. B 79, 239 (1978)
26. J. Preskill, Cosmological production of superheavy magnetic monopoles. Phys. Rev. Lett. 43,
1365 (1979)
27. J. Preskill, Magnetic monopoles. Ann. Rev. Nucl. Part. Sci. 34, 461 (1984)
2005)
30. K. Sato, Cosmological baryon-number domain structure and the first order phase transition of
a vacuum. Phys. Lett. B 99, 66 (1981)
References 241
31. D. Kazanas, Dynamics of the universe and spontaneous symmetry breaking. Astrophys. J.
241, L59 (1980)
32. A.H. Guth, Inflationary universe: a possible solution to the horizon and flatness problems.
Phys. Rev. D 23, 347 (1981)
33. K. Sato, First-order phase transition of a vacuum and the expansion of the universe. Mon.
Not. R. Astron. Soc. 195, 467 (1981)
34. A.R. Liddle, P. Parsons, J.D. Barrow, Formalising the slow-roll approximation in inflation.
35. A.D. Linde, A new inflationary universe scenario: a possible solution of the horizon, flatness,
homogeneity, isotropy and primordial monopole problems. Phys. Lett. B 108, 389 (1982)
36. A. Albrecht, P.J. Steinhardt, Cosmology for grand unified theories with radiatively induced
symmetry breaking. Phys. Rev. Lett. 48, 1220 (1982)
37. J.M. Bardeen, P.J. Steinhardt, M.S. Turner, Spontaneous creation of almost scale-free density
perturbations in an inflationary universe. Phys. Rev. D 28, 679 (1983)
38. A.A. Starobinsky, Dynamics of phase transition in the new inflationary universe scenario and
generation of perturbations. Phys. Lett. B 117, 175 (1982)
39. S.W. Hawking, The development of irregularities in a single bubble inflationary universe.
Phys. Lett. B 115, 295 (1982)
40. A.D. Linde, Scalar field fluctuations in expanding universe and the new inflationary universe
scenario. Phys. Lett. B 116, 335 (1982)
41. A.H. Guth, S.-Y. Pi, Fluctuations in the new inflationary universe. Phys. Rev. Lett. 49, 1110
(1982)
42. A.D. Linde, Chaotic inflation. Phys. Lett. B 129, 177 (1983)
44. A.D. Linde, Can we have inflation with ˝ > 1? JCAP 0305, 002 (2003).
45. S. del Campo, R. Herrera, Extended closed inflationary universes. Class. Quantum Grav. 22,
2687 (2005). [arXiv:gr-qc/0505084]
46. M. Kamionkowski, D.N. Spergel, N. Sugiyama, Small scale cosmic microwave background
anisotropies as a probe of the geometry of the universe. Astrophys. J. 426, L57 (1994).
47. J.R. Gott, Creation of open universes from de Sitter space. Nature 295, 304 (1982)
48. D.H. Lyth, E.D. Stewart, Inflationary density perturbations with ˝ < 1. Phys. Lett. B 252,
336 (1990)
49. B. Ratra, P.J.E. Peebles, CDM cosmogony in an open universe. Astrophys. J. 432, L5 (1994)
50. A. Kashlinsky, I.I. Tkachev, J. Frieman, Microwave background anisotropy in low ˝0
inflationary models and the scale of homogeneity in the universe. Phys. Rev. Lett. 73, 1582
(1994). [arXiv:astro-ph/9405024]
51. N. Sugiyama, J. Silk, The imprint of ˝ on the cosmic microwave background. Phys. Rev.
Lett. 73, 509 (1994). [arXiv:astro-ph/9406026]
52. M. Kamionkowski, B. Ratra, D.N. Spergel, N. Sugiyama, CBR anisotropy in an open
inflation, CDM cosmogony. Astrophys. J. 434, L1 (1994). [arXiv:astro-ph/9406069]
53. B. Ratra, P.J.E. Peebles, Inflation in an open universe. Phys. Rev. D 52, 1837 (1995)
54. M. Bucher, A.S. Goldhaber, N. Turok, An open universe from inflation. Phys. Rev. D 52, 3314
(1995). [arXiv:hep-ph/9411206]
55. K. Yamamoto, M. Sasaki, T. Tanaka, Large angle CMB anisotropy in an open universe in the
one bubble inflationary scenario. Astrophys. J. 455, 412 (1995). [arXiv:astro-ph/9501109]
56. A.D. Linde, Inflation with variable ˝. Phys. Lett. B 351, 99 (1995). [arXiv:hep-th/9503097]
57. A.D. Linde, Toy model for open inflation. Phys. Rev. D 59, 023503 (1999).
58. A.D. Linde, M. Sasaki, T. Tanaka, CMB in open inflation. Phys. Rev. D 59, 123522 (1999).
242 5 Inflation
59. S. del Campo, R. Herrera, Extended open inflationary universes. Phys. Rev. D 67, 063507
(2003). [arXiv:gr-qc/0303024]
60. B. Freivogel, M. Kleban, M. Rodríguez Martínez, L. Susskind, Observational consequences
of a landscape. JHEP 0603, 039 (2006). [arXiv:hep-th/0505232]
61. D. Yamauchi, A. Linde, A. Naruko, M. Sasaki, T. Tanaka, Open inflation in the landscape.
Phys. Rev. D 84, 043513 (2011). [arXiv:1105.2674]
62. M. Joyce, Electroweak baryogenesis and the expansion rate of the universe. Phys. Rev. D 55,
1875 (1997). [arXiv:hep-ph/9606223]
63. J.H. Kung, R.H. Brandenberger, Chaotic inflation as an attractor in initial-condition space.
Phys. Rev. D 42, 1008 (1990)
64. A.D. Dolgov, A.D. Linde, Baryon asymmetry in the inflationary universe. Phys. Lett. B 116,
329 (1982)
65. L.F. Abbott, E. Fahri, M.B. Wise, Particle production in the new inflationary cosmology. Phys.
Lett. B 117, 29 (1982)
66. A.D. Dolgov, D.P. Kirilova, Production of particles by a variable scalar field. Sov. J. Nucl.
Phys. 51, 172 (1990)
67. J.H. Traschen, R.H. Brandenberger, Particle production during out-of-equilibrium phase
transitions. Phys. Rev. D 42, 2491 (1990)
68. L. Kofman, A.D. Linde, A.A. Starobinsky, Reheating after inflation. Phys. Rev. Lett. 73, 3195
(1994). [arXiv:hep-th/9405187]
69. Y. Shtanov, J.H. Traschen, R.H. Brandenberger, Universe reheating after inflation. Phys. Rev.
D 51, 5438 (1995). [arXiv:hep-ph/9407247]
70. D. Boyanovsky, H.J. de Vega, R. Holman, D.S. Lee, A. Singh, Dissipation via particle
production in scalar field theories. Phys. Rev. D 51, 4419 (1995). [arXiv:hep-ph/9408214]
71. D.I. Kaiser, Post inflation reheating in an expanding universe. Phys. Rev. D 53, 1776 (1996).
72. D. Boyanovsky, M. D’Attanasio, H.J. de Vega, R. Holman, D.S. Lee, Linear versus nonlinear
relaxation: consequences for reheating and thermalization. Phys. Rev. D 52, 6805 (1995).
73. S.Yu. Khlebnikov, I.I. Tkachev, Classical decay of inflaton. Phys. Rev. Lett. 77, 219 (1996).
74. S.Yu. Khlebnikov, I.I. Tkachev, The universe after inflation: the wide resonance case. Phys.
Lett. B 390, 80 (1997). [arXiv:hep-ph/9608458]
75. T. Prokopec, T.G. Roos, Lattice study of classical inflaton decay. Phys. Rev. D 55, 3768
(1997). [arXiv:hep-ph/9610400]
76. S.Yu. Khlebnikov, I.I. Tkachev, Resonant decay of Bose condensates. Phys. Rev. Lett. 79,
1607 (1997). [arXiv:hep-ph/9610477]
77. D. Boyanovsky, D. Cormier, H.J. de Vega, R. Holman, A. Singh, M. Srednicki, Scalar
field dynamics in Friedman–Robertson–Walker spacetimes. Phys. Rev. D 56, 1939 (1997).
78. L. Kofman, A.D. Linde, A.A. Starobinsky, Towards the theory of reheating after inflation.
79. B.A. Bassett, S. Liberati, Geometric reheating after inflation. Phys. Rev. D 58, 021302(R)
(1998); Erratum-ibid. 60, 049902(E) (1999). [arXiv:hep-ph/9709417]
80. B.A. Bassett, S. Tsujikawa, D. Wands, Inflation dynamics and reheating. Rev. Mod. Phys. 78,
537 (2006). [arXiv:astro-ph/0507632]
81. P.B. Greene, L. Kofman, A.D. Linde, A.A. Starobinsky, Structure of resonance in preheating
after inflation. Phys. Rev. D 56, 6175 (1997). [arXiv:hep-ph/9705347]
82. B.A. Bassett, D.I. Kaiser, R. Maartens, General relativistic preheating after inflation. Phys.
83. G.N. Felder, L. Kofman, A.D. Linde, Instant preheating. Phys. Rev. D 59, 123523 (1999).
84. M. Kawasaki, K. Kohri, T. Moroi, Big-bang nucleosynthesis and hadronic decay of long-lived
massive particles. Phys. Rev. D 71, 083502 (2005). [arXiv:astro-ph/0408426]
References 243
85. K. Kohri, T. Moroi, A. Yotsuyanagi, Big-bang nucleosynthesis with unstable gravitino

and upper bound on the reheating temperature. Phys. Rev. D 73, 123511 (2006).
86. A.R. Liddle, S.M. Leach, How long before the end of inflation were observable perturbations
produced? Phys. Rev. D 68, 103503 (2003). [arXiv:astro-ph/0305263]
Cambridge, 2009)
89. D.S. Salopek, J.R. Bond, Nonlinear evolution of long wavelength metric fluctuations in
inflationary models. Phys. Rev. D 42, 3936 (1990)
90. A.G. Muslimov, On the scalar field dynamics in a spatially flat Friedmann universe. Class.
Quantum Grav. 7, 231 (1990)
91. P.J. Steinhardt, M.S. Turner, Prescription for successful new inflation. Phys. Rev. D 29, 2162
(1984)
92. A.R. Liddle, D.H. Lyth, COBE, gravitational waves, inflation and extended inflation. Phys.
Lett. B 291, 391 (1992). [arXiv:astro-ph/9208007]
93. E.W. Kolb, S.L. Vadas, Relating spectral indices to tensor and scalar amplitudes in inflation.
94. D.J. Schwarz, C.A. Terrero-Escalante, A.A. García, Higher order corrections to primordial
spectra from cosmological inflation. Phys. Lett. B 517, 243 (2001). [arXiv:astro-ph/0106020]
95. W.H. Kinney, Inflation: flow, fixed points, and observables to arbitrary order in slow roll.
96. A.R. Liddle, Inflationary flow equations. Phys. Rev. D 68, 103504 (2003).
97. E. Ramírez, A.R. Liddle, Stochastic approaches to inflation model building. Phys. Rev. D 71,
123510 (2005). [arXiv:astro-ph/0502361]
98. E.J. Copeland, M.R. Garousi, M. Sami, S. Tsujikawa, What is needed of a tachyon if it is to
be the dark energy? Phys. Rev. D 71, 043003 (2005). [arXiv:hep-th/0411192]
99. S. Dodelson, W.H. Kinney, E.W. Kolb, Cosmic microwave background measurements can dis-
criminate among inflation models. Phys. Rev. D 56, 3207 (1997). [arXiv:astro-ph/9702166]
100. L. Alabidi, D.H. Lyth, Inflation models and observation. JCAP 0605, 016 (2006).
101. C. Wetterich, Kaluza–Klein cosmology and the inflationary universe. Nucl. Phys. B 252, 309
(1985)
102. Q. Shafi, C. Wetterich, Inflation with higher dimensional gravity. Phys. Lett. B 152, 51 (1985)
103. Q. Shafi, C. Wetterich, Inflation from higher dimensions. Nucl. Phys. B 289, 787 (1987)
104. H. Nishino, E. Sezgin, Matter and gauge couplings of N D 2 supergravity in six dimensions.
Phys. Lett. B 144, 187 (1984)
105. A. Salam, E. Sezgin, Chiral compactification on Minkowski S2 of N D 2 Einstein–Maxwell
supergravity in six dimensions. Phys. Lett. B 147, 47 (1984)
106. S. Randjbar-Daemi, A. Salam, E. Sezgin, J.A. Strathdee, An anomaly-free model in six
dimensions. Phys. Lett. B 151, 351 (1985)
107. I.G. Koh, H. Nishino, Towards realistic D D 6, N D 2 Kaluza–Klein supergravity on coset
E7 =SO.12/ Sp.1/ with chiral fermions. Phys. Lett. B 153, 45 (1985)
108. K.-i. Maeda, H. Nishino, Cosmological solutions in D D 6, N D 2 Kaluza–Klein
supergravity: Friedmann universe without fine tuning. Phys. Lett. B 154, 358 (1985)
109. K.-i. Maeda, H. Nishino, An attractor universe in six-dimensional N D 2 supergravity
Kaluza–Klein theory. Phys. Lett. B 158, 381 (1985)
110. J.J. Halliwell, Classical and quantum cosmology of the Salam–Sezgin model. Nucl. Phys. B
286, 729 (1987)
112. K. Freese, J.A. Frieman, A.V. Olinto, Natural inflation with pseudo Nambu–Goldstone
bosons. Phys. Rev. Lett. 65, 3233 (1990)
244 5 Inflation
113. F.C. Adams, J.R. Bond, K. Freese, J.A. Frieman, A.V. Olinto, Natural inflation: particle
physics models, power-law spectra for large-scale structure, and constraints from COBE.
114. J.E. Kim, H.P. Nilles, M. Peloso, Completing natural inflation. JCAP 0501, 005 (2005).
115. M. Sasaki, E.D. Stewart, A general analytic formula for the spectral index of the
density perturbations produced during inflation. Prog. Theor. Phys. 95, 71 (1996).
116. A.A. Starobinsky, Multicomponent de Sitter (inflationary) stages and the generation of
perturbations. Pis’ma Zh. Eksp. Teor. Fiz. 42, 124 (1985) [JETP Lett. 42, 152 (1985)]
117. J. Silk, M.S. Turner, Double inflation. Phys. Rev. D 35. 419 (1987)
118. D.S. Salopek, J.R. Bond, J.M. Bardeen, Designing density fluctuation spectra in inflation.
Phys. Rev. D 40, 1753 (1989)
119. A.R. Liddle, A. Mazumdar, F.E. Schunck, Assisted inflation. Phys. Rev. D 58, 061301 (1998).
120. K.A. Malik, D. Wands, Dynamics of assisted inflation. Phys. Rev. D 59, 123501 (1999).
121. A.A. Coley, R.J. van den Hoogen, Dynamics of multi-scalar-field cosmological models and
assisted inflation. Phys. Rev. D 62, 023517 (2000). [arXiv:gr-qc/9911075]
122. J.M. Aguirregabiria, A. Chamorro, L.P. Chimento, N.A. Zuccalá, Assisted inflation in
Friedmann–Robertson–Walker and Bianchi spacetimes. Phys. Rev. D 62, 084029 (2000).
123. P. Kanti, K.A. Olive, Realization of assisted inflation. Phys. Rev. D 60, 043502 (1999).
124. Z.K. Guo, Y.S. Piao, Y.Z. Zhang, Cosmological scaling solutions and multiple exponential
potentials. Phys. Lett. B 568, 1 (2003). [arXiv:hep-th/0304048]
125. S.A Kim, A.R. Liddle, S. Tsujikawa, Dynamics of assisted quintessence. Phys. Rev. D 72,
043506 (2005). [arXiv:astro-ph/0506076]
126. Z.K. Guo, Y.S. Piao, R.G. Cai, Y.Z. Zhang, Cosmological scaling solutions and cross-
coupling exponential potential. Phys. Lett. B 576, 12 (2003). [arXiv:hep-th/0306245]
127. A. Collinucci, M. Nielsen, T. Van Riet, Scalar cosmology with multi-exponential potentials.
Class. Quantum Grav. 22, 1269 (2005). [arXiv:hep-th/0407047]
128. J. Hartong, A. Ploegh, T. Van Riet, D.B. Westra, Dynamics of generalized assisted inflation.
Class. Quantum Grav. 23, 4593 (2006). [arXiv:gr-qc/0602077]
129. G. Calcagni, A.R. Liddle, Stability of multifield cosmological solutions. Phys. Rev. D 77,
023522 (2008). [arXiv:0711.3360]
130. S. Tsujikawa, General analytic formulae for attractor solutions of scalar-field dark
energy models and their multifield generalizations. Phys. Rev. D 73, 103504 (2006).
131. J.c. Hwang, H. Noh, Cosmological perturbations with multiple scalar fields. Phys. Lett. B
495, 277 (2000). [arXiv:astro-ph/0009268]
132. D. Langlois, F. Vernizzi, Nonlinear perturbations of cosmological scalar fields. JCAP 0702,
017 (2007). [arXiv:astro-ph/0610064]
133. G.I. Rigopoulos, E.P.S. Shellard, B.J.W. van Tent, Nonlinear perturbations in multiple-field
134. M. Sasaki, T. Tanaka, Super-horizon scale dynamics of multi-scalar inflation. Prog. Theor.
Phys. 99, 763 (1998). [arXiv:gr-qc/9801017]
135. T.J. Allen, B. Grinstein, M.B. Wise, Non-gaussian density perturbations in inflationary
cosmologies. Phys. Lett. B 197, 66 (1987)
136. L. Kofman, D.Yu. Pogosyan, Nonflat perturbations in inflationary cosmology. Phys. Lett. B
214, 508 (1988)
137. S. Mollerach, S. Matarrese, A. Ortolan, F. Lucchin, Stochastic inflation in a simple two-field
model. Phys. Rev. D 44, 1670 (1991)
References 245
138. A.D. Linde, V. Mukhanov, Non-Gaussian isocurvature perturbations from inflation. Phys.
139. D.H. Lyth, D. Wands, Generating the curvature perturbation without an inflaton. Phys. Lett.
B 524, 5 (2002). [arXiv:hep-ph/0110002]
140. D.H. Lyth, C. Ungarelli, D. Wands, The primordial density perturbation in the curvaton
scenario. Phys. Rev. D 67, 023503 (2003). [arXiv:astro-ph/0208055]
141. N. Bartolo, S. Matarrese, A. Riotto, Non-Gaussianity in the curvaton scenario. Phys. Rev. D
69, 043503 (2004). [arXiv:hep-ph/0309033]
142. A.D. Linde, Eternal extended inflation and graceful exit from old inflation without Jordan–
Brans–Dicke. Phys. Lett. B 249, 18 (1990)
143. A.D. Linde, Axions in inflationary cosmology. Phys. Lett. B 259, 38 (1991)
144. A.D. Linde, Hybrid inflation. Phys. Rev. D 49, 748 (1994). [arXiv:astro-ph/9307002]
145. S. Mollerach, S. Matarrese, F. Lucchin, Blue perturbation spectra from inflation. Phys. Rev.
D 50, 4835 (1994). [arXiv:astro-ph/9309054]
146. E.J. Copeland, A.R. Liddle, D.H. Lyth, E.D. Stewart, D. Wands, False vacuum inflation with
Einstein gravity. Phys. Rev. D 49, 6410 (1994). [arXiv:astro-ph/9401011]
147. S. Tsujikawa, J. Ohashi, S. Kuroyanagi, A. De Felice, Planck constraints on single-field
148. J.A. Wheeler, W.H. Zurek (eds.), Quantum Theory and Measurement (Princeton University
Press, Princeton, 1983)
149. H. Everett, “Relative state” formulation of quantum mechanics. Rev. Mod. Phys. 29, 454
(1957)
150. J.B. Hartle, Quantum mechanics of individual systems. Am. J. Phys. 36, 704 (1968)
151. B.S. DeWitt, R.N. Graham (eds.), The Many-Worlds Interpretation of Quantum Mechanics
(Princeton University Press, Princeton, 1973)
152. W.H. Zurek, Decoherence, einselection, and the quantum origins of the classical. Rev. Mod.
Phys. 75, 715 (2003)
153. N. Pinto-Neto, G. Santos, W. Struyve, Quantum-to-classical transition of primordial cosmo-
logical perturbations in de Broglie–Bohm quantum theory. Phys. Rev. D 85, 083506 (2012).
[arXiv:1110.1339]
154. E. Joos, H.D. Zeh, C. Kiefer, D. Giulini, J. Kupsch, I.-O. Stamatescu, Decoherence and the
Appearance of a Classical World in Quantum Theory (Springer, Berlin, 2003)
155. M. Schlosshauer, Experimental motivation and empirical consistency in minimal no-collapse
quantum mechanics. Ann. Phys. (N.Y.) 321, 112 (2006). [arXiv:quant-ph/0506199]
156. H.D. Zeh, Emergence of classical time from a universal wave function. Phys. Lett. A 116, 9
(1986)
157. C. Kiefer, Continuous measurement of mini-superspace variables by higher multipoles. Class.
Quantum Grav. 4, 1369 (1987)
158. M. Sakagami, Evolution from pure states into mixed states in de Sitter space. Prog. Theor.
Phys. 79, 442 (1988)
159. J.J. Halliwell, Decoherence in quantum cosmology. Phys. Rev. D 39, 2912 (1989)
160. T. Padmanabhan, Decoherence in the density matrix describing quantum three-geometries
and the emergence of classical spacetime. Phys. Rev. D 39, 2924 (1989)
161. J.P. Paz, S. Sinha, Decoherence and back reaction: the origin of the semiclassical Einstein
equations. Phys. Rev. D 44, 1038 (1991)
162. C. Kiefer, Decoherence in quantum electrodynamics and quantum gravity. Phys. Rev. D 46,
1658 (1992)
163. C. Kiefer, Topology, decoherence, and semiclassical gravity. Phys. Rev. D 47, 5414 (1993).
164. I.G. Moss, Quantum Theory, Black Holes and Inflation (Wiley, Chichester, 1996)
165. L.P. Grishchuk, Amplification of gravitational waves in an isotropic universe. Zh. Eksp. Teor.
Fiz. 67, 825 (1974) [Sov. Phys. JETP 40, 409 (1975)]
166. L.H. Ford, L. Parker, Quantized gravitational wave perturbations in Robertson–Walker
universes. Phys. Rev. D 16, 1601 (1977)
246 5 Inflation
Cambridge, 1982)
168. T.S. Bunch, P.C.W. Davies, Quantum field theory in de Sitter space: renormalization by point
splitting. Proc. R. Soc. Lond. A 360, 117 (1978)
169. R.H. Brandenberger, Quantum fluctuations as the source of classical gravitational perturba-
tions in inflationary universe models. Nucl. Phys. B 245, 328 (1984)
170. M.R. Brown, C.R. Dutton, Energy-momentum tensor and definition of particle states for
Robertson–Walker space-times. Phys. Rev. D 18, 4422 (1978)
171. J. Martin, R.H. Brandenberger, The trans-Planckian problem of inflationary cosmology. Phys.
Rev. D 63, 123501 (2001). [arXiv:hep-th/0005209]
172. N.A. Chernikov, E.A. Tagirov, Quantum theory of scalar fields in de Sitter space-time. Ann.
Poincaré Phys. Theor. A 9, 109 (1968)
173. E.A. Tagirov, Consequences of field quantization in de Sitter type cosmological models. Ann.
Phys. (N.Y.) 76, 561 (1973)
174. J. Géhéniau, C. Schomblond, Fonctions de Green dans l’univers de de Sitter. Acad. R. Belg.
Bull. Cl. Sci. 54, 1147 (1968)
175. C. Schomblond, P. Spindel, Uniqueness conditions for the 1 .x; y/ propagator of the scalar
field in the de Sitter universe. Ann. Poincaré Phys. Theor. A 25, 67 (1976)
176. E. Mottola, Particle creation in de Sitter space. Phys. Rev. D 31, 754 (1985)
177. B. Allen, Vacuum states in de Sitter space. Phys. Rev. D 32, 3136 (1985)
178. B. Allen, A. Folacci, Massless minimally coupled scalar field in de Sitter space. Phys. Rev. D
35, 3771 (1987)
179. R. Bousso, A. Maloney, A. Strominger, Conformal vacua and entropy in de Sitter space. Phys.
Rev. D 65, 104039 (2002). [arXiv:hep-th/0112218]
180. M. Spradlin, A. Volovich, Vacuum states and the S-matrix in dS/CFT. Phys. Rev. D 65,
104037 (2002). [arXiv:hep-th/0112223]
181. D. Polarski, A.A. Starobinsky, Semiclassicality and decoherence of cosmological perturba-
tions. Class. Quantum Grav. 13, 377 (1996). [arXiv:gr-qc/9504030]
182. U.H. Danielsson, Note on inflation and trans-Planckian physics. Phys. Rev. D 66, 023511
(2002). [arXiv:hep-th/0203198]
183. U.H. Danielsson, Inflation, holography and the choice of vacuum in de Sitter space. JHEP
0207, 040 (2002). [arXiv:hep-th/0205227]
184. N. Kaloper, M. Kleban, A. Lawrence, S. Shenker, L. Susskind, Initial conditions for inflation.
JHEP 0211, 037 (2002). [arXiv:hep-th/0209231]
185. U.H. Danielsson, On the consistency of de Sitter vacua. JHEP 0212, 025 (2002).
186. J. Lesgourgues, D. Polarski, A.A. Starobinsky, Quantum to classical transition of cos-
mological perturbations for nonvacuum initial states. Nucl. Phys. B 497, 479 (1997).
187. J. Martin, A. Riazuelo, M. Sakellariadou, Nonvacuum initial states for cosmologi-
cal perturbations of quantum-mechanical origin. Phys. Rev. D 61, 083518 (2000).
188. R.H. Brandenberger, C.T. Hill, Energy-density fluctuations in de Sitter space. Phys. Lett. B
179, 30 (1986)
189. E.D. Stewart, D.H. Lyth, A more accurate analytic calculation of the spectrum of
cosmological perturbations produced during inflation. Phys. Lett. B 302, 171 (1993).
190. E.J. Copeland, E.W. Kolb, A.R. Liddle, J.E. Lidsey, Reconstructing the inflaton poten-
tial: perturbative reconstruction to second order. Phys. Rev. D 49, 1840 (1994).
191. D.H. Lyth, E.D. Stewart, The curvature perturbation in power law (e.g. extended) inflation.
Phys. Lett. B 274, 168 (1992)
192. C. Kiefer, D. Polarski, Why do cosmological perturbations look classical to us? Adv. Sci.
Lett. 2, 164 (2009). [arXiv:0810.0087]
References 247
193. A. Riotto, Inflation and the theory of cosmological perturbations. arXiv:hep-ph/0210162

194. P.J. Steinhardt, Natural inflation, in The Very Early Universe, ed. by G.W. Gibbons, S.W.
Hawking, S.T.C. Siklos (Cambridge University Press, Cambridge, 1983)
195. A. Vilenkin, Birth of inflationary universes. Phys. Rev. D 27, 2848 (1983)
196. A.D. Linde, Eternally existing selfreproducing chaotic inflationary universe. Phys. Lett. B
175, 395 (1986)
197. A.D. Linde, Eternal chaotic inflation. Mod. Phys. Lett. A 01, 81 (1986)
198. A.A. Starobinsky, Stochastic de sitter (inflationary) stage in the early universe. Lect. Notes
Phys. 246, 107 (1986)
199. A.S. Goncharov, A.D. Linde, V.F. Mukhanov, The global structure of the inflationary universe.
Int. J. Mod. Phys. A 2, 561 (1987)
200. M. Aryal, A. Vilenkin, The fractal dimension of the inflationary universe. Phys. Lett. B 199,
351 (1987)
201. K. Nakao, Y. Nambu, M. Sasaki, Stochastic dynamics of new inflation. Prog. Theor. Phys. 80,
1041 (1988)
202. A.D. Linde, D.A. Linde, A. Mezhlumian, From the big bang theory to the theory of a
stationary universe. Phys. Rev. D 49, 1783 (1994). [arXiv:gr-qc/9306035]
203. A.D. Linde, D.A. Linde, A. Mezhlumian, Do we live in the center of the world? Phys. Lett. B
345, 203 (1995). [arXiv:hep-th/9411111]
205. A.D. Linde, Towards a gauge invariant volume-weighted probability measure for eternal
inflation. JCAP 0706, 017 (2007). [arXiv:0705.1160]
206. A. Borde, Geodesic focusing, energy conditions and singularities. Class. Quantum Grav. 4,
343 (1987)
207. A. Vilenkin, Did the universe have a beginning? Phys. Rev. D 46, 2355 (1992)
208. F. Helmer, S. Winitzki, Self-reproduction in k-inflation. Phys. Rev. D 74, 063528 (2006).
209. A. Vilenkin, Quantum fluctuations in the new inflationary Universe. Nucl. Phys. B 226, 527
(1983)
210. A. Vilenkin, Making predictions in eternally inflating universe. Phys. Rev. D 52, 3365 (1995).
211. A. Vilenkin, Unambiguous probabilities in an eternally inflating universe. Phys. Rev. Lett. 81,
5501 (1998). [arXiv:hep-th/9806185]
212. V. Vanchurin, A. Vilenkin, S. Winitzki, Predictability crisis in inflationary cosmology and its
resolution. Phys. Rev. D 61, 083507 (2000). [arXiv:gr-qc/9905097]
213. J. Garriga, A. Vilenkin, Prescription for probabilities in eternal inflation. Phys. Rev. D 64,
023507 (2001). [arXiv:gr-qc/0102090]
214. J. Garriga, D. Schwartz-Perlov, A. Vilenkin, S. Winitzki, Probabilities in the inflationary
multiverse. JCAP 0601, 017 (2006). [arXiv:hep-th/0509184]
215. R. Easther, E.A. Lim, M.R. Martin, Counting pockets with world lines in eternal inflation.
JCAP 0603, 016 (2006). [arXiv:astro-ph/0511233]
216. R. Bousso, Holographic probabilities in eternal inflation. Phys. Rev. Lett. 97, 191302 (2006).
217. R. Bousso, B. Freivogel, I-S. Yang, Eternal inflation: the inside story. Phys. Rev. D 74, 103516
(2006). [arXiv:hep-th/0606114]
(1984)
Phys. Rep. 215, 203 (1992)
220. J.M. Bardeen, Gauge-invariant cosmological perturbations. Phys. Rev. D 22, 1882 (1980)
221. S. Matarrese, S. Mollerach, M. Bruni, Relativistic second-order perturbations of the Einstein-
de Sitter universe. Phys. Rev. D 58, 043504 (1998). [arXiv:astro-ph/9707278]
222. V. Acquaviva, N. Bartolo, S. Matarrese, A. Riotto, Gauge-invariant second-order
perturbations and non-Gaussianity from inflation. Nucl. Phys. B 667, 119 (2003).
248 5 Inflation
223. K.A. Malik, D. Wands, Evolution of second-order cosmological perturbations. Class. Quan-
tum Grav. 21, L65 (2004). [arXiv:astro-ph/0307055]
224. L.F. Abbott, M.B. Wise, Constraints on generalized inflationary cosmologies. Nucl. Phys. B
244, 541 (1984)
225. V.F. Mukhanov, Gravitational instability of the universe filled with a scalar field. Pis’ma Zh.
Eksp. Teor. Fiz. 41, 402 (1985) [JETP Lett. 41, 493 (1985)]
226. M. Sasaki, Large scale quantum fluctuations in the inflationary universe. Prog. Theor. Phys.
76, 1036 (1986)
227. V.F. Mukhanov, Quantum theory of cosmological perturbations in R2 gravity. Phys. Lett. B
218, 17 (1989)
228. D.H. Lyth, What would we learn by detecting a gravitational wave signal in
the cosmic microwave background anisotropy? Phys. Rev. Lett. 78, 1861 (1997).
229. D. Baumann, L. McAllister, A microscopic limit on gravitational waves from D-brane
inflation. Phys. Rev. D 75, 123508 (2007). [arXiv:hep-th/0610285]
230. A. Ortolan, F. Lucchin, S. Matarrese, Non-Gaussian perturbations from inflationary dynamics.
Phys. Rev. D 38, 465 (1988)
231. H.M. Hodges, G.R. Blumenthal, L.A. Kofman, J.R. Primack, Nonstandard primordial
fluctuations from a polynomial inflation potential. Nucl. Phys. B 335, 197 (1990)
232. T. Falk, R. Rangarajan, M. Srednicki, The angular dependence of the three point correlation
function of the cosmic microwave background radiation as predicted by inflationary cosmolo-
gies. Astrophys. J. 403, L1 (1993). [arXiv:astro-ph/9208001]
233. I. Yi, E.T. Vishniac, Inflationary stochastic dynamics and the statistics of large-scale structure.
Astrophys. J. Suppl. Ser. 86, 333 (1993)
234. I. Yi, E.T. Vishniac, Simple estimate of the statistics of large scale structure. Phys. Rev. D 48,
950 (1993)
235. G.I. Rigopoulos, E.P.S. Shellard, B.J.W. van Tent, Simple route to non-Gaussianity in
236. D. Seery, J.E. Lidsey, Primordial non-Gaussianities in single field inflation. JCAP 0506, 003
(2005). [arXiv:astro-ph/0503692]
237. D.H. Lyth, Y. Rodríguez, The inflationary prediction for primordial non-Gaussianity. Phys.
Rev. Lett. 95, 121302 (2005). [arXiv:astro-ph/0504045]
238. X. Chen, M.x. Huang, S. Kachru, G. Shiu, Observational signatures and non-Gaussianities of
general single field inflation. JCAP 0701, 002 (2007). [arXiv:hep-th/0605045]
239. D. Seery, J.E. Lidsey, M.S. Sloth, The inflationary trispectrum. JCAP 0701, 027 (2007).
240. X. Chen, M.x. Huang, G. Shiu, The inflationary trispectrum for models with large non-
Gaussianities. Phys. Rev. D 74, 121301 (2006). [arXiv:hep-th/0610235]
241. F. Arroja, K. Koyama, Non-Gaussianity from the trispectrum in general single field inflation.
Phys. Rev. D 77, 083517 (2008). [arXiv:0802.1167]
242. L. Wang, M. Kamionkowski, The cosmic microwave background bispectrum and inflation.
243. A. Gangui, J. Martin, Cosmic microwave background bispectrum and slow roll inflation. Mon.
Not. R. Astron. Soc. 313, 323 (2000). [arXiv:astro-ph/9908009]
244. A. Gangui, F. Lucchin, S. Matarrese, S. Mollerach, The three-point correlation function of
the cosmic microwave background in inflationary models. Astrophys. J. 430, 447 (1994).
245. J.M. Maldacena, Non-Gaussian features of primordial fluctuations in single field inflationary
models. JHEP 0305, 013 (2003). [arXiv:astro-ph/0210603]
(2004). [arXiv:astro-ph/0405356]
0410, 006 (2004). [arXiv:astro-ph/0407059]
References 249
248. L. Senatore, K.M. Smith, M. Zaldarriaga, Non-Gaussianities in single field inflation and their
optimal limits from the WMAP 5-year data. JCAP 1001, 028 (2010). [arXiv:0905.3746]
249. P. Creminelli, On non-Gaussianities in single-field inflation. JCAP 0310, 003 (2003).
250. S. Gupta, A. Berera, A.F. Heavens, S. Matarrese, Non-Gaussian signatures in the
cosmic background radiation from warm inflation. Phys. Rev. D 66, 043510 (2002).
251. S. Gupta, Dynamics and non-Gaussianity in the weak-dissipative warm inflation scenario.
252. N. Arkani-Hamed, P. Creminelli, S. Mukohyama, M. Zaldarriaga, Ghost inflation. JCAP
0404, 001 (2004). [arXiv:hep-th/0312100]
253. K. Izumi, S. Mukohyama, Trispectrum from ghost inflation. JCAP 1006, 016 (2010).
[arXiv:1004.1776]
254. A. Gangui, J. Martin, M. Sakellariadou, Single field inflation and non-Gaussianity. Phys. Rev.
D 66, 083502 (2002). [arXiv:astro-ph/0205202]
255. S.-J. Rey, Dynamics of inflationary phase transition. Nucl. Phys. B 284, 706 (1987)
256. A. Hosoya, M. Morikawa, K. Nakayama, Stochastic dynamics of scalar field in the inflation-
ary universe. Int. J. Mod. Phys. A 4, 2613 (1989)
257. D.S. Salopek, J.R. Bond, Stochastic inflation and nonlinear gravity. Phys. Rev. D 43, 1005
(1991)
258. I. Yi, E.T. Vishniac, Stochastic analysis of the initial condition constraints on chaotic inflation.
Phys. Rev. D 47, 5280 (1993)
259. A.A. Starobinsky, J. Yokoyama, Equilibrium state of a self-interacting scalar field in the de
Sitter background. Phys. Rev. D 50, 6357 (1994). [arXiv:astro-ph/9407016]
260. J. Martin, M. Musso, Solving stochastic inflation for arbitrary potentials. Phys. Rev. D 73,
043516 (2006). [arXiv:hep-th/0511214]
261. J. Martin, M. Musso, Reliability of the Langevin pertubative solution in stochastic inflation.
Phys. Rev. D 73, 043517 (2006). [arXiv:hep-th/0511292]
262. F. Kühnel, D.J. Schwarz, Large-scale suppression from stochastic inflation. Phys. Rev. Lett.
105, 211302 (2010). [arXiv:1003.3014]
263. H. Haken, Synergetics (Springer, Berlin, 1978)
264. H. Risken, The Fokker–Planck Equation (Springer, Berlin, 1984)
265. S. Matarrese, M.A. Musso, A. Riotto, Influence of super-horizon scales on cosmological
observables generated during inflation. JCAP 0405, 008 (2004). [arXiv:hep-th/0311059]
266. G.I. Rigopoulos, E.P.S. Shellard, The separate universe approach and the evolution of
nonlinear superhorizon cosmological perturbations. Phys. Rev. D 68, 123518 (2003).
267. G.I. Rigopoulos, E.P.S. Shellard, Non-linear inflationary perturbations. JCAP 0510, 006
(2005). [arXiv:astro-ph/0405185]
268. D.H. Lyth, K.A. Malik, M. Sasaki, A general proof of the conservation of the curvature
perturbation. JCAP 0505, 004 (2005). [arXiv:astro-ph/0411220]
269. D. Wands, K.A. Malik, D.H. Lyth, A.R. Liddle, A new approach to the evolution of cosmolog-
ical perturbations on large scales. Phys. Rev. D 62, 043527 (2000). [arXiv:astro-ph/0003278]
270. D.H. Lyth, Y. Rodríguez, Non-Gaussianity from the second-order cosmological perturbation.
271. P.J.E. Peebles, An isocurvature cold dark matter cosmogony. I. A worked example of
evolution through inflation. Astrophys. J. 510, 523 (1999). [arXiv:astro-ph/9805194]
272. P.J.E. Peebles, An isocurvature cold dark matter cosmogony. II. Observational tests. Astro-
273. F. Bernardeau, J.-P. Uzan, Non-Gaussianity in multi-field inflation. Phys. Rev. D 66, 103506
(2002). [arXiv:hep-ph/0207295]
274. F. Bernardeau, J.-P. Uzan, Inflationary models inducing non-Gaussian metric fluctuations.
250 5 Inflation
275. M. Zaldarriaga, Non-Gaussianities in models with a varying inflaton decay rate. Phys. Rev. D
69, 043508 (2004). [arXiv:astro-ph/0306006]
276. K. Enqvist, A. Väihkönen, Non-Gaussian perturbations in hybrid inflation. JCAP 0409, 006
(2004). [arXiv:hep-ph/0405103]
277. D.H. Lyth, Non-Gaussianity and cosmic uncertainty in curvaton-type models. JCAP 0606,
015 (2006). [arXiv:astro-ph/0602285]
278. K.A. Malik, D.H. Lyth, A numerical study of non-Gaussianity in the curvaton scenario. JCAP
0609, 008 (2006). [arXiv:astro-ph/0604387]
279. M. Sasaki, J. Valiviita, D. Wands, Non-Gaussianity of the primordial perturbation in the
curvaton model. Phys. Rev. D 74, 103003 (2006). [arXiv:astro-ph/0607627]
280. K. Enqvist, S. Nurmi, Non-Gaussianity in curvaton models with nearly quadratic potential.
281. D. Seery, J.E. Lidsey, Primordial non-Gaussianities from multiple-field inflation. JCAP 0509,
011 (2005). [arXiv:astro-ph/0506056]
282. G.I. Rigopoulos, E.P.S. Shellard, B.J.W. van Tent, Large non-Gaussianity in multiple-field
283. D.H. Lyth, I. Zaballa, A bound concerning primordial non-Gaussianity. JCAP 0510, 005
(2005). [arXiv:astro-ph/0507608]
284. G.I. Rigopoulos, E.P.S. Shellard, B.J.W. van Tent, Quantitative bispectra from multifield
285. I. Zaballa, Y. Rodríguez, D.H. Lyth, Higher order contributions to the primordial non-
Gaussianity. JCAP 0606, 013 (2006). [arXiv:astro-ph/0603534]
286. F. Vernizzi, D. Wands, Non-Gaussianities in two-field inflation. JCAP 0605, 019 (2006).
287. L. Alabidi, Non-Gaussianity for a two component hybrid model of inflation. JCAP 0610, 015
(2006). [arXiv:astro-ph/0604611]
288. T. Battefeld, R. Easther, Non-Gaussianities in multi-field inflation. JCAP 0703, 020 (2007).
289. D. Seery, J.E. Lidsey, Non-Gaussianity from the inflationary trispectrum. JCAP 0701, 008
(2007). [arXiv:astro-ph/0611034]
290. C.T. Byrnes, M. Sasaki, D. Wands, The primordial trispectrum from inflation. Phys. Rev. D
74, 123519 (2006). [arXiv:astro-ph/0611075]
291. W.H. Kinney, Constraining inflation with cosmic microwave background polarization. Phys.
292. http://cosmologist.info/cosmomc.
293. A. Aguirre, M. Tegmark, Multiple universes, cosmic coincidences, and other dark matters.
JCAP 0501, 003 (2005). [arXiv:hep-th/0409072]
294. D.A. Easson, B.A. Powell, Identifying the inflaton with primordial gravitational waves. Phys.
Rev. Lett. 106, 191302 (2011). [arXiv:1009.3741]
295. M. Bastero-Gil, J. Macias-Pérez, D. Santos, Nonlinear metric perturbation enhancement of
primordial gravitational waves. Phys. Rev. Lett. 105, 081301 (2010). [arXiv:1005.4054]
296. U. Seljak, U.L. Pen, N. Turok, Polarization of the microwave background in defect models.
297. N. Barnaby, R. Namba, M. Peloso, Phenomenology of a pseudo-scalar inflaton: naturally
large non-Gaussianity. JCAP 1104, 009 (2011). [arXiv:1102.4333]
298. L. Sorbo, Parity violation in the cosmic microwave background from a pseudoscalar inflaton.
JCAP 1106, 003 (2011). [arXiv:1101.1525]
299. N. Barnaby, E. Pajer, M. Peloso, Gauge field production in axion inflation: consequences for
monodromy, non-Gaussianity in the CMB, and gravitational waves at interferometers. Phys.
Rev. D 85, 023525 (2012). [arXiv:1110.3327]
300. N. Bartolo, S. Matarrese, M. Peloso, M. Shiraishi, Parity-violating and anisotropic correla-
tions in pseudoscalar inflation. JCAP 1501, 027 (2015). [arXiv:1411.2521]
301. A. Lue, L. Wang, M. Kamionkowski, Cosmological signature of new parity-violating
interactions. Phys. Rev. Lett. 83, 1506 (1999). [arXiv:astro-ph/9812088]
References 251
302. R. Jackiw, S.-Y. Pi, Chern–Simons modification of general relativity. Phys. Rev. D 68, 104012
(2003). [arXiv:gr-qc/0308071]
303. M. Pospelov, A. Ritz, C. Skordis, Pseudoscalar perturbations and polarization of the cosmic
microwave background. Phys. Rev. Lett. 103, 051302 (2009). [arXiv:0808.0673]
304. B. Feng, H. Li, M. Li, X. Zhang, Gravitational leptogenesis and its signatures in CMB. Phys.
305. S. Mercuri, Fermions in Ashtekar–Barbero connections formalism for arbitrary values of the
Immirzi parameter. Phys. Rev. D 73, 084016 (2006). [arXiv:gr-qc/0601013]
306. C.R. Contaldi, J. Magueijo, L. Smolin, Anomalous CMB polarization and gravitational
chirality. Phys. Rev. Lett. 101, 141101 (2008). [arXiv:0806.3082]
307. S. Mercuri, Modifications in the spectrum of primordial gravitational waves induced by
instantonic fluctuations. Phys. Rev. D 84, 044035 (2011). [arXiv:1007.3732]
308. L. Bethke, J. Magueijo, Inflationary tensor fluctuations, as viewed by Ashtekar variables, their
imaginary friends. Phys. Rev. D 84, 024014 (2011). [arXiv:1104.1800]
309. L. Bethke, J. Magueijo, Chirality of tensor perturbations for complex values of the Immirzi
parameter. Class. Quantum Grav. 29, 052001 (2012). [arXiv:1108.0816]
310. A. Gruzinov, Consistency relation for single scalar inflation. Phys. Rev. D 71, 027301 (2005).
311. N. Bartolo, S. Matarrese, A. Riotto, Enhancement of non-Gaussianity after inflation. JHEP
0404, 006 (2004). [arXiv:astro-ph/0308088]
312. N. Bartolo, S. Matarrese, A. Riotto, Evolution of second-order cosmological perturbations
and non-Gaussianity. JCAP 0401, 003 (2004). [arXiv:astro-ph/0309692]
313. N. Bartolo, S. Matarrese, A. Riotto, Gauge-invariant temperature anisotropies and primordial
non-Gaussianity. Phys. Rev. Lett. 93, 231301 (2004). [arXiv:astro-ph/0407505]
314. K. Enqvist, A. Jokinen, A. Mazumdar, T. Multamäki, A. Väihkönen, Non-Gaussianity from
preheating. Phys. Rev. Lett. 94, 161301 (2005). [arXiv:astro-ph/0411394]
315. N. Barnaby, J.M. Cline, Non-Gaussian and nonscale-invariant perturbations from tachyonic
preheating in hybrid inflation. Phys. Rev. D 73, 106012 (2006). [arXiv:astro-ph/0601481]
317. S.W. Hawking, I.G. Moss, Supercooled phase transitions in the very early universe. Phys.
Lett. B 110, 35 (1982)
318. G.W. Gibbons, S.W. Hawking, Cosmological event horizons, thermodynamics, and particle
creation. Phys. Rev. D 15, 2738 (1977)
319. A.A. Starobinsky, Isotropization of arbitrary cosmological expansion given an effective
cosmological constant. Pis’ma Zh. Eksp. Teor. Fiz. 37, 55 (1983) [JETP Lett. 37, 66 (1983)]
320. R.M. Wald, Asymptotic behavior of homogeneous cosmological models in the presence of a
positive cosmological constant. Phys. Rev. D 28, 2118 (1983)
321. L.G. Jensen, J.A. Stein-Schabes, Is inflation natural? Phys. Rev. D 35, 1146 (1987)
322. M. Bruni, S. Matarrese, O. Pantano, A local view of the observable universe. Phys. Rev. Lett.
74, 1916 (1995). [arXiv:astro-ph/9407054]
323. R.H. Brandenberger, J. Martin, Trans-Planckian issues for inflationary cosmology. Class.
324. R.H. Brandenberger, J. Martin, The robustness of inflation to changes in super-Planck-scale
physics. Mod. Phys. Lett. A 16, 999 (2001). [arXiv:astro-ph/0005432]
325. A.A. Starobinsky, Robustness of the inflationary perturbation spectrum to trans-Planckian
physics. Pis’ma Zh. Eksp. Teor. Fiz. 73, 415 (2001) [JETP Lett. 73, 371 (2001)].
326. J. Martin, R.H. Brandenberger, The Corley–Jacobson dispersion relation and trans-Planckian
327. R.H. Brandenberger, J. Martin, On signatures of short distance physics in the cosmic
microwave background. Int. J. Mod. Phys. A 17, 3663 (2002). [arXiv:hep-th/0202142]
328. R. Easther, B.R. Greene, W.H. Kinney, G. Shiu, Generic estimate of trans-Planckian
modifications to the primordial power spectrum in inflation. Phys. Rev. D 66, 023518 (2002).
252 5 Inflation
329. G.L. Alberghi, R. Casadio, A. Tronconi, Trans-Planckian footprints in inflationary cosmol-

ogy. Phys. Lett. B 579, 1 (2004). [arXiv:gr-qc/0303035]
330. J. Martin, R.H. Brandenberger, Dependence of the spectra of fluctuations in inflationary cos-
mology on trans-Planckian physics. Phys. Rev. D 68, 063513 (2003). [arXiv:hep-th/0305161]
331. S. Cremonini, Effects of quantum deformations on the spectrum of cosmological perturba-
tions. Phys. Rev. D 68, 063514 (2003). [arXiv:hep-th/0305244]
332. S. Koh, S.P. Kim, D.J. Song, Gravitational wave spectrum in inflation with nonclassical states.
JHEP 0412, 060 (2004). [arXiv:gr-qc/0402065]
333. S. Shankaranarayanan, L. Sriramkumar, Trans-Planckian corrections to the primor-
dial spectrum in the infrared and the ultraviolet. Phys. Rev. D 70, 123520 (2004).
334. L. Sriramkumar, T. Padmanabhan, Initial state of matter fields and trans-Planckian
physics: can CMB observations disentangle the two? Phys. Rev. D 71, 103512 (2005).
335. R.H. Brandenberger, J. Martin, Back-reaction and the trans-Planckian problem of inflation
revisited. Phys. Rev. D 71, 023504 (2005). [arXiv:hep-th/0410223]
336. R. Easther, W.H. Kinney, H. Peiris, Observing trans-Planckian signatures in the cosmic
microwave background. JCAP 0505, 009 (2005). [arXiv:astro-ph/0412613]
337. R. Brandenberger, X.m. Zhang, The trans-Planckian problem for inflationary cosmology
revisited. arXiv:0903.2065
338. G. ’t Hooft, Naturalness, chiral symmetry, and spontaneous chiral symmetry breaking. NATO
Adv. Study Inst. Ser. B Phys. 59, 135 (1980)
339. R. Barbieri, Looking beyond the standard model: the supersymmetric option. Riv. Nuovo
Cim. 11N4, 1 (1988)
340. J. Polchinski, Effective field theory and the Fermi surface, in Recent Directions in Particle
Theory—From Superstrings and Black Holes to the Standard Model, ed. by J. Harvey, J.
Polchinski (World Scientific, Singapore, 1993). [arXiv:hep-th/9210046]
341. C. Cheung, P. Creminelli, A.L. Fitzpatrick, J. Kaplan, L. Senatore, The effective field theory
of inflation. JHEP 0803, 014 (2008). [arXiv:0709.0293]
342. D. Seery, Infrared effects in inflationary correlation functions. Class. Quantum Grav. 27,
124005 (2010) [arXiv:1005.1649]
343. C.T. Byrnes, M. Gerstenlauer, A. Hebecker, S. Nurmi, G. Tasinato, Inflationary infrared
divergences: geometry of the reheating surface versus ıN formalism. JCAP 1008, 006 (2010).
[arXiv:1005.3307]
344. C.P. Burgess, R. Holman, L. Leblond, S. Shandera, Breakdown of semiclassical methods in
de Sitter space. JCAP 1010, 017 (2010). [arXiv:1005.3551]
345. Y. Urakawa, T. Tanaka, Infrared divergence divergence does not affect the gauge-invariant
curvature perturbation. Phys. Rev. D 82, 121301 (2010). [arXiv:1007.0468]
346. Y. Urakawa, T. Tanaka, Natural selection of inflationary vacuum required by infra-red
regularity and gauge-invariance. Prog. Theor. Phys. 125, 1067 (2011). [arXiv:1009.2947]
347. M. Gerstenlauer, A. Hebecker, G. Tasinato, Inflationary correlation functions without infrared
divergences. JCAP 1106, 021 (2011). [arXiv:1102.0560]
348. W. Xue, X. Gao, R. Brandenberger, IR divergences in inflation and entropy perturbations.
JCAP 1206, 035 (2012). [arXiv:1201.0768]
349. Y. Hosotani, Exact solution to the Einstein–Yang–Mills equation. Phys. Lett. B 147, 44 (1984)
350. D.V. Galt’sov, M.S. Volkov, Yang–Mills cosmology. Cold matter for a hot universe. Phys.
Lett. B 256, 17 (1991)
351. L.H. Ford, Inflation driven by a vector field. Phys. Rev. D 40, 967 (1989)
352. A.B. Burd, J.E. Lidsey, Analysis of inflationary models driven by vector fields. Nucl. Phys. B
351, 679 (1991)
353. M.C. Bento, O. Bertolami, P. Vargas Moniz, J.M. Mourão, P.M. Sá, On the cosmology of
massive vector fields with SO.3/ global symmetry. Class. Quantum Grav. 10, 285 (1993).
References 253
354. A. Golovnev, V. Mukhanov, V. Vanchurin, Vector inflation. JCAP 0806, 009 (2008).
[arXiv:0802.2068]
355. K. Bamba, S. Nojiri, S.D. Odintsov, Inflationary cosmology and the late-time accelerated
expansion of the universe in nonminimal Yang–Mills-F.R/ gravity and nonminimal vector-
F.R/ gravity. Phys. Rev. D 77, 123532 (2008). [arXiv:0803.3384]
356. K. Dimopoulos, M. Karčiauskas, D.H. Lyth, Y. Rodríguez, Statistical anisotropy of the curva-
ture perturbation from vector field perturbations. JCAP 0905, 013 (2009). [arXiv:0809.1055]
357. B. Himmetoglu, C.R. Contaldi, M. Peloso, Instability of anisotropic cosmological solutions
supported by vector fields. Phys. Rev. Lett. 102, 111301 (2009). [arXiv:0809.2779]
358. A. Golovnev, V. Mukhanov, V. Vanchurin, Gravitational waves in vector inflation. JCAP 0811,
018 (2008). [arXiv:0810.4304]
359. E. Dimastrogiovanni, N. Bartolo, S. Matarrese, A. Riotto, Non-Gaussianity and statistical
anisotropy from vector field populated inflationary models. Adv. Astron. 2010, 752670
(2010). [arXiv:1001.4049]
360. K. Dimopoulos, Can a vector field be responsible for the curvature perturbation in the
universe? Phys. Rev. D 74, 083502 (2006). [arXiv:hep-ph/0607229]
361. K. Dimopoulos, Supergravity inspired vector curvaton. Phys. Rev. D 76, 063506 (2007).
[arXiv:0705.3334]
362. K. Dimopoulos, M. Karčiauskas, Non-minimally coupled vector curvaton. JHEP 0807, 119
(2008). [arXiv:0803.3041]
363. S. Yokoyama, J. Soda, Primordial statistical anisotropy generated at the end of inflation. JCAP
0808, 005 (2008). [arXiv:0805.4265]
364. S. Kanno, M. Kimura, J. Soda, S. Yokoyama, Anisotropic inflation from vector impurity.
JCAP 0808, 034 (2008). [arXiv:0806.2422]
365. M. Karčiauskas, K. Dimopoulos, D.H. Lyth, Anisotropic non-Gaussianity from vector field
perturbations. Phys. Rev. D 80, 023509 (2009). [arXiv:0812.0264]
366. M.-a. Watanabe, S. Kanno, J. Soda, Inflationary universe with anisotropic hair. Phys. Rev.
Lett. 102, 191302 (2009). [arXiv:0902.2833]
367. M. Novello, S.E. Perez Bergliaffa, J. Salim, Nonlinear electrodynamics and the acceleration
of the universe. Phys. Rev. D 69, 127301 (2004). [arXiv:astro-ph/0312093]
368. V.V. Kiselev, Vector field as a quintessence partner. Class. Quantum Grav. 21, 3323 (2004).
369. C. Armendáriz-Picón, Could dark energy be vector-like? JCAP 0407, 007 (2004).
370. H. Wei, R.-G. Cai, Interacting vector-like dark energy, the first and second cosmological
coincidence problems. Phys. Rev. D 73, 083002 (2006). [arXiv:astro-ph/0603052]
371. C.G. Boehmer, T. Harko, Dark energy as a massive vector field. Eur. Phys. J. C 50, 423 (2007).
372. J. Beltrán Jiménez, A.L. Maroto, Cosmic vector for dark energy. Phys. Rev. D 78, 063005
(2008). [arXiv:0801.1486]
373. T. Koivisto, D.F. Mota, Vector field models of inflation and dark energy. JCAP 0808, 021
(2008). [arXiv:0805.4229]
374. C. Germani, A. Kehagias, P-nflation: generating cosmic inflation with p-forms. JCAP 0903,
028 (2009). [arXiv:0902.3667]
375. T. Kobayashi, S. Yokoyama, Gravitational waves from p-form inflation. JCAP 0905, 004
(2009). [arXiv:0903.2769]
376. T.S. Koivisto, D.F. Mota, C. Pitrou, Inflation from N-forms and its stability. JHEP 0909, 092
(2009). [arXiv:0903.4158]
377. T.S. Koivisto, N.J. Nunes, Three-form cosmology. Phys. Lett. B 685, 105 (2010).
[arXiv:0907.3883]
378. C. Germani, A. Kehagias, Scalar perturbations in p-nflation: the 3-form case. JCAP 0911, 005
(2009). [arXiv:0908.0001]
379. T.S. Koivisto, N.J. Nunes, Inflation and dark energy from three-forms. Phys. Rev. D 80,
103509 (2009). [arXiv:0908.0920]
254 5 Inflation
380. T.S. Koivisto, F.R. Urban, Three-magnetic fields. Phys. Rev. D 85, 083508 (2012)
[arXiv:1112.1356]
381. T. Banks, Relaxation of the cosmological constant. Phys. Rev. Lett. 52, 1461 (1984)
382. C. Armendáriz-Picón, P.B. Greene, Spinors, inflation, and nonsingular cyclic cosmologies.
Gen. Relat. Grav. 35, 1637 (2003). [arXiv:hep-th/0301129]
383. B. Saha, T. Boyadjiev, Bianchi type I cosmology with scalar and spinor fields. Phys. Rev. D
69, 124010 (2004). [arXiv:gr-qc/0311045]
384. M.O. Ribas, F.P. Devecchi, G.M. Kremer, Fermions as sources of accelerated regimes in
cosmology. Phys. Rev. D 72, 123502 (2005). [arXiv:gr-qc/0511099]
385. B. Saha, Spinor field and accelerated regimes in cosmology. Grav. Cosmol. 12, 215 (2006).
386. B. Saha, Nonlinear spinor field in Bianchi type-I cosmology: inflation, isotropization, and late
time acceleration. Phys. Rev. D 74, 124030 (2006)
387. C.G. Böhmer, D.F. Mota, CMB anisotropies and inflation from non-standard spinors. Phys.
Lett. B 663, 168 (2008). [arXiv:0710.2003]
388. M.O. Ribas, F.P. Devecchi, G.M. Kremer, Cosmological model with non-minimally coupled
fermionic field. Europhys. Lett. 81, 19001 (2008). [arXiv:0710.5155]
389. C.G. Böhmer, Dark spinor inflation: theory primer and dynamics. Phys. Rev. D 77, 123535
(2008). [arXiv:0804.0616]
390. Y.-F. Cai, J. Wang, Dark energy model with spinor matter and its quintom scenario. Class.
391. D. Gredat, S. Shankaranarayanan, Modified scalar and tensor spectra in spinor driven
392. D.G. Caldi, A. Chodos, Cosmological neutrino condensates. arXiv:hep-ph/9903416
393. T. Inagaki, X. Meng, T. Murata, Dark energy problem in a four fermion interaction model.
394. F. Giacosa, R. Hofmann, M. Neubert, A model for the very early universe. JHEP 0802, 077
(2008). [arXiv:0801.0197]
395. S. Alexander, T. Biswas, The cosmological BCS mechanism and the big bang singularity.
Phys. Rev. D 80, 023501 (2009). [arXiv:0807.4468]
396. S. Alexander, T. Biswas, G. Calcagni, Cosmological Bardeen–Cooper–Schrieffer condensate
as dark energy. Phys. Rev. D 81, 043511 (2010); Erratum-ibid. D 81, 069902(E) (2010).
[arXiv:0906.5161]
397. N.J. Popławski, Cosmological constant from quarks and torsion. Ann. Phys. (Berlin) 523, 291
(2011). [arXiv:1005.0893]
398. J.M. Weller, Fermion condensate from torsion in the reheating era after inflation. Phys. Rev.
D 88, 083511 (2013). [arXiv:1307.2423]
399. D.H. Lyth, A. Riotto, Particle physics models of inflation and the cosmological density
perturbation. Phys. Rep. 314, 1 (1999). [arXiv:hep-ph/9807278]
400. G. Aad et al. [ATLAS Collaboration], Observation of a new particle in the search for the
Standard Model Higgs boson with the ATLAS detector at the LHC. Phys. Lett. B 716, 1
(2012). [arXiv:1207.7214]
401. S. Chatrchyan et al. [CMS Collaboration], Observation of a new boson at a mass of 125 GeV
with the CMS experiment at the LHC. Phys. Lett. B 716, 30 (2012). [arXiv:1207.7235]
402. D. Carmi, A. Falkowski, E. Kuflik, T. Volansky, J. Zupan, Higgs after the discovery: a status
report. JHEP 1210, 196 (2012). [arXiv:1207.1718]
403. K.A. Olive et al. [Particle Data Group], Review of particle physics. Chin. Phys. C 38, 090001
(2014)
404. J.J. van der Bij, Can gravity make the Higgs particle decouple? Acta Phys. Polon. B 25, 827
(1994)
405. J.J. van der Bij, Can gravity play a role at the electroweak scale? Int. J. Phys. 1, 63 (1995).
406. J.L. Cervantes-Cota, H. Dehnen, Induced gravity inflation in the standard model of particle
physics. Nucl. Phys. B 442, 391 (1995). [arXiv:astro-ph/9505069]
References 255
407. F.L. Bezrukov, M. Shaposhnikov, The Standard Model Higgs boson as the inflaton. Phys.
Lett. B 659, 703 (2008). [arXiv:0710.3755]
408. A.O. Barvinsky, A.Yu. Kamenshchik, A.A. Starobinsky, Inflation scenario via the Standard
Model Higgs boson and LHC. JCAP 0811, 021 (2008). [arXiv:0809.2104]
409. F. Bezrukov, D. Gorbunov, M. Shaposhnikov, On initial conditions for the hot big bang. JCAP
0906, 029 (2009). [arXiv:0812.3622]
410. J. García-Bellido, D.G. Figueroa, J. Rubio, Preheating in the standard model with the Higgs
inflaton coupled to gravity. Phys. Rev. D 79, 063531 (2009). [arXiv:0812.4624]
411. A. De Simone, M.P. Hertzberg, F. Wilczek, Running inflation in the Standard Model. Phys.
Lett. B 678, 1 (2009). [arXiv:0812.4946]
412. F.L. Bezrukov, A. Magnin, M. Shaposhnikov, Standard Model Higgs boson mass from
inflation. Phys. Lett. B 675, 88 (2009). [arXiv:0812.4950]
413. C.P. Burgess, H.M. Lee, M. Trott, Power-counting and the validity of the classical approxi-
mation during inflation. JHEP 0909, 103 (2009). [arXiv:0902.4465]
414. J.L.F. Barbón, J.R. Espinosa, Naturalness of Higgs inflation. Phys. Rev. D 79, 081302 (2009).
[arXiv:0903.0355]
415. F. Bezrukov, M. Shaposhnikov, Standard model Higgs boson mass from inflation: two loop
analysis. JHEP 0907, 089 (2009). [arXiv:0904.1537]
416. A.O. Barvinsky, A.Yu. Kamenshchik, C. Kiefer, A.A. Starobinsky, C.F. Steinwachs, Asymp-
totic freedom in inflationary cosmology with a non-minimally coupled Higgs field. JCAP
0912, 003 (2009). [arXiv:0904.1698]
417. T.E. Clark, B. Liu, S.T. Love, T. ter Veldhuis, Standard model Higgs boson-inflaton and dark
matter. Phys. Rev. D 80, 075019 (2009). [arXiv:0906.5595]
418. A.O. Barvinsky, A.Yu. Kamenshchik, C. Kiefer, A.A. Starobinsky, C.F. Steinwachs, Higgs
boson, renormalization group, and naturalness in cosmology. Eur. Phys. J. C 72, 2219 (2012).
[arXiv:0910.1041]
419. A.O. Barvinsky, A.Yu. Kamenshchik, C. Kiefer, C.F. Steinwachs, Tunneling cosmological
state revisited: origin of inflation with a nonminimally coupled standard model Higgs inflaton.
Phys. Rev. D 81, 043530 (2010). [arXiv:0911.1408]
420. R.N. Lerner, J. McDonald, Higgs inflation and naturalness. JCAP 1004, 015 (2010).
[arXiv:0912.5463]
421. M. Atkins, X. Calmet, On the unitarity of linearized general relativity coupled to matter. Phys.
Lett. B 695, 298 (2011). [arXiv:1002.0003]
422. C.P. Burgess, H.M. Lee, M. Trott, Comment on Higgs inflation and naturalness. JHEP 1007,
007 (2010). [arXiv:1002.2730]
423. M.P. Hertzberg, On inflation with non-minimal coupling. JHEP 1011, 023 (2010).
[arXiv:1002.2995]
424. D.I. Kaiser, Conformal transformations with multiple scalar fields. Phys. Rev. D 81, 084044
(2010). [arXiv:1003.1159]
425. F. Bezrukov, A. Magnin, M. Shaposhnikov, S. Sibiryakov, Higgs inflation: consistency and
generalisations. JHEP 1101, 016 (2011). [arXiv:1008.5157]
426. L.A. Popa, A. Caramete, Cosmological constraints on Higgs boson mass. Astrophys. J. 723,
803 (2010). [arXiv:1009.1293]
427. M. Atkins, X. Calmet, Remarks on Higgs inflation. Phys. Lett. B 697, 37 (2011).
[arXiv:1011.4179]
428. F. Bauer, D.A. Demir, Higgs–Palatini inflation and unitarity. Phys. Lett. B 698, 425 (2011).
[arXiv:1012.2900]
429. R.N. Lerner, J. McDonald, Distinguishing Higgs inflation and its variants. Phys. Rev. D 83,
123522 (2011). [arXiv:1104.2468]
430. F. Bezrukov, D. Gorbunov, M. Shaposhnikov, Late and early time phenomenology of Higgs-
dependent cutoff. JCAP 1110, 001 (2011). [arXiv:1106.5019]
431. L.A. Popa, Observational consequences of the standard model Higgs inflation variants. JCAP
1110, 025 (2011). [arXiv:1107.3436]
256 5 Inflation
432. K. Nakayama, F. Takahashi, Higgs mass and inflation. Phys. Lett. B 707, 142 (2012).
[arXiv:1108.3762]
433. P.A.R. Ade et al. [Planck Collaboration], Planck 2013 results. XXII. Constraints on inflation.
434. C. Germani, A. Kehagias, New model of inflation with nonminimal derivative cou-
pling of Standard Model Higgs boson to gravity. Phys. Rev. Lett. 105, 011302 (2010).
[arXiv:1003.2635]
435. C. Germani, A. Kehagias, Cosmological perturbations in the new Higgs inflation. JCAP 1005,
019 (2010); Erratum-ibid. 1006, E01 (2010). [arXiv:1003.4285]
436. G.F. Giudice, H.M. Lee, Unitarizing Higgs inflation. Phys. Lett. B 694, 294 (2011).
[arXiv:1010.1417]
437. R.N. Lerner, J. McDonald, Unitarity-violation in generalized Higgs inflation models. JCAP
1211, 019 (2012). [arXiv:1112.0954]
438. R.N. Lerner, J. McDonald, Unitarity-conserving Higgs inflation model. Phys. Rev. D 82,
103525 (2010). [arXiv:1005.2978]
439. K. Nakayama, F. Takahashi, Higgs chaotic inflation in standard model and NMSSM. JCAP
1102, 010 (2011). [arXiv:1008.4457]
440. K. Kamada, T. Kobayashi, M. Yamaguchi, J.’i. Yokoyama, Higgs G-inflation. Phys. Rev. D
83, 083515 (2011). [arXiv:1012.4238]
441. J. Wess, B. Zumino, Supergauge transformations in four dimensions. Nucl. Phys. B 70, 39
(1974)
442. J. Wess, B. Zumino, A Lagrangian model invariant under supergauge transformations. Phys.
Lett. B 49, 52 (1974)
443. A. Salam, J.A. Strathdee, Supergauge transformations. Nucl. Phys. B 76, 477 (1974)
444. S. Ferrara, J. Wess, B. Zumino, Supergauge multiplets and superfields. Phys. Lett. B 51, 239
(1974)
445. J.D. Lykken, Introduction to supersymmetry. arXiv:hep-th/9612114
446. E. Cremmer, J. Scherk, The supersymmetric non-linear -model in four dimensions and its
coupling to supergravity. Phys. Lett. B 74, 341 (1978)
447. B. Zumino, Supersymmetry and Kähler manifolds. Phys. Lett. B 87, 203 (1979)
448. R.L. Arnowitt, P. Nath, B. Zumino, Superfield densities and action principle in curved
superspace. Phys. Lett. B 56, 81 (1975)
449. D.Z. Freedman, P. van Nieuwenhuizen, S. Ferrara, Progress toward a theory of supergravity.
Phys. Rev. D 13, 3214 (1976)
450. S. Deser, B. Zumino, Consistent supergravity. Phys. Lett. B 62, 335 (1976)
451. E. Cremmer, B. Julia, J. Scherk, P. van Nieuwenhuizen, S. Ferrara, L. Girardello, Super-Higgs
effect in supergravity with general scalar interactions. Phys. Lett. B 79, 231 (1978)
452. E. Cremmer, B. Julia, J. Scherk, S. Ferrara, L. Girardello, P. van Nieuwenhuizen, Spontaneous
symmetry breaking and Higgs effect in supergravity without cosmological constant. Nucl.
Phys. B 147, 105 (1979)
453. R. Barbieri, S. Ferrara, D.V. Nanopoulos, K.S. Stelle, Supergravity, R invariance and
spontaneous supersymmetry breaking. Phys. Lett. B 113, 219 (1982)
454. E. Witten, J. Bagger, Quantization of Newton’s constant in certain supergravity theories. Phys.
Lett. B 115, 202 (1982)
455. J. Wess, J. Bagger, Supersymmetry and Supergravity (Princeton University Press, Princeton,
1992)
456. J. Wess, B. Zumino, Superspace formulation of supergravity. Phys. Lett. B 66, 361 (1977)
457. J. Wess, B. Zumino, Superfield Lagrangian for supergravity. Phys. Lett. B 74, 51 (1978)
458. M. Müller, The density multiplet in superspace. Z. Phys. C 16, 41 (1982)
459. N.-P. Chang, S. Ouvry, X. Wu, N D 1 supergravity with nonminimal coupling: a class of
models. Phys. Rev. Lett. 51, 327 (1983)
460. E. Cremmer, S. Ferrara, L. Girardello, A. Van Proeyen, Coupling supersymmetric Yang–Mills
theories to supergravity. Phys. Lett. B 116, 231 (1982)
References 257
461. E. Cremmer, S. Ferrara, L. Girardello, A. Van Proeyen, Yang–Mills theories with local
supersymmetry: Lagrangian, transformation laws and super-Higgs effect. Nucl. Phys. B 212,
413 (1983)
462. J.A. Bagger, Coupling the gauge-invariant supersymmetric non-linear sigma model to
supergravity. Nucl. Phys. B 211, 302 (1983)
463. E.D. Stewart, Inflation, supergravity, and superstrings. Phys. Rev. D 51, 6847, (1995).
464. D.V. Nanopoulos, K.A. Olive, M. Srednicki, K. Tamvakis, Primordial inflation in simple
supergravity. Phys. Lett. B 123, 41 (1983)
465. G.B. Gelmini, D.V. Nanopoulos, K.A. Olive, Finite temperature effects in primordial inflation.
Phys. Lett. B 131, 53 (1983)
466. A.D. Linde, Primordial inflation without primordial monopoles. Phys. Lett. B 132, 317 (1983)
467. B.A. Ovrut, P.J. Steinhardt, Supersymmetry and inflation: a new approach. Phys. Lett. B 133,
161 (1983)
468. R. Holman, P. Ramond, G.G. Ross, Supersymmetric inflationary cosmology. Phys. Lett. B
137, 343 (1984)
469. B.A. Ovrut, P.J. Steinhardt, Inflationary cosmology and the mass hierarchy in locally
supersymmetric theories. Phys. Rev. Lett. 53, 732 (1984)
470. E. Cremmer, S. Ferrara, C. Kounnas, D.V. Nanopoulos, Naturally vanishing cosmological
constant in N D 1 supergravity. Phys. Lett. B 133, 61 (1983)
471. J.R. Ellis, A.B. Lahanas, D.V. Nanopoulos, K. Tamvakis, No-scale supersymmetric standard
model. Phys. Lett. B 134, 429 (1984)
472. J.R. Ellis, C. Kounnas, D.V. Nanopoulos, Phenomenological SU.1; 1/ supergravity. Nucl.
Phys. B 241, 406 (1984)
473. J.R. Ellis, C. Kounnas, D.V. Nanopoulos, No-scale supersymmetric GUTs. Nucl. Phys. B 247,
373 (1984)
474. J.R. Ellis, C. Kounnas, D.V. Nanopoulos, No-scale supergravity models with a Planck mass
gravitino. Phys. Lett. B 143, 410 (1984)
475. N. Dragon, M.G. Schmidt, U. Ellwanger, Sliding scales in minimal supergravity. Phys. Lett.
B 145, 192 (1984)
476. R. Barbieri, E. Cremmer, S. Ferrara, Flat and positive potentials in N D 1 supergravity. Phys.
Lett. B 163, 143 (1985)
477. A.B. Lahanas, D.V. Nanopoulos, The road to no-scale supergravity. Phys. Rep. 145, 1 (1987)
478. G. Gelmini, C. Kounnas, D.V. Nanopoulos, Primordial inflation with flat supergravity
potentials. Nucl. Phys. B 250, 177 (1985)
479. A.S. Goncharov, A.D. Linde, A simple realisation of the inflationary Universe scenario in
SU.1; 1/ supergravity. Class. Quantum Grav. 1, L75 (1984)
480. J.R. Ellis, K. Enqvist, D.V. Nanopoulos, K.A. Olive, M. Srednicki, SU(N,1) inflation. Phys.
Lett. B 152, 175 (1985); Erratum-ibid. B 156, 452 (1985)
481. H. Murayama, H. Suzuki, T. Yanagida, J. Yokoyama, Chaotic inflation and baryogenesis in
supergravity. Phys. Rev. D 50, 2356 (1994). [arXiv:hep-ph/9311326]
482. M. Kawasaki, M. Yamaguchi, T. Yanagida, Natural chaotic inflation in supergravity. Phys.
Rev. Lett. 85, 3572 (2000). [arXiv:hep-ph/0004243]
483. R. Kallosh, A. Linde, New models of chaotic inflation in supergravity. JCAP 1011, 011
(2010). [arXiv:1008.3375]
484. D. Croon, J. Ellis, N.E. Mavromatos, Wess–Zumino inflation in light of Planck. Phys. Lett. B
724, 165 (2013). [arXiv:1303.6253]
485. K. Nakayama, F. Takahashi, T.T. Yanagida, Polynomial chaotic inflation in the Planck era.
Phys. Lett. B 725, 111 (2013). [arXiv:1303.7315]
486. M. Kawasaki, M. Yamaguchi, Supersymmetric topological inflation model. Phys. Rev. D 65,
103518 (2002). [arXiv:hep-ph/0112093]
487. M.B. Einhorn, D.R.T. Jones, Inflation with non-minimal gravitational couplings in supergrav-
ity. JHEP 1003, 026 (2010). [arXiv:0912.2718]
258 5 Inflation
488. S. Ferrara, R. Kallosh, A. Linde, A. Marrani, A. Van Proeyen, Jordan frame supergravity and
inflation in the NMSSM. Phys. Rev. D 82, 045003 (2010). [arXiv:1004.0712]
489. H.M. Lee, Chaotic inflation in Jordan frame supergravity. JCAP 1008, 003 (2010).
[arXiv:1005.2735]
490. S. Ferrara, R. Kallosh, A. Linde, A. Marrani, A. Van Proeyen, Superconformal symmetry,
NMSSM, and inflation. Phys. Rev. D 83, 025008 (2011). [arXiv:1008.2942]
491. I. Ben-Dayan, M.B. Einhorn, Supergravity Higgs inflation and shift symmetry in electroweak
theory. JCAP 1012, 002 (2010). [arXiv:1009.2276]
492. K. Nakayama, F. Takahashi, General analysis of inflation in the Jordan frame supergravity.
JCAP 1011, 039 (2010). [arXiv:1009.3399]
493. M. Arai, S. Kawai, N. Okada, Higgs inflation in minimal supersymmetric SU.5/ GUT. Phys.
Rev. D 84, 123515 (2011). [arXiv:1107.4767]
494. R. Kallosh, A. Linde, Superconformal generalization of the chaotic inflation model 4 4
2
2
R. JCAP 1306, 027 (2013). [arXiv:1306.3211]
495. R. Kallosh, A. Linde, Universality class in conformal inflation. JCAP 1307, 002 (2013).
[arXiv:1306.5220]
496. A.A. Starobinsky, A new type of isotropic cosmological models without singularity. Phys.
Lett. B 91, 99 (1980)
497. V.F. Mukhanov, G.V. Chibisov, Quantum fluctuation and nonsingular universe. Pis’ma Zh.
Eksp. Teor. Fiz. 33, 549 (1981) [JETP Lett. 33, 532 (1981)]
498. J.D. Barrow, A.C. Ottewill, The stability of general relativistic cosmological theory. J. Phys.
A 16, 2757 (1983)
499. A.A. Starobinskiı̆, The perturbation spectrum evolving from a nonsingular initially de Sitter
cosmology and the microwave background anisotropy. Pis’ma Astron. Zh. 9, 579 (1983) [Sov.
Astron. Lett. 9, 302 (1983)]
500. B. Whitt, Fourth-order gravity as general relativity plus matter. Phys. Lett. B 145, 176 (1984)
501. L.A. Kofman, A.D. Linde, A.A. Starobinsky, Inflationary universe generated by the combined
action of a scalar field and gravitational vacuum polarization. Phys. Lett. B 157, 361 (1985)
502. A.A. Starobinsky, H.-J. Schmidt, On a general vacuum solution of fourth-order gravity. Class.
503. K.-i. Maeda, Inflation as a transient attractor in R2 cosmology. Phys. Rev. D 37, 858 (1988)
504. K.-i. Maeda, J.A. Stein-Schabes, T. Futamase, Inflation in a renormalizable cosmological
model and the cosmic no-hair conjecture. Phys. Rev. D 39, 2848 (1989)
505. J. Ellis, D.V. Nanopoulos, K.A. Olive, No-scale supergravity realization of the Starobinsky
model of inflation. Phys. Rev. Lett. 111, 111301 (2013); Erratum-ibid. 111, 129902 (2013).
[arXiv:1305.1247]
506. J. Ellis, D.V. Nanopoulos, K.A. Olive, Starobinsky-like inflationary models as avatars of no-
scale supergravity. JCAP 1310, 009 (2013). [arXiv:1307.3537]
507. S. Cecotti, Higher derivative supergravity is equivalent to standard supergravity coupled to
matter. Phys. Lett. B 190, 86 (1987)
508. S. Cecotti, S. Ferrara, M. Porrati, S. Sabharwal, New minimal higher derivative supergravity
coupled to matter. Nucl. Phys. B 306, 160 (1988)
509. R. Kallosh, A. Linde, Superconformal generalizations of the Starobinsky model. JCAP 1306,
028 (2013). [arXiv:1306.3214]
510. J.A. Casas, Baryogenesis, inflation and superstrings, in International Europhysics Conference
on High Energy Physics, ed. by D. Lellouch, G. Mikenberg, E. Rabinovici (Springer, Berlin,
1999). [arXiv:hep-ph/9802210]
511. J. Ellis, M.A.G. García, D.V. Nanopoulos, K.A. Olive, A no-scale inflationary model to fit
them all. JCAP 1408, 044 (2014). [arXiv:1405.0271]
512. J. Ellis, M.A.G. García, D.V. Nanopoulos, K.A. Olive, Two-field analysis of no-scale
supergravity inflation. JCAP 1501, 010 (2015). [arXiv:1409.8197]
513. J. Ellis, M.A.G. García, D.V. Nanopoulos, K.A. Olive, Phenomenological aspects of no-scale
inflation models. JCAP 1510, 003 (2015). [arXiv:1503.08867]
References 259
514. W. Buchmuller, V. Domcke, K. Kamada, The Starobinsky model from superconformal D-term
inflation. Phys. Lett. B 726, 467 (2013). [arXiv:1306.3471]
515. F. Farakos, A. Kehagias, A. Riotto, On the Starobinsky model of inflation from supergravity.
Nucl. Phys. B 876, 187 (2013). [arXiv:1307.1137]
516. S.V. Ketov, A.A. Starobinsky, Embedding R C R2 inflation in supergravity. Phys. Rev. D 83,
063512 (2011). [arXiv:1011.0240]
517. S.V. Ketov, A.A. Starobinsky, Inflation and non-minimal scalar-curvature coupling in gravity
and supergravity. JCAP 1208, 022 (2012). [arXiv:1203.0805]
518. A.S. Goncharov, A.D. Linde, Chaotic inflation of the universe in supergravity. Zh. Eksp. Teor.
Fiz. 86, 1594 (1984) [JETP 59, 930 (1984)]
519. A.S. Goncharov, A.D. Linde, Chaotic inflation in supergravity. Phys. Lett. B 139, 27 (1984)
520. R. Kallosh, A. Linde, Planck, LHC, and ˛-attractors. Phys. Rev. D 91, 083528 (2015).
[arXiv:1502.07733]
521. R. Kallosh, A. Linde, T. Rube, General inflaton potentials in supergravity. Phys. Rev. D 83,
043507 (2011). [arXiv:1011.5945]
522. R. Kallosh, A. Linde, K.A. Olive, T. Rube, Chaotic inflation and supersymmetry breaking.
Phys. Rev. D 84, 083519 (2011). [arXiv:1106.6025]
523. S. Ferrara, R. Kallosh, A. Linde, M. Porrati, Minimal supergravity models of inflation. Phys.
Rev. D 88, 085038 (2013). [arXiv:1307.7696]
524. R. Kallosh, A. Linde, D. Roest, Superconformal inflationary ˛-attractors. JHEP 1311, 198
(2013). [arXiv:1311.0472]
525. S. Cecotti, R. Kallosh, Cosmological attractor models and higher curvature supergravity.
JHEP 1405, 114 (2014). [arXiv:1403.2932]
526. R. Kallosh, A. Linde, D. Roest, Large field inflation and double ˛-attractors. JHEP 1408, 052
(2014). [arXiv:1405.3646]
527. R. Kallosh, A. Linde, Escher in the sky. C. R. Phys. 16, 914 (2015). [arXiv:1503.06785]
528. D. Roest, M. Scalisi, Cosmological attractors from ˛-scale supergravity. Phys. Rev. D 92,
043525 (2015). [arXiv:1503.07909]
529. A. Linde, Single-field ˛-attractors. JCAP 1505, 003 (2015). [arXiv:1504.00663]
530. J.J.M. Carrasco, R. Kallosh, A. Linde, D. Roest, Hyperbolic geometry of cosmological
attractors. Phys. Rev. D 92, 041301 (2015). [arXiv:1504.05557]
531. J.J.M. Carrasco, R. Kallosh, A. Linde, Cosmological attractors and initial conditions for
532. J.J.M. Carrasco, R. Kallosh, A. Linde, ˛-attractors: Planck, LHC and dark energy. JHEP
1510, 147 (2015). [arXiv:1506.01708]
533. J.A. Casas, C. Muñoz, Inflation from superstrings. Phys. Lett. B 216, 37 (1989)
534. J.A. Casas, J.M. Moreno, C. Muñoz, M. Quirós, Cosmological implications of an anomalous
U(1): inflation, cosmic strings and constraints on superstring parameters. Nucl. Phys. B 328,
272 (1989)
535. P. Binetruy, G.R. Dvali, D-term inflation. Phys. Lett. B 388, 241 (1996).
536. E. Halyo, Hybrid inflation from supergravity D-terms. Phys. Lett. B 387, 43 (1996).
537. T. Matsuda, Successful D-term inflation with moduli. Phys. Lett. B 423, 35 (1998).
538. G. Calcagni, Slow-roll parameters in braneworld cosmologies. Phys. Rev. D 69, 103508
(2004). [arXiv:hep-ph/0402126]
539. G. Calcagni, S. Tsujikawa, Observational constraints on patch inflation in noncommutative
spacetime. Phys. Rev. D 70, 103514 (2004). [arXiv:astro-ph/0407543]
540. G. Calcagni, Braneworld Cosmology and Noncommutative Inflation. Ph.D. thesis, Parma
University, Parma (2005). [arXiv:hep-ph/0503044]
Chapter 6
Big-Bang Problem
Vaccha, the speculative view that the world is eternal. . . that the
world is not eternal. . . that the world is finite. . . that the world is
infinite is a thicket of views, a wilderness of views, a contortion
of views, a vacillation of views, a fetter of views. It is beset by
suffering, by vexation, by despair, and by fever, and it does not
lead to disenchantment, to dispassion, to cessation, to peace, to
direct knowledge, to enlightenment, to Nibbāna.
— Aggivacchagotta Sutta, Majjhima Nikāya 72.14 [1]
Contents
6.1 Spacetimes and Singularities. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262
6.1.1 Globally Hyperbolic Spacetimes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262
6.1.2 Focusing Theorems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265
6.1.3 Classifications of Singularities. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267
6.2 Singularity Theorems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
6.2.1 Hawking–Penrose Theorems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
6.2.2 Borde–Vilenkin Theorems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
6.2.3 Borde–Guth–Vilenkin Theorem. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273
6.2.4 An Undecided Issue. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275
6.3 BKL Singularity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278
6.3.1 Tetrads and Bianchi Models. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278
6.3.2 Kasner Metric. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 280
6.3.3 Generalized Kasner Metric. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282
6.3.4 Mixmaster Dynamics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 284
6.3.5 BKL Conjecture. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295
The big-bang problem is an open issue in theoretical physics. In this and the
next chapters, we will review some of the proposals, few of which successful or
completely satisfactory, advanced to solve it. Let us first explain why the big bang
is a problem at the classical level. We ask ourselves the following question:
Is the big-bang singularity typical in the inflationary universe? (6.1)

DOI 10.1007/978-3-319-41127-9_6
262 6 Big-Bang Problem
We shall give a positive answer thanks to a set of theorems valid in classical

spacetimes (Sect. 6.2). These theorems determine the existence of singularities in
certain spacetimes, but they do not yield any information about their structure.
Classical singularities can be classified (Sect. 6.1.3) and their details studied by
standard techniques. In Sect. 6.3, the example of the chaotic BKL singularity in
anisotropic universes is presented in some detail; it will be further developed in
Sect. 13.9.2.
6.1 Spacetimes and Singularities
In Chap. 2, we have discussed several cosmological solutions of very general char-

acter. In particular, power-law solutions a.t/ / tp are instrumental in understanding
the inflationary expansion and the history of the universe after reheating. They are
singular at the big bang,
lim a.t/ D 0 ; (6.2)

t!tbb
where tbb 0 is the proper-time big bang event. At this point, the laws of
physics break down and the history of the universe we worked out at such a
painstaking length turns out to be based on ill-defined initial conditions. Inspecting
the Friedmann and continuity equations, one sees that the energy density blows up in
the past for ordinary matter fields. One might hope that this pathological behaviour
occurs only in special situations. After all, de Sitter and power-law cosmologies
are idealizations of the real world and one might have inflation work out on a more
complicated but non-singular background a.t/, perhaps to be analyzed via numerical
methods. However, particular non-singular solutions would rely on particular initial
conditions, while general results are desirable to assess “how often” these regular
scenarios are realized. Unfortunately, the big-bang singularity is intrinsic not only
to the Einstein equations, but also to other gravitational dynamics on a wide class
of Lorentzian manifolds. This is the actual reason why the big bang is a serious
problem and not just a glitch of special classes of solutions: it is not special at all,
and is even expected.
6.1.1 Globally Hyperbolic Spacetimes
Before stating the singularity theorems, we need a few definitions (see, e.g., [2] and
the pedagogical introduction [3]). Since we wish to go beyond pure or perturbed
FLRW spacetimes, covariant formalism shall be used.
Many of the following concepts have already been employed, such as the notion
of spacetime. Given a Lorentzian manifold .M; g/, a spacetime is the maximal
6.1 Spacetimes and Singularities 263
manifold M on which the metric g is smooth. We also assumed that all spacetimes
we have been dealing with where time-orientable, which is the case if there exists
a non-vanishing time-like vector field t . This is equivalent to state that all tangent
spaces Tx M can be time-oriented consistently. Time-like vectors v 2 Tx M are
future-directed if they have the same orientation defined by t . In other words,
t v < 0. They are past-directed when the have opposite orientation.
The Lorentz classification of vectors, curves and surfaces also played a role in,
for example, the discussion of the energy conditions (Sect. 2.2.1). Let 2 I R
be a real parameter and W I ! M a curve. A curve is space-like, time-like
or null if its tangent vector d ./=d is, respectively, space-like (positive norm),
time-like (negative norm) or null (zero norm) for all 2 I. Also, a future-directed
(respectively, past-directed) causal curve is a curve such that the tangent vector
d ./=d is either null or time-like future-oriented (respectively, past-oriented)
for all . For a time-like geodesic (the observer), the parameter D t is proper time,
while for a null geodesic it is called affine parameter. The concepts of “past” and
“future” of an event at point x 2 M can be refined progressively. The chronological
future of x 2 M is the set I C .x/ of all points to which x can be connected by a
future-directed time-like curve. If the curve is causal, the set of points is called the
causal future of x and is denoted by J C .x/. Replacing future-directed with past-
directed curves, one obtains the chronological past I .x/ and causal past J .x/
of x. The difference between the causal and the chronological future of x 2 M is
the future light cone E C .x/ D J C .x/ n I C .x/. The past light cone is E .x/ D
J .x/ n I .x/. In particular, for a time-oriented manifold the causal past or future
is the closure of the chronological past or future, J ˙ .x/ D I ˙ .x/. For every pair
x; y 2 M, the set J .x/ \ J C .y/ is the set of all points that can be reached both
from x along a past-oriented time-like curve and from y along a future-oriented
time-like curve.
Globally, a Lorentzian manifold is future (respectively, past) causally simple if
E ˙ .x/ D @I ˙ .x/ ¤ ; for all x 2 M, where @I ˙ .x/ is the boundary of I ˙ .x/. Light
cones and particle horizons in causally simple spacetimes are shown in Fig. 6.1.
Furthermore, .M; g/ is causal if there are no closed causal curves, and is stably
causal if there exists a global time function t W M ! R such that r t is time-like.
In particular, stably causal manifolds are time-orientable and do not admit closed
time-like curves, so that they are causal. Stable causality is a stronger requirement
than causality.
We are finally able to define global hyperbolicity [4, 5], which is an important
ingredient of singularity theorems and of the canonical formulation of classical
gravity. A Lorentzian manifold .M; g/ is globally hyperbolic if:
(i) it is causal, and
(ii) J C .x/ \ J .y/ is compact for all x; y 2 M.
See Fig. 6.2. We can give an alternative definition of global hyperbolicity. This
version is more restrictive and hence less preferable, but we mention it because
it is widely found in the literature. First, we need a few more definitions. A smooth
Fig. 6.1 Past light cone E PO O

of an observer O at point x in PO
a causally simple spacetime.
Particles (dashed world-line)
inside the particle horizon PO
have been observed by O,
while those outside (dotted
world-line) not yet. The x
bottom curve is the space-like _
past infinity E
_
J
Fig. 6.2 The intersection _

y J +(x) J (y)
U
J C .x/ \ J .y/ of the
causal future and past of any
two points in a globally
hyperbolic manifold is a
compact set
future-directed causal curve W .a; b/ ! M (with possibly a D 1 or

b D C1) is future-inextendible if lim !b ./ does not exist. A past-inextendible
past-directed curve is defined analogously. Also, the past domain of dependence of
˙ M is the set D .˙/ of all points x 2 M such that any future-inextendible
causal curve starting at x intersects ˙. The future domain of dependence DC .˙/
is defined analogously. The domain of dependence of ˙ is the set D.˙/ D
D .˙/ [ DC .˙/. Then, a Lorentzian manifold is globally hyperbolic if
(i) it is stably causal, and
(ii) for any a 2 R, the time slice ˙a D t1 .a/ defined as the pre-image of the
global time function t is such that D.˙a / D M.
Global hyperbolicity can also be characterized by Geroch splitting theorem. Let
˙ M be a Cauchy hypersurface in M, i.e., a sub-set intersected exactly once by
every inextendible time-like curve in M. Roughly speaking, the future (and past)
development of M can be pre- (retro-)dicted from data on ˙. Then, we have [6, 7]:
Geroch splitting theorem. Any globally hyperbolic spacetime .M; g/ admits
a smooth space-like Cauchy hypersurface ˙ and, then, it is diffeomorphic to
R ˙.
In other words, a global time function t can be chosen in such a way that each
surface of constant t is a Cauchy surface ˙t and the spacetime topology is M Š
R ˙. The set f˙t g is called a foliation of spacetime.
Cosmology obviously requires a time orientation and a foliation; in previous

chapters we tacitly assumed that spacetime is globally hyperbolic. One can also
extend the concepts of “open” and “closed” universe to non-FLRW spacetimes.
A generic Lorentzian manifold is open if it contains no compact (i.e., without
boundary) edgeless surfaces where any two points cannot be connected by time-
like curves. An alternative but more intuitive definition makes use of a compact
space-like surface. According to this definition, universes previously called “flat”
are open. A spacetime is closed if all its space-like surfaces are compact.
6.1.2 Focusing Theorems
In the presence of a singularity, we expect that all time-like or null geodesics will
focus (i.e., converge) at a point in the past or future. Any time-like vector t obeys
the Raychaudhuri equation (Problem 6.1),
P D ! ! R t t 1 2

C r Pt : (6.3)
D1
Let t D u be a unit time-like congruence, i.e., a family of integral curves such
that1
u r u D 0 D u r u ; u2 :D u u D 1 : (6.4)
When an observer with velocity u follows a time-like congruence, its acceleration

a D uP D u r u is identically zero. Also the last term of (6.3) vanishes.
Convergence of congruences is governed by the focusing theorem, which we now
state for time-like curves.
Focusing theorem 1. Let .M; g/ be a spacetime obeying the time-like conver-
gence condition (2.38), R u u > 0, where u is a time-like congruence of
past-directed geodesics such that its volume expansion satisfies 6 0 < 0
and the vorticity ! vanishes. Then, the congruence must blow up ( !
1) at proper time no greater than t0 D .D 1/=j 0 j in the past.
Proof If vorticity vanishes and the time-like convergence condition holds, the
Raychaudhuri equation for the family of geodesics implies that
P6 1 2
:
D1
1
In Chap. 2, we used the symbol u for a generic unit time-like vector. Here and in the following,
we reserve it for congruences.
Integrating from .t D 0/ 6 0 to .t/, we get the lower bound
1 1 t
> C ; (6.5)
0 D1
where the parameter t is positive and increasing to the past. This means that the
congruence must converge to a caustic ( ! 1) at proper time no greater than
t0 D .D 1/=j 0j in the past.
The theorem implies that all geodesics leaving a point will eventually reconverge
after a finite time. Replacing the expansion condition with > 0 > 0, one gets the
same result for congruences focusing in the future. Employing the Raychaudhuri
equation for null geodesics, one can prove the same theorem almost verbatim for
focusing null congruences, with proper time t replaced by an affine parameter .2
A stronger version of the focusing theorem assumes that the time-like conver-
gence condition is violated everywhere by at least a minimum amount.
Focusing theorem 2. Let .M; g/ be a Lorentzian manifold obeying the
convergence condition
ˇ2
R u u 6 < 0; (6.6)
D1
where ˇ > 0 and u is a unit time-like congruence of past-directed geodesics

such that its volume expansion satisfies 6 0 < ˇ and the vorticity !
vanishes. Then, the congruence must converge to a caustic at proper time no
greater than
D1 0
t0 D arccoth > 0 : (6.7)
ˇ ˇ
Proof From (6.3) and in the absence of shear, the congruence of time-like geodesics
obeys
P> 1
.ˇ 2 2
/:
D1
Integrating as before,

ˇ.t t0 /
6 ˇ coth ;
D1
2
Null congruences will be defined and employed in Sect. 7.7.
so that
< ˇ for t < t0 : (6.8)
The congruence diverges into the past faster than in the previous case, by the amount
given in (6.8). This concludes the proof.
The typical applications of the last result are inflationary universes.
6.1.3 Classifications of Singularities
Let us clarify what we mean by “singularity.” A space-like singularity is a point

where matter is concentrated with infinite density; time-like geodesics, which
represent motions of free particles, cannot be extended beyond that point. A time-
like singularity is a region of infinite curvature where null geodesics (motions of
light rays) end [8].
Then, a spacetime is time-like (respectively, null) geodesically incomplete if
there exist time-like (null) incomplete geodesics. In this case, spacetime is said to
be singular. The requirement of maximal extension in the definition of spacetime
is to avoid the obvious definition of a regular spacetime as a manifold with the
set of singularity points removed. The big bang is both a past space- and time-like
singularity because it marks the beginning (more precisely, the boundary) of time
and space, respectively. It is also a naked singularity, since there is no event horizon
to screen it. Note that these singularities are all physical and should be discriminated
from coordinate singularities, that can be removed by a suitable recharting of the
manifold.
Singularities can be organized in more refined systems. For instance, one speaks
about curvature singularities if at least one of the components of the Riemann or
Weyl tensors diverges or is discontinuous at a point [9]. If it is the Weyl tensor
to be ill behaved, then the singularity is conformal. According to which curvature
invariant displays the worst behaviour, the classification is further specialized. Also,
if the pathological components are bounded, the curvature singularity is oscillatory,
otherwise it is divergent; depending on the model, the big bang can be either
(Sect. 6.3).
Moreover, if objects can fall into the singularity without being torn apart by
tidal forces, the singularity is called weak; if they do not remain intact, it is called
strong [9]. These definitions can be made more precise by invoking only geometric
quantities, thus guaranteeing their validity irrespectively of the physical properties
of the bodies [10–12]. In general, physical singularities are strong. The big bang is,
by definition, a strong curvature singularity.
While the singularity theorems below guarantee geodesic incompleteness under
a set of sufficient conditions, other pathological situations may arise from different
necessary requirements and they must be dealt with separately. For the sake of
completeness, we should mention that in the context of cosmology the big bang may
be not the only singularity one encounters in the evolution of the universe [13–16]. It
is possible, in fact, that certain matter configurations violating the dominant energy
condition (2.56) lead to sudden future singularities (or type II future singularities),
where the fluid pressure P and the Ricci scalar R (as well as its time derivatives)
blow up in a finite proper time t since the big bang [17–20]. The scale factor a.t /,
the Hubble parameter H.t / and the energy density .t / remain finite, although one
has infinite acceleration:
a.t /; H.t /; .t / < 1 ; P; aR ! 1 :
This can even happen not too far from today, with t t0 10 Myr [21], or even
when the DEC is preserved [15]. Sudden future singularities do not entail geodesic
incompleteness and the evolution of the universe may continue through them. They
are not, in fact, of strong type and the surge of tidal forces at t does not destroy all
objects [22].
A more serious case of sudden singularity is the big rip (or type I future
singularity), where the scale factor, the energy density and pressure of the fluid all
go to infinity [23–26]:
a; H; ; P; Ra ! C1 :
This is caused by an exotic matter with effective equation of state with w < 1.
The universe neither collapses into a big crunch (the singularity at the end of a
contraction phase) nor expands forever, but it is torn apart (about 20 Gyr from
now) due to the increase of the energy / a3.1Cw/ with the expansion.
Big rip and sudden future singularities can be brought about not only by non-
conventional matter components but also by modifications of general relativity
(e.g., [27]). Exotic equations of state, where P./ 6/ and the energy density and
pressure are not proportional to each other, can be responsible for other types of
future singularities, where ; P ! C1 for finite scale factor (big freeze, type III)
[14, 28, 29] or finite a, and P but divergent derivatives of H (big brake, type IV)
[14, 30, 31]. Like the big rip, the big freeze is a strong singularity, while the big
brake is a weak one.
Much of future singularities depends on the details of the matter content and on
the nature of dark energy. It is not clear how a complete theory of quantum gravity
would fare with respect to future extremal regimes of curvature and energy and one
might argue that the big-bang problem is a more pressing business to conclude. This
chapter, therefore, concentrates on the latter, leaving further discussion on the fate
of the universe to Chap. 7.
6.2 Singularity Theorems 269
6.2 Singularity Theorems
The presence of the big bang in various scenarios is argued by a series of theorems
developed by Penrose, Hawking and Geroch in the 1960s [32–38] (and their later
extensions [39–41]) and by Borde, Vilenkin and Guth in the 1990s [42–47]. They
hold for classical general relativity and some also describe other singularities such
as those in the interior of black holes.
6.2.1 Hawking–Penrose Theorems
The strategy of singularity theorems is to follow the evolution of a geodesic (that

is, an observer) and see if we can do so indefinitely to the infinite past. All these
theorems share the same structure of hypotheses:
• a local convergence or energy condition;
• a global condition on the causal structure;
• an assumption that gravity is strong enough in some regions to force the
convergence of geodesics therein.
Playing with the strength of each condition, one can obtain different types of
singularity theorems.3
Hawking–Penrose theorem. Let .M; g/ be a spacetime such that:
1. it obeys the time-like convergence condition (2.38), R t t > 0;
2. it is globally hyperbolic;
3. the volume expansion satisfies 6 0 < 0 on a time slice ˙t .
Then .M; g/ is singular, i.e., no past-directed time-like geodesics orthogonal
to ˙t can be extended to proper time greater than t0 D .D 1/=j 0 j to the
past of ˙t .
Sketch of proof Consider a point x 2 M on a time slice ˙t . The first focusing
theorem prescribes that all geodesics from x reconverge to another point y in the
time slice ˙t0 at finite distance t0 t from ˙t . This implies that there is no geodesic of
maximal length. Such a geodesic exists in globally hyperbolic spacetimes which are
geodesically complete. We are thus forced to conclude that .M; g/ is geodesically
incomplete.
Applying the theorem to an expanding universe, this situation describes a
singularity in the past but it is straightforward to extend it to future singularities,
replacing the expansion condition with > 0 > 0. In both cases, after the finite
amount of proper time t0 a free-falling observer ceases to be represented by a point
3
We refer the interested reader to [4] for a complete review of older results and [2] for a concise
proof of the following theorem by Hawking and Penrose [38].
of the Lorentzian manifold. A similar claim holds when the time-like convergence
condition is replaced by (6.6) [45].
Closed time-like curves announce a breakdown of causality and are, perhaps,
more objectionable than singularities. Therefore, the requirement of their non-
existence, implicit in condition 2 of the theorem, seems reasonable. One might
ask, however, if their presence would prevent the formation of singularities. This
is actually not the case [36, 37] as, for instance, in a closed expanding or contracting
universe.
The singularity theorem applies to the region of the Schwarzschild solution inside
the event horizon (neutral non-rotating black holes), to the Milne universe (where,
however, geodesic incompleteness is not associated with a curvature singularity) and
to non-accelerating FLRW expanding spacetimes. In the latter case, one can check
the theorem’s conditions either by looking at (2.70) in D D 4 (R00 D 3H 2 . 1/ >
0) or by employing the Einstein equations to get (2.39) and (2.57). If > 0, then
the SEC is
1 2
w> C ;
3 3 2
which is valid for ordinary matter when 6 0 or when is positive but small.
A small positive cosmological constant, as observed today, would be negligible in
the extreme high-curvature regime near the big bang [38], so that it can be ignored.
The singularity at t D 0 is a physical singularity where the Ricci tensor diverges;
from (2.70) and (2.71),
R00 ! C1 ; R˛ˇ ! C1 :
Notice that the theorem does not apply, in particular, to the following cases:
• Minkowski spacetime, since there is no time slice with expansion bounded away
from zero.
• Anti-de Sitter spacetime ( < 0), which is not globally hyperbolic.
• de Sitter spacetime and expanding inflationary FLRW spacetimes, neither of
which satisfy the time-like convergence condition.
A comment about the last result is in order. So far, we have regarded de Sitter as
an idealization of realistic inflationary cosmologies where the Hubble parameter
H is approximately constant. However, it is important to realize how misleading it
can be to confuse the mathematical de Sitter spacetime with realistic models of the
early universe. In fact, for flat de Sitter the scale factor is a D eHt , the big bang is
in the infinite past at t ! 1 but the Ricci tensor and scalar are all finite there.
From (2.70), (2.71) and (2.72) with constant H and K D 0,
R00 D 3H 2 < 0 ; R˛ˇ ! 0C ; R D 12H 2 :

On the other hand, for power-law inflation ( p 1) the big bang is at t ! 0, where
R00 ! 1 ; R˛ˇ ! 0C ; R ! C1 :
Thus, the singularity in pure de Sitter is not a curvature singularity. It is actually

a coordinate singularity from the point of view of full de Sitter spacetime (four-
dimensional hyperboloid embedded in five-dimensional Minkowski), which is
geodesically complete [4]. For power-law cosmologies, it is a curvature singularity
and only in this case we call it “big bang.”
The Hawking–Penrose theorem is not decisive as far as the big bang issue is
concerned. We have just seen that it does not hold for de Sitter and inflationary
backgrounds. While de Sitter is geodesically complete, we cannot reach the same
conclusion for inflationary spacetimes. The continuity equation (2.76) shows that,
for > 0 and w > 1, the energy density increases backwards in time indefinitely.
In fact, another singularity theorem suggests that the big-bang singularity is typical
also in inflationary cosmologies. A closed trapped surface T is a compact space-
like two-surface such that the two families of null geodesics meeting the surface
orthogonally converge locally at T . Trapped surfaces can be regarded as regions
where gravity is so strong that even light rays cannot escape. They arise during
black-hole formation but also in the early universe when matter is extremely dense.
Then, we have the following [32, 33]
Penrose–Hawking theorem. Let .M; g/ be a spacetime such that:
1. it obeys the null convergence condition (2.34), R n n > 0;
2. there is a non-compact Cauchy surface;
3. there is a closed trapped surface.
Then .M; g/ is null geodesically incomplete.
The theorem does not distinguish between future and past singularities, but we
can see that it finds a physical application in the very early universe. Let us
examine the three hypotheses of the theorem for a universe with D 0 and a
perfectly homogeneous FLRW background. Einstein equations are assumed. First,
the NEC (2.54) is satisfied for inflationary fluids realized by ordinary field theories,
where > 0 and w > 1. In fact, from (2.65)
R n n D 2 T n n D 2 .n @ / > 0 ;
which is a background-independent result. One can see this also by choosing a null
vector n D .1; n˛ /, where n˛ n˛ D 1. Then, from (2.70) and (2.71)
P C 2K
R n n D .D 2/H ;
a2
which is positive definite if the universe does not super-accelerate and the curvature
term is positive or negligible. Second, FLRW spacetimes are globally hyperbolic
and, by Geroch splitting theorem, flat and open universes are naturally foliated
by non-compact Cauchy surfaces ˙t . Third, there are various ways to prove
the existence of trapped surfaces in the early universe [4, 33]. We recall one in
Problem 6.2.
Other singularity theorems can be tailored for closed universes [39, 41] and by
replacing the null energy condition with the WEC and/or the SEC on average [39,
40, 42]. In particular, if the strong energy condition is valid on the average along
all complete causal geodesics, then closed universes in which Einstein equations
hold have an incomplete time-like or null geodesic [40]; suitable focusing theorems
allow one to further extend these results [42].
6.2.2 Borde–Vilenkin Theorems
The eternally inflating scenario of Sect. 5.6.5 attracted particular interest [44–49].
Given a point x 2 B in an inflating region B M, there exists (with non-zero
probability) another point y 2 B in the future of x at a given finite geodesic distance.
Assuming that the boundaries of thermalized regions expand at (almost) the speed of
light, the whole region I .y/ n I .x/ would be non-thermal. Realistically, however,
thermalized regions will form almost surely (i.e., with probability 1) in an infinite
spacetime volume. This scenario satisfies all three hypotheses of the following
theorem.
Borde–Vilenkin theorem. Let .M; g/ be a spacetime such that:
1. it obeys the null convergence condition R n n > 0;
2. it is open;
3. there is at least one point x such that for some point y to the future of x the
volume of I .y/ n I .x/ is finite.
The local convergence condition of this theorem is the same as before. On the other
hand, the restriction on the global causal structure of spacetime is much milder. The
third condition is necessary for inflation to be future-eternal [43]. As a result, almost
all points in the inflating regions will have a singularity in their past.
The global condition on the geometry of spacetime can be relaxed to include also
closed universes. In turn, we need a causality condition prescribing that no light cone
can wrap around to “swallow” the universe. Let .M; g/ be a stably causal Lorentzian
manifold. A past light cone E is called localized if from every spacetime point x …
E there is at least one past-directed time-like curve that does not intersect the cone.
In particular, localized light cones cannot be compact. Minkowski, Schwarzschild,
de Sitter, flat and open FLRW spacetimes and some closed FLRW spacetimes all
have only localized past light cones. Examples of spacetimes with non-localized
light cones are given in [45].
Open and closed spacetimes obeying a set of rather general conditions are null
geodesically incomplete [45]:
Borde theorem. Let .M; g/ be a spacetime such that:
1. it obeys the null convergence condition R n n > 0;
2. it is stably causal and past causally simple;
3. all past light cones are localized;
4. it contains one of the following:
a. a point with a reconverging past light cone (i.e., such that < 0 along
every past-directed null geodesic in the light cone), or
b. a past-trapped surface, or
c. a point x such that for some point y to the future of x the volume of
I .y/ n I .x/ is finite.
This theorem modifies some of the requirements of the Borde–Vilenkin and the
Penrose–Hawking theorem. With respect to the latter and the Hawking–Penrose
theorem, it is also more specific about the location of the would-be singularity (in
the past).
Semi-classical considerations show that the null convergence condition might be
violated in eternally inflating scenarios where quantum fluctuations of the energy-
momentum tensor can produce super-acceleration [48]. However, there is actually a
much stronger singularity theorem for universes which expand sufficiently fast.
6.2.3 Borde–Guth–Vilenkin Theorem
First, we need a new kinematical, local definition of the volume expansion or Hubble
parameter. Consider a Lorentzian manifold .M; g/ and an observer O described
by the unit time-like or null geodesic v ./ D dx =d, embedded in the unit
velocity field u ./ such that (6.4) holds. This congruence may represent, but not
necessarily, a flux of real particles whose world-lines u cross the observer’s path
with zero proper acceleration (Fig. 6.3). Define :D u v > 0. In the time-like
case, this is the usual special-relativistic Lorentz factor between v and u , while in
the null case D dt=d (here the parameter is chosen to increase to the future),
where t is time measured by comoving observers.
At proper time (or affine parameter) and C ı the observer meets,
respectively, the world-line u ./ and u . C ı/. Furthermore, define the unit
space-like vector
v u
w :D p ; w2 D 1 ; u w D 0 : (6.9)
2 C v2
Fig. 6.3 Set-up of

uμ( σ +δσ )
Borde–Guth–Vilenkin
theorem: an observer O in a uμ( σ )
velocity field
δr μ
v μδ
σ
p
The space-like vector ır D 2 C v 2 w ı is orthogonal to the two world-lines
and joins them at equal times

p in their own rest frame. The distance between the two
particles is ır :D jır j D 2 C v 2 ı.
The generalized Hubble parameter is the radial component of the projected

gradient of u along the radial direction:
H :D w w r u : (6.10)
Its interpretation is straightforward. The relative velocity between the two test
particles is ıu D uP ı, where P :D v r . Its radial component is ıur :D w ıu .
Then, H D ıur =ır is the variation of the radial relative velocity of two particles
with respect to their distance at time , computed in the rest frame of one of the
particles. Substituting (6.9) in (6.10), we get
P
H D 2
D FP ; (6.11)
C v2
where we used vP D 0 and

1
; 1 v 2 D 0 (null geodesic)
FD : (6.12)
arctanh ; v 2 D 1 (time-like geodesic)
Notice that the volume expansion D r u of the congruence is nothing but

the
R average over all spatial directions of H . In fact, the average of w w is
d˝D2 w w D ˝D2 ı , where ˝D2 is given in (3.101). Then,

Z
1
d˝D2 H D : (6.13)
˝D2 D1
Therefore, H is the inhomogeneous and anisotropic generalization of the Hubble

parameter H.
We define the averaged expansion condition as
Z e
1
Hav :D dH ./ > 0 ; (6.14)
e i i
where e and i are two reference times such that e is in the future of i . This
excludes long (but not occasional) contracting phases in the past and it does
not require the expansion to be accelerating. The last theorem we review takes
advantage of this general condition [47, 49]:
Borde–Guth–Vilenkin (BGV) theorem. Let .M; g/ be a spacetime where a
congruence u can be continuously defined along any past-directed time-like
or null geodesic v . Let u obey the averaged expansion condition (6.14) for
almost any v . Then .M; g/ is geodesically past-incomplete.
Proof Consider two reference times e > i (e is in the future of i ). From (6.11),
one has
Z e
dH ./ D F. e / F. i / :
i
If the averaged expansion condition holds, then the function F > 0 increases
forward in time (d =d < 0). In particular,
F. e / F. i / C
0 < Hav D D ;
e i e i
where 0 < C < C1 is a positive finite constant. Thus, the proper or affine length
of almost any past-directed geodesic is finite and spacetime is geodesically past-
incomplete.
The averaged expansion condition might not be valid for all observers (for
some, the universe might have contracted during most of its history), hence the
specification “for almost any v ” in the theorem. In particular, the momentum
carried by almost any past-directed causal geodesic is blue-shifted by an infinite
amount in a finite proper time (or affine parameter) interval.
6.2.4 An Undecided Issue
Singularity theorems can be useful but tricky and one should exercise caution in
their interpretation.
On one hand, their failure on a given Lorentzian manifold does not prevent
the manifold to be geodesically incomplete. For instance, the Hawking–Penrose
theorem does not apply to inflation. If we assume that the universe inflates from
the very beginning, then the theorem lefts question (6.1) unanswered. However, one
can envisage a scenario where the universe does not actually begin to expand in
acceleration, thus satisfying the theorem just before inflation. Moreover, the other
theorems clearly show that geodesic incompleteness is manifest under a variety of
local and global conditions. Therefore, in most circumstances a negative result is
due to an unsuitable set of hypotheses.
On the other hand, it is not obvious that the past singularity found in geodesically
incomplete manifolds is the big bang. In fact, the theorems only demonstrate the
existence of incomplete geodesics but they do not guarantee that this feature is
associated with a spacetime singularity, nor are they precise about where in the
past the would-be singularity lies. In other words, to be identified with the cosmic
big bang, geodesic incompleteness should hold for all observers with end-points
at the same time slice t0 D tbb . There is evidence, nonetheless, that at least
in the case of eternal inflation almost all observers satisfy the last condition of
the Borde–Vilenkin theorem, albeit there is no information about the location of
all these singularities. Things get better when the focusing condition (6.8) and
the existence of a global Cauchy hypersurface are assumed, so that all time-like
geodesics emanating orthogonally from the Cauchy surface terminate at t0 [45].
The situation is all the more unclear in inhomogeneous, non-FLRW cosmologies
such as those found in the initial conditions of chaotic inflation, where the perfect-
fluid approximation is almost certainly too optimistic. One would have to look at
dissipative and interacting relativistic fluids [50], which might or might not violate
the hypotheses of the theorems (such as the energy conditions and the existence
of trapped surfaces). Anyway, the BGV theorem applies to a fairly wide class of
cosmological models, without any assumption about homogeneity, isotropy and the
energy conditions. Provided the average expansion condition (6.14) holds, time-like
and null geodesic incompleteness is guaranteed also in extremely inhomogeneous
scenarios. This is an important achievement but, unfortunately, inconclusive like its
predecessors: from geodesic incompleteness one cannot infer that inflating universes
have a unique beginning. On top of that, there are particular scenarios which avoid
also the BGV theorem. Based on arguments inspired by geodesic completeness of
full de Sitter, a geodesically complete eternally inflating spacetime was constructed
in [51, 52].
To summarize, the answer to question (6.1) is: “At the classical level, Yes, there
was a big bang. . . maybe. And there are exceptions.” The big-bang problem exists
as soon as we have a big bang and the evidence collected so far is that likely there
was one. This fact begs for a solution, but there is another reason why we should
prepare ourselves for this task. The singularity theorems show that singularities
plague spacetime manifolds under rather general and reasonable conditions, for
instance in the formation of black holes or in inflationary cosmologies. Maybe it
is not possible to always avoid geodesic incompleteness, or to avoid all singularities
with the same mechanism, but a general solution to the problem (universal in its
qualitative features) would touch upon the very nature of geometry and spacetime
inside black holes and in the early universe and would have deep consequences in
our way of thinking Nature.
6.2.4.1 Bouncing and Cyclic Universes
From a philosophical perspective, the big bang has raised many questions about
the nature of time and its birth, leading to alternative scenarios where the initial
singularity is replaced by a finite event (a bounce) or a series of bounce events
(cyclic universe). A bounce is a moment of the evolution of the universe where, after
a period of contraction, the geometry acquires a minimum volume configuration
where energy density, pressure and curvature do not diverge. After the bounce,
the universe expands. At the semi-classical level, the structure of the perturbations
generated through a single bounce can be more complicated than the standard one in
a monotonically expanding universe; for example, vector modes cannot be neglected
during the contracting phase in contrast to their decaying behaviour in the post-big-
bang phase.
In other models, the Universe experiences a cyclic succession of expansions and
contractions in which a single bounce is just a transitory phase in a wider process of
evolution. This idea is embedded in several ancient cosmogonies (including Hindu
and Theravāda Buddhist traditions4 ) but also modern science tried to implement it
[56] (see [57–64] for early attempts).
Since the big bang is typical of cosmological models in general relativity, the
existence of special bouncing or cyclic solutions does not provide a genuine solution
to the big-bang problem. This led to the study of cosmologies with standard gravity
and alternative forms of matter, such as fermion condensates [65], or with a negative
cosmological constant [66–72]. In the latter case, the dynamics is driven by a
negative via the standard Friedmann equation (in D D 4)
2 jj
H2 D ; (6.15)
3 3
where is made of ordinary dust matter and radiation. The universe has no begin-
ning and undergoes linear asymmetric cycles with constant period due to entropy
exchange between different matter species. Entropy increases monotonically from
cycle to cycle and so does the scale factor. The average expansion mimics one with
acceleration, thus giving rise to a scenario of emergent cyclic inflation.
A cyclic universe can be realized also in models beyond general relativity,
including higher-derivative gravity and scalar-tensor theories [56] and multi-scale
spacetimes where the dynamics takes place in a scale-dependent multi-fractal
4
For Theravāda cosmology and the creation-destruction myth, see Dı̄gha Nikāya 27.10-31 [53],
Majjhima Nikāya 28.7,12 [1], Aṅguttara Nikāya 4:156, 7:66 [54]. A comprehensive account can
be found in the Visuddhimagga, XIII 13, 28–65 [55].
geometry [73]. In the latter case, one has a universe with a finite past but an infinite
number of cycles, which are log-periodic and asymmetric by construction of the
measure. Another possibility, in the presence of a cosmological constant, is to join
cycles of expansions from the big bang to the indefinite de Sitter acceleration by
identifying the conformal 3-surface representing the big bang with the conformal
infinity of the past cycle. This scenario of conformal cyclic cosmology [74–77] gives
rise to ring structures that are still under search in CMB maps [78–80].
It should be stressed that in many cyclic models the average expansion con-
dition (6.14) holds and Hav > 0 for a null geodesic over the cycles. Thus,
the BGV theorem implies that also cyclic scenarios are, in general, geodesically
incomplete. Rather than model-dependent resolutions of the problem, one might
look at altogether different frameworks where classical gravity is heavily modified
by quantum effects and big-bang avoidance is a robust outcome of quantization. In
Chaps. 10, 11, and 13 we shall see candidate models claiming such a feature.
6.3 BKL Singularity
Instead of attempting to address the big-bang issue in quantum gravity, let us

assume, for the time being, that gravity is purely classical and that the initial
singularity exists and is unavoidable. The analytic or semi-analytic behaviour of
solutions to the Einstein’s equations near the singularity [81, 82] has been studied
first in a series of papers [83–93] whose results are collected in the major works
[94, 95] and later in [96–106]. Numerical methods can also be employed [107] (see
[108, 109] for reviews on numerical analyses of singularities).
6.3.1 Tetrads and Bianchi Models
Let .M; g/ be a D-dimensional spacetime manifold M locally equipped with a

metric g . The tangent space Tx M is isomorphic to Minkowski space and we can
define the one-to-one map e W M ! Tx M which sends tensor fields from the
manifold to the Minkowskian tangent space. This map, generally called vielbein
(tetrad or vierbein in the four-dimensional case), is a local reference system for
spacetime, physically representing the gravitational field. Its relation with the metric
g is summarized by the following formulæ:5
g D ab ea eb ; ea ea D ı ; ea eb D ıab ; (6.16)

5
The co-tetrad ea is often denoted as !a in the literature.
6.3 BKL Singularity 279
where both Greek and Latin indices run from 0 to D 1 and transform, respectively,
under general coordinate and local Lorentz transformations. The gravitational field
ea is the transformation matrix between local Minkowski coordinates X a D X a .x/
and arbitrary coordinates x ,
@X a
ea D : (6.17)
@x
Vielbein fields incorporate all the metric properties of spacetime but the converse is
not true. In fact, due to manifest local Lorentz invariance, there are infinitely many
realizations of the local basis reproducing the same metric tensor. This is also the
reason why there are more components in the vielbeins than in the metric field, the
difference being exactly D.D 1/=2, which is the number of free parameters of the
SO.1; D 1/ group representing Lorentz transformations on the Minkowski tangent
space.
A priori, the homogeneous and isotropic assumption of FLRW backgrounds is
too restrictive in the neighborhood of the big bang, but the choice of too general
a background might be intractable. As in many other situations, it is customary
to achieve a compromise and choose some background encoding, say, anisotropies
but not inhomogeneities. Denoting with Latin indices i; j the spatial components of
the tangent space, on a homogeneous manifold M the triad ei ˛ D ei ˛ .t/ is space-
independent. We define D 1 .D 1/-dimensional constant frame vectors li and
D 1 scalars ai such that (no sum over the inert index i)
ei ˛ D ai .t/ li˛ : (6.18)
Then, the metric in synchronous time can be written as
d2 s D dt2 C g˛ˇ dx˛ dxˇ D dt2 C a2i .t/ıij li˛ lˇ dx˛ dxˇ :
j
(6.19)
Three-dimensional homogeneous spaces can be divided according to the Bianchi

classification of three-dimensional real Lie algebras [110–112]. The complete list
of spaces is discussed in [94, 113]; here we mention just a few special cases.
• Bianchi I: homogeneous spacetimes with zero spatial curvature. Isotropic exam-
ples are Euclidean, Minkowski and flat FLRW spacetimes.
• Bianchi V: special cases are homogeneous spacetimes with constant negative
spatial curvature. An isotropic example is open FLRW spacetimes.
• Bianchi IX: special cases are homogeneous spacetimes with constant positive
spatial curvature. An isotropic example is closed FLRW spacetimes.
Some line elements, vacuum and perfect-fluid solutions of Bianchi II, III, IV, VIh ,
VI0 and VIIh can be found in [114].
6.3.2 Kasner Metric
Anisotropic Bianchi I models are important for cosmology. A particular flat metric
is
X
D1
ds2 D dt2 C a2i .t/.dxi /2 : (6.20)
iD1
It describes a homogeneous flat universe characterized by D 1 scale factors ai .t/,

one for each spatial direction expanding at different rates Hi D aP i =ai . In the isotropic
(FLRW) case, ai .t/ D a.t/ for all i D 1; : : : ; D 1. Since li˛ D ı˛i , Latin and Greek
indices are interchangeable.
A vacuum solution in D > 3 is the Kasner metric [115], where
ai .t/ D tpi (6.21)
up to a normalization constant. In fact, the only non-vanishing Levi-Civita compo-

nents in flat Bianchi I are
ij0 D Hi gij ;
j j
i0 D Hi ıi ; (6.22)
and the Ricci tensor reads6
X
D1 X aR i
R00 D P i/ D
.Hi2 C H ; (6.23)
iD1 i
ai
X
Rij D HP i C Hi Hk gij : (6.24)
k
Plugging in the profile (6.21), the vacuum equations R D 0 are solved if
X X
pi D 1 ; p2i D 1 : (6.25)
i i
Therefore, the Kasner exponents pi obey the conditions (6.25), the first defining a
hyperplane and the second a hypersphere SD2 . The intersection of the Kasner plane
and sphere determines the space of solutions pi as a hypersphere SD3 . The covariant
volume has a singularity linear in t, since by the first Kasner condition
6
Sum and product indices range from 1 to D 1 unless stated otherwise. Sums or products with
subscript i < j run both on j and on i < j.
p Y
g D ai .t/ D t : (6.26)
i
The exponents can always be ordered as p1 6 p2 6 6 pD1 6 1 but at least

one inequality must be strict. In fact, isotropic power-law cosmology is not a vacuum
solution (R D 0), since the Kasner conditions cannot be satisfied simultaneously
for pi D p, for all i. Notice also that the Riemann tensor R vanishes on the
solution only when pD1 D 1 and pi D 0 for all i D 1; : : : ; D 2; after a coordinate
transformation, this is actually Minkowski. Barring this case (which we will ignore
from now on), one can show that at least one Kasner exponent is always negative:
squaring the first Kasner condition in (6.25) and using the second, one obtains
X
pi pj D 0 : (6.27)
i<j
Thus, at least one direction is contracting although the total spatial volume increases.
In three spatial dimensions, there is exactly one non-positive Kasner exponent and
the range is
1 2 2
3 6 p1 < 0 ; 0 < p2 6 3 ; 3 6 p3 < 1 : (6.28)
They are equal in pairs only when . p1 ; p2 ; p3 / D .1=3; 2=3; 2=3/ (or
. p1 ; p2 ; p3 / D .0; 0; 1/, which we disregard).
In the presence of matter, the Kasner metric is no longer a solution. However,
near the singularity it is still a good approximation. In fact, consider a perfect fluid

P world-line u and constant barotropic index w. The volume expansion is D
with
i Hi , so that in the absence of shear the continuity equation (2.49) and the spatial
components of (2.52) become
X
1
P C Hi .1 C w/ D 0 ; ui P C 1 C uP i D 0 : (6.29)
i w
The solutions are

Y .1Cw/ .1C w1 /
/ u0 ai ; / ui ; (6.30)
i
where in the second equation the index of ui is inert. The second equation states that
the covariant components ui all have the same magnitude. The largest contravariant
component near the singularity (t ! 0) is uD1 D uD1 =a2D1 , so that

p uD1
u0 D 1 C ui ui :
aD1
Then, solving for uD1 in (6.30) and combining the two results one has
0 1 1Cw
1w
Y 1Cw
@ ai A D t 1w .1pD1 / ;
i¤D1
ui t 1w .1pD1 / ; u0 t 1w .1pD1 /pD1 ;

w w
where we assumed w ¤ 1. For matter obeying the dominant energy condition (2.56)
(radiation, dust, scalar field and so on), the energy density diverges at t D 0 except
in the special case pD1 D 1, as anticipated. From (2.40), and remembering that
p1 < 0,
jT00 j ' .1 C w/u20 t.1CpD1 / TD1

D1
;
1Cw
T11 ' w t 1w .1pD1 / ;
Tii ' .1 C w/ui ui t.1C2pi pD1 / ; i D 2; : : : ; D 2 :
The t D 0 singularity in the energy-momentum tensor is milder than for the Ricci
tensor components (6.23) and (6.24), which scale as R t2 . We conclude that
curvature terms dominate near the big bang and it is not restrictive to neglect matter.
6.3.3 Generalized Kasner Metric
The non-flat generalized Kasner metric [81] is a special case of (6.19) where ai D
tpi . The Ricci tensor reads
X X aR i
R00 D P i/ D
.Hi2 C H ; (6.31)
i i
ai
X
Ri D Rˇ˛ l˛i lˇ D H
j j P i C Hi Hk ıi C .D1/Ri ;
j j
(6.32)
k
where .D1/Ri is the Ricci tensor of spatial slices, built with the purely spatial con-
j
nection ˛ˇ .l/. One can check whether the Kasner profile (6.21) is an approximate
vacuum solution in the limit t ! 0. This happens if the spatial curvature term .D1/Ri
j
can be neglected with respect to the rest, that is, when it does not grow faster than
t2 near the singularity. It turns out that

.D1/
Ri t2.1pijk / ;
j
(6.33)
where
X
pijk :D 2pi C pm D 1 C pi pj pk ; i ¤ j ¤ k ¤ i: (6.34)
m¤i;j;k
In the stability region of the parameter space
pijk > 0 ; (6.35)
the spatial curvature is negligible and one has the desired result. In D > 11
dimensions, there always exists a region wherein the inequality (6.35) is satisfied,
but for D 6 10 this is not possible unless one contradicts the Kasner conditions
[97].7 In that case, one has to impose some constraints on the functions li˛ (and their
derivatives, which are zero in this case) so that (6.35) is enforced.
For instance, in D D 4 one has pijk D 2pi and the dominant term goes as
t2.12p1 / . Define
1 i
.i/ :D l .r li / ; (6.36)
V0
where .r li /˛ D ˛ˇ @ˇ li is the curl of the vector li . The comoving (coordinate)

spatial volume is V0 D j det lj D l1 .l2 l3 /. The spatial curvature is [81] (no sum
over indices)
.3/ i 1
Ri D .i/2 a4i .. j/ a2j .k/ a2k /2 ; i ¤ j ¤ k ¤ i; (6.37)
2.a1 a2 a3 /2
while off-diagonal components vanish. For each component, the dominant term is
.1/2 a21
.1/2 t2.12p1 / ;
2.a2 a3 /2
j
which is negligible in Ri only if
.1/ D 0 : (6.38)
This is the additional constraint advertized in the general case.
7
In nine spatial dimensions, the maximum pijk is zero at p1 D p2 D p3 D 1=3, p4 D D p9 D
1=3.
In Bianchi I spaces, all the .i/ vanish. In Bianchi VIII models, the .i/ are
constant and, all but one, with the same sign. Without loss of generality, one
can choose .1/ D .2/ D .3/ D 1. The volume of spatial slices is infinite
[94]. In Bianchi IX models, all .i/ are constant and of the same sign; we can set
.1/ D .2/ D .3/ D 1. Then, the four-dimensionalQ metric (6.19) represents a
“twisted” ellipsoid [116] of volume V D 16 2 i ai [94] (here V0 D 1), i.e., a
closed universe whose spatial hypersurfaces have non-constant positive curvature.
When all ai are equal (closed FLRW), the curvature is constant.
6.3.4 Mixmaster Dynamics
Since the constants .i/ do not vanish in Bianchi VIII and Bianchi IX spaces, the
generalized Kasner metric with condition (6.38) is an approximate solution. We can
imagine a Bianchi VIII or Bianchi IX universe where the Kasner metric (6.20) plays
the role of a Bianchi I flat background, while spatial curvature terms are regarded
as homogeneous perturbations. Bianchi VIII and IX are geodesically incomplete
[117, 118] and at t D 0 there is a physical (curvature) singularity.
Fix D D 4. Recasting (6.31), (6.32) and (6.37) in terms of the time variable
dt
d :D Q ; (6.39)
i ai
the vacuum equations R D 0 read (here Hi :D a0i =ai D a1

i dai =d )
3
X X
0D Hi0 2 Hi Hj ; (6.40)
iD1 i<j
2
0 D 2Hi0 C a4i . j/ a2j .k/ a2k ; i ¤ j ¤ k ¤ i; (6.41)
which can be combined into the first-order expression

X X X
4 Hi Hj D a4i 2 .i/ . j/ .ai aj /2 : (6.42)
i<j i i<j
These equations can be studied semi-analytically in a neighborhood of the singular-

ity. Initial conditions are set to be Kasner-like at the infinite future,
Hi . D C1/ D pi ; p1 < 0 :
As time evolves (backwards towards the singularity), the perturbation along the i D
1 direction grows, while those corresponding to positive Kasner exponents decay.
Then, at late times, ln t and (6.41) can be approximated as
H10 C 12 a41 ' 0 ; 0

H2;3 12 a41 ' 0 ;
whose positive solutions are

s
2jp1 j p
a1 . / D ; a2;3 . / D e. p2;3 jp1 j/ cosh.2jp1 j / : (6.43)
cosh.2jp1 j /
Near the singularity ( ! 1),
2jp1 j .12jp1 j/
a1 . / ejp1 j ; a2;3 . / e. p2;3 2jp1 j/ ; t e ;
1 2jp1 j
so that asymptotically one has another Kasner epoch
ai tpQi ; (6.44)
where
jp1 j p2 2jp1 j p3 2jp1 j

pQ 1 D > 0; pQ 2 D < 0; pQ 3 D > 0:
1 2jp1 j 1 2jp1 j 1 2jp1 j
(6.45)
The new coefficients stillQobey the Kasner conditions (6.25). Note that the volume
in the new epoch is V D i ai .1 2jp1 j/t.
6.3.4.1 Kasner Epochs
To summarize the physical picture moving towards the singularity, one starts at large
t with a Kasner epoch where space contracts along the ˛ D i D 1 direction. By
effect of the homogeneous perturbation, a1 and a2 reach, respectively, a minimum
and a maximum value, at which there is a transition to another Kasner epoch with
contracting direction ˛ D i D 2. The previously increasing perturbation / a41 now
dies away, while the perturbation / aQ 42 drives the universe to another Kasner epoch,
and so on. p3 , the greatest of the two positive powers in one era, remains positive
also in the next.
It is convenient to parametrize the Kasner exponents with a parameter u > 1
(Fig. 6.4):

u 1
p1 .u/ D D p1 ; (6.46a)
1 C u C u2 u
Fig. 6.4 Kasner exponents

pi .u/ in a four-dimensional
Bianchi I universe

1Cu 1
p2 .u/ D D p 3 ; (6.46b)
1 C u C u2 u

u.1 C u/ 1
p3 .u/ D 2
D p2 ; (6.46c)
1CuCu u
where the rightmost-hand sides show how to extend the parametrization to u 6 1.

Comparing (6.45) and (6.46), the transition rule (in terms of the parameter of the
first epoch) can be recast as
pQ 1 .u/ D p2 .u 1/ ; pQ 2 .u/ D p1 .u 1/ ; pQ 3 .u/ D p3 .u 1/ : (6.47)
Let u D u0 > 1 be the initial value at D C1. Realistic scenarios will not obey a
Kasner power law exactly, so that the only meaningful case is when u0 is irrational.
We can decompose u0 into its integer part n1 D bu0 c (or floor function: the greatest
integer number n1 smaller than u0 ) and a remainder 1 > q0 2 R n Q:
u0 D bu0 c C q0 :
The negative exponent will bounce back and forth between the i D 1 and i D 2
directions for a certain number k1 D 1 C n1 of iterations of the parameter
un D u0 n D n1 C q0 n ; n D 0; 1; : : : ; n1 D k1 1 : (6.48)
When the integer part is exhausted and uk1 1 D q0 < 1, either p1 or p2 is negative
and p3 becomes the smallest positive exponent. This is the beginning of the last
Kasner epoch and uk1 signals the end of a Kasner era (sequence of Kasner epochs),
i.e., a period of time when one of the directions evolves monotonically. p
From (6.43), the maximum value of a1 in a given Kasner epoch is amax D 2jp1 j.
Since jp1 j decreases in successive epochs, the maximal expansion of one direction
decreases in a given era. More precisely, consider the Kasner profile parametrized by
Fig. 6.5 Kasner era made of n1 Kasner epochs. Time flows backwards towards the singularity
to the right. The symbols are explained in the text (Source: adaptation of “Kasner epochs”
by Lantonov – Own work. Licensed under Creative Commons Attribution-Share Alike 3.0 via
Wikimedia Commons [119])
p
p maximum is aQ max D
the parameter u defined in the preceding epoch. The 2jQp1 .u/j.
In the next epoch, the maximum value is aQQ max D 2jQp2 .u/j, so that
s s r
aQQ max jQp2 .u/j jp1 .u 1/j u1
D D D < 1: (6.49)
aQ max jQp1 .u/j jp2 .u 1/j u
The evolution towards the singularity can be represented as in Fig. 6.5 for the
first Kasner era with n1 epochs. We define the anisotropic e-foldings
3
X
Ni :D ln ai ; Ni D j j : (6.50)
iD1
In the second expression, weQnoticed that near the singularity the change in the
normalization of the volume i ai between one epoch and another is negligible for
large j j (small t), so that we can define a global time variable D ln t < 0 and
neglect the gradual decrease (6.49), setting Nimax 0. In particular, Ni 6 0 for
all i. In Fig. 6.5, we denote with j n j the beginning of the .n C 1/-st epoch. All the
minima of N1;2 can be written as
Ni;n D n ın ; i D 1; 2 ;
where 0 < ın < 1, while from (6.50)
N3;n D n .1 ın / :
The discrete evolution of the number of e-foldings is governed by
Ni;nC1 Ni;n
D pi .un / : (6.51)
nC1 n
During the epoch starting at n and ending at nC1 , one of the functions Ni;n (say,
N1;n ) increases from N1;n D n n to N1;nC1 D 0, while the other (N2;n ) decreases
ı
from N2;n D 0 to N2;nC1 D nC1 ınC1 . Combining these expressions with (6.51),
one obtains
1 C un 1 C u0
j nC1 jınC1 D j n jın D j 0 jı0 :
un un
Therefore, the oscillation amplitudes jN˛;n j increase during the whole Kasner
era. The ratio of the amplitude of the last oscillation with respect to the first is
proportional to the “length” of the era if the latter is long,
jNi;k1 1 j k1 C q 0
D ' k1 ; i D 1; 2 :
jNi;0 j 1 C q0
The oscillation amplitude of the scale factors a1;2 grows as a power law,
a1;2 . k1 1 / D Œa1;2 . 0 /k1 .
Still from (6.51), the duration of the Kasner epochs is

1 C un C u2n 1 C un C u2n
j nC1 j D j n j 1 C ı n D j 0 j 1 C .n C 1/ n C ı0 :
un un
The time length of the epochs increases within the era, j nC1 j j n j > j n j j n1 j,
while the total time duration of the era is

1
j k 1 j j 0 j D k1 k1 C q 0 C j 0 jı0 :
q0
Here j k1 j is the beginning of the next era. Notice that it is consistent to assume
instantaneous transitions between Kasner epochs, since the time duration of transi-
tion periods, from (6.43), is long when jp1 j 1 (u 1) and proportional to u.
This should be compared with j nC1 j j n j ' un j n jın un .
6.3.4.2 Kasner Eras and Chaos
After the end of the first Kasner era, according to the reparametrization u ! 1=u
in (6.46) the negative exponent will bounce between the directions i D 3 and
i D 1 or (as in the example of Fig. 6.5) between i D 3 and i D 2. The function
N3 increases from its minimum value N3;k1 D k1 .1 ık1 /, which sets the initial
amplitude for the new series of (wider) oscillations:

1
ıQ0 j Q0 j :D jN3;k1 j D C k1 .k1 C q0 / 1 j 0 jı0 > j 0 jı0 : (6.52)
ı0
If u0 is irrational, there is an infinite sequence of Kasner eras. Asymptotically,

both the order of the transitions between Kasner epochs and the “length” of a Kasner
era (number of Kasner epochs in it) are governed by a random process. In fact,
the initial value of the parameter u for the second era is 1=q0 D n2 C q1 and
its length is n2 . From this conclusion and (6.48), the iterative map governing the
discrete evolution of the parameter un can be written as

un 1 ; un > 2
unC1 D : (6.53)
.un 1/1 ; 1 < un < 2
By induction, the length ki of successive eras are given by the expansion of u0 as a

continuous fraction:
1
u 0 D k1 C : (6.54)
1
k2 C
k3 C : : :
The values of the sequence fkm g are distributed according to stochastic laws [94]
which can be studied with the methods of fractal geometry and chaos theory [120–
127]. In this sense, the singularity is chaotic. This chaotic behaviour has an agile
representation in terms of Misner variables [82, 128]:
p
˝ :D 13 ln.a1 a2 a3 / ; a1;2 D: e˝CˇC ˙ 3ˇ
; a3 D: e˝2ˇC :
(6.55)
One can show that the potential associated with the Bianchi IX cosmology is
n p
U / det.gij / .3/ R / e4˝ e8ˇC 4e2ˇC cosh.2 3 ˇ /
h p io
C2e4ˇC cosh.4 3 ˇ / 1 : (6.56)
Anisotropies are encoded in the two variables ˇ˙ defined by (6.55). The poten-
tial (6.56) is depicted in Fig. 6.6. The dynamics of the universe going towards the
Fig. 6.6 The potential (6.56) of the Bianchi IX mixmaster universe
singularity is described by a particle falling down a potential U with exponentially

steep walls and three sharp corners. As soon as the particle enters a corner and
gets reflected by one of the walls, a Kasner epoch ends and is followed by another
with different expansion. The infinite number of reflections gives rise to a type of
chaotic dynamics known as a cosmological billiard, for which a wide toolbox of
sophisticated techniques is available [82, 128–135] (see [136, 137] for reviews).
These tools have gradually superseded the patchwork-like intuitive method followed
in this section and in the original papers on this type of singularities. We will
describe classical billiards in more detail in Sect. 13.9.2.
The natural stochastic character of the solution has an important consequence.
Given a set of “initial” conditions at a time t > 0, after a few eras evolving towards
the singularity the system loses memory of the chosen initial conditions and admits
a chaotic description. Thus, the solution is stable in a qualitative sense, inasmuch
as the oscillatory regime will be realized in general. However, the details of specific
solutions will be highly sensitive to the initial conditions.
This big-bang model based on a Bianchi VIII or Bianchi IX metric is know
as the Belinsky–Khalatnikov–Lifshitz (BKL) singularity, or also as the mixmaster
universe.8
8
This name was given by Misner [82] after a famous mechanical kitchen mixer, produced by an
American brand of electric home appliances. Reference to this tool is due to the fact that, after a
large number of eras, all parts of the whole universe are in causal contact with one another, along
all directions: the texture of a cream or a dough is homogenized after enough mixing cycles. In
fact, the BKL model was a candidate solution to the horizon problem well before inflation was
proposed. This can be roughly seen from (2.188) and the discussion at the end of Sect. 5.2.2. In
The BKL singularity is space-like: the metric becomes singular in the spatial
dimensions while approaching the big bang. There is also a time-like version with a
similar chaotic behaviour, which we will not discuss [138, 139].
6.3.5 BKL Conjecture
There is good evidence that the mixmaster behaviour is a general property of

solutions of the Einstein equations near the singularity, for fairly arbitrary “initial”
conditions (boundary conditions on a hypersurface in the future of the singularity).
The chaotic singularity is not a peculiar feature of Bianchi VIII and Bianchi IX
models in vacuum. Bianchi I and Bianchi IX universes with Yang–Mills non-
Abelian fields also evolve chaotically, although in a qualitatively different way
[140–142]. Going beyond general relativity and the Einstein–Hilbert action, the
mixmaster behaviour on Bianchi IX anisotropic backgrounds is present in scalar-
tensor theories [143].
The mixmaster universe can be realized also in Bianchi I [144], Bianchi II [145]
and Bianchi VI0 general relativity [146–148] with a magnetic field and a perfect
fluid with jwj 6 1. However, it is not a global attractor and there exist exceptional
solutions which do not display the oscillatory regime near the singularity.
There are other counter-examples of models which do not oscillate. Particular
forms of matter (a massless scalar field or a stiff fluid, w D 1) do lead to
a homogeneous singularity but spoil the mixmaster behaviour [149, 150]. For
D > 10 C 1 general relativity in vacuum, the oscillatory chaotic approach to
the singularity disappears and the general solution is monotonic, contrary to what
happens for 4 6 D 6 10. Therefore, the structure of the singularity is qualitatively
different in theories with 10 or more spatial dimensions [97–105]. This result has
a beautiful mathematical description in terms of the Kac–Moody algebra AED1
defining the BKL oscillatory structure: for D 1 < 10, AED1 is hyperbolic, while
for D 1 > 10 it is not [106]. We will comment in Sect. 12.4, and see more in detail
in Sect. 13.9.2, that p-forms with p > 0 can induce a chaotic evolution in a number
of situations where it would otherwise be suppressed, including in D D 11. String
theory will have to say much about that.
Near the singularity, higher-order curvature terms should become important in
the effective action, so that it is natural to extend the BKL analysis to these models.
The mixmaster universe is also a solution in higher-order D D 4 gravity with
(Riemann)2 contributions in the action [151–154] but it is non-generic because
unstable; on the other hand, there exists a general (stable) solution which is
monotonic, isotropic and non-chaotic. In higher dimensions, it seems unlikely that
a given Kasner era, plays the role of conformal time along the direction expanding (forward
in time) monotonically. Inflation, as we have seen, does much more than addressing the horizon
problem.
the mixmaster universe be realized in fourth-order gravity, since the Kasner metric
is not a Bianchi I solution [155].
Furthermore, one can abandon homogeneous spacetimes and consider less
restrictive backgrounds. The vacuum BKL model can be generalized to the presence
of matter and to inhomogeneous (non-diagonal) metrics as well, where .i/ D
.i/ .x/ are functions of spatial coordinates. In both cases, the only qualitative
difference is that the direction of the oscillation axes changes from one epoch to
another [94, 95, 104, 105]; the BKL singularity is confirmed with other methods for
inhomogeneous backgrounds with barotropic fluids [156]. Generic spacetimes with
no special symmetries have been studied via the orthonormal frame formalism in
four dimensions [157] and with Iwasawa variables [158] and further evidence was
collected that the mixmaster, oscillatory, homogeneous universe is typically realized
near the singularity in 4 6 D 6 10 [157, 159].
We end up with the following conclusion. In a variety of gravitational set-ups
with different total actions and dimensionality, homogeneous (Bianchi) metrics are
a good local approximation of generic spacetimes near the singularity. The dynamics
is either monotonic (in particular, Kasner-like) and non-chaotic, or oscillatory and
(modulo exceptional cases) chaotic. The BKL conjecture elevates the lesson of
all these models to a paradigm. Intuitively, it states that spatially different points
decouple from one another near the big-bang singularity. More precisely:
BKL conjecture. Near a singularity, almost all spacetimes become spatially
homogeneous at every point and time derivatives dominate over spatial
derivatives. The approach towards the singularity is either monotonic or
oscillatory in some of the quantities describing the system.
The sentence “almost all spacetimes” takes into account the existence of excep-
tional cases, while the generic formula “in some of the quantities describing the
system” encompasses both traditional mixmaster scenarios (where spatial directions
oscillate) and others where oscillations occur in other variables (for instance, shear
and color stress in Einstein–Yang–Mills models [141]).
Stronger versions of the conjecture also require that matter be negligible near the
singularity (in this case, the singularity is described by vacuum Bianchi models) and
specify the most general Bianchi metric near the singularity to be of type IX. Since
there are several exceptions to both these requirements, the weaker formulation is
preferable.
The approach in Iwasawa variables is also convenient because it unravels a
surprising feature of oscillatory singularities, with which we conclude the discussion
on the BKL universe. A question which we did not pose so far is what happens
at the big bang: Is there any geometric structure left at the singularity? When the
dynamical evolution is monotonic, the answer is positive. For instance, at least one
direction of a Bianchi I Kasner universe expands when approaching the big bang
and one still has the asymptotic notion of which direction expands or contracts in
the limit ! 1. If the approach is oscillatory, however, we would expect chaos
to erase any information of the metric in the same limit. This is not the case and the
asymptotic geometry at the cosmological singularity, dubbed partially framed flag

[158], is non-trivial.
6.1 Raychaudhuri equation. Derive the Raychaudhuri equation (6.3)

using (2.43) and (3.88) for a time-like vector t .
Solution From (2.43),
1
.r t /.r t / D ! ! C 2
; (6.57)
D1
where we used (2.44), (2.45) and the following relations:
h C h D 0; Pt t C Pt t D 0; ! Pt t C ! Pt t D Pt t Pt t :
Also, from (3.88) one gets
2rŒ r t D R t : (6.58)
Contracting with t ,
R t t D t r r t P D r Pt .r t /.r t / P I
plugging this into (6.57), we obtain (6.3). On a FLRW background, both shear and
vorticity vanish and one gets (2.82) provided the Einstein equations hold.
6.2 Trapped surfaces. Show the existence of trapped surfaces (Sect. 6.2.1)
in a FLRW spacetime.
Solution After the coordinate redefinition

8
Z <arcsinhr ; KD 1
dr
%D p D r; KD0 ;
1 Kr2 :
arcsin r ; K D C1
the spatial FLRW line element (2.2b) in four dimensions becomes

˛
˛ˇ dx dxˇ D d%2 C r2 .%/d˝22 :
Consider a sphere T D S2 of comoving radius % in the slice ˙t . The two families of

past-directed null geodesics orthogonal to S2 intersect the slice ˙t0 for t0 < t in two
two-spheres of radius
Z t0
0 dt00
%˙ .t / D % ˙ D % ˙ Œrp .t0 / rp .t/ ;
t a.t00 /
where rp is, as usual, the particle-horizon radius. The surface area of a two-sphere of
radius % is A D 4a2 r2 .%/, so that both families of null geodesics will be converging
P %˙ .t/ > 0. This happens if
into the past if AŒt;
s
a dr 1 1 dr 1 1
aH C %P D ˙ D ˙ K > 0;
r d% rH r d% rH %2
thus yielding the condition
rH2
KrH2 < 1 : (6.59)
%2
Notice now that the Friedmann equation (2.81) with D 0 can be written as
1 1
2
D 2 K; (6.60)
rH rS
where
s 8
>r ; D 1
3 < H
K
rS :D D rH ; KD0 (6.61)
2 :
< rH ; K D C1
is the comoving Schwarzschild length of the universe filled with a perfect fluid with
positive energy density . Equation (6.59) becomes
% > rS : (6.62)
In each time slice, there is a trapped surface of size greater than the Schwarzschild
length.
References 295
References
1. The Middle Length Discourses of the Buddha. A Translation of the Majjhima Nikāya, transl.
by Bhikkhu Ñān.amoli and Bhikkhu Bodhi (Wisdom, Somerville, 1995)
2. J. Natário, Relativity and singularities —A short introduction for mathematicians. Resenhas
6, 309 (2005). [arXiv:math/0603190]
3. S.W. Hawking, Nature of space and time. arXiv:hep-th/9409195
5. A.N. Bernal, M. Sánchez, Globally hyperbolic spacetimes can be defined as causal instead of
strongly causal. Class. Quantum Grav. 24, 745 (2007). [arXiv:gr-qc/0611138]
6. R.P. Geroch, Domain of dependence. J. Math. Phys. 11, 437 (1970)
7. A.N. Bernal, M. Sánchez, On smooth Cauchy hypersurfaces and Geroch’s splitting theorem.
Commun. Math. Phys. 243, 461 (2003). [arXiv:gr-qc/0306108]
8. K. Tomita, On inhomogeneous cosmological models containing space-like and time-like
singularities alternately. Prog. Theor. Phys. 59, 1150 (1978)
9. G.F.R. Ellis, B.G. Schmidt, Singular space-times. Gen. Relat. Grav. 8, 915 (1977)
10. F.J. Tipler, Singularities in conformally flat spacetimes. Phys. Lett. A 64, 8 (1977)
11. C.J.S. Clarke, A. Królak, Conditions for the occurence of strong curvature singularities. J.
Geom. Phys. 2, 127 (1985)
12. A. Królak, Towards the proof of the cosmic censorship hypothesis. Class. Quantum Grav. 3,
267 (1986)
13. S. Cotsakis, I. Klaoudatou, Future singularities of isotropic cosmologies. J. Geom. Phys. 55,
306 (2005). [arXiv:gr-qc/0409022]
14. S. Nojiri, S.D. Odintsov, S. Tsujikawa, Properties of singularities in (phantom) dark energy
universe. Phys. Rev. D 71, 063004 (2005). [arXiv:hep-th/0501025]
15. C. Cattoën, M. Visser, Necessary and sufficient conditions for big bangs, bounces, crunches,
rips, sudden singularities, and extremality events. Class. Quantum Grav. 22, 4913 (2005).
16. L. Ferńandez-Jambrina, R. Lazkoz, Classification of cosmological milestones. Phys. Rev. D
74, 064030 (2006). [arXiv:gr-qc/0607073]
17. J.D. Barrow, Sudden future singularities. Class. Quantum Grav. 21, L79 (2004).
18. K. Lake, Sudden future singularities in FLRW cosmologies. Class. Quantum Grav. 21, L129
(2004). [arXiv:gr-qc/0407107]
19. J.D. Barrow, More general sudden singularities. Class. Quantum Grav. 21, 5619 (2004).
20. J.D. Barrow, C.G. Tsagas, New isotropic and anisotropic sudden singularities. Class. Quan-
tum Grav. 22, 1563 (2005). [arXiv:gr-qc/0411045]
21. M.P. Da̧browski, T. Denkiewicz, M.A. Hendry, How far is it to a sudden future singularity of
pressure? Phys. Rev. D 75, 123524 (2007). [arXiv:0704.1383]
22. L. Ferńandez-Jambrina, R. Lazkoz, Geodesic behaviour of sudden future singularities. Phys.
Rev. D 70, 121503 (2004). [arXiv:gr-qc/0410124]
23. R.R. Caldwell, A phantom menace? Phys. Lett. B 545, 23 (2002). [arXiv:astro-ph/9908168]
24. A.A. Starobinsky, Future and origin of our universe: modern view. Grav. Cosmol. 6, 157
(2000). [arXiv:astro-ph/9912054]
25. B. McInnes, The dS/CFT correspondence and the big smash. JHEP 0208, 029 (2002).
26. R.R. Caldwell, M. Kamionkowski, N.N. Weinberg, Phantom energy and cosmic doomsday.
27. Y. Shtanov, V. Sahni, New cosmological singularities in braneworld models. Class. Quantum
Grav. 19, L101 (2002). [arXiv:gr-qc/0204040]
28. S. Nojiri, S.D. Odintsov, Final state and thermodynamics of dark energy universe. Phys. Rev.
D 70, 103522 (2004). [arXiv:hep-th/0408170]
29. H. Štefančić, Expansion around the vacuum equation of state: sudden future singularities and
asymptotic behavior. Phys. Rev. D 71, 084024 (2005). [arXiv:astro-ph/0411630]
30. V. Gorini, A. Kamenshchik, U. Moschella, V. Pasquier, Tachyons, scalar fields, and cosmol-
ogy. Phys. Rev. D 69, 123512 (2004). [arXiv:hep-th/0311111]
31. A. Kamenshchik, C. Kiefer, B. Sandhöfer, Quantum cosmology with big-brake singularity.
Phys. Rev. D 76, 064032 (2007). [arXiv:0705.1688]
32. R. Penrose, Gravitational collapse and space-time singularities. Phys. Rev. Lett. 14, 57 (1965)
33. S.W. Hawking, Occurrence of singularities in open universes. Phys. Rev. Lett. 15, 689 (1965)
34. S.W. Hawking, The occurrence of singularities in cosmology. Proc. R. Soc. Lond. A 294, 511
(1966)
35. S.W. Hawking, The occurrence of singularities in cosmology. II. Proc. R. Soc. Lond. A 295,
490 (1966)
36. R.P. Geroch, Singularities in closed universes. Phys. Rev. Lett. 17, 445 (1966)
37. S.W. Hawking, The occurrence of singularities in cosmology. III. Causality and singularities.
Proc. R. Soc. Lond. A 300, 187 (1967)
38. S.W. Hawking, R. Penrose, The singularities of gravitational collapse and cosmology. Proc.
R. Soc. Lond. A 314, 529 (1970)
39. F.J. Tipler, General relativity and conjugate ordinary differential equations. J. Diff. Equ. 30,
165 (1978)
40. F.J. Tipler, Energy conditions and spacetime singularities. Phys. Rev. D 17, 2521 (1978)
41. G.J. Galloway, Curvature, causality and completeness in space-times with causally complete
spacelike slices. Math. Proc. Camb. Philos. Soc. 99, 367 (1986)
42. A. Borde, Geodesic focusing, energy conditions and singularities. Class. Quantum Grav. 4,
343 (1987)
43. A. Vilenkin, Did the universe have a beginning? Phys. Rev. D 46, 2355 (1992)
44. A. Borde, A. Vilenkin, Eternal inflation and the initial singularity. Phys. Rev. Lett. 72, 3305
(1994). [arXiv:gr-qc/9312022]
45. A. Borde, Open and closed universes, initial singularities and inflation. Phys. Rev. D 50, 3692
(1994). [arXiv:gr-qc/9403049]
46. A. Borde, A. Vilenkin, Singularities in inflationary cosmology: a review. Int. J. Mod. Phys. D
5, 813 (1996). [arXiv:gr-qc/9612036]
47. A. Borde, A.H. Guth, A. Vilenkin, Inflationary spacetimes are incomplete in past directions.
Phys. Rev. Lett. 90, 151301 (2003). [arXiv:gr-qc/0110012]
48. A. Borde, A. Vilenkin, Violation of the weak energy condition in inflating spacetimes. Phys.
Rev. D 56, 717 (1997). [arXiv:gr-qc/9702019]
49. A.H. Guth, Eternal inflation and its implications. J. Phys. A 40, 6811 (2007).
50. D. Langlois, F. Vernizzi, Nonlinear perturbations for dissipative and interacting relativistic
fluids. JCAP 0602, 014 (2006). [arXiv:astro-ph/0601271]
51. A. Aguirre, S. Gratton, Steady-state eternal inflation. Phys. Rev. D 65, 083507 (2002).
52. A. Aguirre, S. Gratton, Inflation without a beginning: a null boundary proposal. Phys. Rev. D
67, 083515 (2003). [arXiv:gr-qc/0301042]
53. The Long Discourses of the Buddha. A Translation of the Dı̄gha Nikāya, transl. by M. Walshe
(Wisdom, Somerville, 1995)
54. The Numerical Discourses of the Buddha. A Translation of the Aṅguttara Nikāya, transl. by
Bhikkhu Bodhi (Wisdom, Somerville, 2012)
55. B. Buddhaghosa, Visuddhimagga — The Path of Purification, transl. by Bhikkhu Ñān.amoli
(Buddhist Publication Society, Kandy, 2010), pp. 404–414
56. M. Novello, S.E. Perez Bergliaffa, Bouncing cosmologies. Phys. Rep. 463, 127 (2008).
[arXiv:0802.1634]
References 297
57. R.C. Tolman, On the problem of the entropy of the Universe as a whole. Phys. Rev. 37, 1639
(1931)
58. R.C. Tolman, On the theoretical requirements for a periodic behaviour of the Universe. Phys.
Rev. 38, 1758 (1931)
59. G. Lemaître, L’univers en expansion. Ann. Soc. Sci. Bruxelles A 53, 51 (1933)
60. R.C. Tolman, Relativity, Thermodynamics and Cosmology (Clarendon Press, Oxford, 1934)
61. M.J. Rees, The collapse of the universe: an eschatological study. Observatory 89, 193 (1969)
62. R.H. Dicke, P.J.E. Peebles, The big bang cosmology – enigmas and nostrums, in [63]
63. S.W. Hawking, W. Israel (eds.), General Relativity: An Einstein Centenary Survey (Cam-
bridge University Press, Cambridge, 1979)
64. Ya.B. Zel’dovich, I.D. Novikov, Relativistic Astrophysics. The Structure and Evolution of the
Universe, vol. 2 (University of Chicago Press, Chicago, 1983)
65. S. Alexander, T. Biswas, Cosmological BCS mechanism and the big bang singularity. Phys.
Rev. D 80, 023501 (2009). [arXiv:0807.4468]
66. T. Biswas, Emergence of a cyclic universe from the Hagedorn soup. arXiv:0801.1315
67. T. Biswas, S. Alexander, Cyclic inflation. Phys. Rev. D 80, 043511 (2009). [arXiv:0812.3182]
68. T. Biswas, A. Mazumdar, Inflation with a negative cosmological constant. Phys. Rev. D 80,
023519 (2009). [arXiv:0901.4930]
69. T. Biswas, A. Mazumdar, A. Shafieloo, Wiggles in the cosmic microwave background
radiation: echoes from nonsingular cyclic inflation. Phys. Rev. D 82, 123517 (2010).
[arXiv:1003.3206]
70. T. Biswas, T. Koivisto, A. Mazumdar, Could our universe have begun with ?
arXiv:1105.2636
71. T. Biswas, T. Koivisto, A. Mazumdar, Phase transitions during cyclic inflation and non-
Gaussianity. Phys. Rev. D 88, 083526 (2013). [arXiv:1302.6415]
72. W. Duhe, T. Biswas, Emergent cyclic inflation, a numerical investigation. Class. Quantum
Grav. 31, 155010 (2014). [arXiv:1306.6927]
73. G. Calcagni, Multi-scale gravity and cosmology. JCAP 1312, 041 (2013). [arXiv:1307.6382]
74. R. Penrose, Before the big bang: an outrageous new perspective and its implications for
particle physics. Conf. Proc. C 060626, 2759 (2006)
75. R. Penrose, Cycles of Time: An Extraordinary New View of the Universe (Bodley Head,
London, 2010)
76. V.G. Gurzadyan, R. Penrose, Concentric circles in WMAP data may provide evidence of
violent pre-big-bang activity. arXiv:1011.3706
77. E. Newman, A fundamental solution to the CCC equations. Gen. Relat. Grav. 46, 1717 (2014).
[arXiv:1309.7271]
78. V.G. Gurzadyan, R. Penrose, On CCC-predicted concentric low-variance circles in the CMB
sky. Eur. Phys. J. Plus 128, 22 (2013). [arXiv:1302.5162]
79. A. DeAbreu, D. Contreras, D. Scott, Searching for concentric low variance circles in the
cosmic microwave background. JCAP 1512, 031 (2015). [arXiv:1508.05158]
80. D. An, K.A. Meissner, P. Nurowski, Ring-type structures in the Planck map of the CMB.
arXiv:1510.06537
81. E.M. Lifshitz, I.M. Khalatnikov, Investigations in relativistic cosmology. Adv. Phys. 12, 185
(1963)
82. C.W. Misner, Mixmaster universe. Phys. Rev. Lett. 22, 1071 (1969)
83. E.M. Lifshitz, I.M. Khalatnikov, Problems of relativistic cosmology. Usp. Fiz. Nauk 80, 391
(1963) [Sov. Phys. Usp. 6, 495 (1964)]
84. V.A. Belinskiı̆, I.M. Kahalatnikov, A general solution of the gravitational equations with a
simultaneous fictitious singularity. Zh. Eksp. Teor. Fiz. 49, 1000 (1965) [Sov. Phys. JETP 22,
694 (1966)]
85. L.P. Grishchuk, A.G. Doroshkevich, I.D. Novikov, Anisotropy of the early stages of cosmo-
logical expansion and of relict radiation. Zh. Eksp. Teor. Fiz. 55, 2281 (1968) [Sov. Phys.
JETP 28, 1210 (1969)]
86. V.A. Belinskiı̆, I.M. Khalatnikov, On the nature of the singularities in the general solutions
of the gravitational equations. Zh. Eksp. Teor. Fiz. 56, 1701 (1969) [Sov. Phys. JETP 29, 911
(1969)]
87. E.M. Lifshitz, I.M. Khalatnikov, Oscillatory approach to singular point in the open cosmo-
logical model. Pis’ma Zh. Eksp. Teor. Fiz. 11, 200 (1970) [JETP Lett. 11, 123 (1970)]
88. V.A. Belinskiı̆, E.M. Lifshitz, I.M. Khalatnikov, Oscillatory approach to the singular point in
relativistic cosmology. Usp. Fiz. Nauk 102, 463 (1970) [Sov. Phys. Usp. 13, 745 (1971)]
89. E.M. Lifshitz, I.M. Lifshitz, I.M. Khalatnikov, Asymptotic analysis of oscillatory mode of
approach to a singularity in homogeneous cosmological models. Zh. Eksp. Teor. Fiz. 59, 322
(1970) [Sov. Phys. JETP 32, 173 (1971)]
90. I.M. Khalatnikov, E.M. Lifshitz, General cosmological solution of the gravitational equations
with a singularity in time. Phys. Rev. Lett. 24, 76 (1970)
91. V.A. Belinskiı̆, E.M. Lifshitz, I.M. Khalatnikov, The oscillatory mode of approach to a
singularity in homogeneous cosmological models with rotating axes. Zh. Eksp. Teor. Fiz.
60, 1969 (1971) [Sov. Phys. JETP 33, 1061 (1971)]
92. V.A. Belinskiı̆, E.M. Lifshitz, I.M. Khalatnikov, Construction of a general cosmological
solution of the Einstein equation with a time singularity. Zh. Eksp. Teor. Fiz. 62, 1606 (1972)
[Sov. Phys. JETP 35, 838 (1972)]
93. V.A. Belinskiı̆, I.M. Khalatnikov, E.M. Lifshitz, On the problem of the singularities in the
general cosmological solution of the Einstein equations. Phys. Lett. A 77, 214 (1980)
94. V.A. Belinskiı̆, I.M. Khalatnikov, E.M. Lifshitz, Oscillatory approach to a singular point in
the relativistic cosmology. Adv. Phys. 19, 525 (1970)
95. V.A. Belinskiı̆, I.M. Khalatnikov, E.M. Lifshitz, A general solution of the Einstein equations
with a time singularity. Adv. Phys. 31, 639 (1982)
96. I.M. Khalatnikov, E.M. Lifshitz, K.M. Khanin, L.N. Shchur, Ya.G. Sinai, On the stochasticity
in relativistic cosmology. J. Stat. Phys. 38, 97 (1985)
97. J. Demaret, M. Henneaux, P. Spindel, Non-oscillatory behavior in vacuum Kaluza–Klein
98. J. Demaret, J.L. Hanquin, M. Henneaux, P. Spindel, A. Taormina, The fate of the mixmaster
behaviour in vacuum inhomogeneous Kaluza–Klein cosmological models. Phys. Lett. B 175,
129 (1986)
99. R.T. Jantzen, Symmetry and variational methods in higher-dimensional theories. Phys. Rev.
D 34, 424 (1986); Symmetry and variational methods in higher-dimensional theories: Errata
and addendum. Phys. Rev. D 35, 2034 (1987)
100. Y. Elskens, M. Henneaux, Chaos in Kaluza–Klein models. Class. Quantum Grav. 4, L161
(1987)
101. Y. Elskens, M. Henneaux, Ergodic theory of the mixmaster model in higher space-time
dimensions. Nucl. Phys. B 290, 111 (1987)
102. Y. Elskens, Ergodic theory of the mixmaster universe in higher space-time dimensions. II. J.
Stat. Phys. 48, 1269 (1987)
103. A. Hosoya, L.G. Jensen, J.A. Stein-Schabes, The critical dimension for chaotic cosmology.
Nucl. Phys. B 283, 657 (1987)
104. J. Demaret, Y. De Rop, M. Henneaux, Chaos in non-diagonal spatially homogeneous
cosmological models in spacetime dimensions 6 10. Phys. Lett. B 211, 37 (1988)
105. J. Demaret, Y. De Rop, M. Henneaux, Are Kaluza–Klein models of the universe chaotic? Int.
J. Theor. Phys. 28, 1067 (1989)
106. T. Damour, M. Henneaux, B. Julia, H. Nicolai, Hyperbolic Kac–Moody algebras and chaos
in Kaluza–Klein models. Phys. Lett. B 509, 323 (2001). [arXiv:hep-th/0103094]
107. B.K. Berger, D. Garfinkle, E. Strasser, New algorithm for mixmaster dynamics. Class.
Quantum Grav. 14, L29 (1997). [arXiv:gr-qc/9609072]
108. B.K. Berger, Numerical approaches to spacetime singularities. Living Rev. Relat. 5, 1 (2002)
109. D. Garfinkle, Numerical simulations of general gravitational singularities. Class. Quantum
Grav. 24, S295 (2007). [arXiv:0808.0160]
References 299
110. L. Bianchi, Sugli spazii a tre dimensioni che ammettono un gruppo continuo di movimenti.
On the three-dimensional spaces which admit a continuous group of motions. Soc. Ital. Sci.
Mem. Mat. 11, 267 (1898) [Gen. Relat. Grav. 33, 2157 (2001); Gen. Relat. Grav. 33, 2171
(2001)]
111. A. Krasiński et al., The Bianchi classification in the Schücking–Behr approach. Gen. Relat.
Grav. 35, 475 (2003)
112. W. Kundt, The spatially homogeneous cosmological models. Gen. Relat. Grav. 35, 491 (2003)
113. L.D. Landau, E.M. Lifshitz, The Classical Theory of Fields (Butterworth–Heinemann,
London, 1980)
114. L. Hsu, J. Wainwright, Self-similar spatially homogeneous cosmologies: orthogonal perfect
fluid and vacuum solutions. Class. Quantum Grav. 3, 1105 (1986)
115. E. Kasner, Geometrical theorems on Einstein’s cosmological equations. Am. J. Math. 43, 217
(1921)
1996). [arXiv:gr-qc/0101003]
117. H. Ringström, Curvature blow up in Bianchi VIII and IX vacuum spacetimes. Class. Quantum
Grav. 17, 713 (2000). [arXiv:gr-qc/9911115]
118. H. Ringström, The Bianchi IX attractor. Ann. Henri Poincaré 2, 405 (2001).
119. http://commons.wikimedia.org/wiki/File:Kasner_epochs.svg#mediaviewer/File:Kasner_
epochs.svg
120. J.D. Barrow, Chaos in the Einstein equations. Phys. Rev. Lett. 46, 963 (1981); Erratum-ibid.
46, 1436 (1981)
121. J.D. Barrow, Chaotic behavior in general relativity. Phys. Rep. 85, 1 (1982)
122. D.F. Chernoff, J.D. Barrow, Chaos in the mixmaster universe. Phys. Rev. Lett. 50, 134 (1983)
123. P. Halpern, Chaos in the long-term behavior of some Bianchi-type VIII models. Gen. Relat.
Grav. 19, 73 (1987)
124. N.J. Cornish, J.J. Levin, The mixmaster universe is chaotic. Phys. Rev. Lett. 78, 998 (1997).
125. N.J. Cornish, J.J. Levin, The mixmaster universe: a chaotic Farey tale. Phys. Rev. D 55, 7489
(1997). [arXiv:gr-qc/9612066]
126. A.E. Motter, P.S. Letelier, Mixmaster chaos. Phys. Lett. A 285, 127 (2001).
127. A.E. Motter, Relativistic chaos is coordinate invariant. Phys. Rev. Lett. 91, 231101 (2003).
128. C.W. Misner, Quantum cosmology. I. Phys. Rev. 186, 1319 (1969)
129. C.W. Misner, Minisuperspace, in Magic Without Magic, ed. by J.R. Klauder (Freeman, San
Francisco, 1972)
130. D.M. Chitré, Investigation of Vanishing of a Horizon for Bianchi Type IX (The Mixmaster
Universe). Ph.D. thesis, University of Maryland, College Park (1972)
131. N.L. Balazs, A. Voros, Chaos on the pseudosphere. Phys. Rep. 143, 109 (1986)
132. A. Csordás, R. Graham, P. Szépfalusy, Level statistics of a noncompact cosmological billiard.
Phys. Rev. A 44, 1491 (1991)
133. R. Graham, R. Hübner, P. Szépfalusy, G. Vattay, Level statistics of a noncompact integrable
billiard. Phys. Rev. A 44, 7002 (1991)
134. R. Benini, G. Montani, Frame independence of the inhomogeneous mixmaster chaos via
Misner–Chitré-like variables. Phys. Rev. D 70, 103527 (2004). [arXiv:gr-qc/0411044]
135. J.M. Heinzle, C. Uggla, N. Rohr, The cosmological billiard attractor. Adv. Theor. Math. Phys.
13, 293 (2009). [arXiv:gr-qc/0702141]
138. S.L. Parnovsky, Gravitation fields near the naked singularities of the general type. Physica A
104, 210 (1980)
139. E. Shaghoulian, H. Wang, Timelike BKL singularities and chaos in AdS/CFT. Class.
140. B.K. Darian, H.P. Kunzle, Axially symmetric Bianchi I Yang–Mills cosmology as a dynamical
system. Class. Quantum Grav. 13, 2651 (1996). [arXiv:gr-qc/9608024]
141. J.D. Barrow, J.J. Levin, Chaos in the Einstein–Yang–Mills equations. Phys. Rev. Lett. 80, 656
(1998). [arXiv:gr-qc/9706065]
142. Y. Jin, K.-i. Maeda, Chaos of Yang–Mills field in class A Bianchi spacetimes. Phys. Rev. D
71, 064007 (2005). [arXiv:gr-qc/0412060]
143. R. Carretero-González, H.N. Núñez-Yépez, A.L. Salas-Brito, Evidence of chaotic behavior in
Jordan–Brans–Dicke cosmology. Phys. Lett. A 188, 48 (1994)
144. V.G. LeBlanc, Asymptotic states of magnetic Bianchi I cosmologies. Class. Quantum Grav.
14, 2281 (1997)
145. V.G. LeBlanc, Bianchi II magnetic cosmologies. Class. Quantum Grav. 15, 1607 (1998)
146. V.G. LeBlanc, D. Kerr, J. Wainwright, Asymptotic states of magnetic Bianchi VI0 cosmolo-
gies. Class. Quantum Grav. 12, 513 (1995)
147. B.K. Berger, Comment on the chaotic singularity in some magnetic Bianchi VI0 cosmologies.
148. M. Weaver, Dynamics of magnetic Bianchi VI0 cosmologies. Class. Quantum Grav. 17, 421
(2000). [arXiv:gr-qc/9909043]
149. V.A. Belinskiı̆, I.M. Khalatnikov, Effect of scalar and vector fields on the nature of the
cosmological singularity. Zh. Eksp. Teor. Fiz. 63, 1121 (1972) [Sov. Phys. JETP 36, 591
(1973)]
150. L. Andersson, A.D. Rendall, Quiescent cosmological singularities. Commun. Math. Phys.
218, 479 (2001). [arXiv:gr-qc/0001047]
151. J.D. Barrow, H. Sirousse-Zia, Mixmaster cosmological model in theories of gravity with a
quadratic Lagrangian. Phys. Rev. D 39, 2187 (1989); Erratum-ibid. D 41, 1362 (1990)
152. J.D. Barrow, S. Cotsakis, Chaotic behaviour in higher-order gravity theories. Phys. Lett. B
232, 172 (1989)
153. S. Cotsakis, J. Demaret, Y. De Rop, L. Querella, Mixmaster universe in fourth-order gravity
theories. Phys. Rev. D 48, 4595 (1993)
154. J. Demaret, L. Querella, Hamiltonian formulation of Bianchi cosmological models in
quadratic theories of gravity. Class. Quantum Grav. 12, 3085 (1995). [arXiv:gr-qc/9510065]
155. N. Deruelle, On the approach to the cosmological singularity in quadratic theories of gravity:
the Kasner regimes. Nucl. Phys. B 327, 253 (1989)
156. N. Deruelle, D. Langlois, Long wavelength iteration of Einstein’s equations near a spacetime
singularity. Phys. Rev. D 52, 2007 (1995). [arXiv:gr-qc/9411040]
157. C. Uggla, H. van Elst, J. Wainwright, G.F.R. Ellis, The past attractor in inhomogeneous
158. T. Damour, S. de Buyl, Describing general cosmological singularities in Iwasawa variables.
Phys. Rev. D 77, 043520 (2008). [arXiv:0710.5692]
159. D. Garfinkle, Numerical simulations of generic singuarities. Phys. Rev. Lett. 93, 161101
(2004). [arXiv:gr-qc/0312117]
Chapter 7
Cosmological Constant Problem
What shall we use to fill the empty spaces?

— Roger Waters (Pink Floyd), The Wall
Contents
7.1 The Problem in Field Theory. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302
7.1.1 Spontaneous Symmetry Breaking and Dynamical . . . . . . . . . . . . . . . . . . . . . 302
7.1.2 Zero-Point Energy and Higher Loops. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305
7.1.3 Supersymmetry and Supergravity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 308
7.2 Other Versions of the Problem and Strategies. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310
7.2.1 Broken Symmetries. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310
7.2.2 The 4 Puzzle. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311
7.2.3 UV or IR Problem? Strategies for a Solution. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313
7.3 Quintessence. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 315
7.3.1 Tracking, Freezing and Thawing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 316
7.3.2 Periodic and Power-Law Potentials. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 321
7.3.3 Exponential and Hyperbolic Potentials. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 322
7.3.4 Inverse Power-Law Potential. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 323
7.3.5 Other Potentials. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 324
7.3.6 Quintessence and the Inflaton. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 325
7.3.7 Summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 326
7.4 Scalar-Tensor Theories. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 327
7.4.1 Motivations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 328
7.4.2 Conformal Transformations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 328
7.4.3 Perturbations, Quantum Theory and Extended Inflation. . . . . . . . . . . . . . . . . 332
7.4.4 Cosmological Constant Problem. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333
7.4.5 Experimental Bounds and Chameleon Mechanism. . . . . . . . . . . . . . . . . . . . . . . 334
7.5 Higher-Order and Higher-Derivative Gravity Models. . . . . . . . . . . . . . . . . . . . . . . . . . . . 336
7.5.1 Motivation and Ghosts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 336
7.5.2 General f .R/ Action. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 338
7.5.3 Palatini Formulation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 340
7.5.4 Form of f .R/. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 340
7.5.5 Horndeski Theory and Extensions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 342
(continued)

DOI 10.1007/978-3-319-41127-9_7
302 7 Cosmological Constant Problem
7.6 Other Approaches. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 343

7.6.1 Varying Couplings. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 343
7.6.2 Void Models. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 344
7.6.3 Unimodular Gravity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 346
7.6.4 Analogue Gravity and Condensates. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 348
7.7 Emergent Gravity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351
7.7.1 Rindler Observer and Null Congruences. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 352
7.7.2 Dynamics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 355
7.7.3 Holographic Equipartition. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 356
7.7.4 Cosmological Constant Problem. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 360
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 369
A suspicion shared by some, including the author, is that a satisfactory solution of

the cosmological constant problem would shed much light on the puzzle of quantum
gravity. Quantum field theory, vacuum fluctuations, the microscopic degrees of
freedom of gravity and their coarse graining all converge to this Pandora’s box in
ways that still fascinate even the most consummate expert. About three quarters of
the content of the universe is something whose intimate nature is utterly unknown.
Hundreds of explanations, theories, models, conjectures have been put forward
without successfully convincing the scientist. This chapter is an account, with
neither sad nor happy ending, of the problem and of some of the efforts dedicated
to its solution.
7.1 The Problem in Field Theory
7.1.1 Spontaneous Symmetry Breaking and Dynamical
We saw in Sect. 2.5.3 that the cosmological constant can be interpreted as the
contribution of field vacuum fluctuations to the energy density of the universe.
In quantum field theory, the physics of the vacuum is regulated by the evolution
of particle fields determined by their self and mutual interactions. Therefore, it is
expected that becomes dynamical in realistic models [1, 2]. This is naturally
achieved by the mechanism of spontaneous symmetry breaking from a “false”
to a “true” vacuum, where vacuum configurations corresponding to maxima of a
classical potential are unstable and the system decays to vacua of lower energy. We
saw instances of this in hybrid inflation (Sect. 5.5.3) and, indirectly, in the Higgs
Lagrangian (5.202).
7.1 The Problem in Field Theory 303
The Goldstone model gives the standard example of spontaneous symmetry

breaking. Consider a complex scalar with Lagrangian
L D @ @ "
V. ; "
/; V. ; "
/ D V0 2 j j2 C j j4 ; (7.1)
where 2 ; > 0 and V0 are constants and j j2 D " . The system is invariant under
a U.1/ global transformation ! ei˛ , where 0 6 ˛ < 2 is a real, spacetime-
independent parameter. The vacuum structure reflects this continuous symmetry.
The double-well (or Mexican-hat) potential V has a local maximum (a false vacuum)
at D 0, where V D V0 , and an infinite number of global minima at

D # :D p ei# ; (7.2)
2
labelled by 0 6 # < 2 and where V D Vmin :D V0 4 =.4/. The system is said

to have a U.1/-symmetric true vacuum state. The U.1/ symmetry is “spontaneously
broken” when the scalar falls into one p of the minima. We can choose, for example,
# D 0 and the minimum 0 D = 2. To describe the system near the ground
state, it is convenient to parametrize the field with real scalars and ,
1
.x/ D p Œv C .x/ C i.x/ ; (7.3)
2
p
where v D 2 0 , and expand the Lagrangian (7.1) accordingly:
L D 12 .@/2 12 m2 2 12 .@/2 U.; / : (7.4)
Up to interaction terms U.; /, one can single out the free part and recognize as
a massive scalar with standard kinetic term and mass m2 :D 2v 2 D 22 , and
as a massless scalar called Nambu–Goldstone boson. In general, Nambu–Goldstone
bosons arise whenever a continuous symmetry is spontaneously broken, a result
known as Goldstone’s theorem [3–5]. The model can be generalized to an Abelian
U.1/ gauge symmetry as well as to the non-Abelian local gauge symmetries of
the Standard Model [6–9]. In the Abelian case, the Lagrangian (7.1) is extended to
include a vector boson A with field strength F :D @ A @ A (for instance, the
electromagnetic field),
L D 14 F F D .D /" V. / ; (7.5)
where D D @ C ieA is the covariant derivative with Abelian coupling (electric

charge) e. The system is invariant under a local U.1/ gauge transformation
0
.x/ ! .x/ D ei˛.x/ .x/ ; A .x/ ! A0 .x/ D A .x/ e1 @ ˛.x/ ; (7.6)
where now ˛ is a spacetime-dependent scalar. The vacuum structure (7.2) is

unchanged, since A D 0 in a Lorentz-invariant vacuum. The field decomposition
leading to (7.4) follows the same steps as above except for two differences: the
gauge vector acquires a mass m2A D .ev/2 =2 and the Nambu–Goldstone boson
is not an independent degree of freedom and can be removed by a suitable gauge
transformation (7.6) (unitary gauge). Then, (7.4) is replaced by
L D 14 F F C 12 m2A A A 12 .@/2 12 m2 2 W.; A/ ; (7.7)
up to some interactions W which we do not write here. The generation of a mass for
a gauge boson through spontaneous symmetry breaking is called Higgs mechanism.
Generalizing to non-Abelian groups, the same mechanism applies to the electroweak
sector, where D h is the Higgs boson with mass (5.203). The resulting Lagrangian
is (5.202) with g D .
We may ask now whether the Higgs particle accounts for the observed cosmo-
logical constant [1, 10, 11]. In the semi-classical sense of (2.112), the cosmological
constant is the contribution of vacuum quantum fluctuations of gravity and matter
fields. These fluctuations sum up to a term of the form (2.113), which adds to a
purely classical cosmological constant 0 to give the total vacuum energy
2 hi C 0
vac D hi C 0 D D: 2 D : (7.8)
2
The cosmological vacuum contribution of electroweak interactions is, to first
approximation, hi ' Vmin and the vacuum energy density today reads

.0/ 0 m2h v 2
vac ' C V 0 : (7.9)
2 8
The
p vacuum expectation value v can be determined from its relation v 2 D
1
. 2 GF / with the Fermi constant GF , which is measured from the muon lifetime
[12]: v 2 6:06 104 GeV2 . (From this and the LHC value mh 125 GeV, one
can get the quartic coupling D m2h =.2v 2 / 0:13, which is more difficult to detect
directly [13].) Assuming that V0 exactly cancels the classical cosmological constant
(or that 0 D 0 D V0 identically), one can estimate (7.9) as
.0/
jvac j ' 1:2 108 GeV4 ' 1068 m4Pl ; (7.10)
to be compared with the observed contribution
' 1048 GeV4 ' 10124 m4Pl : (7.11)

Recall that the Higgs cannot play the role of the inflaton as it stands, since its
quartic interaction is too strong. In Sect. 5.11.2, we saw that a non-minimal coupling
with gravity can lead to a viable scenario. However, the principal reason why the
Higgs physics may work for inflation but not for the cosmological constant problem
is that early- and late-time accelerating periods are characterized by very different
energy scales. Equation (7.11) is separated from the GUT scale typical of inflation
by about 12 – 16 orders of magnitude.
7.1.2 Zero-Point Energy and Higher Loops
The vacuum energy density (or zero-point energy) is the eigenvalue of the Hamil-
tonian when acting upon the physical vacuum state, i.e., the partition-function
contribution ZŒ0 of all bubble diagrams; the latter are Feynman diagrams with
no external legs [14]. One can try to fill the gap of about 56 orders of magnitude
between (7.10) and (7.11) by refining the argument leading to (7.9), valid only at
the tree level in perturbation theory. The first correction comes for one-loop bubble
diagrams which, in the absence of interactions, amount to one bubble diagram per
free field. In Sect. 2.5.3, we identified vac with in (2.123), implicitly regarding
as the sum of a classical part 0 plus the vacuum contribution. Here we reconsider
that calculation with more details.
For a free field of mass mi , the zero-point energy is an infinite superposition
q of
harmonic oscillators, given by an integral over all frequencies !.p/ D jpj2 C m2i .
In four dimensions,
Z q Z q
.1/ Ni C1
d3 p Ni C1
vac;i D jpj2 C m2i D dp p2 p2 C m2i ; (7.12)
2 1 .2/3 .2/2 0
where p D jpj and jNi j is the number of one-particle states, positive for bosonic
fields (real scalar: N D 1; complex scalar and real massless vector: N D 2; real
massive vector: N D 3) and negative for fermions (Majorana spinor: N D 2, Dirac
massive spinor: N D 4). Equation (7.12) can be easily checked for a massive
scalar field O on Minkowski background by using the expansion in creation and
annihilation operators of Sect. 5.6.2 for the Bunch–Davies vacuum (5.114),
Z
O .t; x/ D 1 d3 p
p ap ei!.p/ tCipx C a"p ei!.p/ tipx : (7.13)
.2/3 2!.p/
Taking the vacuum expectation value of the energy-density (5.23), 2 D .@t O /2 C

@˛ O @˛ O C m2 O 2 , one obtains (7.12) with N D 1. The check for other fields is
similar [14].
.1/
The zero-point energy vac;i is divergent. To get a finite expression, one can
introduce an ultraviolet momentum cut-off pmax much larger than mi . For each
species, we can then evaluate the integral (7.12) (with domain .0; pmax /) exactly by
an analytic continuation to D dimensions, then expanding for large pmax and setting
D D 4. To keep the expression for the energy density dimensionally correct in
D D 4, we also multiply by a mass scale M. For each species,
Z
M 4D pmax
dD1 p p 2
jpj C m2
2 pmax .2/
D1
Z Z pmax p
M 4D dp
D d˝D2 p D2
p 2 C m2
2 0 .2/D1
1D Z p
M 4D .4/ 2 max p
D D1 dp pD2 p2 C m2
2 0
1D
M 4D .4/ 2 1 D1 DC1 p2max
D mpmax F ;
D1
I I 2 (7.14)
.D 1/ D1 2
2 2 2 m
p 2
DD4 1 4 2 2 m4 em 1
D p C p m C ln C O ; (7.15)
16 2 max max
4 4p2max pmax
where we integrated the solid angle in spherical coordinates to get the area of the
hypersphere SD2 , ˝D2 D 2 .D1/=2 = Œ.D 1/=2, and F (also denoted as 2 F1 )
is the hypergeometric function. Summing over all species i and neglecting all mass
contributions, the result is [15]
X .1/
.1/
vac D vac;i O.1/p4max : (7.16)
i
For a Planck-scale cut-off (2.124),

.1/
vac 1076 GeV4 m4Pl ; (7.17)
an abyss of 124 orders with respect to the observed value (7.11), much worse than
the tree-level estimate (7.10).
What went amiss? First of all, in (7.15) the cut-off pmax has been regarded as
physical, but it is easy to see that the first two terms in the expansion break Lorentz
invariance. In fact, computing the expectation value of the pressure (5.24), one
verifies that hPi ¤ hi. It is then more natural to employ a regularization scheme
preserving Lorentz invariance. One such scheme is dimensional regularization.
Instead of setting D D 4 in (7.14), we first send the cut-off to infinity and then
expand with respect to the parameter D D 4 1:

1D
M 4D .4/ 2 1 D1 DC1 p2max
lim mpmax F ;
D1
I I 2
pmax !C1 .D 1/ D1
2
2 2 2 m

M 4 D2 m D
D D
2.4/ 2 M
2
m4 2 3 m
D 2
EM ln.4/ C ln C O./ ; (7.18)
.8/ 2 M2
where EM 0:577 (Sect. 3.2.3). In the “non-minimal subtraction” renormalization

scheme, one adds to this expression some counter-terms exactly canceling the
divergent term 1= as well as, for convenience, mass-independent contributions
inside the square brackets. Thus, summing over all species, the regularized and
renormalized vacuum energy reads [16]
X Ni m4 m2
.1/
vac D i
2
ln i
2
: (7.19)
i
.8/ M
In contrast with (7.15), this expression is cut-off independent and its size is
determined both by the mass of the particles and by the renormalization scale M.
The latter may be chosen as the average between the graviton energy H0 and
the energy of the photons coming from supernovæ, M 1044 mPl . Taking the
massive particle content of the minimal Standard Model after electroweak symmetry
breaking, one obtains an estimate of the zero-point vacuum density [16]:
.1/
jvac j 2 108 GeV4 1068 m4Pl ; (7.20)
of the same order of magnitude as (7.10). One might hope that the contribution of
field interactions would ameliorate the situation, but this is not the case. Using the
non-perturbative method of the Gaussian effective potential [17–19], one can show
that interactions leave the mass dependence of (7.19) substantially unchanged [14]:
X Ni m4 1 2
mi
vac
int
D i
2 2
ln 2
:
i
.8/ M
As a last stand, we could take the tidal forces of gravity into account: the effect is
negligible [14, 20]. One can see this also by a heuristic argument. Including gravity
and assuming that contributions up to one loop are exactly canceled by 0 , two-loop
diagrams are the next. One can consider a bath of virtual pairs of massive particles–
anti-particles with Compton wave-length D „=m with uniform number-density
distribution n./ / 3 . The gravitational interaction of a pair of such particles is

V./ Gm2 =, yielding an energy density [21]
m6
vac V./ n./ ; (7.21)
m2Pl
not an improvement. Thus, ¤ vac in the non-supersymmetric Standard Model.
7.1.3 Supersymmetry and Supergravity
Supersymmetry itself, rather than loop corrections in the Standard Model, may
lower the hierarchy gap to acceptable values. Here we pick up the thread left at
Sect. 5.12, where we introduced the superpotential W D W. i / and the Kähler
"
potential K D K. i ; i / for the fields i acting as coordinates on a Kähler
manifold.
As a matter of fact, the simplest implementation of supersymmetry overshoots
the target. Thanks to the exact cancellation of vacuum diagrams order by order,
the zero-point energy-momentum density is zero [22]. In models on Minkowski
spacetime, global supersymmetry is realized by values of the fields where the
superpotential is stationary, @i W D 0, so that
hi D Vmin D 0 : (7.22)
One reaches the same conclusion by noting that the zero-point vacuum energy (7.16)
with (7.12) is predicted to be exactly zero when combining the components of super-
multiplets, which all share the same mass. ForP the Wess–Zumino multiplet (5.212)
(two real scalars and one Majorana fermion), i Ni D 1 C 1 2 D 0. A massive
vector multiplet
P has a real massive vector, a real scalar and two Majorana spinors,
giving again i Ni D 3 C 1 2 2 D 0; and so on.
Getting zero instead of something does not solve the problem but sets the
stage for an interesting line of attack. Global supersymmetry is broken at some
scale (2.125) above TeV and the magical cancellations leading to (5.211) and (7.22)
.1/
no longer take place. Equation (7.16) with pmax D MSUSY yields vac & 1064 mPl .
According to the crude estimate (7.21), loop corrections with broken supersymmetry
slightly reduce the vacuum contribution to
vac 1096 m4Pl : (7.23)
This lowers the gap with the value (7.11) down to 30 orders of magnitude, still a
serious fine tuning.
In supergravity, the vacuum structure changes. Supersymmetry becomes local
and is encoded in the condition (5.223). Now the degeneracy of the gauge-invariant
globally supersymmetric vacuum (7.22) is broken by gravity and the minima of the
potential are negative semi-definite:
Vmin D 3 4 jWj2 eK 6 0 : (7.24)
All these vacuum solutions are stable but, in general, only one of them corresponds
to the Minkowski-spacetime vacuum (7.22), provided we fine tune W so that it
vanishes for such a solution. This special vacuum, however, is not the state with
lowest energy [23].
Conversely, one can find non-supersymmetric solutions (Di W ¤ 0) such that
V D 0, but these solutions will not be stable in general. Still, there exist special
classes of Kähler potentials giving rise to positive-definite potentials with a non-
supersymmetric minimum Vmin D 0 without fine tuning. These are the no-scale
models introduced in Sect. 5.12.4 [24]. The matrices G and f in (5.217) and (5.232)
determine the kinetic terms in the Lagrangian density (5.220a) and are therefore
positive definite. This implies that the potential (5.231) is positive semi-definite and
has an absolute minimum for any field configuration such that
VD0 , W1 D const ; Ds W D 0 : (7.25)
This vacuum state fixes the values of the fields ˚ c and Ss but not of T. The
name of these models stems from their featuring only mPl as a mass scale, while
the hierarchy of all the other masses is determined dynamically from the Planck
mass by spontaneous symmetry breaking. If W2 and K did not depend on all
scalars Ss , then for some value sQ of the index s one would have 0 D DQs W D
@W=@SQs C W@K=@S Q Qs Q
D W@K=@S Qs
, which would imply that W D 0. Then, the
supersymmetry condition (5.223) would be trivially obeyed for all i in the vacuum
state (7.25). On the other hand, assuming that the superpotential and the Kähler
potential depend on all scalars Ss , one has W ¤ 0 and the vacuum (7.25) does
break supersymmetry, since for indices c (5.223) is violated: Dc W D W@K=@˚ c D
3WjT C T " hj1 @h=@˚ c ¤ 0.
To summarize, no-scale models predict a non-supersymmetric configuration with
vanishing vacuum energy. The cosmological constant problem is not solved, first
of all because one needs to generate a mechanism to lift the minimum of the
potential to some non-zero value compatible with (7.11). Secondly, one is simply
assuming (5.230) with exactly those coefficients and field dependencies. Not only
are these structures chosen ad hoc at the classical level, but they are not expected to
survive when loop corrections are taken into account. We will come back to these
issues in Chap. 12, where we will see how no-scale models of the form (5.230) and
related lifting mechanisms can be naturally generated in string theory (Sect. 12.3.5).
7.2 Other Versions of the Problem and Strategies
Tree- and loop-level calculations in the minimal Standard Model of particles and
inclusion of supersymmetry confirm the existence of a hierarchy or fine-tuning
problem. This is sometimes called the “old” cosmological constant problem, to
distinguish it from the “new” problem asking why is about the same order of
magnitude as the total matter density m (Sect. 2.6). To these issues, one should add
the coincidence problem as well as the following observations.
7.2.1 Broken Symmetries
Regardless of its value, by itself the cosmological constant is a bizarre object from
the perspective of general relativity as a field theory [24]. Normally, in a field theory
invariant under a group G one can find vacuum solutions symmetric with respect
to some sub-group G0 2 G without fine tuning the parameters. Examples are
provided by the U.1/-invariant vacuum expectation values of doublet scalars in the
SU.2/ ˝ U.1/ electroweak model, or by supergravity models with partial breaking
of supersymmetry.
In the case of plain, non-supersymmetric classical gravity, G includes the
diffeomorphism group of general coordinate transformations. Fixing a background
solution g D gQ of the Einstein equations (2.23) always breaks diffeomorphism
invariance, but some residual symmetry may be left. When D 0, the Minkowski
metric g D is a vacuum solution of the Einstein equations (2.23), R
g R=2 D 0. This vacuum solution carries with it the symmetries of the global
Fermi frame, the Poincaré group of proper Lorentz transformations and translations.
The presence of a non-vanishing cosmological constant forbids such a vacuum
solution (the solution now is de Sitter or anti-de Sitter spacetime, depending on the
sign of ) and, in particular, breaks invariance under spatial and time translations.
Solutions preserving translations would then require a fine tuning on . To put it
differently, one may look for translation-invariant solutions where the gravitational
and all matter fields n are constant. These would obey the equations of motion
p
@L=@ n D 0 and @. g L/=@g D 0 but the latter is incompatible with the
solution L D D = 2 , unless the constant is exactly set to zero.
In yet other words, the cosmological constant is related to a symmetry of the
matter sector (e.g., [25, 26]): in the absence of gravity, a constant shift 0 D const
of the matter Lagrangian
Lm ! Lm C 0 ) T ! T C 0 ı (7.26)
would leave the equations of motion r T D 0 invariant. This is not the case when
a dynamical gravitational field with a non-trivial action is introduced. In particular,
the Einstein’s equations (2.23) are not invariant under (7.26) (the cosmological
7.2 Other Versions of the Problem and Strategies 311
constant is shifted by ! 2 0 ) and we cannot impose (7.26) as a symmetry

of the system. At the level of particle physics, this restriction has no natural
explanation.
7.2.2 The 4 Puzzle
Perhaps no other great question in theoretical physics admits as many formulations

as the cosmological constant problem. As a final piece of information, we recast the
problem in terms of an astounding numerical coincidence purely based on classical
cosmology [27–29].
Consider the Hubble volume in N D 1 gauge
4 3 4
VH D R D : (7.27)
3 H 3H 3
To measure the duration of a cosmological era marked by some initial and final time
t1;2 (which is which depends on the sign of cosmic acceleration), we note that the
number of comoving modes in the comoving volume Vcom D VH =a3 crossing the
Hubble horizon with wave-numbers a.t1 /H.t1 / D k1 6 k 6 a.t2 /H.t2 / D k2 is
Z k2
d3 k 2 rH;2
N12 D Vcom D ln ; (7.28)
k1 .2/3 3 rH;1
which is nothing but the improved number of e-foldings (5.12) up to a numerical

prefactor. Consider now the cosmic history from inflation until today. The evolution
of the proper Hubble horizon is schematically depicted in Fig. 7.1. During both
inflation and late-time acceleration, H is approximately constant. The end of
inflation and the beginning of radiation and dust domination is marked by point
X, while the beginning of the dark-energy era is at point Y. Modes with k < k and
k > kC exit the horizon only once, respectively at point P and Q, and never re-enter.
Modes with k < k < kC exit twice, at A and C, re-entering at some point B during
radiation or dust domination. We are interested in counting the modes k 2 Œk ; kC
which re-entered the Hubble patch during the radiation-matter era XY. We call this
number, characteristic of our universe, Nc :D NXY . It can be calculated from the
condition dk =da D d.aH/=dajY D 0 (the tangent to the curve at Y has unit slope)
and the Friedmann equation (2.81) in the absence of curvature (K D 0) and with
radiation, dust matter and a cosmological constant . Denoting the scale factor at
Y normalized at radiation-matter equality as ˛ D aY =aeq , the slope condition at Y
states that ˛ D ˛./ is solution to the quartic equation 2 4 ˛ 4 ˛ 2 D 0, where
4 :D =eq . Using the Friedmann equation, the result is

1 reh
Nc D ln c./ ; (7.29)
6
ln L
k− k k+
−1 Y C Q
HΛ
−1 P A X
Hinf
LP
aB aY ln a
a− aP a+ aA aX aC aQ
radiation
+ matter late time
inflation
acceleration
dominated
Fig. 7.1 Schematic history of the universe from inflation to late-time acceleration. Here LP D lPl
2
is the Planck length, Hinf D He is the Hubble parameter at the end of inflation and H D =3.
The rest of the symbols are explained in the text (Source: [27], ©2014 World Scientific)
where c./ D 2Œ˛./ C 2=Œ3˛./ C 42 and reh is the energy density at the
end of reheating (for instantaneous reheating, reh D e ). With 2:6 103
(Sect. 2.5.3), one has c 103 . Assuming the energy at the end of inflation
1=4
around the GUT scale, reh 1015 GeV (Sect. 5.9.1), we also have reh =
108
10 . Combining everything and taking into account experimental uncertainties,
we obtain
Nc D 4 (7.30)
up to about one part over 103 . The result is robust with respect to changes of reh
of about one order of magnitude, which is roughly the energy excursion between
instantaneous reheating and less efficient scenarios.
Such a precise determination of this number has no explanation in standard
cosmology: there is no apparent reason why the duration of the radiation-dust era
should be of exactly 4 e-foldings! Barring a fortuitous coincidence or exotic
reheating scenarios, one may wonder whether a yet undiscovered fundamental
principle is being enforced. According to this principle, the cosmological constant
7.2 Other Versions of the Problem and Strategies 313
is completely determined by the thermal properties of the universe at the end of

inflation and at radiation-matter equality, as well as by the duration Nc of the
subsequent thermal history:
D 2 reh c./ e6Nc : (7.31)
It is plausible that, in any theory explaining the relations (7.30) and (7.31),
gravitational and matter degrees of freedom should emerge in a unified way.
7.2.3 UV or IR Problem? Strategies for a Solution
The introduction of supersymmetry in Sect. 7.1.3 attempted to address the cos-

mological constant problem in two steps: first, by modifying the dynamics of the
system, from the Einstein–Hilbert action to the SUGRA action. And, second, by
interpreting as the vacuum energy of the quantum theory. This is an instance
where the cosmological constant problem is regarded as due to some unknown
physics above a certain energy scale (in this case, the scale MSUSY of supersymmetry
breaking). However, there is also a possibility that the root of the problem essentially
lies in the infrared rather than in the ultraviolet. For example, if we do not want
to abandon local perturbative quantum field theory, we can revisit the untested
assumption that it is valid at arbitrary large scales and introduce an IR cut-off of
order of the Hubble horizon. This act has its inspiration in the thermodynamics
of black holes, where the Bekenstein–Hawking entropy-area law [30–33] indicates
that, in the presence of a black hole, ordinary quantum field theory overcounts the
number of degrees of freedom within a given volume and ceases to be valid at large
scales [34] (we will see more of this in Sect. 7.7.3). Similarly, in the presence of a
cosmological horizon it turns out that a UV cut-off must be related to an IR cut-
off in such a way that the fine-tuning problem of Sect. 7.1 disappears, even if
maintains its interpretation as the vacuum energy of a quantum field theory [35].
However, on one hand the IR cut-off is not sufficient to explain the observed
value of and, on the other hand, some UV modification of physics (quantum
gravity?) may still be needed to explain the fundamentals of the overcounting of
degrees of freedom. In this and many other cases, it is therefore difficult, or perhaps
artificial, to reduce solutions of the cosmological constant problem to purely UV or
purely IR effects. This will become most apparent when surveying the landscape of
explanations offered in this chapter and in Chaps. 11 and 13.
Having seen that the cosmological constant cannot be easily interpreted as
the vacuum of a particle field theory, we should look towards other directions.
Intuition, prejudice or experience may lead to very different approaches, sometimes
rediscovered or reformulated in independent frameworks. We limit the discussion to
some broad classes of proposals.
(A) Without abandoning the particle-physics perspective, the cosmological con-

stant can still be the vacuum energy of an effective field theory but in scenarios
more sophisticated that those outlined in Sects. 7.1.1, 7.1.2 and 7.1.3, as, for
instance, in string theory.
(B) The energy-density contribution attributed to the cosmological constant and
the observed acceleration come from dynamical matter fields (Sect. 7.3). Their
full dynamics, not just their vacuum state, is tuned via the free parameters of
the model so that to reproduce the observed energy density and equation of
state. This dynamical approach does not explain why the number in (7.11)
is realized in Nature, but it attempts to combine a reasonable theoretical
framework with a modest tuning of the free parameters of the model. It also
aims to alleviate another aspect of the cosmological problem when is exactly
constant, the coincidence problem discussed in Sect. 2.5.3. Extensions to many-
field configurations are also envisaged.
(C) The culprit is neither the vacuum nor the dynamics of a matter field but
geometry itself (Sects. 7.4 and 7.5.2). This can be made possible if we modify
the Einstein–Hilbert action (2.18) by changing the coupling between gravity
and matter or by including higher-order curvature invariants, or both. The goal
is the same as in class (B), namely, to find a region in the parameter space which
accommodates all observations related to the cosmological constant without
fine tuning.
(D) More complicated matter dynamics, spacetime backgrounds, or deeper mod-
ifications of geometry going beyond the inclusion of higher-derivative terms
may give rise to scenarios which combine some of the features of the previous
classes of models (Sect. 7.6). This general case also includes analogue-gravity
(Sect. 7.6.4) or emergent-gravity scenarios (Sect. 7.7).
(E) A yet undiscovered fundamental principle is in action.
Other classifications are possible [36, 37] but the present one covers virtually all
models where the measured cosmological constant is effectively determined by the
dynamics of geometric and matter degrees of freedom. It was recognized rather
early that matter fields produce a time-dependent contribution tuning with , both
at the classical and at the quantum level [38–40], with particular emphasis on non-
minimally coupled scalars [41–55]. In all these cases, effectively decays in time
from possibly large values, but there are other types of predictions. In one class
of models, the cosmological constant is identically zero, which can be obtained
by requiring the vanishing of graviton and scalar-particle creation in radiation-
dominated universe [56], via Kaluza–Klein compactification [57], via spontaneous
symmetry breaking in modified gravity with non-minimal coupling [58] and in
some quantum-gravity models (Sect. 10.2.4). Still in the same category, we may
include the possibility where is not zero but so strongly suppressed that it vanishes
for all practical purposes (a Euclidean quantum-gravity example is [59]). With the
discovery in the late 1990s of the acceleration of the present universe, models with
exactly D 0 turned out to be, at best, incomplete.
7.3 Quintessence 315
In general, all the above strategies and their specific incarnations can suffer from
two drawbacks when cosmological models are formulated. First, they involve a
constraint on the parameter space of the model, not a prediction of the value (7.11).
Therefore, none of the solutions of the cosmological constant problem falling in the
above categories imposes itself as the solution to the problem. Second, sometimes
they lack a solid theoretical background explaining from first principles why an
action should have such and such form with such and such free parameters. In a
way, these questions are akin to the one we asked for the inflaton: Where does
come from? This issue may be tackled by facing the problem in a candidate
fundamental theory of Nature, such as string theory or one of the many quantum-
gravity approaches. This will be the subject of Sects. 9.4, 10.2.4 and 13.1.2, with
some interesting twist also regarding the issue of the value.
It might also be the case that asking for a unique and unequivocal prediction of
the value (7.11) is an unreasonable request, even in a fundamental theory. When-
ever quantum mechanics or chaos make their appearance, exactly deterministic
predictions become nonsensical and one is entitled only to talk about probabilities.
Canonical quantum cosmology (Sects. 9.4 and 10.2.4) and the string landscape
(Sect. 13.1.2) are precisely examples of that.1 To some, this attitude might look like
an admission of defeat, considering that it almost waves away an unknown that
constitutes 70 % of the observed universe. However, one cannot rule it out only on
the grounds of subjective taste.
On a positive note, there are at least two theories (asymptotic safety, Sect. 11.2;
causal sets, Sect. 11.6) offering a prediction for in agreement with (7.11), while
another (emergent gravity, Sect. 7.7) has the ambition to address the 4 puzzle by
first principles.
7.3 Quintessence
Inspired by the concept of dynamical vacuum in quantum field theory, it is tempting

to consider scenarios with a varying cosmological constant, replacing ! .t/
with no kinetic term [61–66] (a number of time profiles are catalogued in [36]).
The presence of derivatives for in the action, making it a dynamical field, gives
rise to a more flexible dynamics. After early applications with a 3-form [39] and a
non-minimally coupled scalar [41, 42, 50, 51], the attention moved to a minimally
coupled scalar field weakly coupled with the rest of matter [67–81]. As it constitutes
a fifth contribution to the energy density of the universe apart from baryons,
neutrinos, radiation and dark matter, this scalar has been dubbed quintessence
1
An early attempt to explain the cosmological constant in perturbative quantum gravity is [60],
where the bare is compensated by one-loop terms. This case is still of deterministic type, the
cosmological constant being driven to zero by quantum effects.
[71]. Due to inhomogeneous fluctuations, quintessence clusters gravitationally (e.g.,

[82]). This can leave an imprint both on the CMB spectrum and on the evolution of
large-scale structures, which allows one to place constraints via independent data
sets.
A major assumption in quintessence models is that the cosmological constant
is set exactly to zero and that only dynamics contributes to acceleration. In this
respect, the cosmological constant problem in its earliest version is simply assumed
to be solved, as the mechanism leading to D 0 is left unspecified.
7.3.1 Tracking, Freezing and Thawing
The general-relativity action (2.20) with D 0 is assumed. The energy density

and pressure of quintessence are given by (2.79), while the scalar equation of
motion, the continuity equation for matter and the first Friedmann equation are,
respectively, (5.39) also written as
P C 3H.1 C w / D 0 ; (7.32)
and
2
P C 3H.1 C w/ D 0 ; H2 D . C / ; (7.33)
3
where D m C r is the contribution of matter and radiation with effective
barotropic index w D P=. The energy density of quintessence can be integrated
for any potential,
Z
a
da0
.a/ D .Na/ exp 3 .1 C w / ;
aN a0
where aN is some reference scale factor. In the range (5.22), for w approximately
constant one has
a
.a/ ' .Na/ ; 0 < D 3.1 C w / < 6 : (7.34)
aN
Solutions of this form with constant are called scaling [76, 77, 79, 83]. Since
P D 3H P 2 , scaling solutions are such that the kinetic energy of the scalar is a
fixed fraction of the total energy, P 2 = D =3. In the extreme slow-roll regime,
0, while during kination 6.
Scaling behaviour occurs as a response to one or more cosmological fluids
[67, 68]. For a viable evolution, the energy density of the field should be tuned
to decrease at a lower rate than matter (m a3 ) and radiation (r a4 ) in order
to dominate at late times. To classify all possible cases, it is convenient to assume a

constant barotropic index w for the fluid and consider the ratio

/ a3.ww / : (7.35)

Tracking solutions are those with w D w . In this case, quintessence “tracks” in

parallel the evolution of matter or radiation energy density, at least for part of the
cosmic history.
Scaling profiles alleviate the coincidence problem since they are cosmological
attractors: the evolution of the universe is fairly insensitive to the initial conditions.
The problem may be claimed to be solved only if no other fine tuning, for instance
on the mass scale of the potential, is introduced.
If w 6 w, the scaling condition (7.34) occurs when [79]
V V; V
:D D > 1; w 6 w; (7.36)
2V V;2
where V and V are defined in (5.55) and the equality holds for tracking solutions.
In fact, defining X :D .1 C w /=.1 w / D P 2 =.2V/ and X . p/ :D dp ln X=.d ln a/p ,
one sees that

1 X .1/ 4X .2/
D1C w w .1 C w 2w / ;
2.1 C w / 6 C X .1/ Œ6 C X .1/ 2
(7.37)
and the scaling condition is nothing but w const (hence X const, const).
The constancy of w in the scaling regime determines the value of w via (7.37).
Dropping almost vanishing terms in X .1/ and X .2/ , one finds w Œw 2.
1/=.2 1/, valid in the deep radiation- or dust-domination regime. Setting w D 0
does not give the precise value of w today, as the approximation w const is too
coarse to characterize the interpolating solution at late times, since scaling solutions
are not, in general, exact solutions of the equations of motion. To get the correct w ,
one can either integrate the equations of motion numerically for a given potential or
perturb the tracker solution to get semi-analytic formulæ for w . / [84, 85] which
permit to find model-independent constraints on the barotropic index w0 of . In
particular, for scaling solutions 1:19 < w0 D w jtoday < 0:95 [86].
The scaling condition can also be re-expressed in terms of the density parameter
˝ (2.83). Noting that P 2 =H 2 D .3= 2 /˝ .1 C w /, using (7.32) and the definition
of X the following identity holds:
s
V; 2 .1 C w / X .1/
D ˙3 1C ; (7.38)
V 3˝ 6
where the ˙ sign holds when, respectively, P 7 0 (these manipulations are valid
only in regimes where pis monotonic). Away from the extreme slow-roll regime
w 1, jV; =Vj / 1= ˝ . The density parameter ˝ increases in time if w <
w (quintessence dominates at late times), so that the left-hand side must decrease
if one wants to avoid a fine tuning of the initial conditions. In fact, if the left- and
right-hand side of (7.38) had opposite evolution, there would be a time after which,
due to strong cosmological friction, the field freezes at some constant value D f
and behaves just like a cosmological constant. This creeping or freezing regime
[80, 87, 88] dominates the late expansion of the universe. At that point, however,
one would have to tune the initial conditions such that 2 V. f / acquires the observed
value H02 , which would lead us back to the problem. To avoid this situation, we
require that jV; =Vj decreases in time. Since .V; =V/; D .V; =V/2 . 1/, this
condition is satisfied when > 1. When w > w, the scaling condition is 1 .1
w/=Œ2.3 C w/ < < 1 [77, 79].
Realistic solutions should admit time-varying equations of state for quintessence
and matter and, hence, different scaling regimes at different epochs (Problems 7.1
and 7.2 and Figs. 7.2 and 7.3). The late acceleration of the universe is observed,
for instance, via standard candles such as type I supernovæ, objects with a
known luminosity-distance relation. Their proper distance is determined by (2.143)
and (2.156) as a function of the redshift or, equivalently, of the scale factor.
It is therefore useful to parametrize the barotropic index as a function of a.
Equation (7.38) can be recast as
r
dw wP 3˝ p V;
D D 3.1 w2 / ˙ 2
.1 w / 1 C w : (7.39)
d ln a H V
For a given potential, (7.39) evaluated today and (2.120) allow one to plot a
theoretical point in the .w0 ; wa / plane and to compare it with observations. Other
parametrizations than (2.120) are possible [84, 85, 88–91]. One which has attracted
much attention is related to the so-called generalized Chaplygin gas, i.e., any
cosmological fluid obeying and equation of state of the form P D w0 =ˇ , where
ˇ ¤ 1 [92–96]. The special case ˇ D 1 [92, 93] is the original model by
Chaplygin [97]. This class of equations of state can be realized, for a specific
reconstructed potential, by a homogeneous scalar field [92] or by a complex scalar
[93, 94]. Replacing P in the continuity equation (7.32), one obtains .a/ and hence
ˇC1
w .a/ D P .a/= .a/ D w0 = .a/ D w0 Œ.1 C w0 /a3.1Cˇ/ w0 1 [94].
A scaling field slows down near the present as it starts to feel the friction induced
by its own dominance of the energy density, hence its barotropic index decreases
with time (wP < 0). In this setting, quintessence goes through a creeping regime
where w 1 before entering the scaling behaviour [79]. The duration and end of
the freezing regime depend on the parameters of the model. Whenever the scaling
regime is reached after the present epoch, some fine tuning is required, albeit less
severe than for a pure cosmological constant.
Fig. 7.2 (a) Quintessence (solid curve), radiation (dotted curve) and matter (dashed curve) energy
density as a function of redshift for a potential V D M 4Cn n with n D 6, M 5 106 GeV,
for a scaling overshooting solution with initial conditions i D .z D 1012 / D 3:5 106 mPl and
0 3
i D 3:5 10 mPl , where 0 D @N . In the dimensionless variables of Problems 7.1 and 7.2, the
mass and initial conditions read A 60, yi D 105 and y0i D 102 . (b) Quintessence barotropic
index w for the n D 6 model (solid curve) compared with the output for the n D 1 (dotted curve,
M 2 keV, A 0:38) and n D 2 (dashed curve, M 27 MeV, A 0:54) models with same
initial conditions. Today, the barotropic index is w .z D 0; n D 1/ 0:76, w .z D 0; n D
2/ 0:63, w .z D 0; n D 6/ 0:40
Equation (7.39) predicts that the closer w to 1, the closer is wP to zero and the
more indistinguishable is quintessence from a pure cosmological constant. Accord-
ing to observations, the present universe is under strong acceleration (w very close
to 1). Naively, one might try to explain this fact by requiring quintessence to roll
slowly at late times and to obey the SR conditions
V 1 ; jV j 1 : (7.40)
(Incidentally, this would guarantee the stability of the solution according to the
inflationary analysis in Sect. 5.4.3.) The dynamics of quintessence, however, is more
Fig. 7.3 Quintessence barotropic index w for the inverse exponential model (7.49) (dotted curve,
V0 D 0:28crit;0 , yi D 102 ) and the SUGRA-inspired model (7.50) with n D 1 (dashed curve,
M 2 keV, yi D 104 ) and n D 11 (solid curve, M 6 1010 GeV, yi D 103 ), all with
y0i D 102 (in the dimensionless variables of Problems 7.1 and 7.2). Today, the barotropic index
is, respectively w .z D 0/ 0:85; 0:92; 0:82
difficult to assess that of inflation, since it is not potential dominated and (7.40) does
not imply the strong SR approximation (5.19). The SR conditions (7.40) should be
applied with care. When they hold, w . 0:99 [88, 89]. Moreover, while the
inflaton is the only content of the early universe, dark energy is the main component
in the present time but it does not dominate the other fluids completely. These two
factors make a numerical analysis an indispensable tool.
The problem is that solutions with scaling regimes typically predict a negative
barotropic index today, but not too close to 1 for sensible values of the parameters
in the potential or for the simplest potentials. On the other hand, one can get
closer to 1 by allowing a rapid variation of the barotropic index at low redshifts,
which is in disagreement with present observations. Therefore, the simplest scaling
quintessence scenarios are not satisfactory solutions to the problem because an
effective barotropic index w 1 implies a fine tuning in the models.
To overcome these difficulties, one may look for attractor solutions indistinguish-
able from a cosmological constant, i.e., field models such that w 1 at late times
for a modest tuning of the initial conditions. A creeping field is one which sits static
at low energy density until the density of other components drops low enough for
it to become dynamical and, then, starts to move near the present epoch (“thawing
out” [87–89]). When this happens, the barotropic index of the creeper increases
away from 1 (wP > 0). Potentials with tracking regimes always feature a freezing
era and, depending on its onset and ending, it can lead to different observable
predictions for the same potential. Thus, creeping solutions can be found both for
potentials which do not support observable tracking and in scaling models where
the field is frozen until it reaches the attractor in our recent past. In all these cases,
some fine tuning persists although less severe than for a pure cosmological constant.
Thawing solutions, which can be found for most of the potentials below [84, 89], do
not fare better, since the field undergoes a SR phase (7.40) which does simplify the
dynamics but is set by special initial conditions.
7.3.2 Periodic and Power-Law Potentials
This tension can be appreciated by looking at possible candidates for the

quintessence potential. In general, the fine tuning manifests itself either in the initial
conditions of models with potentials more or less motivated by particle physics or
in the specific choice of purely phenomenological potentials. The difference in the
energy scale of the problem makes viable inflaton models unsuitable as dark energy,
inasmuch as their parameters require severe fine tuning [78].
For instance, a simple quadratic potential under thepSR conditions
p (7.40) pis unfit,
since the field would be practically massless, m D V; . 2 V 3H0
1033 eV. The field can acquire a large mass from perturbative radiative corrections
unless some symmetry prevents it (“symmetry protection”). Such is the case for
an ultra-light pseudo-Nambu–Goldstone boson with potential (5.90) [69, 98–102].
The effective mass is then m 2 =f and, setting f mPl , one would need the
rather reasonable value 103 eV. Near the local maximum at D 0, the field
rolls down slowly, an example of “hilltop” dynamics [103]. From (5.91b), it is clear
that the second SR condition in (7.40) is not satisfied unless f & mPl ; this does
not necessarily signal a trans-Planckian problem and viable late-time acceleration
can be obtained even when f . mPl with some theoretical justification [102]. The
potential (5.90) arises also for the axion in string and M-theory [104] (Sect. 13.4)
and in extended (N 2) gauged supergravity, with f of order of the Planck mass
[105].
Neither periodic nor quadratic potentials yield scaling solutions. The power
law (5.81) does give stable scaling profiles if n > 2.3 C w/=.1 w/ but, in this case,
w > w and there cannot be acceleration when the field rolls down the potential.
Still, in the oscillatory phase at the minimum at D 0 the virial theorem produces
a mean equation of state [106]. In fact, if the oscillation frequency is much larger
than the expansion of the universe H, for a potential V / n the average over the
oscillations of the kinetic energy K D P 2 =2 is n=2 times the average of V, so that
hK i hVi n2
hw i D D : (7.41)
hK i C hVi nC2
Consequently, the mean expansion rate of the quintessence energy density is /

a3.1Chw i/ D a6n=.nC2/ . Equation (7.41) is valid classically but oscillations are
bound to stop due to quantum particle production. The latter does not occur if one
assumes that quintessence couples very weakly with the rest of matter. If n 1,
one can reach a cosmological-constant regime while having a tracking attractor at
early times. Unfortunately, in order to have =m D O.1/ at present times one
should tune this ratio initially to the same levels as (7.11).
7.3.3 Exponential and Hyperbolic Potentials
Take also the exponential potential [51, 68, 72, 73, 107] V D V0 e , where
solutions with ! are physically identical upon changing the sign of .
The usual inflationary attractor (2 < 3.1 C w/, w D 1 C 2 =3) cannot
work as dark energy without fine tuning, since in this case the energy density
would have a sizable contribution at nucleosynthesis, thus impeding regular
structure formation [72, 83]. On the other hand, there are tracking attractors for
2 > 3.1 C w/ [83]. Since w D w > 0 and V D V =2 D 2 =2, the resulting
universe does not accelerate in the infinite future but, as a transient regime, today it
does (w .z D 0/ 0:50) under some parameter tuning, although not enough to
account for the observed acceleration.
A much better scenario is achieved with a sum of exponentials [108–113]:
V. / D V0 e0 C V1 e1 ; (7.42)
where the sum can be extended to more terms. If 0 > 5:5 and 0 < 1 < 0:8,
the evolution of is such that the BBN bound is not violated, but observations on
late-time acceleration further require that 0 > 17 and 1 . 0:1 [86]. At early times
the field follows the tracking attractor, while at late times one hits the inflationary
attractor. Today, in the transition between the two regimes where we are approaching
the de Sitter phase but does not dominate yet, w can be close to 1 for a large
range of parameters. If 1 < 0, instead of rolling indefinitely the field falls into a
local minimum, where it acts as an effective cosmological constant.
Potentials of the type (7.42) arise in extended gauged supergravity in four
dimensions [114, 115] and in the dimensional reduction of D D 11 and D D 10
supergravity, which, in turn, can be regarded as the low-energy limit of M- and
string theory (Sect. 12.3.3). Dimensional reduction is the compactification of D-
dimensional spacetime as M D M 4 C, where the .D 4/-dimensional manifold
C is called internal space. The values of the coefficients Vi and i depend on the
geometry and topology of the internal space. Models based on spherical and toroidal
compactifications with static internal space [116–119] predict parameters that do not
lie in the range suitable for dark energy and the BBN bound is violated. Other types
of dimensional reductions are more successful, in primis flux compactifications
(Sect. 12.3.9).
Yet another phenomenological possibility is [120, 121]
V. / D U. / e ; (7.43)
where U is chosen so that the potential exhibits a local minimum. Examples are
U. / D V0 C V1 . 0 /˛ and U. / D V0 C V1 =ŒV2 C . 0 /2 , the latter
arising in string theory where governs the separation of branes via a Yukawa-like
interaction [122]. With natural choices of the parameters ˛ and Vi , the field may
remain trapped in the minimum, thus behaving as an effective at late times and
giving w 1 today, similarly to the model (7.42).
Another class of models, with hybrid properties between the exponential poten-
tial at early times and the power-law potential V / n at late times, is [123]
n2
2
V. / D V0 cosh 1 (7.44)
n
and other variations on the theme such as V. / / .cosh /ˇ ; .sinh /ˇ [124, 125]
and the cosh types arising in extended gauged supergravity [105, 114]. All these
cases avoid the fine tuning of a pure power law or exponential but reintroduce it
from the backdoor, since (7.44) is an ad hoc choice with some crucial assumptions
on the field dependence, the smallness of n and the absence of interactions with
matter.
7.3.4 Inverse Power-Law Potential
Apart from a positive power law and the exponentials, a potential with exact scaling
solutions is the inverse power law [67, 68]
M 4Cn
V. / D n
; M > 0; n > 0: (7.45)
Such potential may arise in dynamical supersymmetry breaking scenarios, as

in super-chromodynamics and its generalizations [126–130]. In that case, at the
classical level one realizes an effective action of the form (5.216) with a flat Kähler
"
potential K D and a superpotential W D M 2Cn=2 1n=2 , where n > 2
depends on the Dynkin indices of the gauge group. It is, however, difficult to
construct a realistic model with global supersymmetry. In fact, while curvature
corrections do not disturb the tracking evolution and radiative quantum corrections
to the superpotential cancel out, the Kähler potential is not protected by the same
"
symmetries and the flat potential K D receives low-energy corrections which
modify the kinetic term of the field. In general, these modifications spoil the scaling
behaviour and make (7.45) hard to embed in a realistic supersymmetric theory [130].
We temporarily ignore this issue and concentrate on the phenomenology.
The SR conditions (7.40) become V D n2 =.2 2 2 / 1 and V D n.n C
1/=. 2 2 / 1, obeyed if nmPl . For mPl , the mass scale in (7.45) is of
order
1
M 0:7 .101 H02 mnC2
Pl
/ nC4 : (7.46)
Using (2.11) and (2.12), for n D 2 one has M D O.10/ MeV, while M D
O.106 / GeV for n D 6. For the power-law expansion (2.93), the solution of (5.39) is
2Cn
1
a 2
2.6p C 3pn n/ 2 p.2Cn/
D t 2Cn D: 0 : (7.47)
M 4Cn n.2 C n/2 aN
The power-law expansion (2.93) is also the background solution of the continuity
equation in (7.33) for / a2=p and, approximately, of (7.33) when = 1.
Consequently, w D 1 C2n=Œ3p.2 Cn/ D 1 Cn.1 Cw/=.2 Cn/, the ratio (7.35)
= / a4=Œ p.2Cn/ increases in time and eventually quintessence dominates the
expansion of the universe. This scaling solution, with w > w and D 1C1=n > 1,
is stable [68, 77]. However, it requires some tuning (either very low initially or
n > 5) as, otherwise, today we would still be in the creeping regime, the scaling
attractor being in the future. Moreover, the barotropic index w in realistic scenarios
(Fig. 7.2) is somewhat smaller than the theoretical value w D 2=.2 C n/ for exact
scaling with dust but it is not very close to 1 for n > 1. Therefore, this model
is unsuitable to explain the present acceleration unless n < 1. In that case, the
observational bound is [86]
n < 0:075 : (7.48)
In Problems 7.1 and 7.2 we work out in detail the numerical solution for the
inverse power-law potential (7.45) with n D 6. This example is experimentally non-
viable, since the predicted barotropic index today is about w 0:40. Still, it
gives a fair idea of the phenomenology of quintessence (Fig. 7.2) and is a practical
example of how to integrate the cosmological equations of motion numerically.
7.3.5 Other Potentials
Phenomenological
P potentials with slightly better properties are the sum of inverse
powers, V D i Vi ni [79] or the inverse exponential [76, 79]
V. / D V0 e V1 ; (7.49)
where is a constant and V1 is either V0 or zero. For > 0, at early times ( 1)

the potential is very steep with large SR parameters 2V ' V 1=. /4 1,
which produces a tracking behaviour during radiation domination. At late times
( 1), (7.49) approaches the inverse power law (7.45) with n D 1 and 4V '
V 2=. /2 1. This model does not need fine-tuning of the initial conditions,
as the field can lie anywhere on the flat part of the potential. However, one has to
fix the value of V0 by hand in order to reproduce the observed dark energy density,
which of course does not explain the cosmological constant. Observationally it is
not favoured either [131], since w .z D 0/ 0:85 (Fig. 7.3). Here is another
negative example illustrating the difficulties of the problem.
Other potentials for quintessence come from supergravity. The fact that, in the
inverse power-law model (7.45), mPl at low redshift indicates that SUGRA
corrections are somewhat unavoidable. The resulting effective potential for the
scalar field is not positive definite [130, 132]: plugging the flat Kähler potential K D
2
(here is real) and the superpotential W D M 2Cn=2 1n=2 into (5.225) yields
2
V. / D e. / M 4Cn n Œ.n 2/2 =4 .n C 1/. /2 C . /4 . The potential (7.45)
is recovered in the limit ! 0. All SUGRA corrections, including the exponential
one, are relevant only at late times, when D O.1/, and they do not affect the
early history of the universe. However, the second term in square brackets is negative
definite and leads to an inconsistent late-time evolution. Various ad hoc choices of
the Kähler potential lead to an effective potential V of the form [130, 132–138]:
˛ . /ˇ
V. / D M 4˛ e : (7.50)
In one of these models, with a string-inspired non-flat Kähler potential, the

quintessence field is one of the compactification moduli (see Sect. 12.3.5) and the
parameters in (7.50) are D 1=2, ˇ D 2 and ˛ 6 11 in order to get a
realistic scale hierarchy [130, 132]. The barotropic index can be seen in Fig. 7.3;
today, w .z D 0/ 0:82. A lower w can be obtained for low n (for instance,
w .z D 0/ 0:92 for n D 1 and w .z D 0/ 0:88 for n D 2) but these cases
have no solid theoretical background.
An index closer to w .z D 0/ 1 can be obtained for other Kähler potentials
such that V acquires a non-zero local minimum. In one such case [133], on the side
of the minimum with 1 (increasing ), V 2=3 , while on the other side
(j j 1, decreasing ) V takes the asymptotic form (7.50) with D .3=2/ˇ ,
˛ D 2=3 and ˇ D 4=3, V 2=3 expŒ.3 =2/4=3 . p
Traditional supersymmetry breaking should happen at energy scales hFi
1010 GeV. In order not to increase the vacuum energy density to unacceptable
levels, the second contribution in (5.225) should compensate the F-term, W
hFi 1 m3=2 2 , where m3=2 is the gravitino mass. As the energy scale of the
superpotential in these models is much smaller than hFi 1 , such corrections can
spoil the quintessence scenario [133]. One way out may be offered by string theory,
in D ¤ 4 models where supersymmetry is unbroken (see [139] and Sect. 13.1.2).
Another possibility is to reconsider the problem in extended gauged supergravities,
where the scale of supersymmetry breaking is much smaller [105, 115].
7.3.6 Quintessence and the Inflaton
Both quintessence and the inflaton are, at some stage of their evolution, slowly
rolling real scalar fields. The culprit of both early- and late-time acceleration could
therefore be the same [68, 107], a possibility named “quintessential inflation”

[136, 140–145]. The potential is tailored to sustain inflation at early times and
acceleration at late times, with a sequence of kination-, radiation- and matter-
domination periods in between. If the scalar field is dynamical throughout the
evolution of the universe, its potential cannot have an absolute minimum at early
times; to this class of profiles there belong runaway potentials with a minimum
at infinity, V. ! 1/ D 0. Also, the scalar field should not decay at the end
of inflation, which forbids the inflaton to be directly coupled with other fields (in
particle-physics jargon, such a field is called “sterile”). A mechanism alternative
to standard reheating must be invoked [140]. In fact, particles can be produced
gravitationally due to changes in the spacetime metric [146, 147] or via preheating,
where particle production occurs via oscillations of the scalar field when it has a
weak non-minimal coupling with matter [145, 148].
Identifying the inflaton with quintessence is not straightforward: any such model
possesses a number of finely tuned features, whose ultimate source is the -problem
[144, 149]. The latter can be stated as follows. As the inflaton potential is very flat,
in order to reach the much lower quintessence energy scales the curvature V; of
the potential must be large towards the end of inflation. However, this implies a
large slow-roll parameter V and a spoiling of scale invariance in the CMB power
spectrum. Among the various requirements, the period of late-time acceleration
must be triggered very recently. The absence of tracking behaviour during matter
domination makes these cosmologies sensitive to the initial conditions. Difficulties
persist if general relativity is extended to a scalar-tensor theory [150]. Quintessential
inflation, however, may be possible with the introduction of further ingredients (such
as non-trivial measures [151], extra dimensions [152] or a curvaton field [144]). A
variant of the scenario may be realized by a complex scalar field, the real part being
the inflaton and the imaginary one quintessence [153].
The -problem will reappear in all its strength in some important models of
supergravity and string cosmology (Sects. 13.2.4 and 13.5). String theory has an
arsenal of tools that can solve this issue as well as the fine-tuning problem of the
initial conditions. This will permit to revive the idea of quintessential inflation and
to realize it in concrete models presented in Chap. 13.
7.3.7 Summary
Quintessence can solve the coincidence problem but, at best, it only restates the old
problem. When the latter is relaxed, some non-negligible amount of parameter
tuning is often present. Models which avoid fine tuning of initial conditions are,
in general, constrained by observations into parameter regimes with no theoretical
motivation. These findings are a practical illustration of the obstacles cosmologists
have been facing in their attempts to address the cosmological constant problem.
7.4 Scalar-Tensor Theories 327
7.4 Scalar-Tensor Theories
A dynamically damped effective can also arise with non-minimally coupled

and multiple scalar fields [41–55]. A scalar field is minimally coupled to gravity
when the only interaction terms are via the covariant derivatives rŒg and the
p
metric determinant weight g. Non-minimal couplings between scalar fields and
curvature invariants give rise to a wide class of models known as scalar-tensor
theories [154–169].
Their key characteristic is that fundamental constants of Nature are neither
fundamental nor constant. For example, in the model
Z
p F. N / N 1 N N N N / C h. N /LN m ;
SN D dD x Ng R !. / N
g @ @ U.
2 2 2
(7.51)
a non-canonical real scalar is coupled both with the Ricci scalar and with the
matter Lagrangian (in the next sub-section we will explain the bars). Then, both the
gravitational coupling G . N / :D G=F. N / and the matter masses m2 . N / :D m2 h. N /
are effectively spacetime-dependent. The special case
!JBD
FDN; !D ; U D 0; D D 4; (7.52)
N
where !JBD is a constant, is known as Jordan–Brans–Dicke theory [154, 155, 157,

158]. General relativity is recovered in the limit !JBD ! 1, provided the trace of the
matter energy-momentum tensor does not vanish, TN ¤ 0 [170–172]. The equations
of motion for the action (7.51) are

NR 1 gN RN F D 2 hTN C TN
N
C rN rN F gN F
N ; (7.53a)
2
N
N N U; N C 1 !; N @N N @ N C F; N R C h; N LN m D 0 ;
! (7.53b)
2 2 2
N
where TN D !@ N @ N C gN LN N is the energy-momentum tensor of the scalar
and the matter energy-momentum tensor is covariantly conserved, rN TN D 0.
Contracting the indices we obtain the trace equation

D N N :
1 F RN D 2 hTN C TN C .D 1/F (7.54)
2
7.4.1 Motivations
A non-minimal coupling F / N 2 (with ! D 1) naturally appears in quantum

field theory on classical curved backgrounds. When quantum corrections to the
minimally-coupled scalar-field action are taken into account, virtual loop processes
of the scalar field produce divergent mass counter-terms, which are exactly canceled
by a N 2 RN contribution [20, 173–177]. The minimal coupling N 2 RN is thus required by
mass renormalization.
A scalar-tensor action can arise also in more exotic situations, for instance
in Kaluza–Klein compactifications [178, 179] of supergravity where the D-
dimensional metric is block diagonal, gMN D .Ng ; ab /, ; D 0; 1; 2; 3,
a; b D 4; : : : ; D 1, and N D det ab [180, 181]. Then, the typical potential
in the Einstein frame (see (7.55) below) is of exponential type (Sect. 7.3.3). The
dilaton of string theory (Sects. 12.1.5 and 13.7.5) is also governed, at lowest order in
the string scale, by the action (7.51) except in the matter sector, where the coupling
is not universal and one cannot reduce it to an overall function h. N /. The equations
P .i/
of motion are easily extended to this case by replacing hLN m ! i hi LNm .
7.4.2 Conformal Transformations
The metric gN is denoted with a bar to highlight that the action (7.51) is presented
in a metric frame where gravity is non-minimally coupled to the scalar field. If
F. N / ¤ 1 and h. N / D 1, N is minimally coupled to matter fields i and the set
of variables .Ng ; N ; i / is called Jordan frame. There exists, however, the Einstein
frame [155] where F. N / D 1 and the scalar field is minimally coupled with the
D-dimensional metric g :
Z
p R 1 D
SD d x g
D
g @ @ V. / C h˝ Lm ; (7.55)
2 2 2
where Lm D LN m Œ˝ 2 g ; i . The two frames are related by a conformal (or Weyl)
transformation of the metric (Problems 7.3, 7.4 and 7.5)
g :D ˝ 2 gN ; (7.56)
where
1
˝ D ˝. / D fFŒ N . /g D2 (7.57)
is a function of spacetime coordinates which is expressed in terms of the scalar field

and D . N / is given by (7.155). Thus, U D ˝ D V. The equations of motion are a
modification of the usual ones in the matter sector:
1
R g R D 2 hT C T ; (7.58a)
2
V; D QhT h; ˝ D Lm ; (7.58b)
where
˝;
Q :D (7.59)
˝
p
and, calling Lm Œ˝ 2 g ; i D ˝ D gLm , we used ıLm =ı D .@Lm =@Ng /
p
ı gN =ı D .@Lm =@g /g ı ln ˝ 2 =ı D gQT. Noting also that
p
2 ıLm 2D 2 ı. N gLNm /
T D p
D ˝ p
D ˝ 2D TN ; (7.60)
g ıg Ng ı gN
from (7.149) one has
r .hT / D QhTr : (7.61)

The total energy-momentum tensor is conserved, r .hT C T / D 0, as required
by general covariance.
In general, a theory can be conformally invariant if no dimensionful coupling
appears in the action. As the Newton constant is dimensionful, a Weyl transfor-
mation will map p an action to another with different characteristics. For instance, if
˝. / D cosh. 2 =6 / in D D 4, the scalar field in the Jordan frame has !. / D 1
(which might be regarded as a canonical kinetic term, were it not for the FR term
which can be integrated by parts to give extra field derivatives). Thus, starting from
a cosmological constant U D = 2 in one frame we can get a hyperbolic potential
V / ˝ D in the other frame, similar to those considered in Sect. 7.3.3. When
combined with a field redefinition ! N . /, different potentials U and form
factors ! in the Jordan frame can lead to a variety of potentials for the scalar field
in the Einstein frame.
On a FRW background with a perfect fluid, (7.53) become
2
F
D
1 H N2C K D h N C N N H FP ; (7.62a)
2 aN 2 D1
2 !
D 1 F; N N PN C U; N D
F; N
!C 2
Œ RN C .D 1/H U
D2 F D2 F

1 F; N 2.D 1/ F; N F; N N PN 2
C !; N C ! C
2 F D2 F 2
h F; N
C Œ.D 1/PN
N h; N PN D 0 ; (7.62b)
D2 F
where we used the trace equation (7.54),
N
D R D
N
1 F 2 D hŒ.D 1/P
N C 1 ! PN 2 DU
2 2
D1
C ŒF; N RN C F; N .D 1/H
N PN C F; N N PN 2 :
2
The continuity equation for matter is
N N C P/
PN C .D 1/H. N D 0: (7.63)
It is important to stress that, in all these equations, dots represent derivatives with
respect to synchronous time Nt, related to time in the Einstein frame by
dt D ˝dNt ; a D ˝ aN : (7.64)
Both relations in (7.64) stem from (7.56) and the relation between line elements
ds2 D ˝ 2 dNs2 .
In the Einstein frame, the first Friedmann equation is (2.74) with D 0 and
total energy density C . The continuity equation (2.76) and the scalar-field
equation (2.80) are augmented by source terms,
P C .D 1/H. C P/ D Œ.D 1/P Q P ; (7.65)

R C .D 1/H P C V; D Œ.D 1/P Q ; (7.66)
as per (7.61) and (7.58b). Here dots are derivatives with respect to t. Compar-
ing (7.63) and (7.65) and taking into account (7.64), it follows that
N C d˝ ;
˝H D H D ˝ D N ; P D ˝ D PN : (7.67)
dt
The change of frame realized by a conformal transformation does not entail
a coordinate transformation, hence it is not a diffeomorphism. The debate about
“which frame is the physical one” has been ongoing for some time [182–185]. In
the original definition of the Jordan–Brans–Dicke theory, h D 1 and the conformal
transformation (7.56) connects the Jordan frame to a frame where measurement
units are changed [158]. On the other hand, setting h˝ D D 1 in (7.55) would
imply that effective masses are constant in both frames, in which case the frame
without bars can be regarded as an Einstein frame with fixed units [185]. The answer
to the above question is conditioned by which pair one chooses: (Jordan frame:
frame with bars where h D 1)–(Einstein frame with running units), (frame with
bars where 1 ¤ h ¤ ˝ D )–(Einstein frame with running units) or (frame with bars
where h D ˝ D )–(Einstein frame with fixed units). For definiteness, we consider
the first pair, i.e., the Jordan frame (h D 1) automatically connected to the Einstein
frame with running units (from now on, Einstein frame in short).
There are some evident differences between the Jordan and the Einstein frame:
(i) the two frames have different field equations and solutions; (ii) only in the
Einstein frame both the scalar-field Hamiltonian and the ADM energy [186] are
positive semi-definite [182, 187]; (iii) in the Jordan frame, the energy-momentum
tensor for matter is conserved, while in the Einstein frame it is not (equation (7.61)).
Similarly, the scalar-field equation (7.58b) in the Einstein frame has a source term;
(iv) by definition, in the Jordan frame matter is minimally coupled with the scalar
field: therefore, massive particles with D-velocity uN D ˝u follow the geodesic
equation uN rN uN D 0 in the Jordan frame but they do not in the Einstein frame,
where u r u D .u u C g /@ ˝=˝ (here we used (7.149)); (v) conformal
transformations encode a change of metric units, as the length of space and time
intervals, as well as the norm of vectors, are clearly affected. Notwithstanding, the
light-cone causal structure of spacetime is maintained, since time-like, space-like
and null vectors are such with respect to both metrics.
At the classical level, for h ¤ ˝ D the physics of the background is the same:
it is simply described differently, with a different interpretation of the phenomena.
Time and length intervals are rescaled as dx ! dNx D ˝dx , masses rescale as
m!m N D ˝ 1 m and measurements (based on time or length or mass ratios) are
unaffected [184, 185, 188–191]. This is no longer true in the Einstein frame with
fixed units, where h D ˝ D and masses are constant as in the frame without bars.
Due to the mass rescaling, the gravitational coupling in the Einstein frame
acquires a field dependence. From this, one might naively conclude that the strong
equivalence principle is violated in the Einstein frame but respected in the Jordan
frame. (The strong equivalence principle requires that all matter fields gravitate in
the same way independently of the location in spacetime.) However, in D D 4
a post-Newtonian calculation shows that the gravitational coupling measured in a
Cavendish experiment is [192, 193]
" #
G F;2N
Geff D 1C (7.68)
F 2! 2 F C 3F;2N
in both frames [191]. The strong equivalence principle is thus violated in both
frames, since the effective Newton’s coupling depends on . The weak equivalence
principle (stating that all forms of neutral matter couple in the same way to gravity,
i.e., test bodies fall with the same acceleration in a gravitational field independently
of their mass and composition) holds classically but is violated in general due to
quantum effects [194–197]. Notice that (7.68) differs from the coupling G D
G=F D G=˝ 2 by less than one part over 104 today [198].
Another example of the classical physical equivalence of the two frames is

provided by the flat FRW metric, conformally equivalent to the Minkowski metric.
In one case cosmology is described by cosmic expansion, in the other by the
dynamics of a scalar field. What ultimately matters is the output of measurements,
which is the same in both frames despite the widely different theoretical explanation.
The redshift of a light signal is given by the ratio of the frequency of the signal
measured at a source S in units of the observer’s reference frame, divided by
the frequency measured by a distant observer O. Using the geodesic equation,
1 C zN D N S =N O D aN O =NaS in the Jordan frame. In the Einstein frame, the units in the
observer’s and in the source reference frame do not coincide, since matter is non-
minimally coupled to gravity. Therefore, one should rescale the source frequency S
in the observer units, yielding 1 C z D .˝O1 =˝S1 /S =O D .˝O1 aO /=.˝S1 aS / D
aN O =NaS D 1 C zN. The observed redshift is then conformal-frame independent [188–
190] and so is the magnitude-redshift relation [189, 190].
7.4.3 Perturbations, Quantum Theory and Extended Inflation
In cosmological linear perturbation theory and in the absence of isocurvature pertur-

bations, the Jordan and Einstein frames are physically equivalent, as the amplitudes
of density perturbations have the same momentum dependence [199, 200]. The
reason is that the Weyl transformation (7.56) is linearized in perturbation theory,
so that quantized fields are rescaled by background factors. In particular, the linear
curvature perturbation on uniform density hypersurfaces is conformally invariant
under (7.56) either in the presence of radiation only or provided the entropy
perturbation between matter and the scalar vanishes [191]. In single-field inflation,
where acceleration is driven by the scalar mode non-minimally coupled to gravity, a
stronger result actually holds: the fully non-linear comoving curvature perturbation
is conformally invariant [201, 202],
RNL D RN NL : (7.69)
On a cosmological background, there is no difference between inflationary observ-

ables calculated in either frame to leading order in the SR parameters [203, 204].
This is no longer the case when other matter components are present apart from the
non-minimally coupled mode [205], such as in multi-field scenarios [202, 206–208].
Related to perturbation theory is the fact that, in general, quantization and the
field-dependent conformal transformation do not commute and the two frames are
physically inequivalent when or g (or both) are treated at the quantum level
[55, 183, 185, 209–214] (see also Sect. 8.1). Scalar-graviton interactions must be
transformed away to allow for a standard field quantization, which points towards
the Einstein frame as a preferred choice.
Thanks to the physical equivalence between the two frames at the leading
perturbative level, non-minimally coupled single-field inflationary scenarios are the
simplest. There, it is sufficient to apply most of the standard results of Chap. 5 in

the Einstein frame and consider the dynamics generated by the potential V; a full
analysis in the Jordan frame is also possible [215, 216]. In Sect. 5.11.2, we already
discussed the case where the scalar field is the Higgs. Models of extended inflation
[217, 218] predict, in general, a low tensor-to-scalar ratio. In terms of the number
of e-foldings, for the Jordan–Brans–Dicke parametrization ns 1 ' 2=Nk and
r D 4.3 C 2!JBD /=Nk2 , which results in the upper bound !JBD < 11:5 at 68%
CL (much larger at the 2 – 3 level) [219]. The prediction (5.235) of Starobinsky
inflation corresponds to !JBD D 0.
7.4.4 Cosmological Constant Problem
How can scalar-tensor theories of gravity address the cosmological constant prob-
lem? By itself, the non-minimal coupling cannot suppress large values to small
ones. In fact, let U D N D = 2 be the energy density of the cosmological
constant in the Jordan frame. In the Einstein frame in four dimensions, the term is
D ˝ 4 N and, if the conformal factor is, say, ˝ D ec , then the cosmological
constant seems to be exponentially suppressed at late times (c ! C1). However,
actual measurements of the energy density are performed with respect to some
standard units ref , which scale as ref D ˝ 4 Nref with respect to the units in the
Jordan frame. The ratio =ref D N =Nref is therefore unchanged [185].
The scalar , however, can act as the quintessence field when endowed with
a non-trivial potential [198, 220–239]. Scaling solutions exist for exponential,
inverse- and (contrary to minimally coupled quintessence) also positive-power-law
potentials U in the Jordan frame [221, 227, 233] and, in general, for U. / / ŒF. /c ,
where c D O.1/ is a constant [223, 225]. Just like for the inflaton, theoretical models
can be compared with observations to reduce the parameter space or, conversely,
one can use experiments to reconstruct the functions in (7.51) [198, 230, 238]. The
interest in these models (dubbed extended or, more generally, coupled quintessence)
is in their ability to generate scalar-field potentials with peculiar properties and an
effective barotropic index smaller than or close to 1. This possibility [238] arises
via the non-minimal coupling with dark matter in the Einstein frame [191, 240]. Let
us set D D 4, no curvature and a dust component, D m , P D 0. In the Einstein
frame, the dark-matter energy density is m D ˝ 4 Nm . An observer assuming the
standard Einstein dynamics would parametrize this contribution as in (2.88), so that
the first Friedmann equation would read
a a
2 0 3 0 3
H2 D m0 C eff ; eff :D C m m0 : (7.70)
3 a a
From (7.63), (7.64) and (7.67),

3 a
4 4 aN 0 ˝0 0 3
m D ˝ Nm D ˝ Nm0 D m0 : (7.71)
aN ˝ a
Combining (7.65) and (7.66), one has @t .m C / C 3H.m C C P / D 0.

Comparing this equation with the effective dark-energy continuity equation Peff C
3H.1 C weff /eff D 0 and plugging in (7.71), one obtains the effective barotropic
index
P w
weff :D D w eff D ˝0 3 : (7.72)
eff
1 C ˝ 1 m0 aa0
Therefore, if ˝ decreases in time then ˝0 =˝ < 1, the denominator is smaller

than 1 and weff < w . For a given choice of parameters in the potential, this
effect lowers the barotropic index of tracking solutions which, as we have seen
in the minimally-coupled cases, fail to reach w 1 today. For a slow-rolling
field, w 1 and weff . 1. In this case, however, one would not fall back
into freezing scenarios where the initial conditions are fine tuned. The intuitive
reason is that gravity now induces an effective potential which is non-trivial even
when the potential U is almost flat. In the ordinary Klein–Gordon equation 0 D
R C 3H P C V; ' R C 3H P , flatness implies a slowly-varying or almost constant
scalar profile. On the other hand, the RN curvature term in (7.53b) (or the source in the
Einstein-frame counterpart (7.66)) fuels the dynamics of even if V is constant, a
mechanism dubbed “R-boost” [241]. Consequently, for a given set of different initial
conditions on the field value, and requiring w0 1 and ˝;0 0:7, extended
quintessence spans a wide range of magnitudes for the initial field density, contrary
to the minimal-coupling case where the density converges to the single value ˝;0
[236]. Thus, the scalar-tensor mechanism can produce an effective equation of state
with ultra-negative barotropic index weff < 1 without invoking ghosts, phantom
matter [242] or any other type of unstable fields (we saw another mechanism, based
on gradients only, in Sect. 5.3.1).
7.4.5 Experimental Bounds and Chameleon Mechanism
Scalar-tensor gravity is subject to stringent experimental bounds coming from post-

Newtonian solar-system tests, equivalence-principle tests and cosmology [243]. For
instance, in the Jordan–Brans–Dicke parametrization solar-system tests require the
global bound [244]
!JBD > 4 104 (95% CL) : (7.73)

For a constant Q, (7.59) gives ˝ D eQ and, since in this case
1
3 C 2!JBD D ; (7.74)
2Q2
for massless quintessence (7.73) corresponds to the constraint jQj < 1:25 103 lPl
on the coupling with dark matter in (7.65) and (7.66). The model is thus very close to
standard general relativity throughout the evolution of the universe. More generally,
at late times the scalar field is driven towards a minimum of the Einstein-frame
coupling h. / ˝ D . / with matter, so that after radiation domination the system
converges to general relativity [245, 246]. A scalar with a non-trivial potential and
a large effective mass m can satisfy local constraints provided its interaction range
1=m is sufficiently short. The value and the variation of Newton’s coupling G,
however, is different in clusters of over-dense regions than at cosmological scales.
Spatial differences between an over-density and the background are typically small,
P
ıG=G 106 ; time differences are even less appreciable, G=G . 1020 – 1019 s1
outside and inside an over-density, respectively [247]. As it turns out, G may
approximately be locally constant in the solar system even if the scalar field
continues evolving. Local bounds should therefore be applied with some caution
to the cosmic evolution.
Another effect to take into account is the chameleon mechanism [248, 249]. Let
D D 4 and h D 1. The non-minimal coupling with matter induces an effective
potential Veff D V. / C ˝ 4 in the Einstein frame. Calling Q :D ˝ 3 the
conserved energy density of non-relativistic matter (dust, P 0), for a constant
Q the effective potential reads
Veff . ; /
Q D V. / C eQ Q : (7.75)
The scalar-field effective mass m2eff ./

Q D Veff; depends on the energy density of
the environment. In general, meff increases with Q and it is possible that the scalar
field is massive enough in regions of high density to respect local constraints on
the equivalence principle and fifth-force effects, while at cosmological scales such
effects become important. Therefore, quintessence may not have yet been detected
in Earth-based experiments because of the high density of the environment. Inside
a compact object, the scalar acquires a constant value minimizing the effective
potential (7.75), with Q being the density in the interior of the object. Assuming a
spherical body of radius r0 , the field grows within a thin shell of width r0 below
the surface, while at distances r > r0 the field emenv r =r behaves as a massive
scalar with mass menv determined by the energy density Qenv of the surrounding
environment. At r r0 , tends to the value minimizing Veff . ; Qenv /. In other
words, the fifth force mediated by the scalar field is mainly generated by the thin
shell and is consequently much weaker than it would be if the whole bulk of the
body contributed.
Since the ratio r0 =r0 is inversely proportional to the Newtonian potential of the
body, massive objects such as planets obey the thin-shell condition r0 =r0 1.
For instance, the chameleon mechanism affects laboratory measurements of G

because r0 =r0 < 103=2 for test masses in a vacuum chamber on Earth. Also,
preservation of equivalence-principle bounds yield, for Earth, r˚ =r˚ < 107 .
In this case, one can show that the scalar field is short-ranged in the atmosphere
(for an inverse power-law potential, m1
atm . 1 – 10 mm) and free already at solar-
system scales: m1
sol . 1 – 10 AU (Astronomical Units), while at cosmological scales
m1
cosmo . 10
1
– 103 pc. In the presence of a thin shell, one can show that the
effectively measured coupling is
r0
Qeff ' 3Q ; (7.76)
r0
which is greatly suppressed near Earth. This parameter replaces Q in all expressions
related to experiments, including (7.74). Therefore, scalar-tensor models can be
made compatible with local and solar-system tests even if they predict a large
Q & O.1/.
7.5 Higher-Order and Higher-Derivative Gravity Models
The scalar field appearing in quintessence and extended quintessence models is an

extra degree of freedom independent of the gravitational sector. When a scalar-
tensor model arises from dimensional reduction, is indeed related to geometry, as
a residual manifestation of the higher-dimensional metric. Another way to generate
a spin-0 degree of freedom from geometry is to replace the Einstein–Hilbert action
with a higher-order expression in the curvature invariants. Although these theories of
modified gravity can ultimately be recast as multi-scalar-tensor models and, hence,
do not really solve the cosmological constant problem, it is attractive to identify the
agent of present cosmic acceleration with gravity itself [250, 251].
7.5.1 Motivation and Ghosts
Consider a quartic Lagrangian containing the curvature invariants R, R R and
R R [252–254]:
1
Lg D 2
R C ˛R2 C ˇR R C R R ; (7.77)
2
where ˛, ˇ and are dimensionful constants. In four dimensions only, the Gauss–
Bonnet term
LGB :D R2 4R R C R R (7.78)

7.5 Higher-Order and Higher-Derivative Gravity Models 337
is topological and does not contributes to the equations of motion. This allows one,
in D D 4 and at the classical level, to rewrite (7.77) only in terms of the Ricci scalar
and the Ricci tensor ( D 0).
Counter-terms of the kind (7.77) are introduced, in four dimensions, in the renor-
malization of matter quantum fields living in a fixed, classical, curved spacetime
[252]. They appear, for instance, in vacuum loop diagrams of an interacting scalar
field and conspire to cancel the ensuing divergences [177]. On a FLRW background,
the renormalized energy-momentum tensor of matter acquires curvature corrections
which lead to an anomalous trace g hT i D b1 R C b2 R R C b3 R2 [255] and
to an effective quartic Lagrangian.
In perturbative quantum gravity (Sect. 8.2), the Einstein–Hilbert Lagrangian
L D R=.2 2/ receives loop corrections in the curvature invariants, coming from
the self-interactions of the graviton [176, 256]. Up to quadratic order (one loop),
the gravitational Lagrangian is of the form (7.77) [176, 256–266]. Fourth-order
quantum gravity (often called Stelle’s theory [257, 258]) is indeed renormalizable
but at the price of loosing S-matrix unitarity, due to the presence of ghost modes
[257, 258]. A ghost is, by definition, a field whose kinetic term is unbounded from
below (roughly speaking, it has the “wrong sign;” see Sect. 11.8.1). It implies a
macroscopic instability which, if not healed, would lead to a breakdown of the
theory, a violation of unitarity and wild particle creation in a time interval of
cosmological length. The same problem arises in a generalization of Stelle’s theory
to D-dimensional spacetimes. Let
X X
Lpoly D ˛2 R C ˛4 R2 C ˇ4 R R C : : : C ˛X R 2 C ˇX .R R / 4
R / 4 C ıX R 2 2 R C : : :
X X
C X .R (7.79)
be a polynomial Lagrangian with at most X derivatives of the metric, where the

first dots include a finite number of extra terms of order < X in the derivatives
of the metric tensor and the second ones indicate a finite number of operators
O.R2 X=23 R/ with the same number of derivatives but higher powers of the
curvature. This theory is renormalizable when X D D (Sect. 8.2) but it also contains
a ghost.
On backgrounds with constant Ricci curvature R D R0 (such as Minkowski
and de Sitter), a generic theory f .R; R R ; R R / can be expanded up to
quadratic order as an Einstein–Hilbert action plus a , an R2 and a Weyl-squared
term C C , where C is the Weyl tensor. There are eight degrees of
freedom associated with this class of actions, including (7.77): two of them are the
usual polarization modes of the spin-2 graviton of general relativity, one corresponds
to a spin-0 massive scalar coming from the R2 term and five more to a massive
spin-2 particle with negative-definite kinetic energy (a ghost), coming from the
Weyl-squared term [257, 258, 267–269]. The massive scalar is a ghost only if the
graviton is a ghost, which is not the case if the leading curvature term is Einstein
gravity.
On these backgrounds with R D R0 , the Weyl ghost can be eliminated if some

quite restrictive conditions on the Lagrangian are satisfied [270]. These strong
selection rules in the space of parameters establish whether a given scalar-tensor
or modified gravity model has ghosts, classical instabilities (when the squared
propagation speed of inhomogeneous perturbative modes is negative definite) and
super-luminal modes (i.e., modes propagating faster than light, which can cause
breaking of causality and an ill-posed Cauchy problem).
No-ghost conditions, classical-stability conditions and sub-luminal conditions
can be generalized to FLRW backgrounds (where the Ricci curvature is time
dependent) and to actions whose higher-order term is either a function of the
Gauss–Bonnet invariant (7.78) [271, 272] or the Gauss–Bonnet term non-minimally
coupled with a scalar field [273–275]. Even when they can be tuned to produce a
viable history of the universe [276, 277], ghosts, super-luminal modes and classical
instabilities arise in a number of models for general initial conditions and during
cosmologically relevant periods. Gauss–Bonnet actions are possibly inadequate, by
themselves, to relax the fine tuning of the cosmological constant problem.
Higher-order gravity admits non-singular cosmological solutions [278–281], but
they are not typical and there are classes of solutions which contain singularities
[254, 282, 283]. Moreover, although changing Einstein’s equations would change
the energy condition of the singularity theorems of Chap. 6 [284], the Borde–
Guth–Vilenkin theorem of Sect. 6.2.3 is independent of the dynamics as long
as the background is inflationary. Therefore, modified-gravity models do not
help to solve the big-bang problem. Yet, modulo stability issues, they do have
cosmological applications to inflation [285] and to quintessence: acceleration is a
typical dynamical feature.
Another modification of standard general relativity consists in including higher
covariant derivatives n R, n D 0; : : : ; N [268, 286–288]. Terms of the form
Rn R can be generated by loop corrections that include massive matter fields,
in the limit of scales much larger than the mass of the fields [286]. In general,
higher-order and higher-derivative actions can arise as effective theories valid up
to the Planck scale, after integrating out the microscopic degrees of freedom of
a more fundamental description of gravity. They also admit an implementation in
supergravity [289, 290].
7.5.2 General f .R/ Action
In quantum field theory, the inclusion of operators made only of the Ricci scalar R is
inconsistent, since one should take into account all tensor invariants that contribute
to the propagator. Nevertheless, to illustrate the main cosmological properties
of higher-order gravity it may be useful to consider the simple case where the
Lagrangian only depends on R [291–296]. More general scenarios of dark energy
include f .R; R R / terms [297]; these give similar results, both in ordinary [295]
and Palatini formalism [298, 299]. The action we start from is thus
Z N
p f .R/
SN D d x Ng
D N
C Lm : (7.80)
2 2
The bars have been introduced to stress that this theory is classically equivalent to
the scalar-tensor action (7.51) with Jordan-frame functions
@f F RN f
FD ; UD ; ! D 0: (7.81)
@RN 2 2
In fact, the system (7.80) is on-shell (i.e., dynamically) identical to the one given
by the Lagrangian L D f . N / C .RN N /F, by varying with respect to the field F.
There is a scalar mode N D RN hidden in (7.80) and two more ( 2 D RN RN ,
N
3 D R R
N ) in the action (7.77) [269]. By expanding the action at second
order in the perturbations on a given background, the kinetic term of the graviton
tensor modes is F h˛ ˇ hˇ ˛ . Taking into account both graviton [300] and scalar
perturbations [301, 302], f .R/ theories are free of ghosts and other instabilities only
if the constraints
@f @2 f
FD > 0; F0 D > 0; (7.82)
@RN @RN 2
are respected on the chosen background before the future de Sitter attractor (if
present) is reached.
The equations of motion in the Jordan frame are
F RN 12 f gN rN rN F C gN F
N D 2 TN : (7.83)
It is more convenient to map this model [294, 295, 303–305] conformally into the
Einstein-frame action (7.55) with h D 1 and, setting ! D 0 in (7.155),
r
p D1
D .D 1/.D 2/ ln ˝ D ln F : (7.84)
D2
p
In particular, Q D 1= .D 1/.D 2/. For general f , the potential in the Einstein
frame is given by (7.81),
.F RN f /F D2
D
VD : (7.85)
2 2
P
N
If f D f .R; N R;
N :::; N then (7.81) is generalized to F D NnD0
N N R/, N n @f =@'n
N nN
and the theory contains N independent scalar modes 'n D R [286–288]. This is
an example of the extra degrees of freedom typically appearing in higher-derivative
theories.
7.5.3 Palatini Formulation
For any given scalar-tensor action with fixed form factors F, ! and U, there exist
two inequivalent definitions of the variational principle for gravity.2 In the second-
order formalism we used so far, the dynamical degrees of freedom are encoded in
the metric g and the action is varied with respect to infinitesimal fluctuations ıg .
In Chap. 9 we will introduce another possibility, the Palatini formalism, where the

metric and the connection are treated as independent variables. While the two
procedures yield the same dynamics for standard general relativity or in vacuum,
they are mutually exclusive in the case of modified gravity with matter [250, 292,
297, 307]. For instance, the Jordan-frame equations of motion (7.83) are replaced
by
p
F RN 12 f gN D 2 TN ; r. / Ng F gN D 0 ; (7.86)
in Palatini formalism. The Jordan–Brans–Dicke theory classically equivalent

to (7.80) has !JBD D 0 in the metric formalism but !JBD D 3=2 in Palatini
formalism [308, 309]. The resulting dynamics is somewhat different but both cases
are well below the lower bound (7.73) and the chameleon mechanism is generally
required [310–314].
7.5.4 Form of f .R/
Various functionals f have been proposed, including (bars are omitted from now on)
the monomial f D Rn [283, 295, 315, 316], the polynomial [291, 317]
f D R C cn R n ; (7.87)
trigonometric, exponential and hyperbolic functions [283, 318, 319] and logarithmic
functions such as f D ln R [320] and f / R C Rn .ln R/m [321], the case n D 1 D
m arising from one-loop quantum corrections in a 2 R scalar field theory [322].
Positive powers of the Ricci scalar are important at high curvature and play a role in
the early universe and inflation [283, 285, 303, 305, 323–325]. Special attention has
been devoted to the polynomial case (7.87) with n D 2 [283, 294, 297, 303, 326,
327],
f D R C c2 R 2 ; (7.88)
2
When the form factors are left unspecified, the two formulations can be mapped one onto the other
by a redefinition of ! [306].
for which the Einstein–frame potential is
1 p.D1/.D2/
.D4/ p D2 2
VD e 1 e D1 ; (7.89)
8c2
the generalization of (5.234) to D dimensions. Here we see the origin of Starobinsky

inflation, the model of accelerating dynamics discussed in Sect. 5.12.4: it is the
Einstein–frame counterpart of the f .R/ model (7.88) and acceleration is fueled by
the quadratic curvature correction. The Lagrangian (7.88) will find another possible
motivation by quantum gravity in Sect. 11.2.2.
For f .R/ given by (7.87) and a dust fluid (w D 0), in the Jordan frame the
effective barotropic index of the universe defined by a D t2=Œ3.1Cw / is w D
1 C 2.2 n/=Œ3.n 1/.2n 1/ [328]. Therefore, the larger n the closer w to
1. A similar conclusion, but with different n dependence, holds also in first-order
formalism, where w D 1 C .w C 1/=n in the Jordan frame in the presence of a
perfect fluid with equation of state P D w [320].
By themselves, positive powers of the Ricci scalar are tightly constrained
throughout the history of the universe [316, 329]. On the other hand, negative powers
of R dominate in low-curvature regimes and can lead to late-time acceleration
[330, 331]. A problem with inverse powers of R is that the Minkowski metric is
not a vacuum solution of the equations of motion, which would invalidate standard
quantum field theory in local inertial frames. The condition that f be analytic in R
(i.e., finite at R D 0) seems therefore fundamental [332]. However, both de Sitter
and anti-de Sitter spacetimes are vacuum solutions and, with some care, quantum
field theory can be defined thereon [20].
The polynomial (7.87) with n < 0 suffers from instabilities in metric formalism
[333] and do not reproduce the correct evolution of the universe [334–336]. These
issues can be overcome either by adopting the Palatini formalism [337–340] or by
assuming, in metric formalism, a phenomenological combination of both positive
and negative powers [341, 342],
f D R C cn .R R0 /n C dl .R R1 /l ; (7.90)
where R0;1 are fixed curvature scales and n; l > 0. This model can fit observations
of the late-time acceleration. In particular, a viable Lagrangian with n D 0 and
R0 D 0 D R1 is
ˇ ˇl !
ˇ R2 ˇ
f .R/ D R R2 1 ˇˇ ˇˇ ; l > 0; (7.91)
R
where > 0 is a dimensionless constant and R2 is some curvature scale. The

corresponding potential (7.81) in the Jordan
p
frame is, in D D 4, of the form
U. / D V0 .; R0 /Œ1 A.; l/.1 e 2 = 6 l=.lC1/
/ . Since Q < 0 but P > 0, then
P P
˝=˝ D Q > 0 and, according to (7.72), weff > w . Then, one cannot mimic
a phantom barotropic index but the de Sitter attractor can be reached nonetheless
[239]. Via the chameleon mechanism, (7.91) is compatible with equivalence-
principle tests provided [239, 314]
l > 1:8 : (7.92)
Another model compatible with observations is [313]
bcH02
f .R/ D R ; (7.93)
d C .cH02 =R/n
where n D 1; 2; 3 and b, c and d lie in a certain range [343].

The quest for viable modified gravity actions is a broad subject; we refer the
reader to the papers cited above, a sample literature such as [343–356] and the
reviews [357–359]. Here we just mention that most of the simplest models with
inverse powers, logarithms or exponentials are ruled out because they fail to obey
certain requirements. Having solutions with an ordinary matter era with a / t2=3
and an adequate late-time acceleration constrains the derivatives @R f and @2R f ,
which single out a general classification for the possible cosmic evolutions of
these models [351]. Linear perturbation theory [302, 360–366] further restricts the
shape of f .R/ via the observation of large-scale structures and the CMB [239, 356].
To date, the base CDM model is compatible with observations and there is no
compelling evidence pointing towards a dynamical dark energy, including that from
a modification of general relativity [367].
Viable f .R/ profiles such as (7.91) and (7.93) can be tailored nonetheless and
the fine tuning of a pure cosmological constant can be relaxed. The modern
challenge to accept is then twofold: observationally, to distinguish viable f .R/ or
other higher-order gravity models from a pure term and, theoretically, to give
phenomenology a solid motivation. Whether quantum-gravity approaches or string
theory can provide such a motivation will be a subject of the next chapters.
7.5.5 Horndeski Theory and Extensions
Attention has been devoted also to Horndeski theory, the most general four-
dimensional scalar-tensor Lagrangian which is higher-order in the derivatives of the
metric and of the scalar field but giving rise to equations of motion which are at most
second order [219, 368–370]. Ghost modes are therefore easily avoided. Horndeski’s
general action includes scalar-tensor theories with second-order kinetic terms, f .R/
models, k-essence [371–375] (where the covariant scalar-field Lagrangian L.X; /
is higher order in the kinetic term X D r r ) and the Galileon model (higher-
order scalar kinetic terms but second-order field equations [376, 377]) and it is
equivalent to generalized Galileon theory [358, 359, 378, 379].
7.6 Other Approaches 343
A yet more general class of models produces equations of motion with higher-
order derivatives but physical degrees of freedom obey second-order dynamics
[380–387]. In this “healthy” extension of Horndeski theory, freedom from ghosts
can be checked via a canonical analysis.
7.6 Other Approaches
Quintessence, scalar-tensor theories and higher-order gravity are all considered

classic approaches to dark energy. However, there exist other scenarios where the
cosmological constant problems are tackled under difference premisses, sometimes
conservative, sometimes involving a deeper conceptual leap beyond standard gen-
eral relativity. In this section we mention only a few, leaving the discussion of other
candidates motivated by quantum gravity or string theory for later.
7.6.1 Varying Couplings
Scalar-tensor theories are a special case of models where all the couplings are
spacetime dependent via a Lorentz scalar field. The phenomenology of such models
can be tightly constrained by cosmological as well as atomic and particle-physics
experiments. For instance, in ordinary electrodynamics the fine-structure constant
˛ D e2 =.„c/ depends on the electron charge e, the Planck constant „ and the
speed of light c. When one or more of these constants are promoted to coordinate-
dependent parameters, one effectively obtains a time-space varying ˛, for which
there are various observational data sets. A possibility for varying ˛ is to keep „
and c constant while allowing for a non-constant electric charge, e ! e.x/, as
in Bekenstein’s model [388, 389]. In other scenarios, the speed of light is made
spacetime dependent [390]:
c ! c.x/ : (7.94)
Varying-speed-of-light (VSL) theories simply correspond to frameworks where

units are adapted with the scales in the dynamics (and, in particular, chosen such
that c varies) [391]. Time and space units are redefined so that differentials scale as
dt ! Œ f .x/a dt, dxi ! Œ f .x/b dxi , where f is a function, a and b are constants and
local Lorentz invariance of the line element requires c.x/ / Œ Rf .x/ba . In particular,
when b D 0 one formally reabsorbs c in the coordinate x0 D dt c.t/, which scales
as a length. With this coordinate, all equations can be made general-covariant and
gauge invariant (in a word, formally identical to the usual ones) provided some
conditions are met. For instance, the field strength of the Abelian electromagnetic
field A is of the form F D .„c=e/f@ ŒeA =.„c/ @ ŒeA =.„c/g and explicit
c dependence disappears if e / „.c/ c. Therefore, VSL models also require, in
general, a varying electric charge and a varying Planck’s constant. They are locally
Lorentz invariant under transformations which look like the usual ones but with a
varying c. Defining the scalar field :D 0 ln.c=c0 /, the total action for a minimal
version of VSL can be written as
Z 4=0
p e !c
S D d4 x g R C e =0
.Lm C LF / @ @
; (7.95)
2 2 2
where 0 and !c are constants. The theory is similar to a scalar-tensor model and the
two are equivalent only when e / „.c/ c D const. A varying c can have an impact
in the history of the early universe and provide a conceptual alternative
R t to standard
inflation. In fact, the comoving particle horizon is modified as rp D 0 dt0 c.t0 /=a.t0 /
and, depending on the profile c.t/, acceleration and nearly scale invariance can
be sustained by a purely geometric effect [392]. For the same reason, the late-
time evolution can be affected by a dynamical cosmological constant with energy
density D c2 .t/= 2 [82, 393–395]. Due to the similarity with scalar-tensor
models and a lack of a compelling theoretical motivation, varying-coupling models
meet with about the same limitations as any other phenomenological explanation
of dark energy. However, if varying couplings were to arise in the effective semi-
classical limit of a fundamental theory of quantum geometry and particle physics,
these models might become strong contenders in explaining the evolution of the
universe.
7.6.2 Void Models
So far, the cosmological principle detailed in Sect. 2.1 has been the main cornerstone
whereupon to found a cosmological theory of the universe. Inhomogeneities have
been taken into account in the form of perturbations around a homogeneous
background, where the energy density of each constituent i is a well-defined average
i over sufficiently large volumes. The cosmological principle is an important statis-
tical assumption and, if the universe was strongly inhomogeneous, the mean values
i would be of little physical significance. Homogeneity has been experimentally
checked up to Hubble-horizon scales but deviations from a FLRW background
become more apparent at very large scales, where super-clusters of galaxies thread
along the outskirts of giant under-dense regions. Thus, such clusters find themselves
surrounded by large voids.
According to the Copernican principle, embedded in the cosmological principle,
the observed isotropy must be the same as seen at any other point in the universe.
However, we have no other reference point than the Earth and one might consider
a non-FLRW background where the observer occupies a somewhat special place
with respect to the surrounding inhomogeneities. In void models [396–401], we live
close to the centre (to avoid a strong dipole effect) of an under-dense region where
the Hubble parameter is larger than in the surroundings. The faster expansion of
the under-dense region produces a velocity gradient between the motion of inner
and outer galaxies. Objects in the near outer region are, for a given redshift z, at a
larger distance from us than in an FLRW universe and, hence, have a lower apparent
magnitude. An observer assuming a homogeneous and isotropic background would
interpret the data as nearby galaxies strongly redshifting away in acceleration.
A concrete example of void is given by the Lemaître–Tolman–Bondi (LTB)
model [402–407]. The local FLRW patch with cosmological constant is replaced
by a spherically symmetric neighborhood with D 0, filled with dust matter and
inhomogeneous in the radial direction, the centre being under-dense. LTB regions
can be patched together through FLRW junctions [408, 409] to obtain a more
realistic large-scale “Swiss cheese” model of voids. If we lived at (or near [406])
the centre of one such void region, we could still fit the isotropic set of redshift-
dependent cosmological observables and explain late-time acceleration without
invoking a non-vanishing cosmological constant [398–400, 410–424], as long as the
void is not too large (size . 1:5 Gpc [418]). In fact, from (2.147) and (2.154) one
can Taylor expand the luminosity distance dL D Œ .z/ around z D 0 and compute
the series coefficients as functions of ˝m0 and ˝0 . Repeating the same procedure
on an LTB background and matching the coefficients of the Taylor expansion, one
finds an effective ˝0 contribution determined by the parameters of the LTB metric.
The assumption that, by sheer coincidence, we should live very close (. 20 Mpc
[422]) to the centre of an under-dense region introduces a fine tuning in void models.
Still, their validity as a viable alternative to dark energy is ultimately assessed by
experiments. There are three main classes of models.
For a local (or small, or minimal) void [399, 400, 416], the under-dense region
has typical redshift size z 0:07–0:1 ( 200–300 h1 Mpc). Large-redshift
supernovæ data rule out this class, unless the outer regions are curved [421].
An off-centre observer can detect TB and EB correlations with a characteristic
multipole dependence, generated by gravitational lensing from the inhomogeneous
matter distribution. However, the signal is rather small and requires high-precision
polarization experiments [425].
Scenarios with large voids can be roughly divided into two categories, one
where the void has a compensating over-dense outer shell (z 0:5 – 1:5, size
1–2 h1 Gpc) and another where the void has no compensating shell and
approaches FLRW only asymptotically (size less sharply defined). With respect to
small voids, large-void scenarios are statistically less favoured by standard structure
formation [426], although some large voids are indeed observed. The details of
specific models as well as of independent data analyses may differ to the point where
it is not completely clear whether large-void models can be rendered compatible
with CMB, baryon acoustic oscillations, BBN, large-scale structure and type I
supernovæ observations [421] or not [418, 419, 422, 423, 427].
Even if, as it seems now likely, void scenarios cannot reconcile the tension
between different data sets, they constitute an important reminder that, no matter
how fascinating the cosmological constant puzzle, there is still a possibility that
(part of) its solution lies within traditional physics.
7.6.3 Unimodular Gravity
By definition, in unimodular gravity some of the components of the metric g

are non-dynamical. This can be implemented in three ways: preserving (in full or
p
almost) general covariance but demanding g to be non-dynamical [24, 428]; as
a partial gauge fixing of general coordinate transformations preserving the volume
p
(so that g transforms as a scalar rather than a density) [429–453]; or replacing
p
the volume weight g with a scalar with some internal symmetries and no kinetic
term [454–457].
p
The case where the determinant of the metric is a trivial scalar g D 1 is
particularly simple. The variation of the action (2.20) with respect to the metric is
p
now done with ı g D 0 instead of (2.21). Since g ıg D 0, only the traceless
part of (2.23) survives as a dynamical equation:

1 1
R g R D 2 T g T : (7.96)
D D
At the classical level, the dynamics of this model of gravity is equivalent to

general relativity: applying the covariant derivative to (7.96) and using the Bianchi
identity (2.32), one obtains the trace (2.26), where is an arbitrary integration
constant. Plugging (2.26) into (7.96), one obtains the Einstein equations (2.23).
p
When g is not globally fixed to 1, the dynamics is that of a scalar-tensor theory.
A justification for this version of unimodular gravity comes from the theory of the
massless spin-2 graviton on Minkowski spacetime [441, 447]. In general, classical
and quantum stability of the graviton is achieved if its dynamics is invariant under
transverse diffeomorphisms or transverse Fierz–Pauli symmetry, defined by (3.90)
with
@ D 0 : (7.97)
The name transverse stems from the fact that k D 0 in momentum space and
the covector is orthogonal to the graviton’s direction of propagation. Here one
is assuming that Minkowski spacetime is described in Cartesian coordinates, so
that the covariant derivatives with respect to the metric are plain derivatives
(as in (3.90) and (7.97)) and D det D 1. (Generalization to an arbitrary
coordinate system is possible [449, 450].) The most general quadratic action
respecting transverse diffeomorphism invariance is
Z Z
p
Sh D d x Lh D
D
dD x Lh ; (7.98)

1 1 ˇ
Lh D @ h @ h @ h @ h ˛@ h@ h C @ h@ h ; (7.99)
2 2 2
where ˛ and ˇ are constants and h D h . Apart from the graviton, the theory
also contains a spin-0 particle, which can be removed by extending the symmetry
group in two alternative ways. In the first, one drops the condition (7.97) and thus
moves from transverse to full diffeomorphisms, in which case one obtains the Fierz–
Pauli Lagrangian [458] at the linear level ((7.99) with ˛ D 1 D ˇ) and Einstein’s
gravity at the non-linear level. In the second, one imposes the Weyl symmetry (the
infinitesimal version of the conformal transformation (7.56))
hQ D h C ˚ ; (7.100)
where ˚ is a scalar. The trace h can thus be changed arbitrarily and, in particular,
set to zero, so that the scalar mode is removed from the spectrum. To have ShN D Sh
and invariance under both transverse diffeomorphisms and conformal symmetry, we
need ˛ D 2=D and ˇ D .D C 2/=D2 in (7.99).
When adding non-linear terms, one combines the background metric and the
graviton into the metric (here Œh D 1, while in (3.1) h is dimensionless)
p
g D C 2 2 h : (7.101)
p p
The metric gO :D j det g= det j1=D g is a tensor density such that Og D ,
equal to 1 in the Cartesian reference frame. This is the non-linear version of the
traceless condition h D 0. At the non-linear level [441, 449, 450], the resulting
theory in the Cartesian frame is unimodular gravity:
Z
1
SgO D dD x RŒOg ; (7.102)
2 2
while in terms of g one has an extra kinetic term for the metric. The equations
of motion are (7.96) with g replaced by gO . The key difference between general
relativity and unimodular gravity is in the choice of the symmetry group.
p
There are some theories which realize the unimodular constraint g D 1
from quite different first principles, for instance by regarding gravity as an emergent
phenomenon born out of more fundamental degrees of freedom or by assuming a
radically altered structure of spacetime geometry. We will describe these approaches
in Sects. 7.7 and 11.6, respectively. While the classical dynamics of general
relativity and unimodular gravity is the same, there are differences in the quantum
theories which depend on the (and disappear in some) formulation of the unimodular
paradigm [451, 452].
In all these scenarios, the cosmological Rconstant is an R integration constant
p
associated with the total D-volume V D dD x g D dD x. An integration
constant can be set at one’s leisure without incurring into the conceptual problem
of the fine tuning of a physical parameter in an action. This constant is unrelated
to the zero-point energy of matter [450]. At the quantum level, the vacuum zero-
point energy of matter Vvac does not gravitate as in general relativity, since the
minimal coupling with the metric is trivial here. Vacuum bubble diagrams are
thereby factored out of the partition function, just as in quantum field theory on
Minkowski spacetime.
Although the magnitude of the vacuum energy as prescribed by particle physics
becomes irrelevant for the cosmological constant problem, a small value such
as (7.11) may be natural in quantum unimodular gravity. In Hamiltonian formalism,
the wave-function of the Universe becomes a superposition of states with different
values of the cosmological constant, which allows for a probabilistic reinterpretation
of the problem (see Sect. 10.2.4). In fact, the Euclidean partition function features
a functional integration over all possible values of [437, 438, 444]:
Z Z Z
N ;
ZD d./ ŒDgŒD eSg Œg; ;
D d./ eSeff ŒNg; ; (7.103)
where ./ is a generic smooth quantum measure weight with an a priori uniform
distribution (i.e., the point D 0 is neither preferred nor disfavoured), the measures
in square brackets are functional measures in terms of the metric and a generic
matter scalar field, and gN and N are the background fields minimizing the effective
action. In vacuum, the background metric is de Sitter, implying in D D 4 that
Z
3 2
ZD d./ exp : (7.104)
8
The contribution D 0 dominates exponentially.
7.6.4 Analogue Gravity and Condensates
A yet different perspective regards the cosmological constant problem as the

reflection of infrared effects. In this case, is determined by macroscopic,
“collective” degrees of freedom describing effective regimes where fundamental
physics intervenes only indirectly. This limit is often called hydrodynamical, as
it entails a long-wave-length coarse graining of the microscopic structure. Such
situation often happens in condensed-matter systems, which in fact provide a
stimulating inspiration to understand gravity in regimes usually inaccessible by
modern experiments. For instance, the excitations in certain superfluid systems
propagate on an effective background which closely resembles a Lorentzian curved
spacetime in a strong-curvature regime. This motivated the study of a number of
models of analogue gravity, where the tools of condensed-matter problems are
applied to astrophysical or cosmological situations [459, 460]. One such example
will be reported in Chap. 14.
In many of these models, a formal analogy with superfluid phases indicates
that is an emergent phenomenon [461–463]. More precisely, the cosmological
constant neither is the energy of the vacuum of a field theory nor is dynamically
determined by a basic scalar field. In some cases, it is related to the grand canonical
energy of the vacuum [464]. In others, it corresponds to the expectation value of a

condensate, whose dynamics governs the energy scale of the problem. A condensate
is a composite object resulting from the binding together of some fundamental
degrees of freedom into an energetically favourable configuration. We mention three
possible condensate scenarios.
(i) In loop quantum gravity and group field theory, the condensate might be
composed by the gravitational degrees of freedom themselves, under some
extreme conditions. These cases can provide an IR resolution or alleviation
of the cosmological constant problem but rooted in a UV modification of
gravity. We postpone a brief discussion of this possibility, which radically
differs from the next two as far as physical interpretation and mathematical
tools are concerned, to Sects. 9.4 and 11.5.
(ii) Gluon condensate. By thermodynamical arguments and without invoking
quantum gravity, one can devise models where is determined by an exten-
sive thermodynamic variable characterized by a Lorentz-invariant quantum
vacuum which is “self-sustained,” i.e., supporting an equilibrium state even
in the absence of an external environment [465, 466]. Such variable can
be realized by the fields of the Standard Model without introducing exotic
particles or modifying general relativity. An example is the gluon condensate

q :D hF a
Fa i, i.e., the vacuum expectation value of the squared QCD
(quantum chromodynamics) field strength [467]. While .q/ D .0/ D 0
in perfect vacuum, a small non-zero value can be obtained in the presence of
thermal matter, also leading to breaking of Lorentz invariance. The effective
cosmological constant term is time dependent, it dominates after the cold-
dark-matter era and it is naturally small (the energy density of a condensate
is typically much smaller than the free vacuum energy of the fields it is made
of). Thus, both the coincidence problem and the old cosmological constant
problem are at least relaxed.
(iii) Fermion condensate forming in the early universe [468–474]. One starts with
ordinary general relativity plus matter. The latter is described by the fermionic
Dirac action with a four-fermion interaction term:
Z
i J5a J5a
S D d4 x e N a ea r C c:c: C ; (7.105)
2 M M2
where ea is the gravitational field introduced in Sect. 6.3.1,3 e D det ea ,

and N are a spinor and its conjugate, “c.c.” indicates the complex conjugate
of the first term, J5a D N 5 a is the axial fermionic current and M is a
mass scale. The interaction term can be decomposed into a scalar interaction
3
First-order formalism is mandatory in order to couple fermions with gravity consistently.
. N /2 plus other contributions which can be ignored for simplicity. Thus, the
interaction reduces to
Z Z
. N /2 2
N / M 2 ;
Sint D d4 x e D d 4
x e . (7.106)
M2 4
where in the second equality we have introduced the auxiliary scalar , which
acts as a mass term for the fermions. On a FLRW background, a non-zero value
for the auxiliary field would signal a cosmological fermionic condensation.
The quantum theory
R associated with thisR system is encodedˇ into the path
integral Z D ŒDŒD ŒD N eiS D: ŒD eiSeff ' eiSeff ˇsp , where one
integrates over the fields, defines the effective action Seff and approximates
the functional integral by the saddle point (mean-field approximation). The
effective action Seff can be evaluated by performing the Gaussian integrals in
terms of the fermionic coordinates and leads to a non-perturbative potential
Veff ./ for the fermion gap. At equilibrium, the system lies at the minimum of
the potential and the gap equation @Veff =@ D 0 determines the mass M./
as a function of and, in turn, as a spacetime function such that M remains
constant. Introducing a chemical potential , which corresponds to having a
non-zero number density n of fermions, the total effective energy density can
be shown to be [472]
2 2
gap D Vmin C n D 2
82 2N C 3 C 2 ln 2 ; (7.107)
32
where M has been reabsorbed by the gap equation and N D O.102 / is a regu-
larization free parameter. (The four-fermion interaction is non-renormalizable
in Minkowski spacetime and the divergence cannot be eliminated. A standard
approach is to interpret the regularization parameter in terms of a finite physical
cut-off scale. On curved spacetimes, the issue of renormalization is still open.)
The energy density (7.107) is called a “gap” because it constitutes a non-
perturbative hiatus in the energy spectrum with respect to the perturbative
vacuum D 0. Via the gap equation, the gap size gap is governed by M.
An analysis of the effective barotropic index weff :D 1 .1=3/d ln jgap j=
d ln a shows that the model can lead to late-time acceleration. At primordial
energy scales M, gap is negative and ensures the existence of a
non-singular bounce point where gap precisely cancels the energy density of
ordinary matter [471]. After the bounce, the gap energy density redshifts away
faster than regular matter and remains sub-dominant as compared to ordinary
matter and radiation. However, once M, we gradually fall into a
regime where M and gap / 4 2 M 2 2 C 4 M 4 . Once the
matter energy density drops down to m M 4 , we enter the present dark-
energy-dominated quasi-de Sitter phase. The details of the scenario and the
size of the effects depend not only on the behaviour of the chemical potential
across the early inflationary stage and the subsequent reheating era [474], but
7.7 Emergent Gravity 351
also on the issues of regularization and renormalization, which have not been
fully assessed yet. In particular, a very small value M D O.1/ meV of the
mass parameter is required. Therefore, one is indeed assuming that a bare O.1/
cosmological constant miraculously cancels by virtue of some mechanism: the
old cosmological constant problem is not solved. It is relaxed, however, since
the cosmological constant is strongly suppressed and the level of fine-tuning is
reduced to just one part over 100, via the choice of N.
7.7 Emergent Gravity
The condensate models briefly described in Sect. 7.6.4 regard as an emergent phe-
nomenon within frameworks where gravity is a fundamental force. One step further
is to consider gravity itself as an effective long-range interaction. An often-invoked
analogy is that of hydrodynamics as a macroscopic model emerging from molecular
physics. The graviton can be conceived as a composite particle or the collective
excitation of some underlying fundamental degrees of freedom. This collective
mode, born from the statistical description of an altogether different microscopic
system, arises through some sort of phase transition. In this purely classical context,
not only the cosmological constant [475] but the whole gravitational sector emerges
from other degrees of freedom and there is no meaning in quantizing gravity per se.
Emergent gravity [476, 477] is a special class of analogue gravity models with
widely and wildly different incarnations. For instance, the linearized graviton can be
obtained as a Goldstone boson associated with the breaking of vacuum degeneracy
in interacting fermionic or scalar field theories, or from a spin-2 particle in flat
space [477–480]. However, it is difficult to recover the fully non-linear gravitational
dynamics from these models. Another idea is to get an effective metric from
some approximation of a classical field theory and an effective dynamics for the
metric from low-energy quantum corrections. A scalar field theory in Minkowski
spacetime produces, when linearized around a given background 0 , an effective
p
metric given by gg :D @2 L=Œ@.@ /@.@ /j D 0 [481, 482]. A dynamics for
the effective metric is then generated automatically by the quantum effective action,
which induces an Einstein–Hilbert term for g at the one-loop level [253, 481, 482].
Other corrections can modify the effective equations of motion in a way deviating
from simple Einstein gravity, by adding higher-order “curvature” and non-minimal
coupling terms.
If gravity was an emergent force, it should be possible to describe its large-scale
dynamics by macroscopic effective degrees of freedom, while near singularities
and in high-curvature regimes the microscopic structure of the fundamental theory
(whatever it may be) should become apparent. A similar transition appears in
thermodynamics: its fundamental substratum is quantum statistical mechanics but
its range of application encompasses macroscopic bodies. Perhaps surprisingly, this
example is not just an analogy. To each vector in spacetimes with horizons, one can
assign thermodynamic potentials: entropy S and temperature T [30–33, 483–486],
energy E, free energy F, enthalpy Q,4 and so on. These potentials obey thermody-
namical laws, which govern the dynamics within the horizon. In the language of
field theory, the potentials are Noether charges and the thermodynamical laws are
the conservation laws for the associated Noether currents. In the context of emergent
gravity, however, Noether charges are derived before the currents.
In general, these thermodynamical laws are equivalent to Einstein’s equations
and reproduce the dynamics of gravity and matter.5 This result holds for generic
backgrounds with a horizon [487–495] and for specific metrics. Examples of
spacetimes with a global horizon are those with spherical symmetry (black holes)
[496–499] and the FLRW universe [499–505] (where the Hubble horizon is
associated with the de Sitter temperature (5.132)). The range of applicability is
much wider than where a global horizon is available; in general, local horizons form
around any observer in constant acceleration. Moreover, the form of the potentials
is dictated by the type of gravity one wishes to obtain in the equations of motion.
Modulo a difference of utmost importance for the cosmological constant problem,
in this way one obtains the dynamics of the Einstein–Hilbert action as well as of
higher-order actions such as f .R/, Gauss–Bonnet and Lanczos–Lovelock gravity.
Here we limit the discussion to ordinary Einstein gravity.
In all these constructions, spacetime is a pre-existing manifold with metric g
and spanned by coordinate charts, as in usual differential geometry; the metric of
such spacetimes, however, is not dynamical and the equations of motion are derived
from thermodynamical laws which are assumed to be fundamental.
One such formulation by Padmanabhan [490–495, 506] elects the horizon of a
local Rindler observer, on which we now make a short digression.
7.7.1 Rindler Observer and Null Congruences
Consider D-dimensional Minkowski spacetime charted by coordinates .T; X; x? /,

where we singled out one spatial coordinate X. In a Rindler frame, time T and the
coordinate X are redefined as
T D % sinh.gt/ ; X D % cosh.gt/ ; (7.108)
4
For vanishing chemical potential, the enthalpy is E F D T S and measures the difference
between the energy and the free energy of a finite-temperature system.
5
An ultra-simplified instance of this mechanism is the recovery of the Friedmann equations with
K D 0 D from Newtonian and thermodynamics considerations. Let D D 4 and consider an
expanding ball of volume V D 4a3 =3 filled with energy (mass) E D M D V . Assuming the
hypothesis of adiabatic expansion (no change in entropy, dS D 0), the first law of thermodynamics
dE CP dV D 0 is equivalent to the continuity equation P C3H. CP/ D 0. On the other hand, the
first Friedmann equation can be interpreted as an energy conservation equation mPa2 =2GmM=a D
0, where the first term is a kinetic energy of a small mass m at distance a from the observer in the
uniform medium and the second term is Newton’s potential.
Fig. 7.4 Two-dimensional

Rindler chart (7.108) (with
g D 1) in coordinates .%; t/
plotted against the
two-dimensional local inertial
frame in Cartesian
coordinates .X; T/. The
dashed line is the X > 0
portion of the light cone
X 2 T 2 D 0 of the local
inertial observer at
.X; T/ D .0; 0/. It
corresponds to the Rindler
horizon (7.110). Some
% D const and t D const lines
are shown for reference.
Rindler observers correspond
to % D const hyperbolæ
where g is a constant. The line element is then
ds2 D dT 2 C dX 2 C dx2? D g2 %2 dt2 C d%2 C dx2? : (7.109)
The two charts are depicted in Fig. 7.4 in the .X; T/ plane. A Rindler observer is at
rest in the Rindler frame and thus corresponds to a hyperbola with constant %. This
observer has constant proper acceleration and g is the proper acceleration at % D 1.
In the following, we set x? D 0. Since there can be no incoming information
from outside the portion of light cone enveloping the Rindler frame, this null
surface acts as a horizon for the Rindler observer. The light cone is defined by
X 2 T 2 D %2 D 0 or, equivalently, by T D ˙X, which corresponds to the two null
hypersurfaces t D ˙1 (infinite future and past). Therefore, the Rindler horizon is
the locus of points
HR .%; t/ D f.%; t/ j % D 0g D f.%; t/ j t D ˙1g : (7.110)

The proper acceleration of a Rindler observer increases as the horizon is approached

asymptotically. One can show that T D g=.2/ is the Davies–Unruh temperature
of the horizon [507].
Presently, we move to the covariant setting of a D D 4 curved spacetime. In
the neighborhood of any point P, one can define a local inertial frame centered at
P and spanned by the Cartesian coordinates .T; X; x? /. A local Rindler observer is
then a Rindler observer charted in this neighborhood, which is called local Rindler
frame. If u is the four-velocity of a local Rindler observer, its acceleration with
respect to the corresponding free-falling observer will be a D uP D u r u . The
Davies–Unruh temperature associated with this observer is then
Na
T D ; (7.111)
2
p
where N is the lapse function and a D ja a j. Notice that T is a local quantity.
One can attach a local Rindler frame to any point in a generic null surface
@V. Therefore, locally @V can be mapped to the Rindler horizon (7.110) of a
local Rindler frame. This null surface is globally described by a family of freely
propagating light rays, a null congruence. An element n of a null congruence6 is a
vector normal to @V such that
n r n D kn ; n2 :D n n D 0 ; (7.112)
where k is a scalar. In a local Rindler frame, k D g is a constant; if k D 0, one has a

null geodesic congruence [508]. We also take a parametrization of the null vectors
n ./ D dx =d, where the affine parameter introduced in Sect. 6.1.1 is defined
by n r D 1:
d
n r D : (7.113)
d
The expansion of the congruence is :D Œr ; n , so that by definition r n D

k C . All these expressions can be compared with their counterparts (6.4), (2.47)
and (2.46) for a time-like congruence.
6
In Chap. 2 we used the symbol n for a generic null vector but here it will denote a congruence.
7.7.2 Dynamics
Having a horizon at hand, one can proceed with the construction of thermodynam-
ical potentials within or at the horizon. In particular, for any null congruence n in
@V, the quantity
Q :D r n r n .r n /2 (7.114)
can be rigorously shown to be the heat or enthalpy density of @V [495]. Equa-

tion (7.114) only depends on the derivatives of the congruence because it is required
to be invariant under covariantly constant translations of the null vectors, just like
the entropy and enthalpy of an elastic solid are invariant under translations. On the
other hand, translation invariance is broken in the presence of matter and the heat
density for matter T n n is quadratic in the null vectors, where T is a symmetric
tensor. The analogy with elastic solids (constancy of the heat coefficient, leading to
r T D 0) and the fact that, for a perfect fluid, the heat density would simply be
T n n D C P immediately identify T with the energy-momentum tensor of
matter fields.
The total heat of the null surface @V is the total heat density integrated over
the surface and constitutes a conserved charge associated with the bulk V whose
boundary is @V. If @V is spanned by embedding coordinates yi and has induced
metric ij (i; j D 1; 2), the total heat is [495]
Z Z
1 2
p
QŒn :D d d2 y Q C 2 T n n : (7.115)
8 1 @V
Demanding that the potential has an extremum for all Rindler observers in
spacetime yields the field equations
ıQ
D 0; (7.116)
ın
where the variation is taken with respect to an arbitrary null vector not necessarily
belonging to @V.
To calculate (7.116), we notice that R n n D n Œr ; r n D Qr . n /.
The last term is a total covariant derivative and contributes with a vanishing
boundary term in (7.115), so that the integrand can be rewritten as QC 2 T n n !
.R C 2 T /n n . Equation (7.116) is then equivalent to [490, 495]

R 2 T n D 0 : (7.117)
Since this equation is valid for all n in spacetime, we can contract (7.117) with
any n . The expression .R 2 T / n n D 0 is invariant if we add a term fg
in the bracket, where f is a function. Using the continuity equation (2.31) and the
contracted Bianchi identity (2.32) on .R C fg 2 T / n n D 0, we find that

f D R=2 C , where is an arbitrary constant. The equations of motion can be
rewritten as

G C g 2 T n n D 0 : (7.118)
Just like, when applied to all free-falling observers, the equivalence principle
determines how gravity interacts with matter in any local inertial frame, so the
maximization (7.116) of the total heat for all local Rindler observers establishes
the dynamics of gravity.
Since the metric g is not varied to obtain Einstein’s equations, the theory
may be regarded as a special case of unimodular gravity (Sect. 7.6.3) where all
the metric components are non-dynamical. The underlying construction is more
involved and differently motivated, since in the present case gravity is not modified
in its symmetry structure but is assumed to be emergent from other degrees of
freedom.
Contrary to (2.23), the equations of motion (7.118) are invariant under the shift
symmetry (7.26) because 0 g n n D 0. Therefore, Padmanabhan’s theory is
not just a reformulation of general relativity in the language of thermodynamics.
The fact that we obtained unimodular gravity rather than Einstein’s seems a rather
general characteristic of the thermodynamic derivation of the equations of motion
[487], through the use of null hypersurfaces.
7.7.3 Holographic Equipartition
Equation (7.118) can be obtained from the entropy S associated with a three-
dimensional bulk V rather than the enthalpy of its boundary [490, 492]. In particular,
one can devise a natural definition for the total energy E in V which is functionally
equal to the heat content of @V [495]. The model is therefore a concrete example of
holography [30, 31, 34, 486, 509, 510], a principle according to which, for a wide
class of gravitational and field theories, the dynamics in the bulk is determined by
the physics (a thermodynamical one, in the present case) at the boundary.7
7
Without invoking thermodynamical arguments, the equations of motion (7.118) can be obtained
in metric formalism by splitting the Einstein–Hilbert Lagrangian LEH D Lbulk C Lsur into a bulk
and a surface term and, then, varying only the surface term with respect to special variations of
the metric encoding a normal displacement to a null surface [511–513]. This is possible thanks
p p
to the holographic relation gLsur D .D=2 1/1 @ Œg @. gLbulk /=@.@ g /. The
same procedure is generalizable to an arbitrary gravitational Lagrangian [514]. The lack of a
thermodynamical interpretation in metric formalism, however, does not explain why gravity is
holographic.
To make holography R morepapparent, we specify surface and bulk degrees of

freedom. Let A D @V d2 y be the area of a generic two-surface @V (not
necessarily null) bounding a region with three-volume V. The boundary of a spatial
region V is compact, while @V is non-compact in the case of a Rindler horizon.
The number of degrees of freedom on @V is defined by counting the number of
elementary areas l2Pl in A:
A
N@V :D : (7.119)
l2Pl
This reproduces the entropy-area law for black holes [30, 32, 33, 165], as
N@V A
SD D 2: (7.120)
4 4lPl
Averaging the local temperature (7.111) over the surface, one gets the average
temperature of @V:
Z
1 p
T@V :D d2 y T : (7.121)
A @V
The number of effective degrees of freedom in the bulk V are then defined as minus
the total gravitational energy E in the bulk divided by the thermodynamical energy:
Z p
E
NV :D 1 ; E :D d3 x h TK ; (7.122)
k T
2 B @V V
where kB is Boltzmann constant, h is the determinant of the induced metric h on

V and

TK :D 2N T 12 g T u u (7.123)
is called Tolman–Komar energy density [515, 516]. The total energy E is defined as
the volume integral of the Tolman–Komar energy density. For a perfect fluid (2.40)
with N D 1, TK D C 3P D .1 C 3w/ is negative definite for a barotropic index
w < 1=3 and positive definite for ordinary matter (w > 1=3). The number of
bulk degrees of freedom takes positive values only for matter driving acceleration.
By simple manipulations of the Noether currents of the system, one can show
that [495]
Z
1
2 d3 x h L D kB T@V .N@V NV / ; (7.124)
V 2
where
1 p
D h .K h K/ (7.125)
2 2
is the momentum conjugate to h ,

K :D 12 Lu h D 1
2 u r h C h r u C h r u (7.126)

is the second fundamental form, K :D K D h K and L is the Lie derivative
with respect to the evolution vector D Nu corresponding to observers with

four-velocity u normal to the constant-t hypersurface. For later use, we notice that
h L D L .h / L h

1 h p i
D 2 2L . h K/ 2N.2 2 / K
2
N p p
D 2 h u @ K C K h h K C 2 2 K

N p
D 2 h u @ K C K K : (7.127)

The left-hand side of (7.124) vanishes for any static spacetime, where an observer
would experience holographic equipartition [517–519]:
NV D N@V : (7.128)
Thermodynamically, it corresponds to the relation E D 2T S. In this case, the

counting of degrees of freedom in the bulk and at the boundary coincide. For a
general dynamical spacetime, the difference between NV and N@V governs the rate
of change of gravitational momentum and the dynamical evolution. Note that, from
what said below (7.123), holographic equipartition can be achieved only in the
presence of dark-energy types of matter.
Equation (7.124) is foliation dependent but is valid for all foliations of spacetime,
hence it carries the same dynamical information as Einstein’s equations. We can
check this for de Sitter cosmology and an FLRW spacetime filled with a perfect
fluid. In the first case, one has holographic equipartition and N@V D NV . We take
V D VH to be the N D 1 Hubble volume (7.27), so that the boundary surface has
an area A D 4R2H D 4H 2 and the temperature is (5.132). With the Boltzmann
constant reinstated,
H
T@V D T D TH D : (7.129)
2kB
Therefore,
2.4/2 4VH .4/2

N@V D ; NV D . C 3P/ D . C 3P/ ; (7.130)
2H2 H 3H 4
with P D and we obtain the first Friedmann equation (2.81) for K D 0
and a pure cosmological constant. The FLRW case is more involved. Comoving
observers follow geodesics with zero acceleration and they do not see a Davies–
Unruh temperature. Therefore, N@V D 0 and
1
kB T@V .N@V NV / D E D . C 3P/V ; (7.131)
2
R p
for any three-volume V D d3 x h. On an FLRW background,
H 3H
K D h ; KD ; (7.132)
N N
so that the left-hand side of (7.124) in synchronous gauge has
6p P
2h L D h.H C H 2 / :
(7.127)
(7.133)
2
Combining (7.131) and (7.133), we get the second Friedmann equation (2.82)
without term,
2
P C H 2 D aR D . C 3P/ :
H (7.134)
a 6
The presence of a preferred time foliation permits to recast the cosmological
dynamics in a way similar but inequivalent to the background-independent equa-
tion (7.124) [520, 521]:
dVH
D l2Pl .N@V NV / ; (7.135)
dt
which corresponds to (7.134) under the replacements (7.27), (7.129) and (7.130).
Here, the degrees of freedom of the spacetime bulk emerge during the dynamical
evolution: as the Hubble horizon and the causal patch VH expand, more bulk modes
become accessible and achieve holographic equipartition.
7.7.4 Cosmological Constant Problem
Having set the main ingredients of the theory, we turn to the problem [25–
29, 495, 521]. The first step towards its solution is (7.118): as in any classical
unimodular gravity model, the cosmological constant in Padmanabhan’s theory is
an integration constant. The most direct consequence of this fact is that gravity
is now insensitive to the value of the cosmological constant. Therefore, if we
could find a fundamental principle fixing the cosmological constant to the observed
value (2.118), the transformation (7.26) would not change it: the shift ! C0
in the matter sector would be reabsorbed by an opposite effect in the gravitational
sector.
The second step is to identify such principle and, with it, to explain the 4
puzzle (7.30). Recall that Nc is the number of modes which became accessible to
our causal patch VH during the transient phase XY (radiation-dust era) between
two quasi-de Sitter regimes. According to the emergent-gravity scenario of (7.128),
this number should coincide with the number of degrees of freedom populating
the Hubble sphere @VH . On the other hand, since the line k in Fig. 7.1 is at
45ı , the expansion rate of the era XY is the same as of the inflationary era PX,
aX =aP D aY =aX . But 4 is precisely the number of degrees of freedom of the
boundary of an elementary Planck ball, N@VPl D .4l2Pl /=l2Pl D 4. Therefore, a
possible solution of the 4 puzzle and of the problem could reside in a realization
of a theoretically and experimentally consistent Planck-scale inflationary regime
within the emergent-gravity paradigm, so that
Nc D N@VPl : (7.136)
Another hint that this theory may be on the right track is the following. Whenever
one has a thermodynamical description of a system, the latter must be constituted
by microscopic degrees of freedom. Therefore, in this particular incarnation of
emergent gravity one is still entitled to look for “quanta of geometry,” although these
will not correspond to graviton modes. A natural bridge between thermodynamics
and a microscopic description of these fundamental degrees of freedom is provided
by statistical mechanics. In general, one can define a partition function Z via a
density of energy states %.E/ such that the thermodynamical potentials such as S and
the free energy F are recovered. In the present scenario, we have defined the surface
degrees of freedom as proportional to the area A of the surface @V which, according
to the entropy-area law, is proportional to the entropy S. In order to reproduce the
entropy-area law, this density of states must be exponential,

E2
%.E/ ' exp 4 2 ; (7.137)
mPl
to lowest order in an E=mPl expansion [522, 523]. Then, the entropy reads S.E/ D
ln % D 4E2 =m2Pl ! S.A/ D A=.4l2Pl / D N@V =4. This result is valid in all
spacetimes with a hypersurface with infinite redshift, such as a black hole. Roughly
speaking, the event horizon “stretches” virtual high-energy field excitations (repre-
senting the quantum interactions of matter with geometry) to sub-Planckian energies
and allows them to become real modes, which then populate thermodynamical
energy levels. Via (7.137), one can show that the enforcement of the entropy-area
law implies the presence of a minimal area 4l2Pl , so that below the Planck scale
spacetime ceases to have the ordinary properties of a continuum [524].8 Since the
cosmological constant (7.31) can be interpreted, in this context, as an energy times
a density of states, there is an indication that
Nc / N@V ;
similar but not quite equal to (7.136). We still lack the details of the theory at the
Planck scale to make this correspondence, and a true solution to the old problem,
more precise.
Note that, in general, a holographic universe prefers a small or vanishing
cosmological constant. The argument is simple and is based on the entropy-area
law (7.120) applied to the whole universe. For a homogeneous and isotropic
universe, the area A in question is the surface A D 4=H 2 of the Hubble horizon
enclosing the causal region, so that for a de Sitter patch S D 3m2Pl =. The
Boltzmann probability distribution goes as the exponential of the entropy [527]:

3m2Pl
P./ / expŒS./ D exp ; (7.138)

which is peaked at D 0. Therefore, it is both conceivable and consistent that a
solution of the 4 puzzle (which implies a small observed cosmological constant)
relies on holography.
7.1 Numerical cosmology and quintessence 1. Write the first Friedmann

equation in (7.33), the scalar-field equation (5.39) and the barotropic index
w D P = in a way both convenient and stable for a numerical code, i.e.,
E 0 D Ef .N; y; z/, where X
in the vector form X E D .z; y/t , z / 0 , y / and Ef
is a two-component vector. Here a prime denotes a derivative with respect to
the number of e-foldings N :D ln a=a0 , where a0 is the scale factor today. All
quantities should be dimensionless.
8
The presence of a minimal length in Padmanabhan’s theory can be inferred also by other
arguments independent of (7.137) [525, 526].
Solution Since the cosmological evolution spans many orders of magnitude in

synchronous time, for numerical work it is useful to switch to the number of e-
folds N D ln.1 C z/, which also coincides (up to a sign) with the redshift.
In particular, near the end of inflation and the beginning of reheating Nreh 6
53 (z < 1023 , equation (2.164)), at big-bang nucleosynthesis NBBN 21
(z 109 , equation (2.164)), at matter-radiation equality Neq 8 (z 3400,
equation (2.90)), at recombination Nrec 7 (z 1100, equation (2.131)), at
dark-energy domination Nacc 0:7 (z 1, equation (2.126)).
In this way, an exponential time span is rendered linear and possible precision
issues are prevented. First, we rewrite the Friedmann equation in (7.33) in terms
of dimensionless densities. Let crit;0 D 3H02 = 2 be the critical density today and
% D =crit;0 a density normalized to the present value of the critical density, not to
be confused with the density parameter (2.83). Then, % D %m C%r , where (see (2.87)
and (2.88))
%r D ˝r;0 e4N ; %m D ˝m;0 e3N : (7.139)

p
Calling x :D H 2 =H02 , y :D = 3 and U :D V=crit;0 , the first Friedmann equation
reads
x D %m C %r C %y ; (7.140a)
0 2
.y /
%y D x C U. y/ ; (7.140b)
2
where we used yP D Hy0 . Since yR D H 2 Œ y00 C x0 y0 =.2x/, using (7.140) the equation
of motion (5.39) for quintessence reads
1
y0 D z ; z0 D .3%m C 2%r /z C 2.3zU C U;y / ; (7.141)
2x
where
%m C %r C U
xD2 : (7.142)
2 z2
The right-hand sides of (7.141) are only functions of N, y and z, as requested. The
barotropic index is

1 . y 0 /2
w D wy D x U. y/ : (7.143)
%y 2
7.2 Numerical cosmology and quintessence 2. Write a portable code

for a general U. Integrate (7.141) numerically for the inverse power-law
potential (7.45) with n D 6 and plot %y and wy for a scaling solution.
Solution The present density parameters have values ˝m;0 0:31 and ˝r;0
5 105 , where we used (2.12), (2.96) and (2.99). (As an input in the numerical
code, we include more digits but all results are unaffected by small changes in the
matter and radiation densities.) Next, the equations of motion should be integrated
forwards in time from before matter-radiation equality. We choose the initial instant
Ni D 28 (z 1012 ), an epoch between the end of reheating and the start of
nucleosynthesis, but modulo convergence issues one can adapt the code to start just
at the end of inflation. At N D Ni , the energy density of quintessence should be
comparable with or lower than the contribution of radiation. Assuming a small
initial velocity, this translates into the relation %y;i ' U . %r;i D %r .Ni /. In
dimensionless variables and constants, the potential (7.45) reads U D A=yn , where
A D M 4Cn . 2 =3/n=2C1 =H02 . From (7.46), we get the order of magnitude that A
should have in order to obtain a viable solution: A ˝ 1. Therefore,
1n
A 45
yi & 10 n : (7.144)
%r;i
For n D 6, we get a rough lower limit yi 108 . Any value around or above this
will do; we set then yi D 105 .
The initial velocity may be set depending on whether we want a scaling or a
creeping asymptotic solution. For scaling attractors, we can consider two qualitative
cases. In the first one (overshooting), the initial energy density %y;i is much greater
than the tracker value. The energy density %y then drops quickly as the solution
overshoots the scaling attractor, which is reached after a transient regime with wy
const. In the second case (undershooting), the field starts in a freezing regime with a
negligible kinetic energy, . y0i /2 U. yi /, and %y is stuck at a constant value. At late
times, it thaws and reaches the scaling attractor. We concentrate on the overshooting
case without loss of generality, setting y0i D 102 (positive, since the field increases
when it rolls down its potential). One can create an undershooting example by fixing
y0i D 0.
Having fixed the initial values yi and y0i , our numerical code should adjust the
normalization A of the potential to yield the solutions which satisfy the boundary
condition today, x.N D 0/ D 1. Although the problem is defined by boundary
conditions at the present time, it is unwise to attempt to integrate backwards from
them, for instance by choosing yi freely while y0i is constrained by the Friedmann
equation at N D 0. This is because strong attractors become repellers in backwards
integration, leading to a rapid growth of numerical instabilities. Instead, we integrate
forwards using a shooting method to adjust the initial conditions to obtain the correct
present properties. The shooting method transforms a boundary-value problem into

an initial-value problem. It consists in having A run over a sensible range, in the
present case around values 0:1 < A < 100. The code runs a loop over A. For each
value of A starting from 0:1, the solution ynum .N/ is found and plugged into (7.142)
at N D 0 to check the value xnum .0/. If the latter is away from 1 above the given
tolerance (about one part over 103 , according to the experimental error in (2.128)),
then A is increased by one step (typically, between 103 and 101 ) and the code is
rerun until a normalization is reached such that jxnum .0/ 1j < 103 .
p Let us implement this strategy in MATHEMATICA. First, we specify the constants
2 =3 (kap3), H0 (H0), ˝r;0 (OMr0) and ˝m;0 (OMm0) associating to them a
numerical value:
kap3 = SqrtŒ8 Pi=.3 .1:22 10^ 19/^ 2/I

H0 = 2:133 0:68 10^ -42I
OMm0 = 0:1426 0:68^ -2I
OMr0 = 2:469 10^ -5 0:68^ -2I
Second, we write down all equations valid for any potential U (dependent on
two parameters A and n) and its y derivative Upr: the radiation and matter
density parameters rhor and rhom, equation (7.141) (eq1, x and eq2) and
equations (7.140b) (rhoQ) and (7.143) (wQ), everything as a function of the number
of e-foldings ne:
rhorŒne_ := OMr0 ExpŒ-4 ne
rhomŒne_ := OMm0 ExpŒ-3 ne
eq1Œne_ := y0 Œne - zŒne

rhomŒne + rhorŒne + UŒ y; ne; A; n
xŒne_; U_; n_ :=
1 - zŒne2 = 2
1
eq2Œne_; U_; Upr_; n_ := z0 Œne + ..3 rhomŒne + 2
2 xŒne; U ; n
rhorŒne/ zŒne + 2 .3 zŒne UŒ y; ne; A; n + UprŒ y; ne; A; n//
rhomŒne + rhorŒne + UŒY; ne; A; n
XŒne_; U_; A_; n_ :=
1 - .Y0 Œne/2 = 2
XŒne; U; Anum; n.Y0 Œne/2
rhoQŒne_; U_; n_ := + UŒY; ne; Anum; n
2

XŒne; U; Anum; n .Y0 Œne/2
wQŒne_; U_; n_ :=
2

- UŒY; ne; Anum; n rhoQŒne; U; n
We defined x twice, once (function x, included in the equations of motion)

exactly as in (7.142) and another (function X) with z ! y0 , the second form being
the one to be evaluated on the numerical solution Y. Notice that the symbols U and
Upr denote functions in the right-hand sides but variables in the left-hand sides, so
that eventually one can plug in different potentials by specifying explicit functional
forms. We do it now for an inverse power law:
UnŒ y_; ne_; A_; n_ := A yŒne-n

UprnŒ y_; ne_; A_; n_ := -n A yŒne-n-1
massŒn_ := .H0^ 2 Anum=kap3^.2 C n//^ .1=.4 C n//
where the last expression relates M to A. At this point, we create a loop with the Do
command for the n D 6 potential:
DoŒ
Y = NDSolveValueŒfeq1Œne == 0; eq2Œne; Un; Uprn; 6 == 0; yŒ-28 == 10^ -5;
zŒ-28 == 0:01g; y; fne; -28; 10g; SolveDelayed -> TrueI
IfŒXŒ0; Un; A; 6 > 1 - 10^ -3 && XŒ0; Un; A; 6 < 1 + 10^ -3; Anum = AI
PrintŒfAnum"=A"; massŒ6"GeV=M"; XŒ0; Un; 6"=x(0)"gI BreakŒ;
fA; 59; 60; :01g
At each new iteration with a given value 59 < A < 60 (the range is guessed a
priori), NDSolveValue is the numerical integration of the differential equations
of motion with given initial conditions at N D 28 (in this example, for the tracking
overshooting solution depicted in Fig. 7.2a), up to some future time N D 10.
The output is the interpolating function Y[ne]. The option SolveDelayed ->
True avoids singularities in the solved form of the equations. The conditional If
checks whether jxnum .0/1j < 103 . If the convergence criterion is not met, the loop
begins anew with a higher value of A, increased by a step A D 0:01. Otherwise,
Break[] interrupts the loop and Print gives some output values, in the above
example the normalization A, the mass M of the potential in GeV and xnum .0/:
f59:75 = A; 5:12595 106 GeV = M; 0:999011 = x(0)g.
It is easy to plot y D rhoQ and w D wQ as in Fig. 7.3 and check that w .z D
0/ D wQŒ0; Un; 6 is 0:39951.
7.3 Conformal transformations 1. Consider the action (7.55) with Lm D 0

and the conformal transformation (7.56), where ˝ D ˝.x/ is a generic scalar
function. Define
O :D @ ln ˝ ; O :D rN O D rN rN ln ˝ ; N ln ˝ ;
O :D O D
(7.145)
where
N D ˝ 2 .D 2/O @N
(7.146)
and indices are raised and lowered with the Jordan metric. Show how the
linear Riemann invariants and (7.55) transform under (7.56).
Solution The inverse metric, measure factor and affine connection (2.15) transform
as
g D ˝ 2 gN ; (7.147)
p p
g D ˝ D Ng ; (7.148)
h i

D N

C 2ı. O/ gN O : (7.149)
From (7.148) one obtains (7.146), while (7.149) and (2.16) yield
R D RN C r ;

r :D 2ıŒ O C 2NgŒ O C 2 ıŒ O O C gN Œ O O C ıŒ gN O O ;
R D RN C r ;

r :D r D .2 D/O gN O C .D 2/ O O gN O O ;
˝ 2 R D RN .D 1/Œ2O C .D 2/O O : (7.150)
After integrating by parts and throwing away boundary terms, the last expression
yields
Z Z p
p
dD x g R D dD x Ng ˝ D2 RN C .D 1/.D 2/O O ;
so that (7.55) becomes

Z N
p R .D 1/.D 2/ 1 N N
SN D dD x Ng ˝ D2 C O O
@ @ ˝ 2
V. /
2 2 2 2 2
(7.151)
in the Jordan frame.
7.4 Conformal transformations 2. (a) Find ˝. / explicitly in (7.151) such

that the kinetic term of the scalar field is canonical. (b) Without specifying
˝, find a field redefinition ! N such that the kinetic term reads
N N N N
.!=2/@ @ , where ! is an arbitrary function of .
Solution
(a) Since O D .˝; =˝/@ , the second and third terms in (7.151) can be
combined into
" #
D2 .D 1/.D 2/ ˝; 2 1 N N
˝ @ @ : (7.152)
2 2 ˝ 2
The coefficient equals 1=2 if the following non-linear differential equation is

satisfied:
.D 1/.D 2/ 2
˝; C ˝ 4D ˝ 2 D 0 :
2
This equation is exactly solved by the profile
8 2 s 39 D2
2
< D2 2 =
˝. / D cosh 4 5 : (7.153)
: 2 .D 1/.D 2/ ;
Then, the action in the Jordan frame is

Z
p 1 D2
SN D dD x Ng ˝ . / N 1 @N @N U. / ;
R (7.154)
2 2 2
where U. / D ˝ D . /V. /.
(b) Imposing the relation
" 2 2 #
.D 1/.D 2/ ˝; N 1 d !
˝ D2
@N N @N N D @N N @N N ;
2 2 ˝ 2 dN 2
one gets
s
Z 2
! .D 1/.D 2/ ˝; N
D dN C
˝ D2 2 ˝
s
Z 2
! .D 1/.D 2/ F; N
D dN C ; (7.155)
F 4 2 F
where F. N / D ˝ D2 . N /.
7.5 Conformal transformations 3. Let

1
˝. / D p (7.156)
cosh 2 =6
in (7.151) with D D 4. Find a transformation of the scalar field ! N such

that the kinetic term of N is canonical. Rewrite SN in terms of N .
p p p p
Solution Since ˝; =˝ D 2 =6 tanh. 2 =6 / D 2 =6 1 ˝ 2 , expres-
sion (7.152) in four dimensions becomes

3 2 1 2 N N 1
2
˝; ˝ @ @ D ˝ 4 @N @N :
2 2
Defining the field

Z r r !
N :D 2 6 2 N2 < 6 ;
d ˝ . /D tanh ; (7.157)
2 6 2
the action (7.151) with D D 4 becomes

Z
p 1 1 N N N N
SN D 4
d x Ng N N N
f . /R @ @ U. / ; (7.158)
2 2 2
where
2 N 2
f . N/ D 1 ; U. N / D ˝ 4 Œ . N / VŒ . N / : (7.159)
6
The energy-momentum tensor of this model has finite matrix entries at every order
of renormalized perturbation theory [173].
References 369
References
1. A.D. Linde, Is the cosmological constant a constant? Pis’ma Zh. Eksp. Teor. Fiz. 19, 320
(1974) [JETP Lett. 19, 183 (1974)]
2. A.D. Linde, Phase transitions in gauge theories and cosmology. Rep. Prog. Phys. 42, 389
(1979)
3. Y. Nambu, Quasi-particles and gauge invariance in the theory of superconductivity. Phys.
Rev. 117, 648 (1960)
4. J. Goldstone, Field theories with “superconductor” solutions. Nuovo Cim. 19, 154 (1961)
5. J. Goldstone, A. Salam, S. Weinberg, Broken symmetries. Phys. Rev. 127, 965 (1962)
6. F. Englert, R. Brout, Broken symmetry and the mass of gauge vector mesons. Phys. Rev. Lett.
13, 321 (1964)
7. P.W. Higgs, Broken symmetries, massless particles and gauge fields. Phys. Lett. 12, 132
(1964)
8. P.W. Higgs, Broken symmetries and the masses of gauge bosons. Phys. Rev. Lett. 13, 508
(1964)
9. G.S. Guralnik, C.R. Hagen, T.W.B. Kibble, Global conservation laws and massless particles.
Phys. Rev. Lett. 13, 585 (1964)
10. J. Dreitlein, Broken symmetry and the cosmological constant. Phys. Rev. Lett. 33, 1243
(1974)
11. S.A. Bludman, M.A. Ruderman, Induced cosmological constant expected above the phase
transition restoring the broken symmetry. Phys. Rev. Lett. 38, 255 (1977)
12. D.B. Chitwood et al. [MuLan Collaboration], Improved measurement of the positive muon
lifetime and determination of the Fermi constant. Phys. Rev. Lett. 99, 032001 (2007).
[arXiv:0704.1981]
13. T. Plehn, M. Rauch, Quartic Higgs coupling at hadron colliders. Phys. Rev. D 72, 053008
(2005). [arXiv:hep-ph/0507321]
14. J. Martin, Everything you always wanted to know about the cosmological constant problem
(but were afraid to ask). C. R. Phys. 13, 566 (2012). [arXiv:1205.3365]
15. Ya.B. Zel’dovich, The cosmological constant and the theory of elementary particles. Sov.
Phys. Usp. 11, 381 (1968)
16. J.F. Koksma, T. Prokopec, The cosmological constant and Lorentz invariance of the vacuum
state. arXiv:1105.6296
17. P.M. Stevenson, Gaussian effective potential. I. Quantum mechanics. Phys. Rev. D 30, 1712
(1984).
18. P.M. Stevenson, Gaussian effective potential. II. 4 field theory. Phys. Rev. D 32, 1389
(1985).
19. P.M. Stevenson, R. Tarrach, The return of 4 . Phys. Lett. B 176, 436 (1986).
Cambridge, 1982)
21. Ya.B. Zel’dovich, Cosmological constant and elementary particles. Pis’ma Zh. Eksp. Teor.
Fiz. 6, 883 (1967) [JETP Lett. 6, 316 (1967)]
22. B. Zumino, Supersymmetry and the vacuum. Nucl. Phys. B 89, 535 (1975)
23. S. Weinberg, Does gravitation resolve the ambiguity among supersymmetry vacua? Phys.
Rev. Lett. 48, 1776 (1982)
25. T. Padmanabhan, Dark energy: mystery of the millennium. AIP Conf. Proc. 861, 179 (2006).
26. T. Padmanabhan, Dark energy and gravity. Gen. Relat. Grav. 40, 529 (2008).
[arXiv:0705.2533]
27. T. Padmanabhan, H. Padmanabhan, Cosmological constant from the emergent gravity
perspective. Int. J. Mod. Phys. D 23, 1430011 (2014). [arXiv:1404.2284]
28. T. Padmanabhan, The physical principle that determines the value of the cosmological
constant. arXiv:1210.4174
29. H. Padmanabhan, T. Padmanabhan, CosMIn: the solution to the cosmological constant
problem. Int. J. Mod. Phys. D 22, 1342001 (2013). [arXiv:1302.3226]
30. J.D. Bekenstein, Black holes and entropy. Phys. Rev. D 7, 2333 (1973)
31. J.D. Bekenstein, Generalized second law of thermodynamics in black hole physics. Phys. Rev.
D 9, 3292 (1974)
32. S.W. Hawking, Particle creation by black holes. Commun. Math. Phys. 43, 199 (1975);
Erratum-ibid. 46, 206 (1976)
33. S.W. Hawking, Black holes and thermodynamics. Phys. Rev. D 13, 191 (1976)
34. G. ’t Hooft, Dimensional reduction in quantum gravity, in Salamfestschrift, ed. by A. Ali, J.
Ellis, S. Randjbar-Daemi (World Scientific, Singapore, 1993). [arXiv:gr-qc/9310026]
35. A.G. Cohen, D.B. Kaplan, A.E. Nelson, Effective field theory, black holes, and the cosmo-
logical constant. Phys. Rev. Lett. 82, 4971 (1999). [arXiv:hep-th/9803132]
36. V. Sahni, A.A. Starobinsky, The case for a positive cosmological -term. Int. J. Mod. Phys.
D 9, 373 (2000). [arXiv:astro-ph/9904398]
37. S. Nobbenhuis, Categorizing different approaches to the cosmological constant problem.
Found. Phys. 36, 613 (2006). [arXiv:gr-qc/0411093]
38. E. Mottola, Particle creation in de Sitter space. Phys. Rev. D 31, 754 (1985)
39. T. Banks, Relaxation of the cosmological constant. Phys. Rev. Lett. 52, 1461 (1984)
40. L.F. Abbott, A mechanism for reducing the value of the cosmological constant. Phys. Lett. B
150, 427 (1985)
41. M. Endō, T. Fukui, The cosmological term and a modified Brans–Dicke cosmology. Gen.
Relat. Grav. 8, 833 (1977)
42. A.D. Dolgov, An attempt to get rid of the cosmological constant, in The Very Early Universe,
ed. by G.W. Gibbons, S.W. Hawking, S.T.C. Siklos (Cambridge University Press, Cambridge,
1983)
43. Y. Fujii, Origin of the gravitational constant and particle masses in a scale-invariant scalar-
tensor theory. Phys. Rev. D 26, 2580 (1982)
44. L.H. Ford, Quantum instability of de Sitter spacetime. Phys. Rev. D 31, 710 (1985)
45. O. Bertolami, Time-dependent cosmological term. Nuovo Cim. B 93, 36 (1986)
46. O. Bertolami, Brans–Dicke cosmology with a scalar field dependent cosmological term.
Fortsch. Phys. 34, 829 (1986)
47. L.H. Ford, Cosmological-constant damping by unstable scalar fields. Phys. Rev. D 35, 2339
(1987).
48. R.D. Peccei, J. Solà, C. Wetterich, Adjusting the cosmological constant dynamically: cosmons
and a new force weaker than gravity. Phys. Lett. B 195, 183 (1987)
49. S.M. Barr, Attempt at a classical cancellation of the cosmological constant. Phys. Rev. D 36,
1691 (1987)
50. C. Wetterich, Cosmologies with variable Newton’s “constant”. Nucl. Phys. B 302, 645 (1988)
51. C. Wetterich, Cosmology and the fate of dilatation symmetry. Nucl. Phys. B 302, 668 (1988)
52. W.-M. Suen, C.M. Will, Damping of the cosmological constant by a classical scalar field.
Phys. Lett. B 205, 447 (1988)
53. Y. Fujii, Saving the mechanism of a decaying cosmological constant. Mod. Phys. Lett. A 04,
513 (1989)
54. E.T. Tomboulis, Dynamically adjusted cosmological constant and conformal anomalies. Nucl.
Phys. B 329, 410 (1990)
55. Y. Fujii, T. Nishioka, Model of a decaying cosmological constant. Phys. Rev. D 42, 361 (1990)
56. L. Parker, Cosmological constant and absence of particle creation. Phys. Rev. Lett. 50, 1009
(1983)
57. V.A. Rubakov, M.E. Shaposhnikov, Extra space-time dimensions: towards a solution to the
cosmological constant problem. Phys. Lett. B 125, 139 (1983)
58. I. Antoniadis, N.C. Tsamis, On the cosmological constant problem. Phys. Lett. B 144, 55
(1984)
References 371
59. S.G. Rajeev, Why is the cosmological constant small? Phys. Lett. B 125, 144 (1983)
60. T.R. Taylor, G. Veneziano, Quenching the cosmological constant. Phys. Lett. B 228, 311
(1989)
61. M. Özer, M.O. Taha, A solution to the main cosmological problems. Phys. Lett. B 171, 363
(1986)
62. M. Özer, M.O. Taha, A model of the universe free of cosmological problems. Nucl. Phys. B
287, 776 (1987)
63. T.S. Olson, T.F. Jordan, Ages of the Universe for decreasing cosmological constants. Phys.
Rev. D 35, 3258 (1987)
64. K. Freese, F.C. Adams, J.A. Frieman, E. Mottola, Cosmology with decaying vacuum energy.
Nucl. Phys. B 287, 797 (1987)
65. M. Reuter, C. Wetterich, Time evolution of the cosmological “constant”. Phys. Lett. B 188,
38 (1987)
66. J.M. Overduin, F.I. Cooperstock, Evolution of the scale factor with a variable cosmological
term. Phys. Rev. D 58, 043506 (1998). [arXiv:astro-ph/9805260]
67. P.J.E. Peebles, B. Ratra, Cosmology with a time variable cosmological constant. Astrophys.
J. 325, L17 (1988)
68. B. Ratra, P.J.E. Peebles, Cosmological consequences of a rolling homogeneous scalar field.
Phys. Rev. D 37, 3406 (1988)
69. K. Coble, S. Dodelson, J.A. Frieman, Dynamical models of structure formation. Phys. Rev.
D 55, 1851 (1997). [arXiv:astro-ph/9608122]
70. M.S. Turner, M.J. White, CDM models with a smooth component. Phys. Rev. D 56, 4439
(1997). [arXiv:astro-ph/9701138]
71. R.R. Caldwell, R. Dave, P.J. Steinhardt, Cosmological imprint of an energy component with
general equation of state. Phys. Rev. Lett. 80, 1582 (1998). [arXiv:astro-ph/9708069]
72. P.G. Ferreira, M. Joyce, Structure formation with a selftuning scalar field. Phys. Rev. Lett. 79,
4740 (1997). [arXiv:astro-ph/9707286]
73. P.G. Ferreira, M. Joyce, Cosmology with a primordial scaling field. Phys. Rev. D 58, 023503
(1998). [arXiv:astro-ph/9711102]
74. G. Huey, L. Wang, R. Dave, R.R. Caldwell, P.J. Steinhardt, Resolving the cosmological
missing energy problem. Phys. Rev. D 59, 063005 (1999). [arXiv:astro-ph/9804285]
75. S.M. Carroll, Quintessence and the rest of the world: suppressing long-range interactions.
76. I. Zlatev, L. Wang, P.J. Steinhardt, Quintessence, cosmic coincidence, and the cosmological
constant. Phys. Rev. Lett. 82, 896 (1999). [arXiv:astro-ph/9807002]
77. A.R. Liddle, R.J. Scherrer, Classification of scalar field potentials with cosmological scaling
solutions. Phys. Rev. D 59, 023509 (1999). [arXiv:astro-ph/9809272]
78. C.F. Kolda, D.H. Lyth, Quintessential difficulties. Phys. Lett. B 458, 197 (1999).
79. P.J. Steinhardt, L. Wang, I. Zlatev, Cosmological tracking solutions. Phys. Rev. D 59, 123504
(1999). [arXiv:astro-ph/9812313]
80. L. Wang, R.R. Caldwell, J.P. Ostriker, P.J. Steinhardt, Cosmic concordance and quintessence.
81. A. de la Macorra, G. Piccinelli, General scalar fields as quintessence. Phys. Rev. D 61, 123503
(2000). [arXiv:hep-ph/9909459]
82. E.J. Copeland, M. Sami, S. Tsujikawa, Dynamics of dark energy. Int. J. Mod. Phys. D 15,
1753 (2006). [arXiv:hep-th/0603057]
83. E.J. Copeland, A.R. Liddle, D. Wands, Exponential potentials and cosmological scaling
solutions. Phys. Rev. D 57, 4686 (1998). [arXiv:gr-qc/9711068]
84. T. Chiba, Slow-roll thawing quintessence. Phys. Rev. D 79, 083517 (2009); Erratum-ibid. D
80, 109902(E) (2009). [arXiv:0902.4037]
85. T. Chiba, Equation of state of tracker fields. Phys. Rev. D 81, 023515 (2010).
[arXiv:0909.4365]
86. T. Chiba, A. De Felice, S. Tsujikawa, Observational constraints on quintessence: thawing,

tracker, and scaling models. Phys. Rev. D 87, 083505 (2013). [arXiv:1210.3859]
87. R.R. Caldwell, E.V. Linder, Limits of quintessence. Phys. Rev. Lett. 95, 141301 (2005).
88. E.V. Linder, Paths of quintessence. Phys. Rev. D 73, 063010 (2006). [arXiv:astro-ph/0601052]
89. R.J. Scherrer, A.A. Sen, Thawing quintessence with a nearly flat potential. Phys. Rev. D 77,
083515 (2008). [arXiv:0712.3450]
90. P.S. Corasaniti, E.J. Copeland, Model independent approach to the dark energy equation of
state. Phys. Rev. D 67, 063521 (2003). [arXiv:astro-ph/0205544]
91. D.K. Hazra, S. Majumdar, S. Pal, S. Panda, A.A. Sen, S.P. Trivedi, Post-Planck dark energy
constraints. Phys. Rev. D 91, 083005 (2015). [arXiv:1310.6161]
92. A.Yu. Kamenshchik, U. Moschella, V. Pasquier, An alternative to quintessence. Phys. Lett. B
511, 265 (2001). [arXiv:gr-qc/0103004]
93. N. Bilic, G.B. Tupper, R.D. Viollier, Unification of dark matter and dark energy: the
inhomogeneous Chaplygin gas. Phys. Lett. B 535, 17 (2002). [arXiv:astro-ph/0111325]
94. M.C. Bento, O. Bertolami, A.A. Sen, Generalized Chaplygin gas, accelerated expansion and
dark energy matter unification. Phys. Rev. D 66, 043507 (2002). [arXiv:gr-qc/0202064]
95. M.C. Bento, O. Bertolami, A.A. Sen, Generalized Chaplygin gas model: dark energy –
dark matter unification and CMBR constraints. Gen. Relat. Grav. 35, 2063 (2003).
96. A.A. Sen, R.J. Scherrer, Generalizing the generalized Chaplygin gas. Phys. Rev. D 72,
063511 (2005). [arXiv:astro-ph/0507717]
97. S. Chaplygin, On gas jets. Sci. Mem. Moscow Univ. Math. 21, 1 (1904)
98. N. Weiss, Possible origins of a small, nonzero cosmological constant. Phys. Lett. B 197, 42
(1987)
99. C.T. Hill, D.N. Schramm, J.N. Fry, Cosmological structure formation from soft topological
defects. Comments Nucl. Part. Phys. 19, 25 (1989)
100. J.A. Frieman, C.T. Hill, R. Watkins, Late-time cosmological phase transitions: particle-
physics models and cosmic evolution. Phys. Rev. D 46, 1226 (1992)
101. J.A. Frieman, C.T. Hill, A. Stebbins, I. Waga, Cosmology with ultralight pseudo Nambu–
Goldstone bosons. Phys. Rev. Lett. 75, 2077 (1995). [arXiv:astro-ph/9505060]
102. L.J. Hall, Y. Nomura, S.J. Oliver, Evolving dark energy with w ¤ 1. Phys. Rev. Lett. 95,
141302 (2005). [arXiv:astro-ph/0503706]
103. S. Dutta, R.J. Scherrer, Hilltop quintessence. Phys. Rev. D 78, 123525 (2008).
[arXiv:0809.4441]
104. K. Choi, String or M theory axion as a quintessence. Phys. Rev. D 62, 043509 (2000).
105. R. Kallosh, A.D. Linde, S. Prokushkin, M. Shmakova, Supergravity, dark energy and the fate
of the universe. Phys. Rev. D 66, 123503 (2002). [arXiv:hep-th/0208156]
106. M.S. Turner, Coherent scalar-field oscillations in an expanding universe. Phys. Rev. D 28,
1243 (1983)
107. B. Spokoiny, Deflationary universe scenario. Phys. Lett. B 315, 40 (1993).
108. T. Barreiro, E.J. Copeland, N.J. Nunes, Quintessence arising from exponential potentials.
109. C. Rubano, P. Scudellaro, On some exponential potentials for a cosmological scalar field as
quintessence. Gen. Relat. Grav. 34, 307 (2002). [arXiv:astro-ph/0103335]
110. A.A. Sen, S. Sethi, Quintessence model with double exponential potential. Phys. Lett. B 532,
159 (2002). [arXiv:gr-qc/0111082]
111. M. Gutperle, R. Kallosh, A.D. Linde, M/string theory, S-branes and accelerating universe.
JCAP 0307, 001 (2003). [arXiv:hep-th/0304225]
112. I.P. Neupane, Accelerating cosmologies from exponential potentials. Class. Quantum Grav.
21, 4383 (2004). [arXiv:hep-th/0311071]
References 373
113. L. Järv, T. Mohaupt, F. Saueressig, Quintessence cosmologies with a double exponential

potential. JCAP 0408, 016 (2004). [arXiv:hep-th/0403063]
114. C.M. Hull, The minimal couplings and scalar potentials of the gauged N D 8 supergravities.
Class. Quantum Grav. 2, 343 (1985)
115. R. Kallosh, A.D. Linde, S. Prokushkin, M. Shmakova, Gauged supergravities, de Sitter space
and cosmology. Phys. Rev. D 65, 105016 (2002). [arXiv:hep-th/0110089]
116. H. Lü, C.N. Pope, p-brane solitons in maximal supergravities. Nucl. Phys. B 465, 127 (1996).
117. I.V. Lavrinenko, H. Lü, C.N. Pope, Fiber bundles and generalized dimensional reduction.
118. M.S. Bremer, M.J. Duff, H. Lü, C.N. Pope, K.S. Stelle, Instanton cosmology and
domain walls from M theory and string theory. Nucl. Phys. B 543, 321 (1999).
119. S.W. Hawking, H.S. Reall, Inflation, singular instantons and eleven-dimensional cosmology.
120. A. Albrecht, C. Skordis, Phenomenology of a realistic accelerating universe using only Planck
scale physics. Phys. Rev. Lett. 84, 2076 (2000). [arXiv:astro-ph/9908085]
121. C. Skordis, A. Albrecht, Planck scale quintessence and the physics of structure formation.
122. G.R. Dvali, S.H.H. Tye, Brane inflation. Phys. Lett. B 450, 72 (1999).
123. V. Sahni, L. Wang, New cosmological model of quintessence and dark matter. Phys. Rev. D
62, 103517 (2000). [arXiv:astro-ph/9910097]
124. L.P. Chimento, A.S. Jakubi, Scalar field cosmologies with perfect fluid in Robertson–Walker
metric. Int. J. Mod. Phys. D 5, 71 (1996). [arXiv:gr-qc/9506015]
125. L.A. Ureña-López, T. Matos, New cosmological tracker solution for quintessence. Phys. Rev.
D 62, 081302 (2000). [arXiv:astro-ph/0003364]
126. T.R. Taylor, G. Veneziano, S. Yankielowicz, Supersymmetric QCD and its massless limit: an
effective lagrangian analysis. Nucl. Phys. B 218, 493 (1983)
127. I. Affleck, M. Dine, N. Seiberg, Dynamical supersymmetry breaking in four dimensions and
its phenomenological implications. Nucl. Phys. B 256, 557 (1985)
128. P. Binétruy, Models of dynamical supersymmetry breaking and quintessence. Phys. Rev. D
60, 063502 (1999). [arXiv:hep-ph/9810553]
129. A. Masiero, M. Pietroni, F. Rosati, SUSY QCD and quintessence. Phys. Rev. D 61, 023504
(2000). [arXiv:hep-ph/9905346]
130. P. Brax, J. Martin, Robustness of quintessence. Phys. Rev. D 61, 103502 (2000).
131. P.-Y. Wang, C.-W. Chen, P. Chen, Confronting tracker field quintessence with data. JCAP
1202, 016 (2012). [arXiv:1108.1424]
132. P. Brax, J. Martin, Quintessence and supergravity. Phys. Lett. B 468, 40 (1999).
133. E.J. Copeland, N.J. Nunes, F. Rosati, Quintessence models in supergravity. Phys. Rev. D 62,
123503 (2000). [arXiv:hep-ph/0005222]
134. P. Brax, J. Martin, A. Riazuelo, Exhaustive study of cosmic microwave back-
ground anisotropies in quintessential scenarios. Phys. Rev. D 62, 103505 (2000).
135. P. Brax, J. Martin, A. Riazuelo, Quintessence with two energy scales. Phys. Rev. D 64,
083505 (2001). [arXiv:hep-ph/0104240]
136. S.C.C. Ng, N.J. Nunes, F. Rosati, Applications of scalar attractor solutions to cosmology.
137. P.S. Corasaniti, E.J. Copeland, Constraining the quintessence equation of state with SnIa data
and CMB peaks. Phys. Rev. D 65, 043004 (2002). [arXiv:astro-ph/0107378]
138. S.A. Bludman, Tracking quintessence would require two cosmic coincidences. Phys. Rev. D
69, 122002 (2004). [arXiv:astro-ph/0403526]
140. P.J.E. Peebles, A. Vilenkin, Quintessential inflation. Phys. Rev. D 59, 063505 (1999).
141. M. Giovannini, Spikes in the relic graviton background from quintessential inflation. Class.
Quantum Grav. 16, 2905 (1999). [arXiv:hep-ph/9903263]
142. M. Peloso, F. Rosati, On the construction of quintessential inflation models. JHEP 9912, 026
(1999). [arXiv:hep-ph/9908271]
143. K. Dimopoulos, J.W.F. Valle, Modeling quintessential inflation. Astropart. Phys. 18, 287
(2002). [arXiv:astro-ph/0111417]
144. K. Dimopoulos, Curvaton hypothesis and the -problem of quintessential inflation, with and
without branes. Phys. Rev. D 68, 123506 (2003). [arXiv:astro-ph/0212264]
145. A.H. Campos, H.C. Reis, R. Rosenfeld, Preheating in quintessential inflation. Phys. Lett. B
575, 151 (2003). [arXiv:hep-ph/0210152]
146. L.H. Ford, Gravitational particle creation and inflation. Phys. Rev. D 35, 2955 (1987)
147. M. Joyce, T. Prokopec, Turning around the sphaleron bound: electroweak baryoge-
nesis in an alternative postinflationary cosmology. Phys. Rev. D 57, 6022 (1998).
148. G.N. Felder, L. Kofman, A.D. Linde, Inflation and preheating in nonoscillatory models. Phys.
Rev. D 60, 103505 (1999). [arXiv:hep-ph/9903350]
149. E.J. Copeland, A.R. Liddle, D.H. Lyth, E.D. Stewart, D. Wands, False vacuum inflation with
Einstein gravity. Phys. Rev. D 49, 6410 (1994). [arXiv:astro-ph/9401011]
150. R.A. Frewin, J.E. Lidsey, On identifying the present day vacuum energy with the potential
driving inflation. Int. J. Mod. Phys. D 2, 323 (1993). [arXiv:astro-ph/9312035]
151. A.B. Kaganovich, Field theory model giving rise to “quintessential inflation” without the
cosmological constant and other fine tuning problems. Phys. Rev. D 63, 025022 (2001).
152. G. Huey, J.E. Lidsey, Inflation, braneworlds and quintessence. Phys. Lett. B 514, 217 (2001).
153. R. Rosenfeld, J.A. Frieman, A simple model for quintessential inflation. JCAP 0509, 003
(2005). [arXiv:astro-ph/0504191]
154. P. Jordan, Formation of the stars and development of the universe. Nature 164, 637 (1949)
155. M. Fierz, Über die physikalische Deutung der erweiterten Gravitationstheorie P. Jordans.
Helv. Phys. Acta 29, 128 (1956)
156. P. Jordan, Zum gegenwärtigen Stand der Diracschen kosmologischen Hypothesen. Z. Phys.
157, 112 (1959)
157. C. Brans, R.H. Dicke, Mach’s principle and a relativistic theory of gravitation. Phys. Rev.
124, 925 (1961)
158. R.H. Dicke, Mach’s principle and invariance under transformation of units. Phys. Rev. 125,
2163 (1962)
159. P.G. Bergmann, Comments on the scalar-tensor theory. Int. J. Theor. Phys. 1, 25 (1968)
160. K. Nordtvedt, Post-Newtonian metric for a general class of scalar-tensor gravitational theories
and observational consequences. Astrophys. J. 161, 1059 (1970)
161. R.V. Wagoner, Scalar-tensor theory and gravitational waves. Phys. Rev. D 1, 3209 (1970)
162. S. Deser, Scale invariance and gravitational coupling. Ann. Phys. (N.Y.) 59, 248 (1970)
163. J. O’Hanlon, Intermediate-range gravity: a generally covariant model. Phys. Rev. Lett. 29,
137 (1972)
164. Y. Fujii, Scalar-tensor theory of gravitation and spontaneous breakdown of scale invariance.
Phys. Rev. D 9, 874 (1974)
165. J.D. Bekenstein, Exact solutions of Einstein-conformal scalar equations. Ann. Phys. (N.Y.)
82, 535 (1974)
166. P. Minkowski, On the spontaneous origin of Newton’s constant. Phys. Lett. B 71, 419 (1977)
167. V. Canuto, S.H. Hsieh, P.J. Adams, Scale-covariant theory of gravitation and astrophysical
applications. Phys. Rev. Lett. 39, 429 (1977)
References 375
168. A. Zee, Broken-symmetric theory of gravity. Phys. Rev. Lett. 42, 417 (1979)
169. Y. Fujii, J.M. Niedra, Solutions of a cosmological equation in the scale invariant scalar-tensor
theory of gravitation. Prog. Theor. Phys. 70, 412 (1983)
170. J.L. Anderson, Scale invariance of the second kind and the Brans–Dicke scalar-tensor theory.
Phys. Rev. D 3, 1689 (1971)
171. N. Banerjee, S. Sen, Does Brans–Dicke theory always yield general relativity in the infinite
! limit? Phys. Rev. D 56, 1334 (1997)
172. V. Faraoni, The ! ! 1 limit of Brans–Dicke theory. Phys. Lett. A 245, 26 (1998).
173. C.G. Callan, S.R. Coleman, R. Jackiw, A new improved energy-momentum tensor. Ann. Phys.
(N.Y.) 59, 42 (1970)
174. D.Z. Freedman, I.J. Muzinich, E.J. Weinberg, On the energy-momentum tensor in gauge field
theories. Ann. Phys. (N.Y.) 87, 95 (1974)
175. D.Z. Freedman, E.J. Weinberg, The energy-momentum tensor in scalar and gauge field
theories. Ann. Phys. (N.Y.) 87, 354 (1974)
176. G. ’t Hooft, M.J.G. Veltman, One loop divergencies in the theory of gravitation. Ann. Poincaré
Phys. Theor. A 20, 69 (1974)
177. L.S. Brown, J.C. Collins, Dimensional renormalization of scalar field theory in curved space-
time. Ann. Phys. (N.Y.) 130, 215 (1980)
178. T. Kaluza, Zum Unitätsproblem in der Physik. Sitz.-ber. Kgl. Preuss. Akad. Wiss. 1921, 966
(1921)
179. O. Klein, Quantentheorie und fünfdimensionale Relativitätstheorie. Z. Phys. 37, 895 (1926)
[Surveys High Energy Phys. 5, 241 (1986)]
180. A.H. Chamseddine, N D 4 supergravity coupled to N D 4 matter and hidden symmetries.
Nucl. Phys. B 185, 403 (1981)
181. P.G.O. Freund, Kaluza–Klein cosmologies. Nucl. Phys. B 209, 146 (1982)
182. G. Magnano, L.M. Sokołowski, Physical equivalence between nonlinear gravity theories
and a general-relativistic self-gravitating scalar field. Phys. Rev. D 50, 5039 (1994).
183. V. Faraoni, E. Gunzig, P. Nardone, Conformal transformations in classical gravitational
theories and in cosmology. Fund. Cosmic Phys. 20, 121 (1999). [arXiv:gr-qc/9811047]
184. É.É. Flanagan, The conformal frame freedom in theories of gravitation. Class. Quantum Grav.
21, 3817 (2004). [arXiv:gr-qc/0403063]
185. V. Faraoni, S. Nadeau, (Pseudo)issue of the conformal frame revisited. Phys. Rev. D 75,
023501 (2007). [arXiv:gr-qc/0612075]
186. R.L. Arnowitt, S. Deser, C.W. Misner, The dynamics of general relativity, in Gravita-
tion: An Introduction to Current Research, ed. by L. Witten (Wiley, New York, 1962).
187. L.M. Sokołowski, Uniqueness of the metric line element in dimensionally reduced theories.
188. C. Armendariz-Picón, Predictions and observations in theories with varying couplings. Phys.
189. R. Catena, M. Pietroni, L. Scarabello, Einstein and Jordan reconciled: a frame-
invariant approach to scalar-tensor cosmology. Phys. Rev. D 76, 084039 (2007).
190. N. Deruelle, M. Sasaki, Conformal equivalence in classical gravity: the example of “veiled”
general relativity. Springer Proc. Phys. 137, 247 (2011). [arXiv:1007.3563]
191. T. Chiba, M. Yamaguchi, Conformal-frame (in)dependence of cosmological observations in
scalar-tensor theory. JCAP 1310, 040 (2013). [arXiv:1308.1142]
192. K. Nordtvedt, Equivalence principle for massive bodies. II. Theory. Phys. Rev. 169, 1017
(1968)
193. T. Damour, G. Esposito-Farèse, Tensor multiscalar theories of gravitation. Class. Quantum
Grav. 9, 2093 (1992)
194. Y.M. Cho, Violation of equivalence principle in Brans–Dicke theory. Class. Quantum Grav.
14, 2963 (1997)
195. L. Hui, A. Nicolis, Equivalence principle for scalar forces. Phys. Rev. Lett. 105, 231101
(2010). [arXiv:1009.2520]
196. C. Armendariz-Picón, R. Penco, Quantum equivalence principle violations in scalar-tensor
theories. Phys. Rev. D 85, 044052 (2012). [arXiv:1108.6028]
197. F. Nitti, F. Piazza, Scalar-tensor theories, trace anomalies, and the QCD frame. Phys. Rev. D
86, 122002 (2012). [arXiv:1202.2105]
198. G. Esposito-Farèse, D. Polarski, Scalar-tensor gravity in an accelerating universe. Phys. Rev.
D 63, 063504 (2001). [arXiv:gr-qc/0009034]
199. N. Makino, M. Sasaki, The density perturbation in the chaotic inflation with non-minimal
coupling. Prog. Theor. Phys. 86, 103 (1991)
200. R. Fakir, S. Habib, W. Unruh, Cosmological density perturbations with modified gravity.
Astrophys. J. 394, 396 (1992)
201. J. Weenink, T. Prokopec, Gauge invariant cosmological perturbations for the nonminimally
coupled inflaton field. Phys. Rev. D 82, 123510 (2010). [arXiv:1007.2133]
202. J.-O. Gong, J.-c. Hwang, W.-I. Park, M. Sasaki, Y.-S. Song, Conformal invariance of curvature
perturbation. JCAP 1109, 023 (2011). [arXiv:1107.1840]
203. D.I. Kaiser, Primordial spectral indices from generalized Einstein theories. Phys. Rev. D 52,
4295 (1995). [arXiv:astro-ph/9408044]
204. D.I. Kaiser, Frame independent calculation of spectral indices from inflation.
205. G. Domènech, M. Sasaki, Conformal frame dependence of inflation. JCAP 1504, 022 (2015).
[arXiv:1501.07699]
206. J. White, M. Minamitsuji, M. Sasaki, Curvature perturbation in multi-field inflation with non-
minimal coupling. JCAP 1207, 039 (2012). [arXiv:1205.0656]
207. J. White, M. Minamitsuji, M. Sasaki, Non-linear curvature perturbation in multi-field inflation
models with non-minimal coupling. JCAP 1309, 015 (2013). [arXiv:1306.6186]
208. T. Qiu, J.-Q. Xia, Perturbations of single-field inflation in modified gravity theory. Phys. Lett.
B 744, 273 (2015). [arXiv:1406.5902]
209. M.J. Duff, Inconsistency of quantum field theory in curved space-time, in Quantum Gravity
2, ed. by C.J. Isham, R. Penrose, D.W. Sciama (Oxford University Press, Oxford, 1981)
210. S.P. de Alwis, Quantization of a theory of 2D dilaton gravity. Phys. Lett. B 289, 278 (1992).
211. R. Fakir, S. Habib, Quantum fluctuations with strong curvature coupling. Mod. Phys. Lett. A
08, 2827 (1993)
212. E. Elizalde, S. Naftulin, S.D. Odintsov, The renormalization structure and quantum equiva-
lence of 2D dilaton gravities. Int. J. Mod. Phys. A 9, 933 (1994). [arXiv:hep-th/9304091]
213. D. Grumiller, W. Kummer, D.V. Vassilevich, Dilaton gravity in two dimensions. Phys. Rep.
369, 327 (2002). [arXiv:hep-th/0204253]
214. D. Grumiller, W. Kummer, D.V. Vassilevich, Positive specific heat of the quantum corrected
dilaton black hole. JHEP 0307, 009 (2003). [arXiv:hep-th/0305036]
215. E. Komatsu, T. Futamase, Complete constraints on a nonminimally coupled chaotic infla-
tionary scenario from the cosmic microwave background. Phys. Rev. D 59, 064029 (1999).
216. S. Tsujikawa, B. Gumjudpai, Density perturbations in generalized Einstein scenarios and
constraints on nonminimal couplings from the cosmic microwave background. Phys. Rev.
D 69, 123523 (2004). [arXiv:astro-ph/0402185]
217. D. La, P.J. Steinhardt, Extended inflationary cosmology. Phys. Rev. Lett. 62, 376 (1989);
Erratum-ibid. 62, 1066 (1989)
218. D. La, P.J. Steinhardt, E.W. Bertschinger, Prescription for successful extended inflation. Phys.
Lett. B 231, 231 (1989)
219. S. Tsujikawa, J. Ohashi, S. Kuroyanagi, A. De Felice, Planck constraints on single-field
References 377
220. C. Wetterich, An asymptotically vanishing time-dependent cosmological “constant”. Astron.

Astrophys. 301, 321 (1995) [arXiv:hep-th/9408025]
221. J.-P. Uzan, Cosmological scaling solutions of nonminimally coupled scalar fields. Phys. Rev.
D 59, 123510 (1999). [arXiv:gr-qc/9903004]
222. T. Chiba, Quintessence, the gravitational constant, and gravity. Phys. Rev. D 60, 083508
(1999). [arXiv:gr-qc/9903094]
223. L. Amendola, Scaling solutions in general nonminimal coupling theories. Phys. Rev. D 60,
043501 (1999). [arXiv:astro-ph/9904120]
224. F. Perrotta, C. Baccigalupi, S. Matarrese, Extended quintessence. Phys. Rev. D 61, 023507
(1999). [arXiv:astro-ph/9906066]
225. R. de Ritis, A.A. Marino, C. Rubano, P. Scudellaro, Tracker fields from nonminimally coupled
theory. Phys. Rev. D 62, 043506 (2000). [arXiv:hep-th/9907198]
226. L. Amendola, Coupled quintessence. Phys. Rev. D 62, 043511 (2000).
227. D.J. Holden, D. Wands, Selfsimilar cosmological solutions with a nonminimally coupled
scalar field. Phys. Rev. D 61, 043506 (2000). [arXiv:gr-qc/9908026]
228. N. Bartolo, M. Pietroni, Scalar-tensor gravity and quintessence. Phys. Rev. D 61, 023518
(2000). [arXiv:hep-ph/9908521]
229. O. Bertolami, P.J. Martins, Nonminimal coupling and quintessence. Phys. Rev. D 61, 064007
(2000). [arXiv:gr-qc/9910056]
230. B. Boisseau, G. Esposito-Farèse, D. Polarski, A.A. Starobinsky, Reconstruction of a scalar-
tensor theory of gravity in an accelerating universe. Phys. Rev. Lett. 85, 2236 (2000).
231. V. Faraoni, Inflation and quintessence with nonminimal coupling. Phys. Rev. D 62, 023504
(2000). [arXiv:gr-qc/0002091]
232. C. Barceló, M. Visser, Scalar fields, energy conditions and traversable wormholes. Class.
Quantum Grav. 17, 3843 (2000). [arXiv:gr-qc/0003025]
233. C. Baccigalupi, S. Matarrese, F. Perrotta, Tracking extended quintessence. Phys. Rev. D 62,
123510 (2000). [arXiv:astro-ph/0005543]
234. S. Sen, T.R. Seshadri, Self interacting Brans–Dicke cosmology and quintessence. Int. J. Mod.
Phys. D 12, 445 (2003). [arXiv:gr-qc/0007079]
235. D.F. Torres, Quintessence, superquintessence, and observable quantities in Brans–Dicke and
nonminimally coupled theories. Phys. Rev. D 66, 043522 (2002). [arXiv:astro-ph/0204504]
236. S. Matarrese, C. Baccigalupi, F. Perrotta, Approaching without fine-tuning. Phys. Rev. D
70, 061301 (2004). [arXiv:astro-ph/0403480]
237. S.M. Carroll, A. De Felice, M. Trodden, Can we be tricked into thinking that w is less than
1? Phys. Rev. D 71, 023525 (2005). [arXiv:astro-ph/0408081]
238. L. Perivolaropoulos, Crossing the phantom divide barrier with scalar tensor theories. JCAP
0510, 001 (2005). [arXiv:astro-ph/0504582]
239. S. Tsujikawa, K. Uddin, S. Mizuno, R. Tavakol, J.’i. Yokoyama, Constraints on scalar-tensor
models of dark energy from observational and local gravity tests. Phys. Rev. D 77, 103009
(2008). [arXiv:0803.1106]
240. S. Das, P.S. Corasaniti, J. Khoury, Superacceleration as signature of dark sector interaction.
241. M. Bruni, S. Matarrese, O. Pantano, A local view of the observable universe. Phys. Rev. Lett.
74, 1916 (1995). [arXiv:astro-ph/9407054]
242. R.R. Caldwell, A phantom menace? Phys. Lett. B 545, 23 (2002). [arXiv:astro-ph/9908168]
243. C.M. Will, The confrontation between general relativity and experiment. Living Rev. Relat.
17, 4 (2014)
244. B. Bertotti, L. Iess, P. Tortora, A test of general relativity using radio links with the Cassini
spacecraft. Nature 425, 374 (2003)
245. T. Damour, K. Nordtvedt, General relativity as a cosmological attractor of tensor scalar
theories. Phys. Rev. Lett. 70, 2217 (1993)
246. T. Damour, K. Nordtvedt, Tensor-scalar cosmological models and their relaxation toward
general relativity. Phys. Rev. D 48, 3436 (1993)
247. T. Clifton, D.F. Mota, J.D. Barrow, Inhomogeneous gravity. Mon. Not. R. Astron. Soc. 358,
601 (2005). [arXiv:gr-qc/0406001]
248. J. Khoury, A. Weltman, Chameleon fields: awaiting surprises for tests of gravity in space.
249. J. Khoury, A. Weltman, Chameleon cosmology. Phys. Rev. D 69, 044026 (2004).
250. S. Capozziello, Curvature quintessence. Int. J. Mod. Phys. D 11, 483 (2002).
251. S. Capozziello, S. Carloni, A. Troisi, Quintessence without scalar fields. Recent Res. Dev.
Astron. Astrophys. 1, 625 (2003). [arXiv:astro-ph/0303041]
252. R. Utiyama, B.S. DeWitt, Renormalization of a classical gravitational field interacting with
quantized matter fields. J. Math. Phys. 3, 608 (1962)
253. A.D. Sakharov, Vacuum quantum fluctuations in curved space and the theory of gravitation.
Dokl. Akad. Nauk SSSR 177, 70 (1967) [Gen. Relat. Grav. 32, 365 (2000)]
254. T.V. Ruzmaikina, A.A. Ruzmaikin, Quadratic corrections to the Lagrangian density of the
gravitational field and the singularity. Zh. Eksp. Teor. Fiz. 57, 680 (1969) [Sov. Phys. JETP
30, 372 (1970)]
255. P.C.W. Davies, S.A. Fulling, S.M. Christensen, T.S. Bunch, Energy-momentum tensor of a
massless scalar quantum field in a Robertson–Walker universe. Ann. Phys. (N.Y.) 109, 108
(1977)
256. B.S. DeWitt, Quantum theory of gravity. II. The manifestly covariant theory. Phys. Rev. 162,
1195 (1967)
257. K.S. Stelle, Renormalization of higher-derivative quantum gravity. Phys. Rev. D 16, 953
(1977)
258. K.S. Stelle, Classical gravity with higher derivatives. Gen. Relat. Grav. 9, 353 (1978)
259. J. Julve, M. Tonin, Quantum gravity with higher derivative terms. Nuovo Cim. B 46, 137
(1978)
260. A. Salam, J.A. Strathdee, Remarks on high-energy stability and renormalizability of gravity
theory. Phys. Rev. D 18, 4480 (1978)
261. E.S. Fradkin, A.A. Tseytlin, Renormalizable asymptotically free quantum theory of gravity.
Nucl. Phys. B 201, 469 (1982)
262. D.G. Boulware, G.T. Horowitz, A. Strominger, Zero-energy theorem for scale-invariant
gravity. Phys. Rev. Lett. 50, 1726 (1983)
263. N.H. Barth, S.M. Christensen, Quantizing fourth-order gravity theories: the functional
integral. Phys. Rev. D 28, 1876 (1983)
264. I.L. Buchbinder, S.D. Odintsov, I.L. Shapiro, Effective Action in Quantum Gravity (IOP,
Bristol, 1992)
265. M. Asorey, J.L. López, I.L. Shapiro, Some remarks on high derivative quantum gravity. Int.
J. Mod. Phys. A 12, 5711 (1997). [arXiv:hep-th/9610006]
266. F.d.O. Salles, I.L. Shapiro, Do we have unitary and (super)renormalizable quantum gravity
below Planck scale? Phys. Rev. D 89, 084054 (2014). [arXiv:1401.4583]
267. A. Hindawi, B.A. Ovrut, D. Waldram, Consistent spin-two coupling and quadratic gravitation.
268. A. Hindawi, B.A. Ovrut, D. Waldram, Nontrivial vacua in higher derivative gravitation. Phys.
Rev. D 53, 5597 (1996). [arXiv:hep-th/9509147]
269. T. Chiba, Generalized gravity and ghost. JCAP 0503, 008 (2005). [arXiv:gr-qc/0502070]
270. A. Núñez, S. Solganik, Ghost constraints on modified gravity. Phys. Lett. B 608, 189 (2005).
271. S.M. Carroll, A. De Felice, V. Duvvuri, D.A. Easson, M. Trodden, M.S. Turner, Cos-
mology of generalized modified gravity models. Phys. Rev. D 71, 063513 (2005).
References 379
272. A. De Felice, M. Hindmarsh, M. Trodden, Ghosts, instabilities, and superluminal propagation

in modified gravity models. JCAP 0608, 005 (2006). [arXiv:astro-ph/0604154]
273. S. Kawai, M.-a. Sakagami, J. Soda, Instability of 1-loop superstring cosmology. Phys. Lett.
B 437, 284 (1998). [arXiv:gr-qc/9802033]
274. S. Kawai, J. Soda, Evolution of fluctuations during graceful exit in string cosmology. Phys.
Lett. B 460, 41 (1999). [arXiv:gr-qc/9903017]
275. G. Calcagni, B. de Carlos, A. De Felice, Ghost conditions for Gauss–Bonnet cosmologies.
Nucl. Phys. B 752, 404 (2006). [arXiv:hep-th/0604201]
276. T. Koivisto, D.F. Mota, Cosmology and astrophysical constraints of Gauss–Bonnet dark
energy. Phys. Lett. B 644, 104 (2007). [arXiv:astro-ph/0606078]
277. T. Koivisto, D.F. Mota, Gauss–Bonnet quintessence: background evolution, large scale struc-
ture and cosmological constraints. Phys. Rev. D 75, 023518 (2007). [arXiv:hep-th/0609155]
278. H. Nariai, On the removal of initial singularity in a big-bang universe in terms of a
renormalized theory of gravitation. I. Examination of the present status and a new approach.
Prog. Theor. Phys. 46, 433 (1971)
279. H. Nariai, K. Tomita, On the removal of initial singularity in a big-bang universe in terms of
a renormalized theory of gravitation. II. Criteria for obtaining a physically reasonable model.
Prog. Theor. Phys. 46, 776 (1971)
280. P.C.W. Davies, Singularity avoidance and quantum conformal anomalies. Phys. Lett. B 68,
402 (1977)
281. M.V. Fischetti, J.B. Hartle, B.L. Hu, Quantum effects in the early universe. I. Influence of
trace anomalies on homogeneous, isotropic, classical geometries. Phys. Rev. D 20, 1757
(1979)
282. K. Tomita, T. Azuma, H. Nariai, On anisotropic and homogeneous cosmological models in
the renormalized theory of gravitation. Prog. Theor. Phys. 60, 403 (1978)
283. J.D. Barrow, A.C. Ottewill, The stability of general relativistic cosmological theory. J. Phys.
A 16, 2757 (1983)
284. S.W. Hawking, R. Penrose, The singularities of gravitational collapse and cosmology. Proc.
R. Soc. Lond. A 314, 529 (1970)
285. A.A. Starobinsky, A new type of isotropic cosmological models without singularity. Phys.
Lett. B 91, 99 (1980)
286. S. Gottlöber, H.-J. Schmidt, A.A. Starobinsky, Sixth-order gravity and conformal transforma-
tions. Class. Quantum Grav. 7, 893 (1990)
287. H.-J. Schmidt, Variational derivatives of arbitrarily high order and multi-inflation cosmologi-
cal models. Class. Quantum Grav. 7, 1023 (1990)
288. D. Wands, Extended gravity theories and the Einstein–Hilbert action. Class. Quantum Grav.
11, 269 (1994). [arXiv:gr-qc/9307034]
289. S. Cecotti, Higher derivative supergravity is equivalent to standard supergravity coupled to
matter. Phys. Lett. B 190, 86 (1987)
290. S. Cecotti, S. Ferrara, M. Porrati, S. Sabharwal, New minimal higher derivative supergravity
coupled to matter. Nucl. Phys. B 306, 160 (1988)
291. B.N. Breizman, V.Ts. Gurovich, V.P. Sokolov, The possibility of setting up regular cosmolog-
ical solutions. Zh. Eksp. Teor. Fiz. 59, 288 (1970) [Sov. Phys. JETP 32, 155 (1971)]
292. H.A. Buchdahl, Non-linear Lagrangians and cosmological theory. Mon. Not. R. Astron. Soc.
150, 1 (1970)
293. H. Nariai, On a phenomenological modification of Einstein’s gravitational Lagrangian. Prog.
Theor. Phys. 51, 613 (1974)
294. P. Teyssandier, P. Tourrenc, The Cauchy problem for the R C R2 theories of gravity without
torsion. J. Math. Phys. 24, 2793 (1983)
295. G. Magnano, M. Ferraris, M. Francaviglia, Nonlinear gravitational Lagrangians. Gen. Relat.
Grav. 19, 465 (1987)
296. H.-J. Schmidt, Comparing self-interacting scalar fields and R C R3 cosmological models.
Astron. Nachr. 308, 183 (1987). [arXiv:gr-qc/0106035]
297. H.A. Buchdahl, Quadratic Lagrangians and Palatini’s device. J. Phys. A 12, 1229 (1979)
298. G. Allemandi, A. Borowiec, M. Francaviglia, Accelerated cosmological models in Ricci

squared gravity. Phys. Rev. D 70, 103503 (2004). [arXiv:hep-th/0407090]
299. A. Borowiec, M. Ferraris, M. Francaviglia, I. Volovich, Universality of Einstein equations for
the Ricci squared Lagrangians. Class. Quantum Grav. 15, 43 (1998). [arXiv:gr-qc/9611067]
300. A. Núñez, S. Solganik, The content of f .R/ gravity. arXiv:hep-th/0403159
301. V. Faraoni, Solar system experiments do not yet veto modified gravity models. Phys. Rev. D
74, 023529 (2006). [arXiv:gr-qc/0607016]
302. Y.-S. Song, W. Hu, I. Sawicki, Large scale structure of f .R/ gravity. Phys. Rev. D 75, 044004
(2007). [arXiv:astro-ph/0610532]
303. K.-i. Maeda, Inflation as a transient attractor in R2 cosmology. Phys. Rev. D 37, 858 (1988)
304. K.-i. Maeda, Towards the Einstein–Hilbert action via conformal transformation. Phys. Rev. D
39, 3159 (1989)
305. J.D. Barrow, S. Cotsakis, Inflation and the conformal structure of higher-order gravity
theories. Phys. Lett. B 214, 515 (1988)
306. T. Koivisto, H. Kurki-Suonio, Cosmological perturbations in the Palatini formulation of
modified gravity. Class. Quantum Grav. 23, 2355 (2006). [arXiv:astro-ph/0509422]
307. M. Ferraris, M. Francaviglia, I. Volovich, The universality of vacuum Einstein equations with
cosmological constant. Class. Quantum Grav. 11, 1505 (1994). [arXiv:gr-qc/9303007]
308. G.J. Olmo, The gravity Lagrangian according to solar system experiments. Phys. Rev. Lett.
95, 261102 (2005). [arXiv:gr-qc/0505101]
309. T.P. Sotiriou, f .R/ gravity and scalar-tensor theory. Class. Quantum Grav. 23, 5117 (2006).
310. B. Li, J.D. Barrow, Cosmology of f .R/ gravity in metric variational approach. Phys. Rev. D
75, 084010 (2007). [arXiv:gr-qc/0701111]
311. I. Navarro, K. Van Acoleyen, f .R/ actions, cosmic acceleration and local tests of gravity.
JCAP 0702, 022 (2007). [arXiv:gr-qc/0611127]
312. T. Faulkner, M. Tegmark, E.F. Bunn, Y. Mao, Constraining f .R/ gravity as a scalar-tensor
theory. Phys. Rev. D 76, 063505 (2007). [arXiv:astro-ph/0612569]
313. W. Hu, I. Sawicki, Models of f .R/ cosmic acceleration that evade solar system tests. Phys.
Rev. D 76, 064004 (2007). [arXiv:0705.1158]
314. S. Capozziello, S. Tsujikawa, Solar system and equivalence principle constraints on f .R/
gravity by the chameleon approach. Phys. Rev. D 77, 107501 (2008). [arXiv:0712.2268]
315. I.W. Roxburgh, Nonlinear Lagrangian theories of gravity. Gen. Relat. Grav. 8, 219 (1977)
316. T. Clifton, J.D. Barrow, The power of general relativity. Phys. Rev. D 72, 103005 (2005).
317. H. Nariai, Gravitational instability of regular model-universes in a modified theory of general
relativity. Prog. Theor. Phys. 49, 165 (1973)
318. P. Zhang, Testing gravity against the early time integrated Sachs–Wolfe effect. Phys. Rev. D
73, 123504 (2006). [arXiv:astro-ph/0511218]
319. G. Cognola, E. Elizalde, S. Nojiri, S.D. Odintsov, L. Sebastiani, S. Zerbini, Class of viable
modified f .R/ gravities describing inflation and the onset of accelerated expansion. Phys. Rev.
D 77, 046009 (2008). [arXiv:0712.4017]
320. G. Allemandi, A. Borowiec, M. Francaviglia, Accelerated cosmological models in first-order
nonlinear gravity. Phys. Rev. D 70, 043524 (2004). [arXiv:hep-th/0403264]
321. S. Nojiri, S.D. Odintsov, Modified gravity with ln R terms and cosmic acceleration. Gen.
Relat. Grav. 36, 1765 (2004). [arXiv:hep-th/0308176]
322. L.H. Ford, D.J. Toms, Dynamical symmetry breaking due to radiative corrections in cosmol-
ogy. Phys. Rev. D 25, 1510 (1982)
323. A.A. Starobinsky, H.-J. Schmidt, On a general vacuum solution of fourth-order gravity. Class.
324. K.-i. Maeda, J.A. Stein-Schabes, T. Futamase, Inflation in a renormalizable cosmological
model and the cosmic no-hair conjecture. Phys. Rev. D 39, 2848 (1989)
325. S. Cotsakis, P.J. Saich, Power-law inflation and conformal transformations. Class. Quantum
Grav. 11, 383 (1994)
References 381
326. H. Weyl, Space, Time, and Matter (Dover, Mineola, 1952)

327. B. Whitt, Fourth-order gravity as general relativity plus matter. Phys. Lett. B 145, 176 (1984)
328. S. Capozziello, F. Occhionero, L. Amendola, The phase-space view of inflation II: fourth
order models. Int. J. Mod. Phys. D 1, 615 (1993)
329. S. Capozziello, S. Nojiri, S.D. Odintsov, A. Troisi, Cosmological viability of f .R/-gravity as
an ideal fluid and its compatibility with a matter dominated phase. Phys. Lett. B 639, 135
(2006). [arXiv:astro-ph/0604431]
330. S.M. Carroll, V. Duvvuri, M. Trodden, M.S. Turner, Is cosmic speed-up due to new
gravitational physics? Phys. Rev. D 70, 043528 (2004). [arXiv:astro-ph/0306438]
331. T. Chiba, 1=R gravity and scalar-tensor gravity. Phys. Lett. B 575, 1 (2003).
332. G.A. Vilkovisky, Effective action in quantum gravity. Class. Quantum Grav. 9, 895 (1992)
333. A.D. Dolgov, M. Kawasaki, Can modified gravity explain accelerated cosmic expansion?
Phys. Lett. B 573, 1 (2003). [arXiv:astro-ph/0307285]
334. G.J. Olmo, Post-Newtonian constraints on f .R/ cosmologies in metric and Palatini formalism.
Phys. Rev. D 72, 083505 (2005). [arXiv:gr-qc/0505135]
335. L. Amendola, D. Polarski, S. Tsujikawa, Are f .R/ dark energy models cosmologically viable?
336. L. Amendola, D. Polarski, S. Tsujikawa, Power-laws f .R/ theories are cosmologically
unacceptable. Int. J. Mod. Phys. D 16, 1555 (2007). [arXiv:astro-ph/0605384]
337. D.N. Vollick, 1=R curvature corrections as the source of the cosmological acceleration. Phys.
338. X. Meng, P. Wang, Modified Friedmann equations in R1 -modified gravity. Class. Quantum
Grav. 20, 4949 (2003). [arXiv:astro-ph/0307354]
339. X. Meng, P. Wang, Cosmological evolution in 1=R-gravity theory. Class. Quantum Grav. 21,
951 (2004). [arXiv:astro-ph/0308031]
340. M. Amarzguioui, Ø. Elgarøy, D.F. Mota, T. Multamäki, Cosmological constraints on
f .R/ gravity theories within the Palatini approach. Astron. Astrophys. 454, 707 (2006).
341. S. Nojiri, S.D. Odintsov, Modified gravity with negative and positive powers of the curvature:
unification of the inflation and of the cosmic acceleration. Phys. Rev. D 68, 123512 (2003).
342. A.W. Brookfield, C. van de Bruck, L.M.H. Hall, Viability of f .R/ theories with additional
powers of curvature. Phys. Rev. D 74, 064028 (2006). [arXiv:hep-th/0608015]
343. Á. de la Cruz-Dombriz, P.K.S. Dunsby, S. Kandhai, D. Sáez-Gómez, Theoretical and
observational constraints of viable f .R/ theories of gravity. Phys. Rev. D 93, 084016 (2016).
[arXiv:1511.00102]
344. G. Cognola, E. Elizalde, S. Nojiri, S.D. Odintsov, S. Zerbini, One-loop f .R/ gravity in de
Sitter universe. JCAP 0502, 010 (2005). [arXiv:hep-th/0501096]
345. J.A.R. Cembranos, The Newtonian limit at intermediate energies. Phys. Rev. D 73, 064029
(2006). [arXiv:gr-qc/0507039]
346. T. Koivisto, Matter power spectrum in f .R/ gravity. Phys. Rev. D 73, 083517 (2006).
347. S. Capozziello, V.F. Cardone, A. Troisi, Dark energy and dark matter as curvature effects.
348. S. Nojiri, S.D. Odintsov, M. Sami, Dark energy cosmology from higher-order, string-inspired
gravity and its reconstruction. Phys. Rev. D 74, 046004 (2006). [arXiv:hep-th/0605039]
349. S. Nojiri, S.D. Odintsov, Modified f .R/ gravity consistent with realistic cosmology: from
matter dominated epoch to dark energy universe. Phys. Rev. D 74, 086005 (2006).
350. S. Nojiri, S.D. Odintsov, Modified gravity and its reconstruction from the universe expansion
history. J. Phys. Conf. Ser. 66, 012005 (2007). [arXiv:hep-th/0611071]
351. L. Amendola, R. Gannouji, D. Polarski, S. Tsujikawa, Conditions for the cosmological
viability of f .R/ dark energy models. Phys. Rev. D 75, 083504 (2007). [arXiv:gr-qc/0612180]
352. I. Sawicki, W. Hu, Stability of cosmological solutions in f .R/ models of gravity. Phys. Rev.
D 75, 127502 (2007). [arXiv:astro-ph/0702278]
353. S. Fay, S. Nesseris, L. Perivolaropoulos, Can f .R/ modified gravity theories mimic a CDM
cosmology? Phys. Rev. D 76, 063504 (2007). [arXiv:gr-qc/0703006]
354. S.A. Appleby, R.A. Battye, Do consistent F.R/ models mimic general relativity plus ? Phys.
Lett. B 654, 7 (2007). [arXiv:0705.3199]
355. A.A. Starobinsky, Disappearing cosmological constant in f .R/ gravity. JETP Lett. 86, 157
(2007). [arXiv:0706.2041]
356. S. Tsujikawa, Observational signatures of f .R/ dark energy models that satisfy cosmological
and local gravity constraints. Phys. Rev. D 77, 023507 (2008). [arXiv:0709.1391]
357. A. De Felice, S. Tsujikawa, f .R/ theories. Living Rev. Relat. 13, 3 (2010).
358. S. Tsujikawa, Modified gravity models of dark energy. Lect. Notes Phys. 800, 99 (2010).
[arXiv:1101.0191]
359. T. Clifton, P.G. Ferreira, A. Padilla, C. Skordis, Modified gravity and cosmology. Phys. Rep.
513, 1 (2012). [arXiv:1106.2476]
360. J.-c. Hwang, Cosmological perturbations in generalized gravity theories: formulation. Class.
Quantum Grav. 7, 1613 (1990)
361. J.-c. Hwang, Perturbations of the Robertson–Walker space: multicomponent sources and
generalized gravity. Astrophys. J. 375, 443 (1991)
362. J.-c. Hwang, H. Noh, Cosmological perturbations in generalized gravity theories. Phys. Rev.
D 54, 1460 (1996)
363. J.-c. Hwang, H. Noh, Gauge-ready formulation of the cosmological kinetic theory in
generalized gravity theories. Phys. Rev. D 65, 023512 (2002). [arXiv:astro-ph/0102005]
364. R. Bean, D. Bernat, L. Pogosian, A. Silvestri, M. Trodden, Dynamics of linear perturbations
in f .R/ gravity. Phys. Rev. D 75, 064020 (2007). [arXiv:astro-ph/0611321]
365. L. Pogosian, A. Silvestri, Pattern of growth in viable f .R/ cosmologies. Phys. Rev. D 77,
023503 (2008); Erratum-ibid. D 81, 049901 (2010). [arXiv:0709.0296]
366. S. Tsujikawa, K. Uddin, R. Tavakol, Density perturbations in f .R/ gravity theories in metric
and Palatini formalisms. Phys. Rev. D 77, 043007 (2008). [arXiv:0712.0082]
367. P.A.R. Ade et al. [Planck Collaboration], Planck 2015 results. XIV. Dark energy and modified
gravity. Astron. Astrophys. 594, A14 (2016). [arXiv:1502.01590]
368. G.W. Horndeski, Second-order scalar-tensor field equations in a four-dimensional space. Int.
J. Theor. Phys. 10, 363 (1974)
369. C. Charmousis, E.J. Copeland, A. Padilla, P.M. Saffin, General second-order scalar-tensor
theory and self-tuning. Phys. Rev. Lett. 108, 051101 (2012). [arXiv:1106.2000]
370. R. Kase, S. Tsujikawa, Cosmology in generalized Horndeski theories with second-order
equations of motion. Phys. Rev. D 90, 044073 (2014). [arXiv:1407.0794]
371. T. Chiba, T. Okabe, M. Yamaguchi, Kinetically driven quintessence. Phys. Rev. D 62, 023511
(2000). [arXiv:astro-ph/9912463]
372. C. Armendáriz-Picón, V.F. Mukhanov, P.J. Steinhardt, Dynamical solution to the problem of
a small cosmological constant and late-time cosmic acceleration. Phys. Rev. Lett. 85, 4438
(2000). [arXiv:astro-ph/0004134]
373. C. Armendáriz-Picón, V.F. Mukhanov, P.J. Steinhardt, Essentials of k-essence. Phys. Rev. D
63, 103510 (2001). [arXiv:astro-ph/0006373]
374. A. Melchiorri, L. Mersini, C.J. Ödman, M. Trodden, The state of the dark energy equation of
state. Phys. Rev. D 68, 043509 (2003). [arXiv:astro-ph/0211522]
375. S. Tsujikawa, M. Sami, A unified approach to scaling solutions in a general cosmological
background. Phys. Lett. B 603, 113 (2004). [arXiv:hep-th/0409212]
376. D.B. Fairlie, J. Govaerts, Universal field equations with reparametrization invariance. Phys.
377. A. Nicolis, R. Rattazzi, E. Trincherini, Galileon as a local modification of gravity. Phys. Rev.
D 79, 064036 (2009). [arXiv:0811.2197]
378. C. Deffayet, X. Gao, D.A. Steer, G. Zahariade, From k-essence to generalised Galileons. Phys.
Rev. D 84, 064039 (2011). [arXiv:1103.3260]
References 383
379. T. Kobayashi, M. Yamaguchi, J.’i. Yokoyama, Generalized G-inflation: inflation with the most
general second-order field equations. Prog. Theor. Phys. 126, 511 (2011). [arXiv:1105.5723]
380. J. Gleyzes, D. Langlois, F. Piazza, F. Vernizzi, Healthy theories beyond Horndeski. Phys. Rev.
Lett. 114, 211101 (2015). [arXiv:1404.6495]
381. C. Lin, S. Mukohyama, R. Namba, R. Saitou, Hamiltonian structure of scalar-tensor theories
beyond Horndeski. JCAP 1410, 071 (2014). [arXiv:1408.0670]
382. J. Gleyzes, D. Langlois, F. Piazza, F. Vernizzi, Exploring gravitational theories beyond
Horndeski. JCAP 1502, 018 (2015). [arXiv:1408.1952]
383. X. Gao, Hamiltonian analysis of spatially covariant gravity. Phys. Rev. D 90, 104033 (2014).
[arXiv:1409.6708]
384. A. De Felice, S. Tsujikawa, Inflationary gravitational waves in the effective field theory of
modified gravity. Phys. Rev. D 91, 103506 (2015). [arXiv:1411.0736]
385. S. Tsujikawa, Possibility of realizing weak gravity in redshift space distortion measurements.
Phys. Rev. D 92, 044029 (2015). [arXiv:1505.02459]
386. F. Arroja, N. Bartolo, P. Karmakar, S. Matarrese, The two faces of mimetic Horndeski
gravity: disformal transformations and Lagrange multiplier. JCAP 1509, 051 (2015).
[arXiv:1506.08575]
387. F. Arroja, N. Bartolo, P. Karmakar, S. Matarrese, Cosmological perturbations in mimetic
Horndeski gravity. JCAP 1604, 042 (2016). [arXiv:1512.09374]
388. J.D. Bekenstein, Fine structure constant: is it really a constant? Phys. Rev. D 25, 1527 (1982)
389. J.D. Bekenstein, Fine structure constant variability, equivalence principle and cosmology.
390. J. Magueijo, New varying speed of light theories. Rep. Prog. Phys. 66, 2025 (2003).
391. J. Magueijo, Covariant and locally Lorentz invariant varying speed of light theories. Phys.
Rev. D 62, 103521 (2000). [arXiv:gr-qc/0007036]
392. J. Magueijo, Speedy sound and cosmic structure. Phys. Rev. Lett. 100, 231302 (2008).
[arXiv:0803.0859]
393. A. Albrecht, J. Magueijo, A time varying speed of light as a solution to cosmological puzzles.
394. J.D. Barrow, J. Magueijo, Solutions to the quasi-flatness and quasilambda problems. Phys.
395. B.A. Bassett, S. Liberati, C. Molina-París, M. Visser, Geometrodynamics of variable-speed-
of-light cosmologies. Phys. Rev. D 62, 103518 (2000). [arXiv:astro-ph/0001441]
396. K. Tomita, Bulk flows and CMB dipole anisotropy in cosmological void models. Astrophys.
J. 529, 26 (2000). [arXiv:astro-ph/9905278]
397. K. Tomita, Distances and lensing in cosmological void models. Astrophys. J. 529, 38 (2000).
398. M.-N. Célérier, Do we really see a cosmological constant in the supernovae data? Astron.
399. K. Tomita, A local void and the accelerating universe. Mon. Not. R. Astron. Soc. 326, 287
(2001). [arXiv:astro-ph/0011484]
400. K. Tomita, Analyses of type Ia supernova data in cosmological models with a local void. Prog.
Theor. Phys. 106, 929 (2001). [arXiv:astro-ph/0104141]
401. M.-N. Célérier, The accelerated expansion of the universe challenged by an effect of the
inhomogeneities. A review. New Adv. Phys. 1, 29 (2007). [arXiv:astro-ph/0702416]
402. G. Lemaître, The expanding universe. Ann. Soc. Sci. Bruxelles Ser. I Sci. Math. Astron. Phys.
A 53, 51 (1933) [Gen. Relat. Grav. 29, 641 (1997)]
403. R.C. Tolman, Effect of imhomogeneity on cosmological models. Proc. Natl. Acad. Sci. 20,
169 (1934) [Gen. Relat. Grav. 29, 935 (1997)]
404. H. Bondi, Spherically symmetrical models in general relativity. Mon. Not. R. Astron. Soc.
107, 410 (1947)
405. M. Hossein Partovi, B. Mashhoon, Toward verification of large-scale homogeneity in
cosmology. Astrophys. J. 276, 4 (1984)
406. N.P. Humphreys, R. Maartens, D.R. Matravers, Anisotropic observations in universes with
nonlinear inhomogeneity. Astrophys. J. 477, 47 (1997). [arXiv:astro-ph/9602033]
407. N. Mustapha, C. Hellaby, G.F.R. Ellis, Large-scale inhomogeneity versus source evolution:
can we distinguish them observationally? Mon. Not. R. Astron. Soc. 292, 817 (1997).
408. S. Khakshournia, R. Mansouri, Dynamics of general relativistic spherically symmetric dust
thick shells. Gen. Relat. Grav. 34, 1847 (2002). [arXiv:gr-qc/0308025]
409. T. Biswas, A. Notari, ‘Swiss-cheese’ inhomogeneous cosmology and the dark energy
problem. JCAP 0806, 021 (2008). [arXiv:astro-ph/0702555]
410. J.W. Moffat, Cosmic microwave background, accelerating universe and inhomogeneous
cosmology. JCAP 0510, 012 (2005). [arXiv:astro-ph/0502110]
411. J.W. Moffat, Late-time inhomogeneity and acceleration without dark energy. JCAP 0605, 001
(2006). [arXiv:astro-ph/0505326]
412. H. Alnes, M. Amarzguioui, O. Grøn, Inhomogeneous alternative to dark energy? Phys. Rev.
D 73, 083519 (2006). [arXiv:astro-ph/0512006]
413. D.J.H. Chung, A.E. Romano, Mapping luminosity-redshift relationship to Lemaître–Tolman–
Bondi cosmology. Phys. Rev. D 74, 103507 (2006). [arXiv:astro-ph/0608403]
414. K. Enqvist, T. Mattsson, The effect of inhomogeneous expansion on the supernova observa-
tions. JCAP 0702, 019 (2007). [arXiv:astro-ph/0609120]
415. K. Enqvist, Lemaître–Tolman–Bondi model and accelerating expansion. Gen. Relat. Grav.
40, 451 (2008). [arXiv:0709.2044]
416. S. Alexander, T. Biswas, A. Notari, D. Vaid, Local void vs dark energy: confrontation with
WMAP and type Ia supernovae. JCAP 0909, 025 (2009). [arXiv:0712.0370]
417. J. García-Bellido, T. Haugbølle, Confronting Lemaître–Tolman–Bondi models with observa-
tional cosmology. JCAP 0804, 003 (2008). [arXiv:0802.1523]
418. J. García-Bellido, T. Haugbølle, Looking the void in the eyes—the kinematic
Sunyaev–Zeldovich effect in Lemaître–Tolman–Bondi models. JCAP 0809, 016 (2008).
[arXiv:0807.1326]
419. J.P. Zibin, A. Moss, D. Scott, Can we avoid dark energy? Phys. Rev. Lett. 101, 251303 (2008).
[arXiv:0809.3761]
420. S. February, J. Larena, M. Smith, C. Clarkson, Rendering dark energy void. Mon. Not. R.
Astron. Soc. 405, 2231 (2010). [arXiv:0909.1479]
421. T. Biswas, A. Notari, W. Valkenburg, Testing the void against cosmological data: fitting CMB,
BAO, SN and H0 . JCAP 1011, 030 (2010). [arXiv:1007.3065]
422. A. Moss, J.P. Zibin, D. Scott, Precision cosmology defeats void models for acceleration. Phys.
Rev. D 83, 103515 (2011). [arXiv:1007.3725]
423. M. Zumalacárregui, J. García-Bellido, P. Ruiz-Lapuente, Tension in the void: cosmic rulers
strain inhomogeneous cosmologies. JCAP 1210, 009 (2012). [arXiv:1201.2790]
424. R. de Putter, L. Verde, R. Jimenez, Testing LTB void models without the cosmic microwave
background or large scale structure: new constraints from galaxy ages. JCAP 1302, 047
(2013). [arXiv:1208.4534]
425. H. Goto, H. Kodama, The gravitational lensing effect on the CMB polarisation anisotropy in
the -LTB model. Prog. Theor. Phys. 125, 815 (2011). [arXiv:1101.0476]
426. P. Hunt, S. Sarkar, Constraints on large scale inhomogeneities from WMAP-5 and SDSS:
confrontation with recent observations. Mon. Not. R. Astron. Soc. 401, 547 (2010).
[arXiv:0807.4508]
427. P.A.R. Ade et al. [Planck Collaboration], Planck intermediate results. XIII. Constraints on
peculiar velocities. Astron. Astrophys. 561, A97 (2014). [arXiv:1303.5090]
428. M. Henneaux, C. Teitelboim, The cosmological constant and general covariance. Phys. Lett.
B 222, 195 (1989)
429. J.L. Anderson, D. Finkelstein, Cosmological constant and fundamental length. Am. J. Phys.
39, 901 (1971)
430. J. Rayski, The problems of quantum gravity. Gen. Relat. Grav. 11, 19 (1979)
References 385
431. J.J. van der Bij, H. van Dam, Y.J. Ng, The exchange of massless spin-two particles. Physica
A 116, 307 (1982)
432. A. Zee, Remarks on the cosmological constant paradox, in High Energy Physics: Proceedings
of the 20th Orbis Scientiae, 1983, ed. by S.L. Mintz, A. Perlmutter (Plenum, New York, 1985)
433. W. Buchmüller, N. Dragon, Einstein gravity from restricted coordinate invariance. Phys. Lett.
B 207, 292 (1988)
434. W. Buchmüller, N. Dragon, Gauge fixing and the cosmological constant. Phys. Lett. B 223,
313 (1989)
435. W.G. Unruh, Unimodular theory of canonical quantum gravity. Phys. Rev. D 40, 1048 (1989)
436. W.G. Unruh, R.M. Wald, Time and the interpretation of canonical quantum gravity. Phys.
Rev. D 40, 2598 (1989)
437. Y.J. Ng, H. van Dam, Possible solution to the cosmological constant problem. Phys. Rev. Lett.
65, 1972 (1990)
438. Y.J. Ng, H. van Dam, Unimodular theory of gravity and the cosmological constant. J. Math.
Phys. 32, 1337 (1991)
439. A.N. Petrov, On the cosmological constant as a constant of integration. Mod. Phys. Lett. A
06, 2107 (1991)
440. E. Álvarez, Can one tell Einstein’s unimodular theory from Einstein’s general relativity? JHEP
0503, 002 (2005). [arXiv:hep-th/0501146]
441. E. Álvarez, D. Blas, J. Garriga, E. Verdaguer, Transverse Fierz–Pauli symmetry. Nucl. Phys.
B 756, 148 (2006). [arXiv:hep-th/0606019]
442. E. Álvarez, A.F. Faedo, J.J. López-Villarejo, Ultraviolet behavior of transverse gravity. JHEP
0810, 023 (2008). [arXiv:0807.1293]
443. B. Fiol, J. Garriga, Semiclassical unimodular gravity. JCAP 1008, 015 (2010).
[arXiv:0809.1371]
444. L. Smolin, Quantization of unimodular gravity and the cosmological constant problems. Phys.
Rev. D 80, 084003 (2009). [arXiv:0904.4841]
445. E. Álvarez, R. Vidal, Weyl transverse gravity (WTDiff) and the cosmological constant. Phys.
Rev. D 81, 084057 (2010). [arXiv:1001.4458]
446. D. Blas, M. Shaposhnikov, D. Zenhäusern, Scale-invariant alternatives to general relativity.
Phys. Rev. D 84, 044001 (2011). [arXiv:1104.1392]
447. E. Álvarez, The weight of matter. JCAP 1207, 002 (2012). [arXiv:1204.6162]
448. A. Eichhorn, On unimodular quantum gravity. Class. Quantum Grav. 30, 115016 (2013).
[arXiv:1301.0879]
449. C. Barceló, R. Carballo-Rubio, L.J. Garay, Unimodular gravity and general relativity from
graviton self-interactions. Phys. Rev. D 89, 124019 (2014). [arXiv:1401.2941]
450. C. Barceló, R. Carballo-Rubio, L.J. Garay, Absence of cosmological constant problem in
special relativistic field theory of gravity. arXiv:1406.7713
451. A. Padilla, I.D. Saltas, A note on classical and quantum unimodular gravity. Eur. Phys. J. C
75, 561 (2015). [arXiv:1409.3573]
452. R. Bufalo, M. Oksanen, A. Tureanu, How unimodular gravity theories differ from general
relativity at quantum level. Eur. Phys. J. C 75, 477 (2015). [arXiv:1505.04978]
453. A. Basak, O. Fabre, S. Shankaranarayanan, Cosmological perturbation of unimodular gravity
and general relativity are identical. Gen. Relat. Grav. 48, 123 (2016). [arXiv:1511.01805]
454. E.I. Guendelman, A.B. Kaganovich, Principle of nongravitating vacuum energy and some of
its consequences. Phys. Rev. D 53, 7020 (1996)
455. E.I. Guendelman, A.B. Kaganovich, Gravitational theory without the cosmological constant
problem. Phys. Rev. D 55, 5970 (1997). [arXiv:gr-qc/9611046]
456. E.I. Guendelman, Scale invariance, new inflation and decaying lambda terms. Mod. Phys.
Lett. A 14, 1043 (1999). [arXiv:gr-qc/9901017]
457. E.I. Guendelman, A.B. Kaganovich, Dynamical measure and field theory models free of the
cosmological constant problem. Phys. Rev. D 60, 065004 (1999). [arXiv:gr-qc/9905029]
458. M. Fierz, W. Pauli, On relativistic wave equations for particles of arbitrary spin in an
electromagnetic field. Proc. R. Soc. Lond. A 173, 211 (1939)
459. G.E. Volovik, Superfluid analogies of cosmological phenomena. Phys. Rep. 351, 195 (2001).
460. C. Barceló, S. Liberati, M. Visser, Analogue gravity. Living Rev. Relat. 14, 3 (2011)
461. G.E. Volovik, Vacuum energy and cosmological constant: view from condensed matter. J.
Low Temp. Phys. 124, 25 (2001). [arXiv:gr-qc/0101111]
462. G.E. Volovik, Cosmological constant and vacuum energy. Ann. Phys. (Berlin) 14, 165 (2005).
463. G.E. Volovik, Vacuum energy: myths and reality. Int. J. Mod. Phys. D 15, 1987 (2006).
464. G. Jannes, G.E. Volovik, The cosmological constant: a lesson from the effective gravity of
topological Weyl media. JETP Lett. 96, 215 (2012). [arXiv:1108.5086]
465. F.R. Klinkhamer, G.E. Volovik, Self-tuning vacuum variable and cosmological constant.
Phys. Rev. D 77, 085015 (2008). [arXiv:0711.3170]
466. F.R. Klinkhamer, G.E. Volovik, Dynamic vacuum variable and equilibrium approach in
cosmology. Phys. Rev. D 78, 063528 (2008). [arXiv:0806.2805]
467. F.R. Klinkhamer, G.E. Volovik, Gluonic vacuum, q-theory, and the cosmological constant.
Phys. Rev. D 79, 063527 (2009). [arXiv:0811.4347]
468. D.G. Caldi, A. Chodos, Cosmological neutrino condensates. arXiv:hep-ph/9903416
469. T. Inagaki, X. Meng, T. Murata, Dark energy problem in a four fermion interaction model.
470. F. Giacosa, R. Hofmann, M. Neubert, A model for the very early universe. JHEP 0802, 077
(2008). [arXiv:0801.0197]
471. S. Alexander, T. Biswas, The cosmological BCS mechanism and the big bang singularity.
Phys. Rev. D 80, 023501 (2009). [arXiv:0807.4468]
472. S. Alexander, T. Biswas, G. Calcagni, Cosmological Bardeen–Cooper–Schrieffer condensate
as dark energy. Phys. Rev. D 81, 043511 (2010); Erratum-ibid. D 81, 069902(E) (2010).
[arXiv:0906.5161]
473. N.J. Popławski, Cosmological constant from quarks and torsion. Ann. Phys. (Berlin) 523, 291
(2011). [arXiv:1005.0893]
474. J.M. Weller, Fermion condensate from torsion in the reheating era after inflation. Phys. Rev.
D 88, 083511 (2013). [arXiv:1307.2423]
475. S. Finazzi, S. Liberati, L. Sindoni, Cosmological constant: a lesson from Bose–Einstein
condensates. Phys. Rev. Lett. 108, 071101 (2012). [arXiv:1103.4841]
476. B.L. Hu, Can spacetime be a condensate? Int. J. Theor. Phys. 44, 1785 (2005).
477. L. Sindoni, Emergent models for gravity: an overview of microscopic models. SIGMA 8,
027 (2012). [arXiv:1110.0686]
478. H.C. Ohanian, Gravitons as Goldstone bosons. Phys. Rev. 184, 1305 (1969)
479. D. Atkatz, Dynamical method for generating the gravitational interaction. Phys. Rev. D 17,
1972 (1978)
480. S. Deser, Gravity from self-interaction redux. Gen. Relat. Grav. 42, 641 (2010).
[arXiv:0910.2975]
481. C. Barceló, S. Liberati, M. Visser, Analog gravity from field theory normal modes? Class.
482. C. Barceló, M. Visser, S. Liberati, Einstein gravity as an emergent phenomenon? Int. J. Mod.
Phys. D 10, 799 (2001). [arXiv:gr-qc/0106002]
483. P.C.W. Davies, Scalar particle production in Schwarzschild and Rindler metrics. J. Phys. A
8, 609 (1975)
484. W.G. Unruh, Notes on black hole evaporation. Phys. Rev. D 14, 870 (1976)
Phys. Rev. D 15, 2752 (1977)
486. J.D. Bekenstein, Universal upper bound on the entropy-to-energy ratio for bounded systems.
Phys. Rev. D 23, 287 (1981)
References 387
487. T. Jacobson, Thermodynamics of space-time: the Einstein equation of state. Phys. Rev. Lett.
75, 1260 (1995). [arXiv:gr-qc/9504004]
488. T. Padmanabhan, Gravity and the thermodynamics of horizons. Phys. Rep. 406, 49 (2005).
489. C. Eling, R. Guedens, T. Jacobson, Non-equilibrium thermodynamics of spacetime. Phys.
Rev. Lett. 96, 121301 (2006). [arXiv:gr-qc/0602001]
490. T. Padmanabhan, A. Paranjape, Entropy of null surfaces and dynamics of spacetime. Phys.
Rev. D 75, 064004 (2007). [arXiv:gr-qc/0701003]
491. D. Kothawala, T. Padmanabhan, Thermodynamic structure of Lanczos–Lovelock field equa-
tions from near-horizon symmetries. Phys. Rev. D 79, 104020 (2009). [arXiv:0904.0215]
492. T. Padmanabhan, A physical interpretation of gravitational field equations. AIP Conf. Proc.
1241, 93 (2010). [arXiv:0911.1403]
493. T. Padmanabhan, D. Kothawala, Lanczos–Lovelock models of gravity. Phys. Rep. 531, 115
(2013). [arXiv:1302.2151]
494. K. Parattu, B.R. Majhi, T. Padmanabhan, Structure of the gravitational action and its relation
with horizon thermodynamics and emergent gravity paradigm. Phys. Rev. D 87, 124011
(2013). [arXiv:1303.1535]
495. T. Padmanabhan, General relativity from a thermodynamic perspective. Gen. Relat. Grav. 46,
1673 (2014). [arXiv:1312.3253]
496. T. Padmanabhan, Classical and quantum thermodynamics of horizons in spherically symmet-
ric space-times. Class. Quantum Grav. 19, 5387 (2002). [arXiv:gr-qc/0204019]
497. A. Paranjape, S. Sarkar, T. Padmanabhan, Thermodynamic route to field equations in
Lanczos–Lovelock gravity. Phys. Rev. D 74, 104015 (2006). [arXiv:hep-th/0607240]
498. D. Kothawala, S. Sarkar, T. Padmanabhan, Einstein’s equations as a thermodynamic identity:
the cases of stationary axisymmetric horizons and evolving spherically symmetric horizons.
Phys. Lett. B 652, 338 (2007). [arXiv:gr-qc/0701002]
499. M. Akbar, R.-G. Cai, Thermodynamic behavior of field equations for f .R/ gravity. Phys. Lett.
B 648, 243 (2007). [arXiv:gr-qc/0612089]
500. A.V. Frolov, L. Kofman, Inflation and de Sitter thermodynamics. JCAP 0305, 009 (2003).
501. R.-G. Cai, S.P. Kim, First law of thermodynamics and Friedmann equations of Friedmann–
Robertson–Walker universe. JHEP 0502, 050 (2005). [arXiv:hep-th/0501055]
502. G. Calcagni, de Sitter thermodynamics and the braneworld. JHEP 0509, 060 (2005).
503. M. Akbar, R.-G. Cai, Friedmann equations of FRW universe in scalar-tensor gravity, f .R/
gravity and first law of thermodynamics. Phys. Lett. B 635, 7 (2006). [arXiv:hep-th/0602156]
504. M. Akbar, R.-G. Cai, Thermodynamic behavior of Friedmann equations at the apparent
horizon of the FRW universe. Phys. Rev. D 75, 084003 (2007). [arXiv:hep-th/0609128]
505. R.-G. Cai, L.-M. Cao, Unified first law and thermodynamics of the apparent horizon in the
FRW universe. Phys. Rev. D 75, 064008 (2007). [arXiv:gr-qc/0611071]
506. T. Padmanabhan, The atoms of space, gravity and the cosmological constant. Int. J. Mod.
Phys. D 25, 1630020 (2016). [arXiv:1603.08658]
507. T. Padmanabhan, Thermodynamics of horizons: a comparison of Schwarzschild, Rindler and
de Sitter spacetimes. Mod. Phys. Lett. A 17, 923 (2002). [arXiv:gr-qc/0202078]
508. T.M. Adamo, C.N. Kozameh, E.T. Newman, Null geodesic congruences, asymptotically flat
space-times and their physical interpretation. Living Rev. Relat. 15, 1 (2012).
509. L. Susskind, The world as a hologram. J. Math. Phys. 36, 6377 (1995).
510. R. Bousso, A covariant entropy conjecture. JHEP 9907, 004 (1999). [arXiv:hep-th/9905177]
511. T. Padmanabhan, The holography of gravity encoded in a relation between entropy, horizon
area and action for gravity. Gen. Relat. Grav. 34, 2029 (2002). [arXiv:gr-qc/0205090]
512. T. Padmanabhan, Holographic gravity and the surface term in the Einstein–Hilbert action.
Braz. J. Phys. 35, 362 (2005). [arXiv:gr-qc/0412068]
513. T. Padmanabhan, A new perspective on gravity and the dynamics of space-time. Int. J. Mod.
Phys. D 14, 2263 (2005). [arXiv:gr-qc/0510015]
514. A. Mukhopadhyay, T. Padmanabhan, Holography of gravitational action functionals. Phys.
Rev. D 74, 124023 (2006). [arXiv:hep-th/0608120]
515. R.C. Tolman, On the use of the energy-momentum principle in general relativity. Phys. Rev.
35, 875 (1930)
516. A. Komar, Covariant conservation laws in general relativity. Phys. Rev. 113, 934 (1959)
517. T. Padmanabhan, Entropy of static spacetimes and microscopic density of states. Class.
518. T. Padmanabhan, Equipartition of energy in the horizon degrees of freedom and the
emergence of gravity. Mod. Phys. Lett. A 25, 1129 (2010). [arXiv:0912.3165]
519. T. Padmanabhan, Surface density of spacetime degrees of freedom from equipartition law in
theories of gravity. Phys. Rev. D 81, 124040 (2010). [arXiv:1003.5665]
520. T. Padmanabhan, Emergence and expansion of cosmic space as due to the quest for
holographic equipartition. arXiv:1206.4916
521. T. Padmanabhan, Emergent perspective of gravity and dark energy. Res. Astron. Astrophys.
12, 891 (2012). [arXiv:1207.0505]
522. T. Padmanabhan, Quantum structure of space-time and black hole entropy. Phys. Rev. Lett.
81, 4297 (1998). [arXiv:hep-th/9801015]
523. T. Padmanabhan, Event horizon: magnifying glass for Planck length physics. Phys. Rev. D
59, 124012 (1999). [arXiv:hep-th/9801138]
524. M. Arzano, G. Calcagni, Black-hole entropy and minimal diffusion. Phys. Rev. D 88, 084017
(2013). [arXiv:1307.6122]
525. D. Kothawala, T. Padmanabhan, Entropy density of spacetime as a relic from quantum gravity.
Phys. Rev. D 90, 124060 (2014). [arXiv:1405.4967]
526. D. Kothawala, T. Padmanabhan, Entropy density of spacetime from the zero point length.
Phys. Lett. B 748, 67 (2015). [arXiv:1408.3963]
527. P. Hořava, D. Minic, Probable values of the cosmological constant in a holographic theory.
Phys. Rev. Lett. 85, 1610 (2000). [arXiv:hep-th/0001145]
Chapter 8
The Problem of Quantum Gravity
Keating: Why do I stand up here? Anybody?

Charlie: To feel taller.
Keating: No! (Keating rings the bell on his desk with his foot.)
Thank you for playing, Mr. Dalton. I stand upon my desk to
remind yourself that we must constantly look at things in a
different way.
— Tom Schulman, Dead Poets Society (1989)
Contents
8.1 Do We Need to Quantize Gravity?. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 390
8.2 Perturbative Quantum Gravity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 392
8.2.1 Supergravity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 397
8.2.2 Effective Field Theory. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 397
8.2.3 Resummed Quantum Gravity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 398
8.3 Approaches to Quantum Gravity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 400
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 402
In Chap. 6, we examined the big-bang problem in general relativity and in generic

expanding cosmologies. All the arguments therein are classical and assume that
spacetime can be described by a Lorentzian manifold, including at arbitrarily small
scales. A fascinating possibility is that classical singularities are resolved in a
quantum setting. In fact, at the classical level a singularity may emerge because there
is no geometrical obstruction for matter to be compressed in arbitrarily small regions
of space and for spacetimes to be distorted to arbitrarily high curvature. However,
if gravity was a quantum interaction, at sufficiently small scales or large curvature
a continuum manifold picture might be invalid and replaced by a very different
structure preventing the formation of pathological configurations. In particular, the
hope is that “quanta of geometry” would not allow themselves to be packed into
regions smaller than a certain critical size, as it happens in loop quantum gravity. Or,
like in string theory, if one was able to set geometry on equal footing with particle

DOI 10.1007/978-3-319-41127-9_8
390 8 The Problem of Quantum Gravity
matter fields and to unify all forces of nature in the dynamics of extended objects,
all ultraviolet divergences would be smoothed out because there would be no point-
wise interactions. Or, again, a certain “minimal length scale” might be implemented
to achieve a similar result.
There are great differences between these scenarios. Technically, their formal-
ization in a rigorous framework turns out to be a formidable task and one may ask
whether we need to quantize gravity at all and, if we make the attempt, why it is so
difficult. This is the subject of the present chapter.
8.1 Do We Need to Quantize Gravity?
The resolution of issues such as the big-bang and the cosmological constant
problems could lie in a fully quantum realm but, to be pedantic, we do not observe
phenomena that, as far as we can tell, necessarily require to quantize gravity. Then,
we should seriously contemplate the eventuality that this is not the right track to
follow and that gravity is an intrinsically classical interaction [1, 2]. In such a
case, only matter fields would be quantized (as dictated by experimental evidence)
and embedded in a classical, general relativistic background. Then, the Einstein
equations would exactly be of the form (2.112), which we repeat here:
G D 2 hT i : (8.1)
Angular brackets denote the expectation value of the total matter energy-momentum
tensor on a quantum state. At this point, we should start devising experiments to
detect deviations of Lorentzian quantum field theory or quantum mechanics from the
predictions based on a flat Minkowski background (e.g., by taking the Schrödinger
equation in curved space [3]). So, how far can we maintain the view that gravity is
classical? Are there arguments against that?
According to the Einstein equations (8.1), a matter state jm; gi cannot be
the linear combination of matter eigenstates jmi ; gi i describing a mass mi in a
classical metric gi , because there is only one spacetime for all states. Therefore,
the superposition principle for matter wave-functions does not hold. In doing so, we
face a problem concerning the theory of measurement. Imagine a matter state jm; gi
such that the total mass of matter is either m1 or m2 , with probability 1=2. Write the
linear superposition
jm1 ; g1 i C jm2 ; g2 i
jm; gi D p :
2
If the wave-function jm; gi collapsed during the measurement of the gravitational

field, as in the Copenhagen interpretation, an experiment would measure either m1 or
m2 and the right-hand side of the Einstein equations (8.1) would not be covariantly
8.1 Do We Need to Quantize Gravity? 391
conserved [4].1 If, instead, we adopted the Everett interpretation and assumed that
jm; gi does not collapse, then (8.1) would predict that any measurement on matter
gives a mass equal to the average .m1 C m2 /=2. This is not the case experimentally
[6], thus favouring the hypothesis that (8.1) is an approximation and gravity is
intrinsically quantum. The result is not conclusive because one can avoid the
above problems by devising more complicated ways to couple classical gravity
with quantum matter fields, for instance by changing the representation of matter
quantum operators and the Hilbert space of the theory [7, 8]. In general, due to
back-reaction of the gravitational field, quantum mechanics governing particles on
a classical curved background is non-linear. This can be subject to experimental
tests [9].
A theoretical objection against classical gravity is the following [10].2 Consider
an action SŒg; dependent on the metric and some matter fields generically denoted
as . Consider also a suitably regular field redefinition
gN D gN .g ; / ; N D N .g ; / ;
so that the action
N g; N D SŒg;
SŒN
describes the same physics as S at the classical level. At the quantum level, if
all fields (including g ) are quantized the two theories are equivalent on shell
(i.e., after using the equations of motion) order by order in perturbation theory,
although they differ as far as individual Feynman diagrams and off-shell S-matrix
elements are concerned. This is because the on-shell S-matrix is invariant under field
redefinitions. However, if gravity is purely classical only matter fields are quantized
and the two theories are physically inequivalent. The intuitive reason is that internal
graviton lines, which are essential to maintain the on-shell equivalence, are now
absent. An example is the minimally-coupled massless scalar field theory
Z
p R 1
SŒg; D d4 x g g @ @ : (8.2)
2 2 2
1
Another Gedankenexperiment along similar lines is given in [5].
2
Much of the criticism of [10] was actually directed towards quantum field theories in curved space
even as approximated models, but one can take the milder view that (8.1) has a limited validity,
depending on the circumstances [4, 11].
At the one-loop level, UV divergences are removed if one adds a certain counter-
term S [12]. The classical theory (8.2) is equivalent to the non-minimal action
Z N2 1
p 1
N g; N D
SŒN d4 x Ng RN N
g
@ N @ N (8.3)
2 2 12 2
via a conformal transformation (Problem 8.1). One-loop finiteness of this theory

requires a counter-term S. N When graviton internal lines are taken into account, on
N
shell we have S D S. However, when only the scalar field is quantized one finds
that S ¤ SN [10].
The conclusion is that the same classical theory could be written in infinitely
many different ways and one would have to invoke a criterion selecting one frame
among all the others. This may be problematic, but the existence of such a criterion
is not altogether unreasonable: for instance, one could impose positivity of energy
and choose the Einstein frame as the frame where the fundamental theory is defined
[13].
Other objections have been moved to rule out classical gravity but, after careful
scrutiny, to date none of them has been found to be strong enough. Classical gravity
can be made plausible under certain interpretations and applications of quantum
mechanics and, ultimately, the issue lies with experimental evidence [14]. The
beautiful economy of thought of unification theories is very appealing, also because
it has been instrumental in the discovery of new physics in the past. Hereafter the
working hypothesis is maintained that gravity is quantum, but one should keep an
open mind about other possibilities as in Sect. 7.7.
8.2 Perturbative Quantum Gravity
The most natural attempt to unify the gravitational interaction and particle physics
consists in regarding gravity as a particle field and quantizing it. This can be done
either in canonical formalism [15] or via the perturbative background-field method
in covariant quantization [16–18]. In the second case, one treats the graviton as a
small perturbation on a background,
g D gQ C h : (8.4)
The background is typically chosen as Minkowski, gQ D . The squared Planck

length plays the role of coupling constant, 2 / G lD2 Pl
. Its engineering
dimension (2.19), however, is positive for spacetime dimensionality larger than or
equal to 3 and the ensuing perturbation theory is not renormalizable, meaning that
the S-matrix diverges at a given finite order.
8.2 Perturbative Quantum Gravity 393
A standard power-counting argument is sufficient to understand qualitatively the

relation between coupling dimensionality and renormalization properties of a theory
[19, 20]. Consider the classical action
Z X
SD dD xL ; LD ˛d Od ; (8.5)
M d>0
where ˛d and Od are, respectively bare coupling constants and local operators with
scaling dimension
Œ˛d D D d ; ŒOd D d : (8.6)
Here D is the topological dimension of the manifold M whereon the field theory
lives. More generally, it is the scaling dimension of the measure defining the action.
The expressions (8.6) are imposed so that S is dimensionless in „ D 1 units.
Operators are classified according to their scaling dimension:
• An operator Od is relevant if d < D. Then Œ˛d > 0.
• An operator Od is marginal if d D D. Then Œ˛d D 0.
• An operator Od is irrelevant if d > D. Then Œ˛d < 0.
For example, in the scalar field theory given by (2.59), (2.60), (2.61) and (2.62), in
four dimensions the operators n with n 6 3 are relevant, the kinetic and quartic
terms are marginal, while the operators n with n > 5 are irrelevant.
When constructing perturbation theory, one must take into account all possible
gauge-inequivalent interactions ˛Q d OQ d order by order in the effective low-energy
action. Some couplings diverge (˛Q d ! 1) when the regulator in the given
regularization scheme is removed. However, if the operator OQ d associated with one
of these couplings is already present at the tree level (OQ d D Od for some d), one can
absorb the divergence into an effective coupling which is defined to be finite when
the regulator is removed:
˛d Od C ˛Q d Od D: ˛deff Od ; j˛deff j < 1 :
Contrary to the bare coupling ˛d , the effective coupling is what one measures
physically. If this procedure works order by order, the theory is said to be
perturbatively renormalizable and, hence, physically predictive. This means that
the number of physical couplings we must measure at any perturbative order is
finite. In renormalizable theories, high-momentum modes only shift the bare values
of the couplings and high-energy effects (i.e., of heavy particles with mass above
the energy cut-off) are under complete control [21]. This result is known as the
decoupling theorem.
The way operators enter the effective action is governed by the dimension of
the bare couplings. In the renormalization group picture, the physical action stems
from the bare action by integrating out momentum modes greater than a certain
energy cut-off scale M, and then removing the cut-off. In terms of the dimensionless
constants
d D ˛d M dD ; Œd D 0 ; (8.7)
the operators scale as

Z p dD
dD x ˛d Od d ; (8.8)
M
where p is the proper (physical) momentum. Therefore:

• Relevant operators are important at low energies (p=M 1).
• Marginal operators are equally important at all scales. Their detailed behaviour is
not obvious and they can be, case by case, either marginally relevant or irrelevant.
• Irrelevant operators become important in the ultraviolet (p=M 1) but, contrary
to what the name suggests, some of them can also alter macroscopic physics. If
they are important also for the IR physics, they are called dangerous.
Since there is a finite number of relevant operators and, typically, also of marginal
operators, macroscopic physics is described only by few observables.
If divergences are present, they correspond to local operators of dimensionality
increasing with the order of the perturbative expansion. Suppose the bare action
S contains only relevant operators; then, only a finite number of relevant oper-
ators (those which do not appear in S) will enter the effective action and any
divergence will be absorbed in the finite number of couplings f˛d g. For instance,
in electromagnetism the electron mass and charge have non-negative dimension
in natural units and one can formally absorb the divergences just in these two
coupling constants, which are then determined by experiments. Conversely, if even
one irrelevant operator appears in S one can construct new irrelevant operators at
each order. Explicit calculations can determine whether their couplings are finite or
not. If they diverge, the perturbative approach loses predictive power because we
can absorb all the divergences only by adding an infinite number of operators to the
action.
A theory is said to be power-counting renormalizable if
Œ˛d > 0 (8.9)
for all couplings ˛d . This condition is not sufficient to guarantee that the theory be
renormalizable in the sense of the full renormalization group flow, but it provides a
good guiding principle in many situations. If a model is not power-counting renor-
malizable, then it will likely be non-renormalizable unless remarkable divergence
cancellations happen. In the case of gravity, these cancellations do happen [22] and
explicit calculations are necessary to settle the issue.
The condition that a theory have a good UV behaviour in the absence of irrelevant
operators can be understood by looking at the superficial degree of divergence
of a Feynman diagram. Consider a one-particle-irreducible (1PI) Feynman sub-
graph with L loops, I internal propagators and V vertices. The superficial degree
of divergence ı is the canonical dimension of all these contributions: given a UV
energy cut-off M, the divergent part of the diagram scales as M ı . If ı D 0, one
has at most superficial logarithmic divergences and the theory is power-counting
renormalizable. When ı < 0 for every sub-diagram in a Feynman graph, the graph
is convergent; if only a finite number of Feynman diagrams diverge superficially,
the theory is power-counting super-renormalizable.
We can count divergences in the case of the scalar field theory (2.59), (2.60),
(2.61) and (2.62). Each loop integral over momenta gives ŒdD p D D, while
the propagator G.Q p2 / D 1=. p2 C m2 / has ŒG. Q p2 / D 2. For the scalar
field theory (2.59), (2.60), (2.61) and (2.62), interaction vertices do not carry
dimensionality and, overall, ı D DL 2I. Since I > L, the maximum superficial
degree of divergence can be L.D 2/. L is the number of independent momenta,
given by I minus the number of relations they satisfy among themselves: these are
V 1 (one for each vertex, up to the total momentum conservation), so that
L D I V C1: (8.10)
This result is often called Euler’s theorem for graphs. With only mass and a n n
interaction, for each vertex there are n lines and we get nV D N C 2I, where N is the
number of external legs in the diagram. Replacing L and I with these expressions,
one obtains

D
ı D DL 2I D D Œn V 1 N; (8.11)
2
where we used (2.62). This formula can be derived also by dimensional arguments
(Problem 8.2).
If N is the maximum power in (2.61), the superficial degree of divergence is
ı D ŒN .1 V/. For the theory to be power-counting renormalizable, it must be
ŒN > 0, implying
2D
N6 :
D2
In two dimensions, ı does not depend on the number of external legs (N is

unconstrained) and, the greater the number of vertices, the more convergent the
diagram. In four dimensions, the 4 theory is renormalizable while higher powers
of are responsible for an infinite number of divergent diagrams. In general, ı is
bounded by the dimension of operators which already appear in the bare action.
The coupling constant governing the perturbative expansion of graviton field
theory is l2Pl . This is not immediately obvious from the Einstein–Hilbert action, but
we already have almost all the ingredients sufficient to understand this result. Let us
revert to units where Œh D 0. As we have seen in Chaps. 3 and 5, the quadratic
kinetic term of the action (3.24)
p for h is canonical if one defines the graviton
as the field (5.138), u h= G mPl h, with scaling dimension Œu D 1 (the
polarization indices are omitted here). The next-to-leading terms are irrelevant
operators. In fact, borrowing (9.65) and (9.66) from Chap. 9, one can see that the
classical Lagrangian for u contains not only the kinetic term K 2 uP 2 but also
interactions of the form
1 1
h.@h/2 l2Pl u.@u/2 ; .h@h/2 l2Pl .u@u/2 ; (8.12)
G G
where @ is a spatial derivative. Contrary to the polynomial scalar field theory of the
previous example, also vertices contribute to the superficial degree of divergence,
each with a factor of 2. In particular, for any 1PI Feynman diagram there can be up
to 2V extra factors of internal momenta, so that ı 6 DL C 2.V I/ D .D 2/L C 2.
Apart from this difference, the power-counting argument applies also to gravity,
which is not renormalizable due to the presence of the dangerously irrelevant
operators (8.12). Intuitively, the cause why these operators are dangerous despite
the small value of the Planck length is the strong equivalence principle: the coupling
strength with gravity is the same for all forms of matter and energy, including gravity
itself. As soon as a curved background notably affects a matter quantum field theory,
it does so also in the gravitational sector and at all spacetime scales.
Looking more in detail at the perturbative diagrammatic expansion, it turns out
that, in the absence of supersymmetry, the ultraviolet properties of the theory are
in agreement with the power-counting argument. Pure vacuum Einstein gravity in
four dimensions is perturbatively non-renormalizable at two loops, meaning that
its S-matrix diverges at that order on shell [23–25]. This is due to the presence
of dimension-six operators made of three Riemann invariants. In vacuum, the only
non-vanishing contribution is

OQ 6 D R R R
; ˛Q 6 ! 1 : (8.13)
Inclusion of matter does not improve the scenario. On the contrary, divergences
appear already at the one-loop level in the presence of a massive scalar field [12],3
Dirac fermions [26] and Yang–Mills fields [27].
Stelle’s higher-order polynomial theory (7.79), on the other hand, is renormaliz-
able. An elementary power counting shows that the maximal superficial degree of
divergence of a Feynman graph is ı D D .D X/.V I/ D D C .D X/.L 1/.
For X D D, the theory is renormalizable since the maximal divergence is ı D D.
Then, all the infinities can be absorbed in the operators already present in the
Lagrangian (7.79). Unfortunately, as we saw in Sect. 7.5.1, the propagator contains
at least one ghost (i.e., a state of negative norm) that marks a violation of unitarity.
3
A non-minimal coupling between gravity and the scalar can cancel some of the divergences but
not all [12].
8.2.1 Supergravity
In Sect. 5.12, we briefly introduced N D 1 supergravity models in four dimensions.

Generalization to N supercharges is also possible, in which case (5.207) becomes
"
fQAa ; Qb B g D 2. /ab P ıBA and fQAa ; QBb g D ab X AB, where A D 1; : : : ; N , ab is
the two-index Levi-Civita symbol and the scalars X AB are called central charges. We
do not enter into details and only quote some results obtained in SUGRA scenarios.
The minimal D D 4, N D 1 supergravity model is non-renormalizable at three
(respectively, one) loops in the absence (presence) of matter [28]. N D 8 extended
supergravity fares better. In fact, in four dimensions and in vacuum this theory is
finite at the four-loop level and promising evidence justifies the conjecture that
N D 8 SUGRA be perturbatively finite [29–37], although there is disagreement
on such an optimistic view [38] and the theory may have a divergence at seven
loops [39–41]. In any case, the improved UV behaviour of the theory is due both
to supersymmetry and to a class of diagrammatic cancellations typical of gravity
which lower its degree of divergence [22]. These cancellations are not accounted for
by traditional power-counting arguments and can be made manifest only in certain
representations of the quantum amplitudes.
8.2.2 Effective Field Theory
Quantum fluctuations involve all energy scales. For this reason, loop corrections
to tree-level interactions carry information on all such scales, as they involve
integration on all momenta p. This information is confined to the shifting of a finite
number of bare parameters in renormalizable theories, while in non-renormalizable
ones high-energy scales affect infinitely many parameters and, consequently, most
types of experimental measurements. Such is the case of gravity, as we saw in
this chapter. However, it is possible to treat it as a consistent effective field theory
(e.g., [42]) even when the usual perturbative expansions are non-predictive [43–
47] (see [48, 49] for reviews). To achieve this result, one reorganizes the theory
by separating the known low-energy quantum effects from the unknown high-
energy contributions via a momentum expansion (i.e., grouping terms with the same
number of derivatives). The graviton is nothing but the residual low-energy degrees
of freedom of the effective theory.
Under this perspective, the series “R2 C R3 C R4 C : : : ” of Riemann curvature
operators is regarded as an expansion in a low-curvature, low-energy regime. By
definition, higher-order operators are sub-dominant at such scales and the series
can be truncated at any given order without the necessity to know the form of the
full expansion (if any) or to worry about ghosts (which would be artifacts of the
truncation). For instance, the quartic Lagrangian (7.77) would be interpreted as an
effective quantum description of gravity valid at scales much larger than the Planck
scale, where the parameters ˛, ˇ and are all O.l2Pl / D O.„/.
A quantum field theory of gravity is thus obtained at low curvature and low
energies, where classical dynamics is modified by well-defined and well-behaved
quantum corrections. A fully calculable example is the one-loop correction to the
Newtonian potential between two masses m1;2 at distance r [45, 50]:

Gm1 m2 G.m1 C m2 / 41 l2Pl
˚.r/ ' 1C3 C ; (8.14)
r r 10 r2
where the middle term [51–53] is the classical relativistic correction. The O.r3 /
term is expected to arise in the low-curvature, low-energy limit of quantum gravity,
whatever the form of the theory at high energies.
For consistency, solutions to higher-order equations of motion should be trun-
cated at lowest order in the couplings (˛, ˇ and in the example (7.77)), while
solutions which are non-perturbative in such couplings must be discarded [54, 55].
When applied to the models of Sect. 7.5, this prescription virtually excludes most
of their interesting applications to cosmology, since corrections to the Einstein–
Hilbert dynamics are negligible at inflationary as well as dark-energy scales. Such
corrections scale as .=Pl /n , with n > 0 and Pl (see (10.69) and below
(11.101)). Therefore, the problem of quantum gravity is not whether gravity admits
a consistent quantum description (it does, at an effective level) but to understand the
deep UV structure of the theory. Getting access to the high-curvature, high-energy
limit would also pave the way to phenomenology beyond the standard cold big-bang
CDM model and, hopefully, to the resolution of its problems.
8.2.3 Resummed Quantum Gravity
An approach complementary to effective field theory is resummed quantum gravity

[56–62]. The latter is based on the Feynman expansion of perturbative quantum
gravity (where the graviton h is in the decomposition (8.4) of the metric)
with matter fields [16] and resums all the propagators á la Dyson [63] in the
loop expansion of the theory. For instance, for a massive scalar particle in four
dimensions the propagator is modified as
00
Bg . p/ 2
Q p2 / D 1 N p2 / D e 2 p2 p
G. ! G. ; B00g . p/ D ln C 1 ;
p 2 C m2 p 2 C m2 8 2 m2
(8.15)
and loop diagrams of the graviton self-energy with matter loops feature G. N p2 /
Q 2
instead of the bare scalar propagator G. p /. Remarkably [58, 60], the inclusion of all
radiative corrections at all orders and the rearrangement of the IR divergences tame
the UV divergences that appear in the finite-loop calculation where resummation is
not performed [12] and counter-terms are not added. In other words, after the IR
resummation, an eventual UV regulator can be removed and the UV regime remains

finite order by order in the loop expansion.
In this approach, the graviton propagator with matter loops is strongly damped in
the UV. At one loop in the resummed series and to leading order in the momentum
p, the gauge-invariant graviton propagator acquires a correction of the form
1 1
2
! 2 ; (8.16)
p p C p4 = 2
where depends on the masses of the matter fields [56]. Without resummation,
would be infinity and the one-loop propagator would diverge [12, 57] but, for the
Standard-Model masses, 0:21mPl [60]. When Fourier transforming (8.16), the
corresponding correction to Newton’s potential is [56]
Gm1 m2
˚.r/ ' .1 e r / : (8.17)
r
Note the difference with respect to the result (8.14). In the effective-field-theory
case, the final expression for Newton’s potential is valid at large distances r D
O.1/r=lPl 1, i.e., at low energies (momenta lower than the Planck mass); the
corrective IR term in (8.14) is perturbative in 1=r. On the other hand, the resummed-
gravity approach is focussed on the UV regime of the theory and (8.17) is the
one-loop result valid at all distances r > 1= , even near the Planck length. The
relationship between the effective-field-theory and the resummed-gravity approxi-
mation is similar to that between, respectively, the use of chiral perturbation theory
for soft strong interactions at large distance and the use of perturbative quantum
chromodynamics for hard strong interactions [58]. Both approaches to perturbative
quantum gravity recover the classical potential for r lPl and they should agree at
some intermediate scale.
An interesting property of (8.17) is that Newton’s potential vanished in the UV
and, apparently, the gravitational interaction becomes weakly couple. This stems
from the one-loop propagator (8.16) and the associated effective Newton’s coupling
G0
Geff . p/ D : (8.18)
1 C p2 = 2
This is no longer a constant in the momentum and its value can change with the
probed energy scale. We will come back to this important point in Sect. 11.2.1.
8.3 Approaches to Quantum Gravity
The lack of a UV-finite perturbative formulation of quantum gravity, the non-

conclusive nature of finiteness proofs in the N D 8 supersymmetric extension
(together with the fact that we have not observed supersymmetry yet), the limi-
tations of the effective-field-theory approach and some criticism [64, 65] on the
decomposition (8.4) are all motivations for turning our attention to other proposals
related to quantum gravity.4 We can roughly divide them in three categories.
In the first, gravity is on a completely different status with respect to matter
fields. A conservative but not very satisfactory point of view is that it may be
just a classical force which does not admit a quantum formulation, contrary to the
other interactions. A somewhat more rewarding possibility is that gravity is not a
fundamental interaction but an emergent phenomenon, as detailed in Sect. 7.7.
The second class of frameworks is constituted by the “theories of everything,”
which aim to describe both gravity and matter in a unified, consistent framework.
Among the theories of everything, we count:
• Supergravity, where all fields are components of a super-multiplet.
• String theory [67–69], which we will meet in Chap. 12. The fundamental objects
are strings, propagating in a D-dimensional target spacetime. Different vibration
modes of a string correspond to particles; gravity emerges from the modes of
closed strings. The end-points of open strings live on lower-dimensional extended
objects called Dp-branes, whose dynamics is highly non-perturbative. There
exist several formulations of the theory, most of which supersymmetric. Self-
consistency requirements impose the number of spatial dimensions to be larger
than three: D1 D 25 in the bosonic string and D1 D 9 in the supersymmetric
versions. The low-energy limit of the supersymmetric string is supergravity.
Non-perturbative dualities between all these theories suggest that they are just
different aspects of a bigger and yet elusive framework, exotically called M-
theory (Sect. 12.4).
• Group field theory [70–77] (Sect. 11.5). Field theories on group manifolds are
higher-dimensional generalizations of matrix models, characterized by Feyn-
man diagrams having the combinatorial structure of D-dimensional simplicial
complexes. The Feynman amplitudes of these theories are spin-foam models,
based on the same simplicial complexes, so that group field theories potentially
represent a truly unified framework for both loop quantum gravity and simplicial
approaches. The interpretation is that group field theories define the covariant
dynamics of quantum spacetime as a sum over Feynman diagrams, interpreted
as discrete spacetimes of arbitrary topology, each weighted by a purely algebraic
and discrete version of a quantum-gravity path integral.
4
For a nice historical overview of ideas in quantum gravity, we refer the reader to [66].
8.3 Approaches to Quantum Gravity 401
All these theories have several problems. On one hand, they include a number
of ingredients which are phenomenologically unnecessary (e.g., supersymmetry,
large or infinite particle spectra, extra spatial dimensions) or yet unclear (e.g.,
individuation of physical observables, definition of semi-classical and continuum
limits). These ingredients can be accommodated to fit with experiments but the task
is challenging. Both the study of the theoretical structure of these theories and their
predictive power advance rather slowly despite the concentrated efforts by a large
part of the scientific community.5
On the other hand, it is difficult to devise experiments unravelling a funda-
mentally quantum structure of the gravitational force. This is due to an inevitable
combination of technological limitations and the way we are presently asking the
questions. The third group of theories, which we introduce now, is no exception
[72]. There, unification is abandoned6 in favour of tackling the less ambitious but
still difficult problem of how to quantize the gravitational sector alone. The way
general relativity is fused with quantum mechanics notably differs from the theories
of everything, mainly because greater emphasis is placed on diffeomorphisms (in
string theory, these symmetries are not fundamental and stem from a much larger
gauge group). To this category there belong:
• Loop quantum gravity (LQG), where gravity is quantized in canonical formalism
and quanta of space are described by states called spin networks [64, 78–80].
The spectra of area and volume operators are discrete and bounded from below.
A path-integral evolution of spin networks is spin foams [81–84] (Sect. 11.4).
• Asymptotic safety, also known as quantum Einstein gravity [85–90], a non-
perturbative renormalization group approach realizing asymptotic safety in a
traditional field-theory setting (Sect. 11.2).
• Causal dynamical triangulations (CDT) [91–93], a Lorentzian path-integral
formulation of quantum gravity where the integral is performed over piecewise
flat 4-geometries (Sect. 11.3). This is an example of approach where spacetime
geometry is fundamentally continuous but discretized on a lattice to regularize
infinities; the limit of zero lattice spacing is eventually taken.
• Causal sets [94–99], where the texture of spacetime is fundamentally discrete
(Sect. 11.6).
Other approaches (non-commutative spacetimes, non-local gravity and models with
dimensional flow) will be discussed in Sects. 11.7, 11.8 and 11.9.
Contrary to supergravity and string theory, in group field theory and the rest of
the approaches the notion of “quantizing gravity” is replaced by that of “quantizing
spacetime.” In the first case, (8.4) singles out a particular background whereupon
5
To give two instances, it has become apparent that string theory is not a model of Nature but,
rather, a framework wherein to construct such a model, just in the same way “quantum field theory”
stands to the Standard Model of particles. Group field theory is much less developed than any of
the other models and its status as a viable theory of everything is unclear.
6
Notice, however, that there exist arguments against non-unified frameworks of quantum-gravity
(Sect. 13.2.2).
the perturbation h is the graviton particle. On the other hand, in any of the above
non-perturbative theories the decomposition (8.4) is somewhat artificial and the
fundamental objects are quanta of spacetime itself rather than particle quanta.
8.1 Conformal transformations. Find the transformation mapping (8.2)

into (8.3).
Solution From Problems 7.3, 7.4 and 7.5, it is easy to see that
r r !
2 2 1 ND 6 2
gN D ˝ g ; ˝ D p ; tanh :
cosh 2
2 =6 2 6
8.2 Power-counting renormalizability. Derive (8.11) with a dimensional

argument.
Solution A diagram with N external lines can be generated by a N N term, so

that its scaling dimension is ŒN . On the other hand, with only the n interaction
term available the divergent part of the diagram scales as nV M ı , where M is the UV
cut-off. Therefore, we have ŒN D Œn V C ı and (8.11) is recovered.
References
1. C. Møller, The energy-momentum complex in general relativity and related problems, in Les
Théories Relativistes de la Gravitation, ed. by A. Lichnerowicz, M.A. Tonnelat (CNRS, Paris,
1962)
2. L. Rosenfeld, On quantization of fields. Nucl. Phys. 40, 353 (1963)
3. L. Parker, Path integrals for a particle in curved space. Phys. Rev. D 19, 438 (1979)
4. L.H. Ford, Gravitational radiation by quantum systems. Ann. Phys. (N.Y.) 144, 238 (1982)
5. R.M. Wald, General Relativity (The University of Chicago Press, Chicago, 1984)
6. D.N. Page, C.D. Geilker, Indirect evidence for quantum gravity. Phys. Rev. Lett. 47, 979 (1981)
7. B. Mielnik, Generalized quantum mechanics. Commun. Math. Phys. 37, 221 (1974)
8. T.W.B. Kibble, S. Randjbar-Daemi, Nonlinear coupling of quantum theory and classical
gravity. J. Phys. A 13, 141 (1980)
References 403
9. S. Carlip, Is quantum gravity necessary? Class. Quantum Grav. 25, 154010 (2008).
[arXiv:0803.3456]
10. M.J. Duff, Inconsistency of quantum field theory in curved space-time, in Quantum Gravity 2,
ed. by C.J. Isham, R. Penrose, D.W. Sciama (Oxford University Press, Oxford, 1981)
12. G. ’t Hooft, M.J.G. Veltman, One loop divergencies in the theory of gravitation. Ann. Poincaré
Phys. Theor. A 20, 69 (1974)
13. V. Faraoni, E. Gunzig, P. Nardone, Conformal transformations in classical gravitational
theories and in cosmology. Fund. Cosm. Phys. 20, 121 (1999). [arXiv:gr-qc/9811047]
14. S. Boughn, Nonquantum gravity. Found. Phys. 39, 331 (2009). [arXiv:0809.4218]
15. B.S. DeWitt, Quantum theory of gravity. I. The canonical theory. Phys. Rev. 160, 1113 (1967)
16. R.P. Feynman, Quantum theory of gravitation. Acta Phys. Polon. 24, 697 (1963) [Reprinted
in 100 Years of Gravity and Accelerated Frames, ed. by J.-P. Hsu, D. Fine (World Scientific,
Singapore, 2005)]
17. B.S. DeWitt, Quantum theory of gravity. II. The manifestly covariant theory. Phys. Rev. 162,
1195 (1967)
18. B.S. DeWitt, Quantum theory of gravity. III. Applications of the covariant theory. Phys. Rev.
162, 1239 (1967)
19. P. Ramond, Field Theory: A Modern Primer (Westview Press, Boulder, 1997)
20. S. Weinberg, The Quantum Theory of Fields, vol. I (Cambridge University Press, Cambridge,
1995)
21. T. Appelquist, J. Carazzone, Infrared singularities and massive fields. Phys. Rev. D 11, 2856
(1975)
22. Z. Bern, J.J. Carrasco, D. Forde, H. Ita, H. Johansson, Unexpected cancellations in gravity
23. M.H. Goroff, A. Sagnotti, Quantum gravity at two loops. Phys. Lett. B 160, 81 (1985)
24. M.H. Goroff, A. Sagnotti, The ultraviolet behavior of Einstein gravity. Nucl. Phys. B 266, 709
(1986)
25. A.E.M. van de Ven, Two loop quantum gravity. Nucl. Phys. B 378, 309 (1992)
26. S. Deser, P. van Nieuwenhuizen, Nonrenormalizability of the quantized Dirac–Einstein system.
Phys. Rev. D 10, 411 (1974)
27. S. Deser, H.-S. Tsao, P. van Nieuwenhuizen, One-loop divergences of the Einstein–Yang–Mills
system. Phys. Rev. D 10, 3337 (1974)
28. S. Deser, J.H. Kay, K.S. Stelle, Renormalizability properties of supergravity. Phys. Rev. Lett.
38, 527 (1977)
29. Z. Bern, L.J. Dixon, R. Roiban, Is N D 8 supergravity ultraviolet finite? Phys. Lett. B 644, 265
(2007). [arXiv:hep-th/0611086]
30. Z. Bern, J.J. Carrasco, L.J. Dixon, H. Johansson, D.A. Kosower, R. Roiban, Three-
loop superfiniteness of N D 8 supergravity. Phys. Rev. Lett. 98, 161303 (2007).
31. Z. Bern, J.J.M. Carrasco, L.J. Dixon, H. Johansson, R. Roiban, Manifest ultraviolet behavior
for the three-loop four-point amplitude of N D 8 supergravity. Phys. Rev. D 78, 105019 (2008).
[arXiv:0808.4112]
32. N. Arkani-Hamed, F. Cachazo, J. Kaplan, What is the simplest quantum field theory? JHEP
1009, 016 (2010). [arXiv:0808.1446]
33. Z. Bern, J.J. Carrasco, L.J. Dixon, H. Johansson, R. Roiban, The ultraviolet behavior of N D 8
supergravity at four loops. Phys. Rev. Lett. 103, 081301 (2009). [arXiv:0905.2326]
34. R. Kallosh, The ultraviolet finiteness of N D 8 supergravity. JHEP 1012, 009 (2010).
[arXiv:1009.1135]
35. N. Beisert, H. Elvang, D.Z. Freedman, M. Kiermaier, A. Morales, S. Stieberger, E7.7/
constraints on counterterms in N D 8 supergravity. Phys. Lett. B 694, 265 (2010).
[arXiv:1009.1643]
36. R. Kallosh, E7.7/ symmetry and finiteness of N D 8 supergravity. JHEP 1203, 083 (2012).
[arXiv:1103.4115]
37. R. Kallosh, N D 8 counterterms and E7.7/ current conservation. JHEP 1106, 073 (2011).
[arXiv:1104.5480]
38. T. Banks, Arguments against a finite N D 8 supergravity. arXiv:1205.5768
39. M.B. Green, J.G. Russo, P. Vanhove, String theory dualities and supergravity divergences.
JHEP 1006, 075 (2010). [arXiv:1002.3805]
40. J. Björnsson, M.B. Green, 5 loops in 24=5 dimensions. JHEP 1008, 132 (2010).
[arXiv:1004.2692]
41. J. Björnsson, Multi-loop amplitudes in maximally supersymmetric pure spinor field theory.
JHEP 1101, 002 (2011). [arXiv:1009.5906]
42. S. Weinberg, Effective field theory, past and future. Proc. Sci. CD09, 001 (2009).
[arXiv:0908.1964]
43. J.F. Donoghue, Leading quantum correction to the Newtonian potential. Phys. Rev. Lett. 72,
2996 (1994). [arXiv:gr-qc/9310024]
44. J.F. Donoghue, General relativity as an effective field theory: the leading quantum corrections.
45. N.E.J. Bjerrum-Bohr, J.F. Donoghue, B.R. Holstein, Quantum gravitational corrections to
the nonrelativistic scattering potential of two masses. Phys. Rev. D 67, 084033 (2003);
Erratum-ibid. D 71, 069903(E) (2005). [arXiv:hep-th/0211072]
46. N.E.J. Bjerrum-Bohr, J.F. Donoghue, B.R. Holstein, Quantum corrections to the Schwarzschild
and Kerr metrics. Phys. Rev. D 68, 084005 (2003); Erratum-ibid. D 71, 069904(E) (2005).
47. N.E.J. Bjerrum-Bohr, J.F. Donoghue, P. Vanhove, On-shell techniques and universal results in
quantum gravity. JHEP 1402, 111 (2014). [arXiv:1309.0804]
48. J.F. Donoghue, The effective field theory treatment of quantum gravity. AIP Conf. Proc. 1483,
73 (2012). [arXiv:1209.3511]
49. J.F. Donoghue, B.R. Holstein, Low energy theorems of quantum gravity from effective field
theory. J. Phys. G 42, 103102 (2015). [arXiv:1506.00946]
50. I.B. Khriplovich, G.G. Kirilin, Quantum long-range interactions in general relativity. Zh. Eksp.
Teor. Fiz. 125, 1219 (2004) [JETP 98, 1063 (2004)]. [arXiv:gr-qc/0402018]
51. A. Einstein, L. Infeld, B. Hoffmann, The gravitational equations and the problem of motion.
Ann. Math. 39, 65 (1938)
52. A. Eddington, G. Clark, The problem of n bodies in general relativity theory. Proc. R. Soc.
Lond. A 166, 465 (1938)
53. Y. Iwasaki, Quantum theory of gravitation vs. classical theory. Fourth-order potential. Prog.
Theor. Phys. 46, 1587 (1971)
54. J.Z. Simon, Higher-derivative Lagrangians, nonlocality, problems, and solutions. Phys. Rev. D
41, 3720 (1990)
55. J.Z. Simon, Stability of flat space, semiclassical gravity, and higher derivatives. Phys. Rev. D
43, 3308 (1991)
56. B.F.L. Ward, Quantum corrections to Newton’s law. Mod. Phys. Lett. A 17, 2371 (2002).
57. B.F.L. Ward, Are massive elementary particles black holes? Mod. Phys. Lett. A 19, 143 (2004).
58. B.F.L. Ward, Massive elementary particles and black holes. JCAP 0402, 011 (2004).
59. B.F.L. Ward, Planck scale cosmology in resummed quantum gravity. Mod. Phys. Lett. A 23,
3299 (2008). [arXiv:0808.3124]
60. B.F.L. Ward, An estimate of in resummed quantum gravity in the context of asymptotic
safety. Phys. Dark Univ. 2, 97 (2013)
61. B.F.L. Ward, Running of the cosmological constant and estimate of its value in quantum
general relativity. Mod. Phys. Lett. A 30, 1540030 (2015). [arXiv:1412.7417]
References 405
62. B.F.L. Ward, Einstein–Heisenberg consistency condition interplay with cosmological con-
stant prediction in resummed quantum gravity. Mod. Phys. Lett. A 30, 1550206 (2015).
[arXiv:1507.00661]
63. D.R. Yennie, S.C. Frautschi, H. Suura, The infrared divergence phenomena and high-energy
processes. Ann. Phys. (N.Y.) 13, 379 (1961)
65. T.P. Singh, T. Padmanabhan, Notes on semiclassical gravity. Ann. Phys. (N.Y.) 196, 296 (1989)
66. C. Rovelli, Notes for a brief history of quantum gravity, in Recent Developments in
Theoretical and Experimental General Relativity, Gravitation and Relativistic Field Theo-
ries, ed. by V.G. Gurzadyan, R.T. Jantzen, R. Ruffini (World Scientific, Singapore, 2002).
68. B. Zwiebach, A First Course in String Theory (Cambridge University Press, Cambridge, 2009)
70. L. Freidel, Group field theory: an overview. Int. J. Theor. Phys. 44, 1769 (2005).
71. D. Oriti, The group field theory approach to quantum gravity, in [72]. [arXiv:gr-qc/0607032]
72. D. Oriti (ed.), Approaches to Quantum Gravity (Cambridge University Press, Cambridge,
2009)
73. D. Oriti, The microscopic dynamics of quantum space as a group field theory, in [74].
[arXiv:1110.5606]
74. G.F.R. Ellis, J. Murugan, A. Weltman (eds.), Foundations of Space and Time (Cambridge
75. D. Oriti, Group field theory as the second quantization of loop quantum gravity. Class.
76. A. Baratin, D. Oriti, Ten questions on group field theory (and their tentative answers). J. Phys.
Conf. Ser. 360, 012002 (2012). [arXiv:1112.3270]
77. S. Gielen, L. Sindoni, Quantum cosmology from group field theory condensates: a review.
SIGMA 12, 082 (2016). [arXiv:1602.08104]
arXiv:gr-qc/0110034
79. T. Thiemann, Quantum gravity: from theory to experimental search. Lect. Notes Phys. 631,
412003 (2003). [arXiv:gr-qc/0210094]
80. A. Ashtekar, J. Lewandowski, Background independent quantum gravity: a status report. Class.
Quantum Grav. 21, R53 (2004). [arXiv:gr-qc/0404018]
81. D. Oriti, Spacetime geometry from algebra: spin foam models for non-perturbative quantum
gravity. Rep. Prog. Phys. 64, 1489 (2001). [arXiv:gr-qc/0106091]
82. A. Perez, Spin foam models for quantum gravity. Class. Quantum Grav. 20, R43 (2003).
83. C. Rovelli, A new look at loop quantum gravity. Class. Quantum Grav. 28, 114005 (2011).
[arXiv:1004.1780]
84. A. Perez, The spin-foam approach to quantum gravity. Living Rev. Relat. 16, 3 (2013)
85. M. Niedermaier, The asymptotic safety scenario in quantum gravity: an introduction, Class.
86. M. Niedermaier, M. Reuter, The asymptotic safety scenario in quantum gravity. Living Rev.
Relat. 9, 5 (2006)
87. M. Reuter, F. Saueressig, Functional renormalization group equations, asymptotic safety and
quantum Einstein gravity. arXiv:0708.1317
88. A. Codello, R. Percacci, C. Rahmede, Investigating the ultraviolet properties of gravity
with a Wilsonian renormalization group equation. Ann. Phys. (N.Y.) 324, 414 (2009).
[arXiv:0805.2909]
89. D.F. Litim, Renormalisation group and the Planck scale. Philos. Trans. R. Soc. Lond. A 369,
2759 (2011). [arXiv:1102.4624]
90. M. Reuter, F. Saueressig, Asymptotic safety, fractals, and cosmology. Lect. Notes Phys. 863,
185 (2013). [arXiv:1205.5431]
91. R. Loll, The emergence of spacetime, or, quantum gravity on your desktop. Class. Quantum
Grav. 25, 114006 (2008). [arXiv:0711.0273]
92. J. Ambjørn, J. Jurkiewicz, R. Loll, Causal dynamical triangulations and the quest for quantum
gravity, in [74]. [arXiv:1004.0352]
93. J. Ambjørn, A. Görlich, J. Jurkiewicz, R. Loll, Nonperturbative quantum gravity. Phys. Rep.
519, 127 (2012). [arXiv:1203.3591]
94. R.D. Sorkin, Spacetime and causal sets, in Relativity and Gravitation: Classical and Quantum,
ed. by J.C. D’Olivo, E. Nahmad-Achar, M. Rosenbaum, M.P. Ryan, L.F. Urrutia, F. Zertuche
(World Scientific, Singapore, 1991)
95. D.D. Reid, Introduction to causal sets: an alternate view of spacetime structure. Can. J. Phys.
79, 1 (2001). [arXiv:gr-qc/9909075]
96. J. Henson, The causal set approach to quantum gravity, in [72]. [arXiv:gr-qc/0601121]
97. J. Henson, Discovering the discrete universe. arXiv:1003.5890
98. S. Surya, Directions in causal set quantum gravity, in Recent Research in Quantum Gravity,
ed. by A. Dasgupta (Nova Science, Hauppauge, 2011). [arXiv:1103.6272]
99. F. Dowker, Introduction to causal sets and their phenomenology. Gen. Relat. Grav. 45, 1651
(2013)
Chapter 9
Canonical Quantum Gravity
Tirem-me daqui a metafísica!

— Fernando Pessoa, Lisbon Revisited (1923)
Spare me metaphysics!
Contents
9.1 Canonical Variables in General Relativity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 408
9.1.1 First-Order Formalism and Parity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 408
9.1.2 Hamiltonian Analysis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 414
9.1.3 Ashtekar–Barbero Variables. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 424
9.1.4 ADM Variables. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 426
9.2 Wheeler–DeWitt Equation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 428
9.2.1 Superspace and Quantization. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 429
9.2.2 Semi-classical States. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 432
9.2.3 Boundary Conditions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 437
9.3 Some Features of Loop Quantum Gravity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 439
9.4 Cosmological Constant Problem. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 442
9.4.1 Chern–Simons State. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 443
9.4.2 as a Condensate?. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 447
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 459
Letting aside temporarily the possibility to unify all forces in a single consistent
framework, a question one can ask oneself is whether quantum mechanics can
resolve a particular singularity, such as the big bang or that inside a black hole. With
this goal in mind, we examine an approach to gravity based on the Hamiltonian
formalism and its cosmological applications. Section 9.1 is devoted to classical
canonical gravity, where the so-called Schwinger, Ashtekar–Barbero and ADM
variables are defined. Quantization in ADM variables is discussed in Sect. 9.2,
where the Wheeler–DeWitt equation is introduced for the first time. This equation
will be the starting point for tackling the big-bang problem in Chap. 10.

DOI 10.1007/978-3-319-41127-9_9
408 9 Canonical Quantum Gravity
9.1 Canonical Variables in General Relativity
For the purpose of quantization, the Lagrangian formalism can be used in path-
integral approaches, but the Hamiltonian formalism is more transparent regarding
the choice of classical canonical variables (from which quantum operators are then
constructed). Historically, the first attempt to carry out the program of Hamiltonian
quantization was via the second-order Arnowitt–Deser–Misner (ADM) formalism
[1–3], but eventually it has become clear that first-order Ashtekar–Barbero variables
[4–6] are more convenient. We will present the Hamiltonian analysis in first-order
formalism, later sketching the results for the ADM decomposition. Throughout
this chapter, we specialize to four spacetime dimensions and do not include
supersymmetry (see [7–10] for supergravity Hamiltonian formalism, [11–16] for its
applications to quantum cosmology and Sect. 13.9.3 for supersymmetric canonical
quantum cosmology in string theory).
9.1.1 First-Order Formalism and Parity
The fundamental object of first-order gravity is the local frame, or gravitational

field, or tetrad ea , which we introduced in Sect. 6.3.1. Since parity symmetry is
by definition a frame symmetry, we will briefly discuss it here before moving to
Hamiltonian formalism. Chosen a frame ea .x/ in the internal (i.e., tangent) space
at a point x in a spacetime manifold M, one can consider an internal Lorentz
ba :
ea ! ea 0 D ba eb : (9.1)
This transformation may be proper or improper, according to the sign of the

Jacobian determinant
det D ˙1 : (9.2)
Improper transformations (det D 1) include time reversal and spatial reflec-
tions. In four dimensions, there are four such reflections: those flipping only one
spatial direction and one flipping all of them. An even number of reflections
corresponds to a proper Lorentz rotation and, for the same reason, an odd com-
bination of time reversal and reflections is just a time reversal plus a rotation.
In the following, we shall define a parity transformation by a single spatial
reflection.
In general relativity, parity is a well-defined operation on the internal space
tangent at every point of the manifold M: it changes the orientation of a frame.
9.1 Canonical Variables in General Relativity 409
Since, in general, it is not possible to define a global frame, i.e., a spatially constant
tetrad, parity is a local symmetry. Conversely, in homogeneous spacetimes one can
choose one and the same frame for every point in M or along particle geodesics:
this is the so-called Fermi frame [17–20]. In this case, which includes the obvious
examples of Newtonian physics, (quantum) field theory and FLRW cosmology,
parity flips the orientation of all the copies of the same frame at all points in M
and the whole universe is mirrored in its reflected image. The notion of parity in
Minkowski or FLRW spacetimes is a special case of the one in general relativity.
An immediate consequence is that reflection operations can be rigorously defined
only in internal space, so that parity properties should apply only to the projected
tensors. For instance, one can ask whether a vector v is a polar or pseudo-vector
only after considering its projection v a :D ea v on the local frame. The vielbein
internal basis changes orientation under reflection, so that if v a is a polar internal
vector, then one simply says that “v is a polar vector.” In this sense, “internal
parity” and “parity” are synonyms.
Consider a proper scalar field .x/. By definition, it is invariant under internal
Lorentz transformations, including parity. On the other hand, a pseudo-scalar '.x/
changes sign under reflection and can be defined as the weight-zero scalar
.4/
e
'.x/ :D " .x/ ; " :D D ˙1 ; (9.3)
j.4/ej

where .4/e :D det.ea / is the determinant of the tetrad. ' is a frame-dependent object
and, by definition, it makes sense only after choosing a frame. An internal Lorentz
transformation x ! x0 D x yields
' 0 .x/ D '.x0 / D .det / '.x/ : (9.4)
An internal scalar is not necessarily a spacetime scalar (any spacetime tensor T
is a collection of internal scalars) but a spacetime (pseudo-)scalar is also an internal
(pseudo-)scalar. Also, while a shift ! C 0 of a proper scalar with a constant is
still a scalar field, a shifted pseudo-scalar is no longer a pseudo-scalar, as .'C'0 /0 D
' C '0 , unless '0 D " 0 be a frame-dependent constant.
Equation (9.4) is commonly adopted as the definition of pseudo-scalars. The
quantity " is a constant whose presence is typically “hidden” in pseudo-tensors,
because the choice of frame (and its orientation) is often left implicit.
Let us consider a proper internal totally anti-symmetric tensor abcd . Under
reflections, it transforms as
0a0 b0 c0 d0 D aa0 bb0 cc0 dd0 abcd :
On the other hand, the fully anti-symmetric tensor
abcd :D "abcd (9.5)

is an internal pseudo-tensor, i.e.,
a0 0 b0 c0 d0 D .det /aa0 bb0 cc0 dd0 abcd :
If
0123 D 3210 D 1 (9.6)
and the other usual permutation rules follow, this is the Levi-Civita symbol, which
can be characterized either as an internal pseudo-tensor or as a spacetime pseudo-
scalar. The Levi-Civita tensor differs from the “symbol” by an extra weight factor
j.4/ej.
The determinant of the tetrad can be expressed as
.4/ 1
eD abcd ea eb ec ed D abcd ea0 eb1 ec2 ed3 :
4Š
A property of the symbol with n indices is
a1 a2 an a1 a2 an D nŠ ;
that is, the number of permutations of a set with n elements. Also,
a1 ak akC1 an a1 ak bkC1 bn D kŠ.n k/ŠıŒakC1

b
kC1
: : : ıabnn ;
where anti-symmetrization on multiple indices is done by subtracting odd index

permutations to even permutations and dividing by the factorial of the number of
indices. For n D 3, we get two relations which we shall use later:
0 0 j0 0 0 0
ijk ij k D 2ıŒ j ıkk ; ijk ijk D 2ıkk : (9.7)
9.1.1.1 Volume Form
The Einstein–Hilbert action for gravity has a counterpart in first-order Palatini

formalism. Let us discuss the terms in (2.18) one by one, ignoring the cosmological
constant.
p
In second-order formalism, the volume element is g d4 x. In terms of the
p
gravitational field, g D Œ e . In the real field R, one has g D j.4/ej and
.4/ 2
we can define the volume form
ea ^ eb ^ ec ^ ed :D j.4/ej abcd d4 x ; (9.8)
where ea :D ea dx and (9.8) defines the wedge product ^. The volume form is a
pseudo-tensor.
Before moving on, a remark is in order. The first-order Hamiltonian formalism

is actually an independent formulation of general relativity. On one hand, in this
case the existence of a metric is not even required. A metric g is well defined
only if .4/e ¤ 0.1 Here we assume that the gravitational sector is non-degenerate, so
that it is possible to define a Lagrangian theory. On the other hand, one can define
the volume form with .4/e rather than with its absolute value as in (9.8). This step
is far from being innocuous for at least two reasons. First, it changes the parity
properties of fields and canonical variables in the Hamiltonian formulation. Second,
in view of an operational definition of physical volume and quantum mechanical
measurements, the orientation of a frame should not make any difference in such
measurements. This is all the more important in general relativity, as there is no
universal frame and different observers can choose their own orientation. Then, one
can show that only (9.8) is a meaningful choice for the volume form; we shall give
an explicit argument later (Problem 9.1).
9.1.1.2 Lorentz Spin Connection and Curvature
According to the Palatini formulation, we introduce the Lorentz-valued spin

connection ! ab as an independent field. Its curvature is
Rab :D 2@Œ ! ab C 2! acŒ ! cb ; (9.9)
where Latin indices are lowered via the internal metric. The curvature is invariant
under the exchange of the index pairs and ab. At this stage, the spin connection is

unrelated to the Levi-Civita connection but the analogue of the Riemann tensor
is simply
R D ea eb Rab : (9.10)
Quite often in the literature, the language of differential forms is used to write in
compact notation expressions with explicit spacetime components. One defines
! ab :D ! ab dx (9.11)
and the exterior derivative
d! ab :D @ ! ab dx ^ dx : (9.12)
1
Thanks to this, first-order quantization is fit for describing quantum gravitational systems at scales
where the concept of metric and smooth spacetime can break down.
The curvature of ! is
Rab D d! ab C ! ac ^ !cb D 12 Rab dx ^ dx D 12 ec ^ ed Rabcd : (9.13)
9.1.1.3 Torsion
All spacetimes we have considered so far were manifolds with vanishing torsion
tensor T a , defined in the generalized Bianchi identity
RabŒ eb D D.!/Œ T a ¤ 0 ; (9.14)
where D is the covariant derivative operator made with the spin connection ! ab .
Intuitively, torsion characterizes how tangent planes (moving frames) twist around
a geodesic. Certain fields (such as ˇ in Problem 9.1) or fermions [21–24] can act
as sources for torsion.2 In that case, one writes the Lorentz spin connection as a
torsionless part !N ab (the Ricci connection, which obeys the homogeneous structure
equation (9.49) we will see later) plus the contortion 1-form Kab [27]:
! ab D !N ab C Kab ; (9.15)
where the contortion tensor
Kab D ea eb K ; K D K ; (9.16)
is related to the torsion T D T by
K D 12 .T T T / : (9.17)
It is particularly convenient to split torsion into its irreducible components in

accordance with the Lorentz group [28–31]:
1 1
T D T g T g S C q ; (9.18)
3 6
where
T :D T (9.19)
2
Spinors cannot be coupled directly to the metric but they are easily accommodated in the vielbein
formulation [25, 26].
is the trace vector,
S :D T (9.20)
is the pseudo-trace axial vector and the anti-symmetric tensor q is such that
q D 0 D q .
9.1.1.4 Hilbert–Palatini Action
The Hilbert–Palatini action can be obtained formally by replacing the Ricci scalar
in the Einstein–Hilbert action (2.18) with the spin connection curvature scalar

R.e; !/ :D ea eb Rab :
Z Z
1 1
SHP D d4 x j.4/ej ea eb Rab D abcd ea ^ eb ^ Rcd ; (9.21)
2 2 4 2
which is a proper scalar. Often the notation Rab D abcd Rcd =2 is used.
9.1.1.5 Holst and Nieh–Yan Terms
The Hilbert–Palatini action can be generalized by adding terms which do not affect
the classical dynamics. These terms will be involved in the theory of canonical
quantum gravity and loop quantum cosmology.
The first of these terms was introduced by Holst [32] and it reads
Z Z
1 1
SH :D d4 x j.4/ej ea eb abcd Rcd D ea ^ eb ^ Rab ; (9.22)
4 2 2 2
where is a real constant called the Barbero–Immirzi parameter [6, 33, 34]. Often
the notation ˇ :D 1= is used. The Holst term violates parity at the level of
the action but the classical dynamics is unchanged. In fact, the integrand in (9.22)
vanishes because of the Bianchi cyclic identity RabŒ eb D 0. This identity is also
called second Cartan structure equation and is obtained when solving the dynamics.
For this reason the Holst term, which is not topological, is said to vanish “on half
shell,” i.e., when half of the dynamical equations have been taken into account.
This situation is unusual in Lagrangian theories, where one would expect
that non-dynamical terms be topological. Indeed, the Holst modification is not
completely general and it constitutes only one of the two terms composing a
well-known topological density called the Nieh–Yan 4-form [23, 35–39]:

Z
1
SNY D d4 x j.4/ej ab T a T b ea eb abcd Rcd
4 2
Z
1 a
D T ^ Ta ea ^ eb ^ Rab
2 2
Z
1
D d .ea ^ T a / : (9.23)
2 2
The Nieh–Yan density is linear in the curvature and contains a torsion-torsion term.
Contrary to the Holst term, it is a total divergence, as the last line shows.
Topological invariants become dynamical when their coupling constant is pro-
moted to a field. In Problem 9.1, we study the general case where 1= D ˇ !
ˇ.x/ [40–44], since it is an example of a gravitational theory with torsion which is
dynamically equivalent to a torsion-free model.
9.1.2 Hamiltonian Analysis
At the classical level, we can ignore the Holst or Nieh–Yan term and consider
torsion-free gravity minimally coupled with a real scalar field (bars omitted):
Z Z
1 ab 1
SD dt L D d4 x j.4/ej e e R @
@ V. / : (9.24)
2 2 a b 2
Here the field is left arbitrary and we study the dynamics as a constrained
Hamiltonian system.3
In torsion-free (pseudo-)Riemannian manifolds, the requirement that the spin
connection realizes parallel transport of angles and lengths translates into the
compatibility equation. In second-order formalism, it states that the metric is
covariantly constant, r g D 0.
Systems endowed with gauge symmetries are characterized by a phase-space
structure defined by a set of relations called constraints. Different phase-space
points describe the same solution if they are related by a gauge transformation.
Also, some of the canonical variables are not dynamical and the phase space is
larger than the space of physical degrees of freedom. Both these occurrences are
translated into geometric surfaces by the constraints. Constraint surfaces are sub-
sets of phase-space points whose intersections correspond to dynamical solutions.
Examples of constrained systems are the parametrized non-relativistic point particle
3
The scalar can be identified with the Barbero–Immirzi field, in which case D Q̌ is a pseudo-
scalar and V. / D 0. One can check that the Hamiltonian analysis of (9.24) is consistent with the
one starting from the fundamental action (9.136) (Problems 9.1 and 9.2).
[3], the relativistic particle [45, 46], Yang–Mills theory on curved space and, last but
not least, gravity.
Following the Dirac procedure [47–49], we calculate the first- and second-class
constraints of the theory (9.24). The latter can be easily solved and the system turns
out to be characterized by a set of first-class constraints which reflect rotational and
diffeomorphism gauge freedom.
We assume spacetime M to be globally hyperbolic, so that we can define
a Cauchy surface ˙, the normal u to the surface and the first fundamental
form (2.41). Let t D t .x/ be the time-flow vector field on M 3 x satisfying
t r t D t @ t D 1. Neither t nor t can be interpreted in terms of physical mea-
surements of time, since the metric is, in fact, an unknown dynamical field. The
time-flow vector field generates a one-parameter group of diffeomorphisms, known
as embedding diffeomorphisms, xt W R ˙ ! M, defined as xt .x/ :D x.t; x/. This
allows to represent spacetime as a smooth deformation of the three-dimensional
Cauchy surfaces ˙ into a one-parameter family of three-dimensional Cauchy

surfaces ˙t . These are described by the parametric equations xt D xt .x/, where
t denotes the hypersurface at different “times.”
A general parametrization can be obtained by introducing the normal and
tangential components of the vector field t .x/ with respect to ˙. Namely, we define
N :D t u ; N :D h t ; (9.25)
respectively called the lapse function and the shift vector. As a consequence, we
have
t .t; x/ D @t x .t; x/jx.t;x/Dxt .x/ D N.t; x/ u .t; x/ C N .t; x/ :
By acting with a Wigner boost on the local basis, we can rotate it so that its zeroth
component is made parallel, at each point of ˙, to the normal vector u , i.e., u D
e0 , implying that the local boost parameter ei 0 vanishes at each point of ˙:
ei 0 D 0 :
The requirement that this particular choice of the orientation of the local basis
be preserved along the evolution fixes the so-called Schwinger or time gauge, the
net result being that the action no longer depends on the boost parameters. Also,
the local symmetry group is reduced from the initial SO.1; 3/ to SO.3/, which
encodes the spatial rotational symmetry. It can be demonstrated that fixing the time
gauge into the action does not affect the consistency of the canonical analysis, this
procedure being equivalent to a canonical gauge fixing.
9.1.2.1 Step 1: Decomposing the Action in the .3 C 1/ Splitting
The action (9.24) can be written as follows (all quantities are torsion free):
Z
1 3 .4/
SD dtd x j ej 2e0 ei R0i C ei ej Rij
2 2
Z
1
dtd3 x j.4/ej @ @ C V
2
Z
1 t N 0i
D 2 dtd3 x j.4/ej 2 ei R C ei ˛ ej R˛ ij
2 N
Z
3 .4/ 1
dtd x j ej .h u u /@ @ C V
2
Z
1 N ˛ .3/ ij
D 2 dtd3 x jej K N ˛ ei .3/R˛ 0i
j
e e R˛ C 2KŒ˛ i
2 i j
h i
C e˛i KP ˛i C ! 0i @˛ t @˛ t ! i ! ik˛ .t ! k / C .t ! ik /K˛k
Z
3 1 P 1
C dtd x jej . N ˛ @˛ /2 N @˛ @˛ C V ;
2N 2
where P D t @ , .4/e D Ne, e D det ei˛ is the determinant of the triad, .3/R˛
ij
is the
ij
curvature of the spin connection ! ˛ with all indices spatial and
K˛i :D ! 0i˛ (9.26)
is the extrinsic curvature. The following notation has been used: t ! i D t ! 0i ,
t ! ik D t ! ik . Greek indices ˛; ; : : : from the beginning of the alphabet and
Latin indices i; j; : : : from the middle of the alphabet run from 1 to 3 and denote,
respectively, components transforming under spatial diffeomorphisms and local
spatial rotations. The three-dimensional Levi-Civita symbol is defined as in (9.6)
and we will often make use of the relation (9.7). The absolute value of e is often
omitted in the literature; here we keep it in order to avoid confusion when studying
how the action and canonical variables transform under parity.
Now, remembering the definition of the Lie derivative operator (3.39) on a vector,
one has
Z
1
S D 2 dtd x jej e˛i Lt K˛i D˛ t ! i C .t ! ik /K˛k
3

N
2N ˛ ei DŒ˛ K i C e˛i ej .3/R˛ ij C 2KŒ˛
j
i
K CS ; (9.27)
2
where
Z
3 1 1
S D dtd x jej .Lt N ˛ @˛ /2 N @˛ @˛ C V : (9.28)
2N 2
We introduced the so.3/-valued torsion-free covariant derivative D˛ , which can be

written as
D˛ v i D @˛ v i C ijk ˛j v k (9.29)
on a gauge vector, where
1 jk
˛i :D ijk !˛jk ; !˛jk D i ˛i : (9.30)
2
Notice the minus sign, which stems from the definition (9.6). In much of the
literature the other convention 0123 D 1 with upper indices is in use and the minus
signs are absorbed into the Levi-Civita symbol.
The curvature of is defined as
1
Ri˛ :D 2@Œ˛ i C ijk ˛j k D ijk .3/R˛ jk
: (9.31)
2
ij
Later we will discover that ! ˛ is the Ricci spin connection, which depends on the
triad field. The canonical analysis will eventually show that it is not an independent
variable but we must keep it as such for the time being.
9.1.2.2 Step 2: Defining Canonical Variables and Symplectic Structure
The next step is the definition of the momenta conjugate to the fundamental
variables. Since the Lagrangian is singular, we expect a set of primary constraints
to appear. In particular, the only non-vanishing momenta are those conjugate to K˛i
and :
ıS
K˛i W Ei˛ :D 2 D jeje˛i ; (9.32a)
ıLt K˛i
ıS jej P
W ˘ :D D . N ˛ @˛ / : (9.32b)
ıLt N
All the others vanish identically, i.e.,
e˛i W P˛i D 0 ; (9.33a)

˛i W ˘i˛ D 0 ; (9.33b)
t !i W ˘i D 0 ; (9.33c)
t ! ij W ˘ij D 0 ; (9.33d)
˛
N W ˛ D 0 ; (9.33e)
N W 0 D 0 : (9.33f)
The field Ei˛ is a weight-1 proper vector called densitized triad or, in analogy
with Maxwell theory, gravielectric field. The pair .K; E/ is called Schwinger
variables [50].
Once a set of canonical fields and conjugate momenta .Qn ; Pn / is established,
one can define the equal-time Poisson bracket between two functionals f and g as

f ŒQn .t; x/; Pn .t; x/; gŒQn .t; x0 /; Pn .t; x0 /
X ıf ıg ıf ıg

:D : (9.34)
n
ıQn .t; y/ ıPn .t; y/ ıPn .t; y/ ıQn .t; y/
This writing is only formal for functionals wherein spatial derivatives act on the
canonical variables. In that case, one can define smeared functionals
Z
F.t/ :D d3 xf .x/f ŒQn .t; x/; Pn .t; x/ ;
Z
G.t/ :D d3 x0 g .x0 /gŒQn .t; x0 /; Pn .t; x0 / ;
where f ;g are arbitrary space-dependent tensors, and calculate the Poisson bracket
fF; Gg.
The phase space is equipped with the symplectic structure
n o
ˇ
K˛i .t; x/; Ej .t; x0 / D 2 ı˛ˇ ıji ı.x; x0 / ; (9.35a)
˚
.t; x/; ˘ .t; x0 / D ı.x; x0 / ; (9.35b)
n o
e˛i .t; x/; Pˇ .t; x0 / D ıˇ˛ ıi ı.x; x0 / ;
j j
(9.35c)
n o
ˇ
˛i .t; x/; ˘j .t; x0 / D ı˛ˇ ıji ı.x; x0 / ; (9.35d)
˚
t ! i .t; x/; ˘k .t; x0 / D ıki ı.x; x0 / ; (9.35e)
˚
t ! ij .t; x/; ˘kl .t; x0 / D ıŒki ıl ı.x; x0 / ;
j
(9.35f)
˚
N.t; x/; 0 .t; x0 / D ı.x; x0 / ; (9.35g)
˚ ˛
N .t; x/; ˇ .t; x0 / D ıˇ˛ ı.x; x0 / : (9.35h)
9.1.2.3 Step 3: Primary Constraints
As one can immediately note, in none of the above conjugate momenta except ˘
is there the temporal Lie derivative of any of the fundamental variables. Therefore,
in principle, all momenta but one should be considered as primary constraints,
that is to say, relations encoding the fact that not all the canonical variables are
independent. Thus, the following set of primary constraints has to be imposed:
.K/ ˛
Ci :D Ei˛ jeje˛i 0 ; (9.36a)
.e/ i
C˛ :D P˛i 0; (9.36b)
. / ˛
Ci :D ˘i˛ 0 ; (9.36c)
Ci :D ˘i 0 ; (9.36d)
Cij :D ˘ij 0 ; (9.36e)
C.N/ :D 0 0 ; (9.36f)
C˛ :D ˛ 0 ; (9.36g)
where the symbol indicates weak equalities valid only on the constraint sur-
face, i.e., the manifold in phase space defined by the imposition of the primary
constraints. Expressions which vanish weakly do not vanish identically throughout
phase space. In contrast, strong equalities D hold on all phase space.
9.1.2.4 Step 4: Dirac Hamiltonian
Having calculated the conjugate momenta, we can now perform a Legendre trans-
formation and extract the canonical Hamiltonian. Since the latter is not uniquely
determined because of the primary constraints (generically denoted as Cm ), we write
the Dirac Hamiltonian:
Z
3 1 ˛
HD D d x E Lt K˛ C ˘ Lt C Cm L
i m
2 i
Z (
1
D d x N ˛ H˛ C NH C 2 Ei˛ D˛ t ! i .t ! ik /K˛k
3

C.K/i˛ .K/Ci˛ C .e/˛i .e/C˛i C . /i˛ . /Ci˛ C i Ci

)
C ij Cij C C.N/ C ˛ C˛ ;
where m are arbitrary functions,
2
H˛ :D E DŒ˛ K i C ˘ @˛ (9.37)
2 i
is the super-momentum and
1
˛ ij j
H :D 2
E E k Rk˛ 2KŒ˛
i
K CH (9.38a)
2 jej i j
is the super-Hamiltonian (or scalar constraint),4 where
1 1
H :D ˘ 2 C jej@˛ @˛ C jejV. / : (9.38b)
2jej 2
For autonomous systems (no explicit time dependence), the time evolution of a
functional f Œe; K; ; N; : : : of the canonical variables is governed by the Hamilton
equation
fP D f f ; HD g : (9.39)
9.1.2.5 Step 5: Secondary and First-Class Constraints
The primary constraints (9.36) have been imposed at a given time t. As a consistency
requirement, the Dirac canonical procedure calculates the Poisson brackets between
the primary constraints and the Dirac Hamiltonian. In fact, the constraints (9.36)
must be (weakly) constant in time,
CP m D fCm ; HD g 0 : (9.40)
For example, a momentum which is zero on a hypersurface ˙t must vanish also

on any other surface ˙t0 . If (9.40) is not realized on the primary surface for some
value of the Lagrange multipliers m , then it must be constrained to vanish. In this
way, secondary constraints are generated which determine the secondary constraint
surface on the phase space [51].
4
The context should be clear enough to avoid confusion with the Hubble parameter H in conformal
time.
The Poisson brackets between the Dirac Hamiltonian and the first two primary
constraints (9.36a) and (9.36b) do not generate any secondary constraint. In fact,
they can be set to zero by choosing the Lagrange multiplier .K/˛i suitably.
For the other primary constraints, one gets
2 fCi ; HD g D Bi :D D˛ Ei˛ ; (9.41)

2 . / ˛ ˛j N
f Ci ; HD g D ijk E t ! N K E @
k k k
jej
N jk
CN ˛ ijk E j K k i D Ej˛ Ek ; (9.42)
jej
2 fCij ; HD g D K˛Œ j Ei˛ ; (9.43)
fC˛ ; HD g D H˛ ; (9.44)
fC.N/ ; HD g D H : (9.45)
Note that all these equations are to be regarded as a shorthand for relations involving
smeared functionals. For instance, define the smeared super-momentum and super-
Hamiltonian as
Z Z
dŒN ˛ :D d3 x N ˛ H˛ ; HŒN :D d3 x NH ; (9.46)
and the smeared constraints

Z Z
CŒ˛ :D d3 x0 ˛ C˛ ; CŒ :D d3 x0 C.N/ :
Then, the last two equations are actually (time dependence omitted)
fCŒ˛ ; HD g D fCŒ˛ ; dŒN ˇ g

Z Z
3 ı
D 0 d y d3 x0 ˛ .x0 /˛ .x0 /
ı˛ .y/
Z
ı
˛ d3 xN ˇ .x/Hˇ .x/
ıN .y/
Z
D d3 yd3 x0 d3 x˛ .x0 /ı.x0 ; y/ı.x; y/H˛ .x/
Z
D d3 x˛ .x/H˛ .x/
D dŒ˛ ; (9.47)
fCŒ; HD g D HŒ : (9.48)
To keep notation light, we will always write constraint formulæ in their non-smeared
version.
Since the spin connection is torsion-free, the triad obeys the homogeneous
structure equation
DŒ˛ ei D 0 : (9.49)
This expression is an equation of motion in the Lagrangian approach. Its derivation

in time-gauge canonical formalism is not obvious since it comes from a second-class
constraint which, in this case, is identically zero. Without fixing the time gauge, this
second-class constraint is non-trivial and vanishes if (9.49) holds weakly.
As a consequence of (9.49), ˛i is the spatial torsion-free so.3/ spin connection
2˛i ŒE D ijk E j r˛ Ek ; (9.50)
where the covariant derivative r˛ contains the Christoffel symbol, in turn expressed
as a function of Ei˛ (see [52] for the explicit expression). While in second-order
formalism (9.50) is the definition of the spin connection, in first-order formalism the
spin connection is not an independent variable and (9.50) is a result of the constraint
analysis. In particular, this implies that the Lagrange multiplier . /i˛ is determined
by the equation of motion of ˛i ŒE.
On the other hand, the dynamical equations of the canonical variables t ! i ,
t ! ij , N and N ˛ are completely arbitrary, since each of their Poisson brackets with
the full Dirac Hamiltonian is proportional to the associated Lagrange multiplier
m (the same is true for the equations of motion of e˛i and ˛i but, as argued
above, their Lagrange multipliers are no longer arbitrary). Therefore, at this point
a useful simplification of the canonical system of constraints can be naturally
provided and consists in treating the above sub-set of canonical variables directly as
Lagrange multipliers. This gauge fixing could have been done at the very beginning
by discriminating, by an educated guess, dynamical variables from Lagrange
multipliers. However, the Dirac procedure does not give us any hint about this
classification ab initio, so that we have preferred to follow the general analysis and
to arrive at this conclusion after having calculated the set of primary and secondary
constraints.
Equation (9.41) is solved because of the compatibility equation; then, D˛ Ei˛ D
0 strongly. Equations (9.42), (9.43), (9.44) and (9.45) do not contain Lagrange
multipliers m and do not vanish on the primary surface. Hence they have to be
considered as secondary constraints.
Equation (9.42) has been arranged to isolate three terms. The first includes some
of the new Lagrange multipliers and can be made to vanish by definition. The second
is proportional to (9.43) and it vanishes weakly. The third term is strongly equal to
zero as it is nothing but the compatibility equation. Overall, (9.42) is redundant with
other constraints and it will be ignored from now on.
The left-hand side of (9.43) suggests to define
Ri :D ij k K˛j Ek˛ : (9.51)
Equations (9.43), (9.44) and (9.45) (i.e., (9.37), (9.38) and (9.51)) must be imposed
to vanish weakly, in which case they are called first-class constraints. In general,
first-class functions of canonical variables are those whose Poisson brackets with
every constraint vanish weakly. The Poisson bracket of two first-class constraints,
which we omit to report here, is a linear superposition of first-class constraints.
Moreover, it is a general result of constrained Hamiltonian systems that first-class
primary constraints generate small gauge transformations [51] (see Problem 9.4).
The “Dirac conjecture” postulates that also first-class secondary constraints are
gauge generators.
Functions which are not first class (and so admit a non-zero Poisson bracket with
at least one constraint) are second class. After the gauge fixing, we are left with no
second-class constraints.
To summarize, the initial complicated system of constraints has been reduced to
a set of seven first-class constraints,
Ri 0 ;
H˛ 0 ; (9.52)
H 0;
which reflect the gauge freedom of the physical system. In fact, Ri 0 and
H˛ 0 establish, respectively, invariance under rotations of the local spatial
basis and spacetime diffeomorphisms within the three-surfaces ˙. The super-
momentum and super-Hamiltonian constraints correspond, respectively, to the 0˛
and 00 components of Einstein’s equations. The Hamiltonian constraint both
encodes invariance under time reparametrizations and generates the dynamics (time
evolution) of the system. Symmetry and dynamics are thus entangled.
A standard counting of degrees of freedom (d.o.f.) shows that, before any gauge
fixing,

# of physical # of canonical # of 2nd-class
2 D
d.o.f. variables constraints

# of 1st-class
2 :
constraints
The number of second-class constraints is even, as it must be in order for them to

be completely solved. Once the Dirac bracket [51] is defined for the system, the
second-class constraints can be considered as strong equations and the original

variables are reintroduced in the theory by using a simple identity.
After gauge fixing, the system is completely described by the so.3/-valued
extrinsic curvature K˛i and its momentum Ei˛ , together with the field and its
momentum ˘ , for a total of 9 C 9 C 1 C 1 D 20 canonical variables. Then,
the physical degrees of freedom on the phase space are
20 0 2 7 D 6 ;
specifically four corresponding to the two polarizations of the graviton and two
associated with the scalar field.
9.1.3 Ashtekar–Barbero Variables
The Ashtekar–Barbero variables are defined as
Ai˛ :D K˛i C ˛i ; (9.53)

1
P˛i :D Ei˛ ; (9.54)
where is the Barbero–Immirzi parameter and the internal indices i; j; : : : live in

the su.2/ Lie algebra. For quantization purposes, it is desirable that (9.54) defines
a one-to-one mapping between the triad and the momentum P˛i . This requires to
be a frame-independent constant. In Problem 9.3, it is shown that the symplectic
structure is non-canonical if one tries to define the pair .A; P/ with the Barbero–
Immirzi field ˇ.x/ instead of ˇ0 D 1= . ˇ0 can arise as the expectation value of
the Barbero–Immirzi field.
The symplectic structure in the new variables is
n o
Ai˛ .t; x/; Pj .t; x0 / D 2 ı˛ ıji ı.x; x0 / ;
(9.55)
˚ 0 0
.t; x/; ˘ .t; x / D ı.x; x / ;
while the other brackets vanish. Since (9.53) and (9.54) define a canonical trans-
formation of variables (a symplectomorphism), one can apply it directly to the
constraints without repeating the canonical analysis.
The first-class constraints can be easily rearranged. Equations (9.41) and (9.51)
are, respectively, the polar (boost) and axial part of the covariant Gauss tensor,
governing frame Lorentz transformations (small gauge transformations). The strong
and weak equations Bi D 0 and Ri 0 can be combined into the Gauss constraint
Gi :D Ri C Bi = (Problem 9.4):
Gi :D D˛ P˛i D @˛ P˛i C ij k Aj˛ P˛k ; (9.56)
where D is the covariant derivative with respect to Ai˛ .

The curvature of the Ashtekar–Barbero connection is
i
F˛ˇ :D 2@Œ˛ Aiˇ C ijk Aj˛ Akˇ : (9.57)
This is related to (9.31) by the formula
Ri˛ˇ D F˛ˇ
i
C ijk K˛j Kˇk 2 DŒ˛ Kˇ
i
: (9.58)
Ignoring the matter part, (9.37) can be rewritten as
ˇ ˇ
2 H˛ D 2Pi DŒ˛ Aiˇ 2Pi DŒ˛ ˇi
ˇ ˇ ˇ j ˇ ˇ
D Pi F˛ˇ
i
C ijk Pi Aj˛ Akˇ 2 ijk Pi KŒ˛ Akˇ Pi Ri˛ˇ ijk Pi ˛j ˇk
ˇ
Pi F˛ˇ
i
K˛i Ri ;
where in the third (weak) equality we made use of the Bianchi identity Ei Ri˛ D 0
and of the rotation constraint (9.51). Then, dropping the last term the super-
momentum constraint is
1 ˇ i
H˛ P F C ˘ @˛ : (9.59)
2 i ˛ˇ
Exploiting the compatibility condition and the rotation constraint, the super-
Hamiltonian constraint is, up to a weakly vanishing term,
2 ˛ ˇ
Pi Pj h i
ij 2 j
H k F˛ˇ
k
2.1 C /KŒ˛
i
Kˇ C H ; (9.60)
2 2 jej
where
p
jej D j det Pj : (9.61)
The Hamiltonian constraint is non-linear in the connection (via the curvature) and
in the extrinsic curvature.
Note that Ai˛ does not have definite parity and that the way improper Lorentz
transformations act on the Hamiltonian constraint is not transparent. This is a
problem intrinsic to the choice of the action.
Loop quantum gravity makes extensive use of the Ashtekar–Barbero connection
and its conjugate momentum, which allow to reformulate gravity almost as a gauge
theory and facilitate a well-defined quantization. The main difference with respect to
gauge theories is in the fact that the Hamiltonian constraint is not only dynamical but
it also generates time reparametrizations. Moreover, it contains quadratic extrinsic
curvature terms, which decouple from the theory when
D i : (9.62)
In this case, the Ashtekar connection A is complex and it is said to be (anti-)self-

dual. The (anti-)self-dual theory is greatly simplified but, unfortunately, it turns out
that it is difficult to quantize, the reason being that the gauge group becomes the non-
compact group SL.2; C/. Therefore, most of the progress in LQG is based upon the
real Ashtekar–Barbero connection.
9.1.4 ADM Variables
The ADM variables were the first to be used in canonical quantization [53, 54]. As
we have seen, the foliation into spatial hypersurfaces ˙t defines the lapse function
and shift vector (9.25). The corresponding metric is
2
N C N N N˛
g D ; (9.63)
Nˇ h˛ˇ
whose inverse is

1 1 N˛
g D : (9.64)
N2 N ˇ N 2 h˛ˇ N ˛ N ˇ
The geometric meaning of the metric components is the same as before. Given the
normal u D .N; 0; 0; 0/ toRthe hypersurface ˙t , the lapse N relates coordinate
time t with proper time D dt N on curves orthogonal to ˙t . The shift vector
vanishes in comoving coordinates. The projected spatial components of the second
fundamental form (7.126) are the extrinsic curvature K˛ˇ D K˛ˇ .t; x/ D ei.˛ Kˇ/i of
the hypersurface ˙t with respect to the embedding manifold M:

1 1P
K˛ˇ D h˛ˇ r.˛ Nˇ/ : (9.65)
N 2
In terms of (9.65), the Einstein–Hilbert action becomes

Z p
1
S D 2 dtd3 x h N .3/R C K˛ˇ K ˛ˇ K 2 C S ; (9.66)
2
where K :D K˛˛ is the trace of the extrinsic curvature and h is the determinant of
the 3-metric. This result can be found either via a direct computation or from (9.27).
We
R have pimplicitly thrown away the York–Gibbons–Hawking boundary term S@ /
3
@M d x h K mentioned in Sect. 2.2 [55, 56].5 In a quantum setting, however (as
when looking for instantonic solutions in Euclidean signature), the boundary piece
may be important since the classical equations of motion do not hold. It is customary
to add its opposite in the definition of the fundamental action, so that it cancels away
and (9.66) remains the same.
Define the DeWitt “metric of metrics” (or super-metric) [54]
G˛ˇ ı :D h˛. hı/ˇ 12 h˛ˇ h ı ; (9.67)
and its inverse
G ˛ˇ ı
:D h˛. hı/ˇ h˛ˇ h ı :
Note that the inverse of G is not G with indices raised by h˛ˇ . The canonical
momenta in the gravity sector are defined as
ıS 1 p 1 p
˛ˇ :D D 2 h G ˛ˇ ı K ı D h .K ˛ˇ h˛ˇ K/ ; (9.68a)
ı hP ˛ˇ 2 2 2
ıS
0 :D D 0; (9.68b)
ı NP
ıS
˛ :D D 0: (9.68c)
ı NP ˛
Here we used the time derivativeP D t r rather than the Lie derivative Lt , but this
difference is immaterial due to the simple form of the action (9.66). Equation (9.68a)
is the spatial version of (7.125).
The total Dirac Hamiltonian is
Z
HD D d3 x.N ˛ H˛ C NH C 0 NP C ˛ NP ˛ / ; (9.69)
5
A derivation of the York–Gibbons–Hawking boundary term can be found in [57]. Within first-
order formalism, the boundary term is discussed in [58, 59].
where
H˛ D 2r ˇ ˛ˇ C ˘ @˛ ; (9.70)
and
p
2 2 h
H D p ˛ˇ G˛ˇ ı ı
2 .3/
RCH : (9.71)
h 2
More explicitly, the last expression is
p
h
H D 2 K ˛ˇ K˛ˇ K 2 .3/R C H : (9.72)
2
The classical constraints are the primary relations
0 0 ; ˛ 0 ; (9.73)
and the momentum and Hamiltonian constraints
H˛ 0 ; H 0: (9.74)
The latter are first-class constraints (Problem 9.5). After imposing second-class
constraints and skimming out Lagrange multipliers, (9.69) is reduced to
Z
HD D d3 x .N ˛ H˛ C NH/ : (9.75)
9.2 Wheeler–DeWitt Equation
Once the phase space of a Hamiltonian system has been identified, it can be
quantized according to the Dirac prescription. The first step is extremely important
and consists in finding suitable phase-space coordinates .Q; P/ (e.g., Ashtekar–
Barbero or ADM variables). Not all possible choices will lead to a well-defined
quantization. In the case of gravity, for example, the ADM variables turn out to be a
natural but also ill-suited choice, while the Ashtekar–Barbero connection, although
less intuitive from the point of view of classical metric spacetimes, allows for greater
progress.
9.2 Wheeler–DeWitt Equation 429
The second step is to identify an algebra of functionals f ŒP; Q with well-defined

Poisson brackets, closed under Poisson brackets (meaning that the bracket f f ; gg
of two elements f and g belong to the algebra) and separating distinct points in
phase space (i.e., if Q Q0 and P P0 under a small gauge transformation,
then f ŒQ0 ; P0 D f ŒQ; P). The third step is to find an irreducible representation
of the algebra on a Hilbert space, that is to say, a mapping f ! fO from elements
of the algebra to operators on a Hilbert space. Poisson brackets are mapped into
commutators, complex conjugation to adjointness, and so on. Only at this point can
one deal with the quantum constraints.
We will not follow this detailed agenda but we will introduce some basic concepts
first in ADM and then in Ashtekar–Barbero variables, with special reference to
quantum cosmology in the next chapter. The Planck constant „ is restored.
9.2.1 Superspace and Quantization
The DeWitt metric (9.67) can be regarded as the metric of a six-dimensional

manifold with pseudo-Riemannian structure [54]. Indices can be collected in pairs
to define the metric GAB D G.˛ˇ/. ı/ , A D 1; : : : ; 6, and the elements hA as h˛ D h˛˛ ,
˛ D 1; 2; 3, h4 D 21=2 h23 , h5 D 21=2 h31 , h6 D 21=2 h12 . In the Minkowski case
h˛ˇ D ı˛ˇ , it is easy to solve the characteristic equation det.G 1/ D 0 and to
find that G has signature .; C; ; C/. This can be extended to a metric of fields
accounting for all the physical degrees of freedom of the problem. In the case of a
matter scalar field, for example, one can define the block-diagonal metric

1 4 2 G AB 0
G :D p ; (9.76)
h 0 1
which appears in the Hamiltonian constraint. G is defined over the infinite-

dimensional space Riem .˙t / of all Riemannian 3-metrics h˛ˇ .t; x/ and scalar-field
configurations .t; x/ at each point x 2 ˙t M in a spatial hypersurface with
fixed topology:
˚
Riem .˙t / :D h˛ˇ .t; x/; .t; x/ j x 2 ˙t : (9.77)
Since diffeomorphisms define equivalence classes of physical geometries, the actual

configuration space is the so-called superspace6
S.˙t / :D Riem .˙t / n Diff0 .˙t / ; (9.78)
6
This has nothing to do with the superspace of supersymmetry of Sect. 5.12.
where Diff0 is the space of maps realizing infinitesimal foliation-preserving diffeo-

morphisms.
The superspace (9.78) should run over all the canonical coordinates QA (conju-
gate to their momenta PA ), including the lapse and the shift. We have not included
N and N˛ because they factor out as Lagrange multipliers. It is instructive to see this
explicitly. Consider the extended superspace SQ D S [ fN.t; x/; N˛ .t; x/ j x 2 ˙t g.
In the Dirac quantization, SQ is regarded as the Hilbert space whereon we define
functionals Œh˛ˇ ; ; N; N˛ of the canonical variables. The classical momenta are
promoted to functional derivative operators
ı
PA ! POA :D i„ : (9.79)
ıQA
In particular,
ı
˛ˇ ! O ˛ˇ :D i„ ; (9.80a)
ıh˛ˇ
ı
˘ ! ˘O :D i„ ; (9.80b)
ı
ı
0 ! O 0 :D i„ ; (9.80c)
ıN
ı
˛ ! O ˛ :D i„ : (9.80d)
ıN˛
The quantum version of a constraint C.Q; P/, denoted with a hat, is a composite
operator acting on the Hilbert space of functionals . It is made of the above
momentum operators and of multiplication operators on S corresponding to the
canonical variables:
h˛ˇ ! hO ˛ˇ :D h˛ˇ ; ! O :D : (9.80e)
The classical Poisson bracket is replaced by the commutator bracket as in standard

quantum mechanics,
i
f; g ! Œ; : (9.81)
„
Each non-smeared constraint equation
O D CŒ
C O Q;
O P
O D0 (9.82)
is an infinite set of relations, one for each spatial point in the foliation ˙t . The
way the operators (9.80) are arranged in the definition of the quantum constraints
gives rise to inequivalent orderings. The operator ordering problem is a source of
ambiguity which, typically, cannot be removed in canonical approaches. In general,

a classical constraint is of the form
X Y
CD Cn ; Cn D An;m ;
n m
where An;m are functions of the canonical variables. Take, for instance, terms like
Al B. When quantizing, we have an infinite number of options, as many as the ways
to resolve the classical identity:
8
ˆ
Âbl BO
ˆ
ˆ
ˆ
.A/
< O l BO
AB!
l O
B.A/ O l : (9.83)
ˆ
ˆ
ˆ O
ˆ .A/ lq O A/
B. O q
:̂
:::
Sometimes, independent arguments can select one or more choices over the others;
below we will see an example.
The functional obeying the quantum version of the constraints is often called the
wave-function of the Universe. The primary constraints (9.73) are
ı ı
O 0 D i„ D 0; ˛ D i„ D 0; (9.84)
ıN ıN˛
implying that does not depend on any of the non-dynamical variables:
D Œh˛ˇ ; :
From now on, we drop SQ and refer to S as the superspace.

Choosing derivative operators to act to the right, the momentum constraint is

ı ı
HO ˛ D i„ @˛ 2r ˇ ˛ˇ D 0: (9.85)
ı ıh
One can show that this condition is satisfied if is invariant under coordinate
transformations x ! x0 in the hypersurface ˙t [60]. This is in agreement with
the definition of superspace (9.78) and the classical interpretation of the momentum
constraint as the generator of diffeomorphisms on ˙t .
The quantum dynamics is governed by the Hamiltonian constraint, which has
come to be known as the Wheeler–DeWitt (WDW) equation [54, 61]. With derivative
operators acting to the right of G, it reads

„2
O D p ı2 1 ı2
H 2 2 G˛ˇ ı C C U.h˛ˇ ; / D 0 ; (9.86)
h ıh˛ˇ ıh ı 2ı 2
where the effective potential is
.3/
p R 1 ˛ˇ
U.h˛ˇ ; / D h 2 C h @˛ @ˇ C V. / : (9.87)
2 2
With a different operator ordering where G is in between the momenta A [54, 62],
the WDW equation becomes

O D 1 „2 rA r A C U.hA ; / D 0 ;
H (9.88)
2
where
1 p
rA r A :D p @A G GAB @B (9.89)
G
is the Laplace–Beltrami operator in the super-metric. The operator ordering ambi-

guity is removed if one requires invariance of the WDW equation under a conformal
transformation of the super-metric G and the lapse function [63, 64]. Then, in (9.88)
one must add an O.„2 / operator proportional to the scalar curvature R of the n-
dimensional superspace [54], with conformal coupling:
n2
rA r A ! rA r A R: (9.90)
4.n 1/
We should bear in mind that gravity is not perturbatively renormalizable when

the three-metric h˛ˇ is elected as the fundamental field, so that we expect that
the quantum constraints in ADM variables be invalid, or receive corrections, when
sufficiently small volumes are considered. Therefore, the interpretation of should
be as the wave-function of a large universe. The adjectives “small” and “large”
are intentionally used in a loose sense. The understanding is that these universes
are large compared to Planck-size patches but small enough to justify a quantum
treatment.
9.2.2 Semi-classical States
In quantum mechanics and quantum field theory on classical spacetimes, the path
integral or transition amplitude [65, 66] gives, after squaring, the probability for a
given quantum state at the initial time ti to evolve into another state at the final time
tf . A general-covariant, background independent theory of gravitation does not have
a preferred time coordinate by which to define “initial” and “final” but, nevertheless,
it may be possible to define a notion of path integral. In particular, the most general
solution of the WDW equation on a spatial hypersurface ˙ with induced metric h˛ˇ
and a generic matter content h can be formally written as the path integral [56, 67]
Z
i
Œh˛ˇ ; h D ŒDgŒD e „ SŒg ; ; (9.91)
where the functional integration is over all Lorentzian 4-geometries and matter
fields acquiring the values h˛ˇ and h on ˙. The measure is, in general, ill-
defined and the integral diverges. One can restrict the sum to geometries with
Euclidean signature and perform a Wick rotation t ! it, so that one replaces
iS with the Euclidean action I, S ! ˙iI, and the integrand is e˙I=„ (the sign
depends on various considerations [68, 69]). However, this does not guarantee the
convergence of expression (9.91), which is too general to be manageable. In order
to find semi-classical solutions of the WDW equation (9.88), we can employ the
Wentzel–Kramers–Brillouin (WKB) approximation [70–73]. This approximation
finds applications in quantum mechanics, inflationary cosmology and cosmological
perturbation theory [74–77] and in the WDW equation [78–82].
Classically, the WKB method can be illustrated as follows. Consider an ordinary
differential equation in one dimension,
X
M
cm .t/ @m
t f .t/ D 0 ; cM .t/ D c 1 ; t > t0 :
mD0
We look for perturbative solutions of the form

" #
1X n
N
f .t/ ' fWKB D exp " fn .t/ ; N > M 1; (9.92)
" nD0
in the formal limit " ! 0. “Formal” means that, eventually, the expansion parameter
p finite and both c and " can be set to be O.1/.
remains
If " 1= t is the inverse of a large time interval t over which the system
evolves slowly, the expansion (9.92) is called adiabatic. For example, plugging this
Ansatz up to order N D 1 into the second-order equation
Œc@2t C ! 2 .t/ f D 0 ;
where ! 2 2 R can also be negative, one finds

c P 2 c c c R
0' 2
f0 C "fP1 C fR0 C "fR1 C ! 2 .t/ D 2 fP02 C f0 C 2fP0 fP1 C ! 2 .t/ ;
" " " "
where we have neglected the sub-leading term fR1 C fP12 . The leading-order term must
match ! 2 . Setting c D "2 , this yields
Z t
fP02 D ! 2 .t/ ) f0 .t/ D ˙i dt0 !.t0 / ;
t0
while the next-to-leading term vanishes if
fR0
fP1 D ) f1 .t/ D 12 ln !.t/ ;
2fP0
up to integration constants. Collecting these results, the approximated WKB

solution is
Rt 0 0 Rt 0 0
1 pi dt !.t / pi dt !.t /
fWKB .t/ D p b C e c t0 C b e c t0 ; (9.93)
!.t/
where b˙ are constants. Depending on the sign of ! 2 , at any time slice t the WKB
solution can be regarded as a superposition of plane waves or exponentials.
p
In quantum mechanical systems, the expansion parameter is " „ and pure-
phase WKB solutions are regarded as semi-classical. In this approximation, (9.91)
is peaked at a classical trajectory .h˛ˇ .x/; h .x// and one can neglect integration
over all possible 4-geometries and field configurations. Consider the Hamiltonian
constraint (9.88) and the trial wave-function
X X 1
D n D 0;n e „ .iSn In / ; (9.94)
n n
where 0;n are slowly-varying complex amplitudes and Sn and In are real-valued
functionals. We can assume that each mode n obeys the WDW equation separately.
Then, denoting with r covariant gradients in superspace (for a superspace scalar,
rA D @A ),

0 D 12 „2 r 2 C U n
1
D 12 r „2 r0;n C „0;n r.iSn In / e „ .iSn In /
1
D e „ .iSn In / 0;n 12 X1 C iX2 „X3 12 „2 r 2 0;n ; (9.95)
where the dot denotes the scalar product in superspace and
X1 D .rSn /2 .rIn /2 C 2U ; (9.96a)

X2 D rSn rIn ; (9.96b)
1 2
X3 D r0;n r.iSn In / C 2 0;n r .iSn In / : (9.96c)
Ignoring the O.„2 / term, the solution of (9.95) is found after setting
X1 D 0 ; X2 D 0 ; X3 D 0 :
Equation (9.96c) determines the real and imaginary part of 0;n once Sn and In are
known. These are fixed by (9.96a) and (9.96b).
In analogy with the quantum mechanical particle in a potential, if In D 0 one
can interpret the single mode n eiSn =„ as the wave-function of a post-big-
bang, sufficiently large universe in the classical oscillatory regime with Lorentzian
action Sn . If Sn D 0, the exponential wave-function n eIn =„ represents a
universe in the classically forbidden region across the potential, reachable through
a tunneling process described by the Euclidean action In . This suggests that we can
go to classically admissible regions in phase space if superspace gradients of the
imaginary part of the action are negligible with respect to the real part,
.rSn /2 .rIn /2 : (9.97)
Dropping the subscript n, we then consider a single semi-classical WKB mode

of (9.94) of the form7
i
D 0 e „ S : (9.98)
At this point we face a problem. In general, having dropped the exponential

factor eI=„ , this state is not square integrable. This implies that it does not
belong to the physical Hilbert space of square-integrable distributions and that
the probabilistic interpretation associated with wave-functions breaks down. For
the usual Schrödinger equation of non-relativistic quantum mechanical systems,
thanks to the time-evolution operator i„@t one can define Gaussian states, wave-
packets with a probability density j j2 peaked about a classical trajectory. In
this sense, Gaussian states are regarded as semi-classical. Then, one can easily
relate Gaussian and semi-classical WKB states, thus showing explicitly that semi-
classical WKB states are approximations of square-integrable functionals [80, 83].
In gravity this is not possible in a forthright way, since there is no explicit
time dependence in the quantum Hamiltonian constraint. WKB wave-functions
annihilate the Hamiltonian constraint and are thus “energy eigenstates with zero
energy.” Therefore, it is not clear in which sense semi-classical WKB states of
quantum gravity are approximations of physical states. In fact, there seems to be no
physical reason why j j2 should be zero at the boundary of superspace. On the other
hand, WKB states can give sensible descriptions of what a semi-classical universe
could look like and, a posteriori, it is reasonable to regard them as approximations
7
Very often in the literature, WKB wave-functions are defined as (9.98) (I 0), in which case
“WKB state” and “semi-classical WKB state” are one and the same thing. Here we keep the
distinction.
of yet-unknown (but hopefully existing) states in the full theory. In the absence of
tools performing the appropriate checks, the use of WKB states in quantum gravity
is ambiguous and requires a certain dose of faith, as well as caution in accepting the
ensuing results. In Sect. 11.5.2, we will see a remarkable example where such tools
are indeed available: in group field theory, WKB-like states can be constructed as
approximations of known normalizable states.
Having said that, let us study (9.98). The expression X1 D 0 becomes the
Einstein–Hamilton–Jacobi equation [78, 84]
.rS/2 C 2U D GAB rA SrB S C 2U D 0 : (9.99)
Comparing this expression with the classical constraint, one can identify the
classical momenta A with the gradient of the real action S,
A D @A S : (9.100)
Sometimes these are called Hamilton–Jacobi momenta. Define now the vector field
d
:D r A SrA : (9.101)
ds
Taking the superspace derivative of (9.99) and plugging in (9.100), one gets the
geodesic equation
0 D @A S@A @C S C 12 @A S@B S@C GAB C @C U

dC
D C 12 A B @C GAB C @C U : (9.102)
ds
The parameter s in (9.101) is called WKB time [85]. The signature of the DeWitt
metric already suggested the possibility to choose an intrinsic time coordinate in
superspace [86]. This definition depends on the choice of a solution h˛ˇ of the
Einstein equations but not on the choice of coordinates. However, the definition
of the vacuum state of the theory does depend on coordinate transformations, as it
happens in quantum field theories on curved spacetime [80].
Equation X3 D 0 in the classically allowed region (9.97) simply becomes
d0 1
C 0 r 2 S D 0 ; (9.103)
ds 2
yielding
Z
1
0 D A exp dsr 2 S ; (9.104)
2
where A 2 C is an integration constant. Summarizing, the semi-classical WKB

state (9.98) is
Z
i 1
D A exp S dsr 2 S : (9.105)
„ 2
This wave-function is characterized by the precise correlation (9.100) between
conjugate variables [80, 83]. It is the analogue of a wave-packet peaked about a
classical particle trajectory in ordinary quantum mechanics. However, a superposi-
tion of WKB states is no longer semi-classical and, as in the case of inflationary
perturbations in Sect. 5.6.1, one could invoke a decoherence process [87].
Semi-classical states can be useful to understand how to track down quantum-
gravity effects in the empirical world. However, classical gravity is already too
weak to affect particle physics in ordinary laboratory set-ups and WDW canonical
quantum gravity corrections are even smaller. To give a quantum mechanical
example, these corrections alter the Schrödinger equation and, in particular, modify
the spectral lines of hydrogen-type atoms; the effect is unobservable [81]. In this
respect, as we have already emphasized in early occasions, cosmology of the early
universe and black-hole physics are generally regarded as more promising test
grounds for quantum-gravity models.
9.2.3 Boundary Conditions
The Wheeler–DeWitt equation does not encode a unique quantum dynamics and
one must specify boundary conditions for the wave-function of the Universe (9.91),
fixing a pre-classical geometry. Various boundary conditions have been put forward,
chiefly the Hartle–Hawking no-boundary proposal [88] and the Vilenkin tunneling
proposal [89], among others [90–92].
The Hartle–Hawking no-boundary proposal assumes that the fundamental for-
mulation of (9.91) is in Euclidean signature and that Lorentzian spacetimes emerge
from Riemannian quantum configurations only in certain cases. Second, the path
integral in (9.91) is estimated over complex 4-geometries such that the Lorentzian
universe emerges from compact Riemannian configurations [62, 68, 88, 93–105].
Thus, the “initial” boundary of the path integral is a Riemannian manifold with
no boundary at all, while the final one is the present universe. A typical initial
configuration is half the 4-sphere S4 , “sewn” to de Sitter spacetime [89]. Contrary
to initial expectations, the no-boundary prescription does not fix the solution HH
of the WDW equation uniquely [96, 97]. A density matrix approach describing a
microcanonical ensemble of cosmological models generalizes the vacuum state of
the Hartle–Hawking proposal to a quasi-thermal ensemble [106–108].
The Vilenkin tunneling proposal pictures the nucleation of a bubble universe
from nothing by quantum tunneling [68, 69, 89, 98, 99, 109–113]. The Lorentzian
path integral (9.91) is summed over compact 4-geometries with the restriction to
solutions V with only “outgoing” modes at singular boundaries of superspace. For
a WKB mode eiSn , the “outgoing” condition is with respect to WKB time (9.101)
Fig. 9.1 The no-boundary (thin solid curve) and the real part of the tunneling (dashed curve)
wave-functions in a cosmological simplified setting described in Sect. 10.2.1, for D 1 D H. The
no-boundary wave-function is damped through the superspace potential barrier U (thick curve),
while the other tunnels through it
and requires that the superspace vectors rSn point out of superspace at the
boundary. Lorentzian configurations beginning from degenerate .3 C 1/-geometries
are thus excluded, since the nucleation point is non-singular; still, allowed solutions
can end at a singularity. A cosmological example [89, 109] illustrates the point.
The solution of the Friedmann equation (2.81) for a closed universe (K D 1)
and a cosmological constant is (2.111). The universe bounces at t D 0 after an
infinite period of contraction. The cosmological constant is, more generally, the
false vacuum of a scalar field (i.e., a local maximum of its potential), which later
decays into true vacuum. Analytic continuation t ! it leads to the Hawking–Moss
instanton [56, 114]
a.t/ D H 1 cos.Ht/ ; (9.106)
where jtj < =.2H/ and which describes the four-sphere S4 of radius H 1 .8 As
in quantum mechanics, the solution of the classical Euclidean equations represents
tunneling from a barrier at t D 0. In this simplified setting, the path integral (9.91)
is peaked about the de Sitter instanton and the wave-function V tunnels through
the superspace potential barrier U. In the Hartle–Hawking proposal, on the other
hand, the wave-function HH is damped through the barrier, the universe does
bounce at t D 0 and the evolution is symmetric in time. Also, while the no-
boundary prescription yields real wave-functions, the tunneling proposal in general
does not. Figure 9.1 shows the general behaviour of HH and Re.V /, in a typical
cosmological model which we will describe in Sect. 10.2.1.
8
For the inquisitive reader, we notice that the profile (9.106) is periodic with period 1=TH D 2=H.
Its inverse is precisely the de Sitter temperature (5.132) giving the size of a field fluctuation. An
explanation of this fact can be found in [115, 116].
9.3 Some Features of Loop Quantum Gravity 439
9.3 Some Features of Loop Quantum Gravity
Modulo some important differences, a superspace can be defined also for Ashtekar–
Barbero variables. In this case, geometry is not necessarily associated with a metric
and the first-order superspace can include also a degenerate sector of triads and
connections not corresponding to metric structures. Moreover, the quotient space is
taken over the group SU.2/ of small gauge transformations (this is isomorphic to
the universal cover of SO.3/, Spin.3/ Š SU.2/).
The connection itself cannot be promoted to a well-defined operator; only the
holonomy of the connection can be consistently quantized. The holonomy along an
oriented path e is an element of the SU.2/ group defined as
Z
he :D P exp d e˛ ./Ai˛ ./ i ; (9.107)
e
where e˛ ./ D dx˛ =d is the tangent vector along the path parametrized by
and i , i D 1; 2; 3, are the three generators of the su.2/ algebra in irreducible j-
representation (of dimension 2j C 1). Expanding the exponential, the path ordering
P cancels the factor 1=nŠ in the Taylor series, so that for a path of length e
X Z e
C1 Z e Z e Y
n
he D 1 C d1 d2 : : : dn e˛ .k /Ai˛ .k / i : (9.108)
nD1 0 1 n1 kD1
Holonomies describe how the connection is transported along any given path. If
the path is closed, he is called a loop. In the original formulation of LQG, the
kinematical Hilbert space was spanned by a basis of loops, hence the name of the
theory.
For technical convenience, in the quantization of the Hamiltonian constraint
one can fix the representation to be the fundamental one, i.e., j D 1=2. This
choice may be justified also by theoretical considerations,9 which however are not
compelling and can be bypassed. In this representation, one introduces the Pauli
matrices (5.208), so that the generators are
i ij k k ıij
i D ; i j D 12 ; (9.109)
2i 2 4
9
In the high-j case, the Hamiltonian constraint is a difference equation of higher-than-second
order. This may lead to an enlargement of the physical Hilbert space and, as a consequence, to
the presence of solutions with incorrect large-volume limit [117]. Even if this were not the case,
there is evidence (in 2 C 1 dimensions the proof is actually complete) that LQG has a well-defined
continuum limit to quantum field theory only in the fundamental representation of the gauge group
[118].
where 12 is the 2 2 identity matrix. We will not fix the spin label j unless stated
otherwise.
The quadratic Casimir invariant C2 in representation j is given by the algebraic
relation
X
i i D C2 . j/12 D j. j C 1/12 : (9.110)
i
The variable conjugate to the holonomy is not the triad by its flux through an
arbitrary two-surfaces ˙:
Z
EŒ˙ D d2 y E˛i n˛ i ; (9.111)
˙
where n˛ is the normal to ˙. These are interpreted as elementary areas.

The spatial volume of a given space can be expressed as
Z p
VD d3 x j det Ej : (9.112)
The determinant (9.61) of the densitized triad appears in the Hamiltonian con-
straint (9.60) as an inverse power. In quantizing inverse powers of the volume, one
faces the problem that naive inverse operators .V/ O s , with s > 0, are not well-
defined. As a standard procedure, one rewrites inverse powers of the densitized
triad using classical Poisson brackets involving positive powers of V [119, 120].
For instance, in (9.60) we want to rewrite the term
Ei˛ Ej ij
k F˛k :
2 2 jej
Starting from (9.8) and the property (9.7), one can prove Thiemann identity
Ei˛ Ej ˇV
1q
ıV q
ijk p D 2 ˛
j det Ej q ıEˇ
k
4 X 0
D 2
V 1q ˛ ˇ
eiˇ tr. k hi0 fh1
i0 ; V g/ ;
q
(9.113)
ql0
i0
where we introduced an ambiguity parameter q 2 R and in the second equality we

used the expression hk ıh1 ˛
k =ıA˛ D l0 ek i , where l0 is the length of the holonomy
i
path.
The field strength F˛k can be manipulated as follows. One considers the loop
hij :D hi hj h1 1
i hj on a closed oriented square path whose edges have length l0
and are labelled by spin indices i; j; i; j. Then, from (9.108) for a closed path and
9.3 Some Features of Loop Quantum Gravity 441
applying Stokes theorem one gets [121, 122]
tr. k hij /
F˛k D 2ei˛ ej lim : (9.114)
l0 !0 l20
The field strength is a local function of spacetime coordinates and is obtained in the
limit where the area of the plaquette ij tends to zero. Overall,
Ei˛ Ej 4 V 1q X 0
tr. k hij / iji tr. k hi0 fh1
ij
k F˛k D lim 3 i0 ; V g/ :
q
2 2 jej 4 l0 !0 ql
0 i;j;k;i0
(9.115)
The kinematical states of LQG are spin networks, graphs in an embedding space
whose edges e are labeled by spin quantum numbers je . Edges meet at nodes in
sets of three; at each node, the labels of group elements are governed by a map
SU.2/ ˝ SU.2/ ! SU.2/, called intertwiner. Physical states are based upon
the kinematical Hilbert space of the theory and are annihilated by the quantum
constraints. We leave this analysis to specialized literature [52, 123, 124], limiting
this section to the barest details we shall need when discussing the cosmological
version of the LQG quantization scheme (Sect. 10.3).
A key feature of loop quantum gravity, which is not an assumption but a
consequence of the full theory, is that the spectrum of the area operator is bounded
from below by the Planck scale. In particular, a direct calculation shows that the
lowest area eigenvalue on any gauge-invariant state is [124, 125]
p
Q Pl :D 2 3 `2Pl : (9.116)
This result can be roughly illustrated as follows. A given two-dimensional

R surface ˙
is divided into elementary cells ˙e on which a flux Ei Œ˙e D ˙e dy1 dy2 n˛ .y/Ei˛ .y/
is defined, where y1;2 are coordinates over ˙e and n˛ is the normal to the surface.
The classical area of ˙ is given by the surface integral of the induced two-metric
and the sumpof the areas of the ˙e . Each area is determined by the fluxes through it,
Ae D 1 Ei Œ˙e Ei Œ˙e . Upon quantization, each ˙e is pierced by the edge of an
“adapted” spin network (i.e., the edge starts at ˙e , or is embedded in ˙e , or it has
empty intersection with ˙e ). The quantum number determines the spectrum Ae of
the area operator AO associated with an elementary plaquette intersecting only one
edge e:
p
Ae D `2Pl je . je C 1/ : (9.117)
The geometrical size of the plaquette changes only when the latter intersects another
edge, thus increasing in quantum jumps. Therefore, the action of two flux operators
over an adapted edge gives the quadratic Casimir invariant of the algebra su.2/,
which takes discrete values and is non-zero in the fundamental representation
je D 1=2. Multiplying (9.117) by a solid-angle factor of 4 yields (9.116). Although

this is only a kinematical result, it makes one hope that physical singularities such
as those found inside black holes and at the birth of the Universe be resolved in the
quantum setting of LQG.
The value of can be constrained by the computation of the entropy of
non-rotating black-hole isolated horizons [126–128]. Matching the LQG result
with Bekenstein–Hawking’s semi-classical area law fixes the Barbero–Immirzi
parameter to be [128]
0:2375 : (9.118)
This is the only available determination of and it would be highly desirable to

obtain further input from LQG calculations performed in independent physical set-
tings. In loop quantum cosmology, the Barbero–Immirzi parameter is left arbitrary,
its value affecting (but in a mild way) only the energy scale of the quantum bounce.
Note, however, that the black-hole area law is recovered also in group field theory,
a powerful Lagrangian extension of canonical quantum gravity (Sect. 11.5), but for
any value of the Barbero–Immirzi parameter [129]. The difference is due to the
fact that in group field theory one uses large superpositions of graphs (condensate
states), while in LQG one picks a very special choice of states, the eigenstates of the
black-hole horizon area.
Loop quantum gravity and other approaches of discrete gravity have often
inspired the following naive objection. One starts from a theory defined on a
smooth manifold where classical general relativity applies. Then, one quantizes
and hopes to obtain something like a continuum limit for geometry. Is this not
a snake biting its tail? The answer is No and is exactly the same as for ordinary
quantum mechanics. When we say that “a system is quantized” we do not mean that
a classical physical system is transformed into something else. Rather, quantization
means to define a set of operators on a Hilbert space (or a tower of spaces, if we
consider also the kinematical spaces instrumental to the construction of the physical
Hilbert space) in a certain representation. Everything which happens before we
complete the construction of these operators and Hilbert space is a mathematical
framework. For instance, the embedding manifold of LQG is just an auxiliary
object with no physical significance. Only after quantizing can we start talking
about physics, because the assumption is that the interaction we are studying is
intrinsically quantum.
9.4 Cosmological Constant Problem
Canonical quantum gravity in the ordinary Wheeler–DeWitt quantization faces the

cosmological constant problem from a probabilistic point of view. Results in this
direction can be obtained in a cosmological setting, where the number of degrees
of freedom of the full theory is reduced and the problem becomes tractable. We
9.4 Cosmological Constant Problem 443
still need to introduce many ingredients of this construction and we postpone

the discussion to Sect. 10.2.4. Here we concentrate on -related aspects of loop
quantum gravity and its modifications. At the time of writing, loop quantum gravity
has not been fully formulated in the presence of a cosmological constant, which
deforms the internal group to a quantum group and therefore complicates the
structure of the theory; attempts to relate such quantum-group structure to the one
in spin foams are fairly recent [130, 131]. In parallel, proposals to solve or relax the
problem in LQG have been preliminary but, interestingly, none of them invokes
symmetry reduction to a cosmological background. Results in this direction have not
been supported by much study [132–134] and are to be considered as speculative
but they provide a case, among recent others, of cross-fertilization between quantum
gravity and ideas taken from condensed matter physics. This section can be skipped
on a first reading.
9.4.1 Chern–Simons State
Spin-network solutions of the constraints with a cosmological constant have been

sketched in [132]. Their small number suggests that the state of the Universe should
have been initially one with D 0. Such indication is not incompatible with the
current notion that well-defined solutions of the constraints with a cosmological
constant are difficult to find. LQG solutions of the full theory are, in fact,
incompletely understood in the presence of a term.
For simplicity, we work in complex Ashtekar variables, such that the connection
is (9.53) with D i. In particular, in the absence of matter, the classical
ˇ
Hamiltonian constraint (9.60) simplifies to H D ŒP˛i Pj =.2 2 jej/ k F˛ˇ
ij k
, which
is remindful of a gauge theory. A formal solution of the quantum constraints in
vacuum is the Chern–Simons (or Kodama) state [135–139]:
Z
3m2Pl
CS ŒA D C exp YCS (9.119)
2 M3
in Lorentzian spacetime, where A is the self-dual connection, M3 is the three-

dimensional spatial sub-manifold and YCS is the Chern–Simons form [140]

1 2
YCS :D tr A ^ dA C A ^ A ^ A
2 3

D d3 x ˛ˇ Ai˛ @ˇ A i C 13 ijk Ai˛ Aˇ Ak :
j
(9.120)
The trace is taken in the adjoint representation, where the su.2/ algebra generators
are fijk D ijk . The normalization factor N is independent of both the triad and the
connection but it can depend on the topology, as it will be soon demanded.
The Chern–Simons state is not only a WKB state, but also a genuine ground state
of the theory, inasmuch as by adding a matter sector its weakly-coupled excitations
reproduce standard quantum field theory on de Sitter background [141], while
linearizing the quantum theory one recovers long-wave-length gravitons on de Sitter
[139]. However, there are several issues associated with this state. CS is solution to
the constraints only for a special operator ordering such that the super-Hamiltonian
is not Hermitian. Moreover, it depends on the self-dual Ashtekar connection, not on
the real Ashtekar–Barbero connection (9.53) by which the actual theory is defined.
This makes the whole construction of the state (9.119) rather formal, even when
re-expressed in terms of holonomies. On top of that, CS is not normalizable. This
is expected because states satisfying the constraints are generally non-normalizable
in the inner product on the kinematical Hilbert space. An immediate consequence
is that it does not make sense to calculate the probability jCS j2 that the Universe
realizes a configuration with D 0, a datum which could have been compared
with the arguments about spin-network solutions [132]. Last, CS violates CPT
symmetry, which may result in negative energy levels and a Lorentz violation [142].
Eventually, these problems may be solvable. Hermiticity is a requirement only
for excited states in a quantum mechanical theory and it is not necessary in
vacuum gravity, where one is interested only in the ground state. Next, in the
self-dual formulation there are many perturbative self-consistency checks which
can be performed on the mathematical structure emerging from the Chern–Simons
state [138, 143]. Further, the latter can be generalized to a real Immirzi parameter
(real connection) [144–146]. This state solves all the quantum constraints, is ı-
normalizable (i.e., h j i D ı) just like momentum eigenstates in quantum
mechanics (since de Sitter space is unbounded, the corresponding wave-function
cannot be normalizable as h j i D 1) and invariant under large-gauge and
CPT transformations (it violates CP and T separately). Last, many key features of
quantum mechanics and quantum field theory were discovered even before a Fock
or Hilbert space were defined. Therefore, although there is no rigorously known
kinematical Hilbert space for self-dual gravity with Lorentzian signature, one can
still work in complex variables and make generic predictions about the quantum
theory.
The Chern–Simons wave-function can give a useful insight in non-perturbative
properties of the theory. Since it is a WKB state, even if one does not believe it to
lie in the physical Hilbert space of the full theory, at some level a true quantum state
close to the de Sitter ground state must be approximated by CS reasonably well.
Taking these caveats on board, one may illustrate how to relate the problem to the
topological structure of the theory.
As mentioned in Sect. 7.2.1, the full gauge group of gravity G is made of
diffeomorphisms, but it contains also the group G of small as well as large gauge
transformations in internal space. In the case of the complex connection formulation
[4, 5], G D SU.2/. Under a local gauge transformation, the Ashtekar connection
transforms as
A ! A0 D gAg1 g1 dg ; g.x/ 2 G : (9.121)

Let G0 G be the sub-group of small gauge transformations, local transforma-

tions continuously connected to the identity. Its elements are of the form g0 D
expŒi i i .x/, where i are the generators of the su.2/ algebra and i .x/ are
some functions on a spatial slice of the four-manifold. Pure gauge configurations
Ao D g1 dg are equivalent to the flat gauge A D 0. Invariance of the exponent
in (9.119) under small gauge transformations requires the integration manifold
M3 to have no boundary. It turns out that M3 must have constant curvature in
the semi-classical limit, selecting the 3-sphere S3 as the only classically viable
alternative [143].
On the other hand, large gauge transformations are unitary transformations
g.x/ which are not homotopic to the identity but tend to the identity at large x,
limjxj!1 g.x/ D 1. All directions in Euclidean space are identified at infinity, so
that g is a mapping from M3 Š S3 to S3 . The different ways in which the sphere
S3 can be continuously mapped onto itself are summarized by the homotopy group
3 .S3 / D Z. Therefore, each disconnected component Gn G is labelled by an
integer n, called winding number, generated out of a pure gauge connection Ao in
that component,
Z
I.Ao / 1
w.Ao / D 2
D tr.Ao ^ Ao ^ Ao /; (9.122)
24 24 2 S3
where I is the Cartan–Maurer invariant [147]. The winding number characterizes

how many times g.x/ winds around the non-contractible 3-sphere in the SU.2/
internal space as x ranges over S3 in space. If g D g0 2 G0 , the winding number is
zero, while all large gauge transformations fall into homotopy classes gn generated
by the n D 1 transformation: gn .x/ D Œg1 .x/n . Then, the quotient G=G0 is
isomorphic to the homotopy sequence of integer-valued winding configurations
f.g1 /n j n 2 Zg.
This is the origin of topological vacua in quantum Yang–Mills theory
[148, 149] and gravity [150, 151]. On each disconnected component of G, there
lives an independent physical sector (an inequivalent quantum theory) with ground
state ŒA D hAjni. When promoted to a quantum operator, a unitary large gauge
transformation jumps from a sector to another, acting as gO 1 jni D jn C 1i. By
definition, physical observables are invariant under the full gauge group, so that gO 1
must commute with the Hamiltonian and they share the same basis of eigenstates,
chosen so that the eigenvalue of gO 1 on a state is a pure phase ei , 0 6 6 2. In
other words, under a gauge transformation as in (9.121), ŒA0 D ŒA if g is small
(g D g0 ), while ŒA0 D ein ŒA if g D gP
n is large. This implies that the quantum
vacuum is a superposition of states j i D n ein jni.
In the case of quantum chromodynamics, the sector appears as parity-breaking

terms for the gauge field and the fermions. In particular, the classical Yang–Mills
Lagrangian
1 X i i
LF D F F ; i
F :D @ Ai @ Ai C f ijk Aj Ak ; (9.123)
4 i
receives a correction of the form

Q X
i Q i
LF ! LF F F ; (9.124)
32 2 i
where FQ i D Fi

=2. Classical dynamics is unaffected, since the new term is
topological, i.e., it can be written as a total derivative. At the quantum level, however,
topological terms do change the dynamics. But then (9.124) creates a problem, since
there is no experimental evidence that chromodynamics violates the CP symmetry.
To solve this “strong CP problem,” the QCD partition function is augmented by a
U.1/ symmetry, namely, the Peccei–Quinn symmetry corresponding to invariance
under a rotation by the angle [152, 153]. This procedure corresponds to promoting
the Q parameter to a field , a light particle called the axion [154, 155]. The
Lagrangian (9.124) is modified by Q ! Q and augmented by a kinetic term
/ @ @ . Instanton effects spontaneously break the U.1/ symmetry associated
with the axion by introducing a potential with a minimum at ' Q , thus canceling
or suppressing the CP-violating term.
For self-dual gravity, something similar happens. At the level of the action, a
sector is, on solutions to the field equations, realized by adding the self-dual
Pontryagin term L Q / i Q R RQ , which is topological. In the classical super-

Hamiltonian, there appear parity-violating terms at linear and quadratic order in Q

[156]. At the quantum level, theRChern–Simons R form transforms non-trivially under
large gauge transformations, as YCS ! YCS C 4 2 w:
CS ŒA ! CS ŒA0 D ein CS ŒA ; n 2 Z; (9.125)
where
6 2 m2Pl
i D : (9.126)

The parameter is indeed a phase in the Euclidean case, where instantonic solutions
are considered. Invariance of the state under large gauge transformations can be
achieved by setting the normalization constant in (9.119) to be C D eiw.A / [143,
o
157]. The dependence is absorbed in the state normalization but reappears in the
inner product, giving inequivalent quantum probabilities.
9.4.2 as a Condensate?
At this point, in analogy with the Peccei–Quinn mechanism in QCD, we modify

the theory and promote to an evolving functional .A/ [133, 134]. The Peccei–
Quinn mechanism (and its pseudo Nambu–Goldstone boson) has been proposed
to explain the observed cosmological constant in quantum field theory [158–160],
but in the present context it is directly related to gravitational quantum degrees
of freedom. The operation ! .A/ is recognizable, thanks to (9.126), as a
deformation of the topological sector of the quantum theory: ! .A/. A large
gauge transformation (9.125) in the Chern–Simons wave-function is regarded as
a U.1/ rotation, so that by allowing to vary as a function of the connection
we explicitly break the U.1/ symmetry. Deforming leads to an altogether new
theory, which reduces to ordinary loop quantum gravity when the deformation is
very weak (perturbative limit). An expansion of the new ground state and of the
total Hamiltonian in a perturbative parameter must yield, at leading order, just LQG
with a cosmological constant. This is sufficient to justify the application of the
deformation procedure to a constraint and a to state of the usual theory, obtaining
some new constraint annihilating a new state. In other words, one modifies the
super-Hamiltonian constraint by adding a counter-term breaking gauge invariance.
The Chern–Simons state will also be deformed because, according to (9.125)
and (9.126), enters in the phase of the wave-function. As the Chern–Simons state
lives in the connection space, we assume that in the deformed state depends
only on A and not on the triad operator.
The new state turns out to be compatible with all the constraints only in two
cases: either .A/ is invariant under small gauge transformations or det E D 0,
leading to a degenerate sector of gravity with rkE 6 2 (rk denotes the rank of the
triad). The special case rkE D 1 gives rise to an intriguing structure. Classically,
this configuration has been studied by Jacobson [161]. Geometry amounts to a
collection of parallel gravitational lines where the triad Ei˛ has rank 1 and vanishes
elsewhere. Spatial diffeomorphism invariance can be partially fixed by spreading
the gravitational lines along the z direction. Then, the only non-zero component of
the electric field is Eiz . Solving the classical Gauss constraint and fixing the gauge,
one ends up with Aiz D 0, A3a D A3a .xa /, Ai¤3 a D Ai¤3
a .t; z/, Ei¤3 D 0 D Ei and
z a
E3 D E3 .x /, where a D 1; 2 and the last equality stems from the Gauss constraint
z z a
and the equation of motion coming from the super-Hamiltonian:

PAi˛ D i ijk Eˇj F˛ˇ
k
C ˛ˇ E k
; (9.127)
2
where the presence of a cosmological constant is, at this stage, irrelevant. The
residual freedom in the choice of the transverse coordinates xa can be used to fix
E3z D 1. The equation of motion (9.127) for the transverse-transverse components
of the connection is AP ia D i i3j @z Aja , which can be written as a .1 C 1/-dimensional
Dirac equation at every point xa in the transverse plane [133, 134]:

0 P C z
@z D 0; (9.128)
where
01
iA11
B A1 C
:D B 2 C
@ A2 A (9.129)
1
iA22

and are the Dirac matrices in the Dirac basis (compare with (5.209))

0 0 0 ˛ 0 ˛
D ; D : (9.130)
0 0 ˛ 0
The causal structure of the degenerate sector is, from (9.128), that of a world-
sheet, a .1 C 1/-dimensional spacetime. Rotating back in the target space, (9.128)
can be put in covariant form as @ D 0. One can take several differently
oriented gravitational lines and patch them together at their boundaries [161].
The emerging classical picture would be that of two-dimensional world-sheets
interacting at the edges, where the gravitational field may be non-degenerate.
However, at the classical level one lacks a model for this interaction, as well
as a physical interpretation of the world-sheet network and fermionic degrees of
freedom.
Both naturally emerge at the quantum level when the deformed constraints
act upon the Chern–Simons state. To see this, consider that we are working in a
semi-classical approximation, so that the counter-term in the quantum Hamiltonian
constraint can be interpreted as a deformation also of the classical equation of
motion (9.127), which reads
Z
ı ln .A/
AP i˛ D i ijk Eˇj F˛ˇ
k
C ˛ˇ E k
˛ˇ YCS : (9.131)
2 M3 ıA k
In Jacobson’s degenerate sector, this equation can be manipulated as in the classical

case. To proceed, one has to make an Ansatz for the functional .A/. We choose
. / D 0 exp. t /, where 0 D O.1/ is a constant and t D .1 ; 2 ; 3 ; 4 /
depends on A and is, for the time being, arbitrary. Then, (9.131) yields 0 P C
z
@z C im x D 0, where is given by (9.129) and m :D i t x z @z is an
effective mass. Imposing the condition D x , the field obeys the Dirac
equation
0 P C z
@z C im D 0: (9.132)
Classically the mass vanishes, as t D 0. However, after quantization the field

components become Grassmann variables and the symmetry-breaking mechanism
described at the beginning of this section comes into effect. In order to get quantities
with a well-defined Lorentz structure, t should actually be related to N :D " 0 .
This is realized if D c :D i y , i.e., if is equal to its charge conjugate
(Majorana fermion). Then, the connection components must satisfy the conditions
A21 D A1 2 1
2 , A2 D A1 . In particular,
t
D t x D N 5 z D: jz5 , where
5 :D 0 x y z
i (see Problem 9.1) and becomes
D 0 exp.jz5 / : (9.133)
The cosmological constant encodes the imprint of an axial vector current j˛5 ,
5
associated with a chiral transformation ! ei and not conserved in the
presence of the effective mass
m D i N 5
@z : (9.134)
This is yet another reminder that the deformation process affects the topological
sector and the CP symmetry of the theory.
Before applying (9.133) to the cosmological constant problem, its physical inter-
pretation must be sharpened. Equations (9.132) and (9.134) are the starting point of
a scenario which has loop quantum gravity as an effective limit. One begins with a
spinor field in two dimensions and canonically quantizesP it as a fermion. Expand
it in discrete one-dimensional momentum space, D k; eikz .2Ek /1=2 .ck uk C
"
ck vk /, where D ˙ is the spin, Ek is the energy, c and c" are annihilation and
"
creation operators obeying a fermionic algebra fck ; ck0 0 g D ıkk0 ı 0 and u and v
are spinorial functions. Due to (9.134), there appear four-fermion non-local (i.e.,
dependent on non-coincident points) correlations in the Hamiltonian, of the form
X " "
Vkk0 ck ck0 0 ck ck0 0 ; (9.135)
;k;k0
where Vkk0 is a function of the momenta determined by the effective mass (9.134).
Spin models of the form (9.135), generically called Fermi-liquid theories, are
employed in condensed matter physics to describe fermionic systems with many-
body interactions at sufficiently low temperature (see, e.g., [162–164] for introduc-
tory reviews). In the mean-field approximation Vkk0 D g Dconst, the interaction
is effectively independent of the momenta. This is the Bardeen–Cooper–Schrieffer
(BCS) model of superconductivity [165–170]. Fermions with opposite spin can
interact non-locally in N pairs at a given energy level, lower than the Fermi energy
EF of the free-fermion sea. Let Epair be the binding energy of the two-body bound
state in vacuum. At weak coupling (small binding energy Epair EF ), Cooper pairs
are weakly bound by an attractive potential and may overlap, forming a condensate
with superconducting properties. This lowest-energy state jBCSi is regarded as
the true vacuum of the theory, whose energy is separated from the perturbative-
vacuum energy by a mass gap e1=g . At strong coupling (large binding energy
Epair EF ), pairs are tighter and do not overlap, so that they can be treated as
a gas of non-interacting bosons, which is expected to undergo a Bose–Einstein
condensation (BEC) [171–177]. The transition between the two regimes (called
BEC-BCS crossover [178, 179]) is smooth and well described by a weak-field
approximation.
One can check that the gravitational analogues of the BCS wave-function jBCSi
and mass gap are the Chern–Simons state and the cosmological constant. As
we will see in Chap. 12, a world-sheet is invariant under conformal transformations
and one can build a conformal field theory (CFT) on it. The quantum degenerate
sector of the present gravity model is described by world-sheets interacting at their
boundary. Indeed, one can place a CFT on each of these world-sheets. However,
conformal invariance is broken at the boundary by a marginal operator encoding
the interaction, which results in a deformed CFT. It turns out that this deformed
CFT is fermionic in nature, as the classical picture suggests, and describes a Fermi
liquid. On the other hand, its geometric and algebraic structures reproduce those of
a quantum (or framed) spin network (e.g., [139]), which is the natural environment
wherein to embed the Chern–Simons state. This and other pieces of evidence lead
to the conjecture that the degenerate sector has a direct interpretation in terms
of framed spin networks, so that the .1 C 1/-dimensional theory holographically
generates the full four-dimensional spacetime via quantum interactions.
The edges of quantum spin networks are two-dimensional tubular surfaces and
vertices are promoted to punctured two-spheres. BCS levels are mapped to edges
of quantum spin networks, interacting at nodes via a BCS coupling. The BCS
interaction is the cornerstone of the construction of three-dimensional geometry
from quantum spin networks. Classically, it describes scattering of Jacobson electric
lines at their end-points; these world-sheets, patched together at their edges, span
the three-dimensional space, giving rise to the geometric sector which was lost
in the free-field picture. The screening charges at a given node have an intuitive
picture as the sites activated in an area measurement, i.e., when a classical area
intersects the spin network. In condensed matter physics, the Fermi sea is a ground
state of uncorrelated electron pairs whose Fermi energy is higher that the BCS pair-
correlated state. In quantum gravity, pair correlation can be regarded as a process
of quantum decoherence. An abstract spin network is the gravity counterpart of an
unexcited Fermi sea. As soon as an area measurement is performed on the state,
N edges (as many as the number of Cooper pairs) of a given node are activated
and the system relaxes to a lower-energy vacuum corresponding to the selection of
one of the area eigenstates in a wave superposition. In a sense, measuring quantum
geometry means counting Cooper pairs. This picture [134] is actually more general
than the BCS case and gravity possibly allows for a dual description in terms of a
more generic non-locally interacting Fermi liquid.
A natural 3 C 1 embedding structure arises from the quantum interaction of
electric lines, but it is still an open issue how to precisely recover the usual structure
of LQG, i.e., the bosonic statistics of connection variables living in 3C1 dimensions
from the fermions describing the 1 C 1 quantum degenerate sector. Perhaps, the
physics of the BEC-BCS crossover could play an important role. If this scenario
could be clarified, one would benefit from a possible relaxation of the cosmological
constant problem. In fact, (9.133) can address the smallness problem in terms
of a condensate with vacuum expectation value (with respect to the deformed
Chern–Simons state ) h jz5 i D O.102 /. In the perturbative regime (small values
of the connection, jh jz5 ij 1), hi ' 0 .1 h jz5 i/ D O.1/. In the non-
perturbative regime, the effective mass becomes important and the cosmological
constant, supposing h jz5 i to be positive definite for large connection values, becomes
exponentially small. In this case, the smallness of the cosmological constant would
be regarded as a large-scale non-perturbative quantum mechanism similar to quark
confinement.
9.1 Barbero–Immirzi field and torsion 1. Consider the Hilbert–Palatini

action (9.21) augmented by the Nieh–Yan term (9.23), but with 1= D ˇ !
ˇ.x/ now inside the integral:
S D SHP C SNY;ˇ
Z
1
D 2 d4 x j.4/ej ea eb
2

ˇ
Rab C ab T T abcd Rcd : (9.136)
2
Plug (9.15), (9.17) and (9.18) into the action and obtain an expression in terms
of the irreducible torsion components. Vary the action with respect to these
components and pull back the resulting equations into the action to obtain an
effective action. Is ˇ a scalar o a pseudo-scalar field?
Solution The Hilbert–Palatini piece becomes

Z
1
SHP D d4 x j.4/ej ea eb RN ab C Kac Kcb Kac Kcb
2 2
Z
1 4 .4/ N ab 1 2 1
D 2 d x j ej ea eb R C S S T T C q q ;
2 24 3 2
(9.137)
where in the first line we have dropped out a total divergence and quantities with an
overbar are torsion-free. In particular, rN is the torsionless and metric-compatible
covariant derivative. The Nieh–Yan contribution is rather simple, as a number of
cancellations take place between the Holst and the torsion-torsion term:
Z
1
SNY;ˇ D d4 x j.4/ejˇ rN S : (9.138)
4 2
By varying the action with respect to the irreducible components of torsion S , T

and q , we obtain, respectively,
@ ˇ 16 S D 0 ; T D 0 ; q D 0 : (9.139)
After reinserting the solutions above into (9.136), we get the effective action
Z
1 3
Seff D d4 x j.4/ej ea eb RN ab @ ˇ@ ˇ : (9.140)
2 2 2
Therefore, the system is equivalent to Hilbert–Palatini torsion-free gravity plus a

canonical massless scalar field
r
Q 3 Q D 1:
ˇ :D ˇ; Œˇ (9.141)
2 2
Notice that Seff preserves parity because ˇ is a pseudo-scalar. To see this, project

the first equation in (9.139), @a ˇ :D ea @ ˇ D Sa =6. @a is a polar internal vector,
but the pseudo-vector component of the torsion Sa D abcd T bcd is defined via the
Levi-Civita symbol. Therefore, ˇ is a pseudo-scalar. If we had chosen the volume
form with .4/e instead of j.4/ej, the classical decomposition of torsion according to
the Lorentz group would have lost its meaning.
Another way to reach the same conclusion is by adding fermions to the action.
Then, the Barbero–Immirzi field obeys a massless Klein–Gordon equation with a
fermionic axial bilinear as a source [44]:
ˇ / N 5
; (9.142)
5
where the matrix is defined as
5 i
D abcd a b c d
: (9.143)
4Š
Since a parity operation would change only one of the a ’s, 5 changes sign and
hence it is a pseudo-scalar. As spinors are internal scalars, this yields the desired
result. With the volume form without absolute value of the determinant, one would
have been obliged to change the definition of 5 , but this would ultimately spoil
the usual notion of parity in quantum field theory. Since the physics on Minkowski
spacetime is a collection of local experiments conducted in a given frame, the notion
of parity must be consistent. We are then forced to assume (9.8) and it then follows
that this choice is the only self-consistent one when matter is included in the theory.
9.2 Barbero–Immirzi field and torsion 2. Perform the Hamiltonian anal-

ysis of S D SHP C SNY;ˇ given by (9.137) and (9.138). Set 2 D 1. Fix the
tensor q to zero since it is non-dynamical and does not contribute to the
torsion tensor, as it was clear from (9.139). Setting q D 0 does not affect
the generality of the formulation and it has the advantage of simplifying the
canonical analysis, which is rather involved for 3tensors.
Solution Following step by step the calculation of Sect. 9.1.2, the action S D SHP C
SNY;ˇ can be written as follows:
Z
1
SD dt d x j ej 2e0 ei R0i C ei ej Rij
3 .4/
2

1 1 2
S @ ˇ C S S TT
2 24 3
Z
1 t N 0i
D dt d3 x j.4/ej 2 ei R C ei ˛ ej R˛ ij
2 N

1 1 2
C.h u u / S @ ˇ C S S T T
2 24 3
Z
N ˛
D dt d3 x jej e e .3/R˛ ij C 2KŒ˛ K N ˛ ei .3/R˛ 0i
i j
2 i j
h i
Ce˛i KP ˛i C ! 0i @˛ t @˛ t ! i ! ik˛ .t ! k / C .t ! ik /K˛k
1 1 N
C .u S/ˇP .u S/N ˛ @˛ ˇ S˛ @˛ ˇ
4 4 4

1 1 1 1
CN .u T/2 .u S/2 C S˛ S˛ T˛ T ˛
3 48 48 3
Z

D dt d x jej e˛i Lt K˛i D˛ t ! i C .t ! ik /K˛k
3
N ˛ .3/
2N ˛ ei DŒ˛ K i C
j
e e R˛ ij
C 2KŒ˛
i
K
2 i j
1 1 N
C .u S/Lt ˇ .u S/N ˛ @˛ ˇ S˛ @˛ ˇ
4 4 4

1 1 1 1
CN .u T/2 .u S/2 C S˛ S˛ T˛ T ˛ ;
3 48 48 3
where we have omitted the bars for torsionless geometrical objects, we fixed q to
zero and the following notation have been used: u S D u S , u T D u T .
The only non-zero momenta are the conjugate to K˛i (equation (9.32a)) and to ˇ:
ıS 1
ˇ W ˘.ˇ/ :D D jeju S : (9.144)
ıLt ˇ 4
All the other momenta vanish identically, as given by (9.33) and
uS W ˘ .S/ D 0 ; (9.145a)

uT W ˘ .T/ D 0 ; (9.145b)
˛
S W ˘˛.S/ D 0; (9.145c)
T˛ W ˘˛.T/ D 0 : (9.145d)
The phase space is equipped with the symplectic structure (9.35) with replaced
by ˇ and ˘ ! ˘.ˇ/ , plus the Poisson brackets
˚
u S.t; x/; ˘ .S/ .t; x0 / D ı.x; x0 / ; (9.146a)
˚
u T.t; x/; ˘ .T/ .t; x0 / D ı.x; x0 / ; (9.146b)
n o
S˛ .t; x/; ˘ .S/ .t; x0 / D ı ˛ ı.x; x0 / ; (9.146c)
n o
T ˛ .t; x/; ˘ .T/ .t; x0 / D ı ˛ ı.x; x0 / : (9.146d)
The primary constraints are (9.36) and
1
C.ˇ/ :D ˘.ˇ/ jeju S 0 ; (9.147a)
4
C.b/ :D ˘ .b/ 0 ; b D S; T ; (9.147b)
C˛.b/ :D ˘˛.b/ 0 ; b D S; T : (9.147c)
Note that, contrary to an ordinary scalar field, the momentum of the Barbero–
Immirzi field defines a primary constraint, as it does not contain Lie derivatives.
The Dirac Hamiltonian reads
Z

HD D d3 x Ei˛ Lt K˛i C ˘.ˇ/ Lt ˇ C m Cm L
Z (

D d3 x N ˛ H˛ C NH C Ei˛ D˛ t ! i .t ! ik /K˛k
C.K/i˛ .K/Ci˛ C C.ˇ/ C .e/˛i .e/C˛i C . /i˛ . /Ci˛ C i Ci C ij Cij

)
X h i
.N/ ˛ .b/ .b/ ˛ .b/
CC C C˛ C C C .b/ C˛ ; (9.148)
bDS;T
where the super-momentum and super-Hamiltonian are, respectively,
H˛ :D 2Ei DŒ˛ K i C ˘.ˇ/ @˛ ˇ (9.149)
and
1 ˛ ij k 1
˘2
j
H :D Ei Ej k R˛ 2KŒ˛
i
K C
2jej 3jej .ˇ/

1 ˛ 1 ˛ 1 2 1 ˛
Cjej T˛ T S˛ S .u T/ C S @˛ ˇ : (9.150)
3 48 3 4
The analysis of the constraints runs as in the torsion-free case, with slight mod-
ifications. The Poisson bracket between the Dirac Hamiltonian and the primary
constraint (9.147a) does not generate any secondary constraints and it can be set to
zero by suitably choosing the Lagrange multiplier .S/ . The secondary constraints
are (9.41), (9.42), (9.43), (9.44) and (9.45) and
fC.S/ ; HD g D 0 ; (9.151)

.S/ 1 1
fC˛ ; HD g D Njej @˛ ˇ S˛ ; (9.152)
4 6
2
fC˛.T/ ; HD g D NjejT˛ ; (9.153)
3
2
fC.T/ ; HD g D Njeju T : (9.154)
3
Equation (9.151) is a consequence of the fact that u S disappears from the Dirac
Hamiltonian, so that its momentum is preserved by the Hamiltonian flow. Then we
can set ˘ .S/ D 0 strongly, as it vanishes initially. The dynamical equations of the
canonical variables S˛ , T ˛ and u T are completely arbitrary, since they depend
only on the associated Lagrange multipliers m . Therefore, they can be considered

as new Lagrange multipliers, which are easily calculated in order for the secondary
constraints (9.152), (9.153) and (9.154) to vanish:
uT D 0; T˛ D 0 ; S˛ D 6@˛ ˇ ; (9.155)
in agreement, after projection, with (9.139). Accordingly, ˇ must be a pseudo-scalar.

Plugging these expressions into (9.150), we get
1 ˛ ij k 1 2 3
C jej @˛ ˇ@˛ ˇ :
j
HD Ei Ej k R˛ 2KŒ˛
i
K C ˘.ˇ/ (9.156)
2jej 3jej 4
Notice that the system possesses a shift symmetry ˇ ! ˇ C ˇ0 which is absent in

the Holst case. Four out of the six physical degrees of freedom correspond to the
two polarizations of the graviton, while the remaining two are associated with the
pseudo-scalar field ˇ.
9.3 Barbero–Immirzi field. Recast the first-class constraints of the previous

problem in terms of the variables
1
AQ i˛ :D K˛i C ˛i ; PQ ˛i :D ˇEi˛ ; (9.157)
ˇ
where ˇ is the Barbero–Immirzi field. Show that the symplectic structure

generated by AQ and PQ is not canonical.
Solution Equation (9.56) remains formally the same. On the other hand, (9.58) with
non-constant ˇ becomes
1 i j k 2 2 i
Ri˛ D F˛i 2
jk K˛ K C DŒ˛ K i C 2 KŒ˛ @ ˇ :
ˇ ˇ ˇ
Then, the super-momentum (9.149) can be rewritten as
1
H˛ ˘Q .ˇ/ @˛ ˇ C Pi F˛i C 2 K˛i Pi @ ˇ ; (9.158)
ˇ
where
1
˘Q .ˇ/ :D ˘.ˇ/ C Ei˛ K˛i : (9.159)
ˇ
The weak equality stems, as before, from the use of the rotation constraint (9.51).
The super-Hamiltonian is

P˛i Pj ij 1 j
H F
k ˛
k
2 1 C K i
Œ˛ K
2ˇ 2 jej ˇ2
2
1 1 3
C ˘Q .ˇ/ C 2 P˛i K˛i C jej @˛ ˇ@˛ ˇ : (9.160)
3jej ˇ 4
Although the constraints include couplings between @ˇ and the other variables, this
does not signal a transition from the Einstein to a Jordan frame, since the physical
metric is still the same.
It is immediately clear that the naive generalization (9.157) of the Ashtekar–
Barbero connection does not lead to a canonical algebra [180]. Due to the mixing
of matter and gravitational degrees of freedom in A, the symplectic structure in the
new variables is non-canonical:
˚ i 1 ı i
AQ ˛ .t; x/; ˘Q .ˇ/ .t; x0 / D Ej ˛ ¤ 0 : (9.161)
ˇ ıEj
There is another way to state this result. The rotation constraint and the saturated
compatibility condition combine into the Gauss constraint (9.56). Taking the
Poisson bracket of the Gauss constraint with itself, one can see that the algebra
of gauge rotations does not close.
One should justify the definition (9.53) and explain the relation between the
constant ˇ0 and the Barbero–Immirzi field. If ˇ0 was regarded as the expectation
value of the ˇ field, for instance at a local minimum of an effective potential, then
its parity properties would be irrelevant as long as the state in the inner product is an
eigenstate of the parity operator. This is indeed the case, as the expectation value of
a quantum field is always tacitly defined as an operation which fixes the orientation
of local frames. On the other hand, if ˇ0 was interpreted as the asymptotic value of ˇ
at a suitable boundary of the ambient manifold, then ˇ0 would be a pseudo-constant.
However, ˇ0 is a proper constant and the second interpretation cannot hold.
9.4 Small gauge transformations. Let j .x/ be an internal 3-vector

defining the generating functional
Z
GŒ :D
j
d3 xj Gj ; (9.162)
called smeared Gauss constraint. Show that GŒ generates infinitesimal

gauge transformations of the gravielectric field and of the Ashtekar–Barbero
connection. For a variable X, such transformations are given by ıX D fX; Gg.
Solution Taking the Poisson bracket between P˛i and the smeared constraint G, we
have
Z
˚ ˇ
ıP˛i D P˛i .x/; GŒj D P˛i .x/; d3 x0 j jk l Akˇ Pl
(9.56)
Z
ˇ
D 2 ıˇ˛ ıik d3 xj jk l Pl ı.x; x0 / D 2 j .x/ij l P˛l .x/ :
(9.55)
(9.163)
Therefore, the (smeared) Gauss constraint generates gauge rotations of the gravi-
electric field. The Poisson bracket with the connection reads
Z
1 ˚ i 1 3 0 ˇ j l k ˇ
ıAi˛ D A .x/; GŒj
D A i
.x/; d x P @
j ˇ j
C A P
2 ˛ 2 ˛ jk ˇ l

D ı˛ˇ ıji @ˇ j C ıli j jk l Akˇ D @˛ i C jk i Ak˛ j
D D˛ i : (9.164)
This is the expected form of a small gauge transformation of the connection.
9.5 Smeared constraints. The total Hamiltonian in ADM variables (9.69)

can be written as
Z
HD D HŒN C dŒN C ˛
d3 x. 0 NP C ˛ NP ˛ / ; (9.165)
where the first two terms are the smeared constraints (9.46). In order to show
that the super-momentum and super-Hamiltonian constraints are first class,
we must check that
P ˛ D fdŒN ˛ ; HD g 0 ;
dŒN P
HŒN D fHŒN; HD g 0 :
Calculate these Poisson brackets. Can the result give a hint about why gravity
is difficult to quantize?
Sketch of solution: The only non-trivial Poisson brackets to compute are
fdŒN ˛ ; HŒNg ; fdŒN ˛ ; dŒM ˇ g ; fHŒN; HŒMg :
The smeared constraints are convenient because derivatives can be transferred to

the smearing functions N and N ˛ . In all the above brackets, non-derivative terms
References 459
commute. With a calculation similar to that of (9.47), one can see that [49]
fdŒN ˛ ; HŒNg D HŒN ˇ @ˇ N 0 ; (9.166a)

fdŒN ˛ ; dŒM ˇ g D dŒN @ M ˛ M @ N ˛ 0 ; (9.166b)
˛ˇ
fHŒN; HŒMg D dŒh .N@ˇ M M@ˇ N/ 0 : (9.166c)
All these constraints are satisfied on the secondary constraint surface. Equa-
tions (9.166a) and (9.166b) specify how the smeared Hamiltonian and momentum
constraints transform under spatial diffeomorphisms. The last expression is the
only equation featuring the canonical variable h˛ˇ in the smearing function. When
quantizing the system, canonical variables (and, in particular, h˛ˇ ) become operators
on a Hilbert space, but, from (9.166c), it is clear that the commutator ŒH; O H
O on
a wave-functional will vanish only after a suitable operator ordering. The same
obstruction occurs in the Ashtekar–Barbero phase space, but it can be overcome.
The curious reader might enjoy the fact that the commutator of the Hamiltonian
constraint with itself is strongly zero in the so-called ultra-local gravity [181–183].
References
1. R.L. Arnowitt, S. Deser, C.W. Misner, Dynamical structure and definition of energy in general
relativity. Phys. Rev. 116, 1322 (1959)
2. R.L. Arnowitt, S. Deser, C.W. Misner, Canonical variables for general relativity. Phys. Rev.
117, 1595 (1960)
3. R.L. Arnowitt, S. Deser, C.W. Misner, The dynamics of general relativity, in Gravita-
tion: An Introduction to Current Research, ed. by L. Witten (Wiley, New York, 1962).
4. A. Ashtekar, New variables for classical and quantum gravity. Phys. Rev. Lett. 57, 2244
(1986)
5. A. Ashtekar, New Hamiltonian formulation of general relativity. Phys. Rev. D 36, 1587 (1987)
6. J.F. Barbero, Real Ashtekar variables for Lorentzian signature space-times. Phys. Rev. D 51,
5507 (1995). [arXiv:gr-qc/9410014]
7. S. Deser, J.H. Kay, K.S. Stelle, Hamiltonian formulation of supergravity. Phys. Rev. D 16,
2448 (1977)
8. E.S. Fradkin, M.A. Vasiliev, Hamiltonian formalism, quantization and S matrix for supergrav-
ity. Phys. Lett. B 72, 70 (1977)
9. M. Pilati, The canonical formulation of supergravity. Nucl. Phys. B 132, 138 (1978)
10. P.D. D’Eath, The canonical quantization of supergravity. Phys. Rev. D 29, 2199 (1984);
Erratum-ibid. D 32, 1593 (1985)
11. P.D. D’Eath, Supersymmetric Quantum Cosmology (Cambridge University Press, Cambridge,
2005)
12. P. Vargas Moniz, Quantum Cosmology – The Supersymmetric Perspective. Lect. Notes Phys.
803, 1 (2010); Lect. Notes Phys. 804, 1 (2010)
13. T. Damour, P. Spindel, Quantum supersymmetric cosmology and its hidden Kac–Moody
structure. Class. Quantum Grav. 30, 162001 (2013). [arXiv:1304.6381]
14. T. Damour, P. Spindel, Quantum supersymmetric Bianchi IX cosmology. Phys. Rev. D 90,
103509 (2014). [arXiv:1406.1309]
15. P. Vargas Moniz, Supersymmetric quantum cosmology: a ‘Socratic’ guide. Gen. Relat. Grav.
46, 1618 (2014)
16. P. Vargas Moniz, Quantum cosmology: meeting SUSY. Springer Proc. Math. Stat. 60, 117
(2014)
17. E. Fermi, Sopra i fenomeni che avvengono in vicinanza di una linea oraria. Atti Accad. Naz.
Lincei Cl. Sci. Fis. Mat. Nat. Rend. 31, 21-51-101 (1922)
18. B. Bertotti, Fermi’s coordinates and the principle of equivalence, in Enrico Fermi, ed. by C.
Bernardini, L. Bonolis (Springer, Berlin, 2004)
19. F.K. Manasse, C.W. Misner, Fermi normal coordinates and some basic concepts in differential
geometry. J. Math. Phys. 4, 735 (1963)
20. N. Ashby, B. Bertotti, Relativistic effects in local inertial frames. Phys. Rev. D 34, 2246
(1986)
21. L. Freidel, D. Minic, T. Takeuchi, Quantum gravity, torsion, parity violation and all that. Phys.
Rev. D 72, 104002 (2005). [arXiv:hep-th/0507253]
22. A. Perez, C. Rovelli, Physical effects of the Immirzi parameter. Phys. Rev. D 73, 044013
(2006). [arXiv:gr-qc/0505081]
23. S. Mercuri, Fermions in Ashtekar–Barbero connection formalism for arbitrary values of the
Immirzi parameter. Phys. Rev. D 73, 084016 (2006). [arXiv:gr-qc/0601013]
24. M. Bojowald, R. Das, Fermions in loop quantum cosmology and the role of parity. Class.
25. S. Deser, P. van Nieuwenhuizen, Nonrenormalizability of the quantized Dirac–Einstein
system. Phys. Rev. D 10, 411 (1974)
26. E. Cartan, Leçons sur la Théorie des Spineurs, vol. II (Hermann, Paris, 1938)
27. F.W. Hehl, P. von der Heyde, G.D. Kerlick, J.M. Nester, General relativity with spin and
torsion: foundations and prospects. Rev. Mod. Phys. 48, 393 (1976)
28. I.L. Buchbinder, S.D. Odintsov, I.L. Shapiro, Effective Action in Quantum Gravity (IOP,
Bristol, 1992)
29. J.D. McCrea, Irreducible decompositions of nonmetricity, torsion, curvature and Bianchi
identities in metric-affine spacetimes. Class. Quantum Grav. 9, 553 (1992)
30. J.A. Helayel-Neto, A. Penna-Firme, I.L. Shapiro, Conformal symmetry, anomaly and
effective action for metric-scalar gravity with torsion. Phys. Lett. B 479, 411 (2000).
31. S. Capozziello, G. Lambiase, C. Stornaiolo, Geometric classification of the torsion tensor in
space-time. Ann. Phys. (Berlin) 10, 713 (2001). [arXiv:gr-qc/0101038]
32. S. Holst, Barbero’s Hamiltonian derived from a generalized Hilbert–Palatini action. Phys.
Rev. D 53, 5966 (1996). [arXiv:gr-qc/9511026]
33. G. Immirzi, Real and complex connections for canonical gravity. Class. Quantum Grav. 14,
L177 (1997). [arXiv:gr-qc/9612030]
34. C. Rovelli, T. Thiemann, Immirzi parameter in quantum general relativity. Phys. Rev. D 57,
1009 (1998). [arXiv:gr-qc/9705059]
35. H.T. Nieh, M.L. Yan, An identity in Riemann–Cartan geometry. J. Math. Phys. 23, 373 (1982)
36. O. Chandía, J. Zanelli, Topological invariants, instantons and chiral anomaly on spaces with
torsion. Phys. Rev. D 55, 7580 (1997). [arXiv:hep-th/9702025]
37. H.T. Nieh, A torsional topological invariant. Int. J. Mod. Phys. A 22, 5237 (2007)
38. S. Mercuri, From the Einstein–Cartan to the Ashtekar–Barbero canonical constraints, passing
through the Nieh–Yan functional. Phys. Rev. D 77, 024036 (2008). [arXiv:0708.0037]
39. G. Date, R.K. Kaul, S. Sengupta, Topological interpretation of Barbero–Immirzi parameter.
Phys. Rev. D 79, 044008 (2009). [arXiv:0811.4496]
40. L. Castellani, R. D’Auria, P. Frè, Supergravity and Superstrings: A Geometric Perspective,
vol. 1 (World Scientific, Singapore, 1991)
41. V. Taveras, N. Yunes, Barbero–Immirzi parameter as a scalar field: K-inflation from loop
quantum gravity? Phys. Rev. D 78, 064070 (2008). [arXiv:0807.2652]
42. A. Torres-Gomez, K. Krasnov, Remarks on Barbero–Immirzi parameter as a field. Phys. Rev.
D 79, 104014 (2009). [arXiv:0811.1998]
References 461
43. G. Calcagni, S. Mercuri, Barbero–Immirzi field in canonical formalism of pure gravity. Phys.
Rev. D 79, 084004 (2009). [arXiv:0902.0957]
44. S. Mercuri, V. Taveras, Interaction of the Barbero–Immirzi field with matter and pseudoscalar
perturbations. Phys. Rev. D 80, 104007 (2009). [arXiv:0903.4407]
45. C. Teitelboim, Quantum mechanics of the gravitational field. Phys. Rev. D 25, 3159 (1982)
46. J.B. Hartle, K.V. Kuchař. Path integrals in parametrized theories: the free relativistic particle.
Phys. Rev. D 34, 2323 (1986)
47. P.A.M. Dirac, Generalized Hamiltonian dynamics. Can. J. Math. 2, 129 (1950)
48. P.A.M. Dirac, Generalized Hamiltonian dynamics. Proc. R. Soc. Lond. A 246, 326 (1958)
49. P.A.M. Dirac, The theory of gravitation in Hamiltonian form. Proc. R. Soc. Lond. A 246, 333
(1958)
50. J. Schwinger, Quantized gravitational field. Phys. Rev. 130, 1253 (1963)
51. M. Henneaux, C. Teitelboim, Quantization of Gauge Systems (Princeton University Press,
Princeton, 1994)
arXiv:gr-qc/0110034
53. J.A. Wheeler, Geometrodynamics and the issue of the final state, in Relativity, Groups and
Topology, ed. by C. DeWitt, B.S. DeWitt (Gordon and Breach, New York, 1964)
55. J.W. York, Role of conformal three-geometry in the dynamics of gravitation. Phys. Rev. Lett.
28, 1082 (1972)
Phys. Rev. D 15, 2752 (1977)
57. A. Guarnizo, L. Castañeda, J.M. Tejeiro, Boundary term in metric f .R/ gravity: field equations
in the metric formalism. Gen. Relat. Grav. 42, 2713 (2010). [arXiv:1002.0617]
58. A. Ashtekar, Lectures on Non-perturbative Canonical Gravity (World Scientific, Singapore,
1991)
59. A. Ashtekar, J. Engle, D. Sloan, Asymptotics and Hamiltonians in a first order formalism.
Class. Quantum Grav. 25, 095020 (2008). [arXiv:0802.2527]
60. P.W. Higgs, Integration of secondary constraints in quantized general relativity. Phys. Rev.
Lett. 1, 373 (1958); Erratum-ibid. 3, 66 (1959)
61. J.A. Wheeler, Superspace and the nature of quantum geometrodynamics, in Battelle Rencon-
tres: 1967 Lectures in Mathematics and Physics, ed. by C. DeWitt, J.A. Wheeler (Benjamin,
New York, 1968)
62. S.W. Hawking, D.N. Page, Operator ordering and the flatness of the universe. Nucl. Phys. B
264, 185 (1986)
63. J.J. Halliwell, Derivation of the Wheeler–DeWitt equation from a path integral for minisuper-
space models. Phys. Rev. D 38, 2468 (1988)
64. I. Moss, Quantum cosmology and the self-observing universe. Ann. Poincaré Phys. Theor. A
49, 341 (1988)
65. R.P. Feynman, A.R. Hibbs, Quantum Mechanics and Path Integrals: Emended Edition (Dover,
Mineola, 2010)
66. M. Chaichian, A. Demichev, Path Integrals in Physics (IOP, Bristol, 2001)
67. S.W. Hawking, The path-integral approach to quantum gravity, in General Relativity: An
Einstein Centenary Survey, ed. by S.W. Hawking, W. Israel (Cambridge University Press,
Cambridge, 1979)
68. A.D. Linde, Quantum creation of an inflationary universe. Zh. Eksp. Teor. Fiz. 87, 369 (1984)
[Sov. Phys. JETP 60, 211 (1984)]; Quantum creation of the inflationary universe. Lett. Nuovo
Cim. 39, 401 (1984)
69. A. Vilenkin, Wave function discord. Phys. Rev. D 58, 067301 (1998). [arXiv:gr-qc/9804051]
70. H. Jeffreys, On certain approximate solutions of linear differential equations of the second
order. Proc. Lond. Math. Soc. s2-23, 428 (1925)
71. G. Wentzel, Eine Verallgemeinerung der Quantenbedingungen für die Zwecke der Wellen-
mechanik. Z. Phys. A 38, 518 (1926)
72. H.A. Kramers, Wellenmechanik und halbzahlige Quantisierung. Z. Phys. A 39, 828 (1926)
73. L. Brillouin, La mécanique ondulatoire de Schrödinger; une méthode générale de résolution
par approximations successives. C. R. Acad. Sci. 183, 24 (1926)
74. D.S. Salopek, J.R. Bond, J.M. Bardeen, Designing density fluctuation spectra in inflation.
Phys. Rev. D 40, 1753 (1989)
75. M. Nagasawa, J. Yokoyama, Phase transitions triggered by quantum fluctuations in the
inflationary universe. Nucl. Phys. B 370, 472 (1992)
76. J. Martin, D.J. Schwarz, WKB approximation for inflationary cosmological perturbations.
77. R. Casadio, F. Finelli, M. Luzzi, G. Venturi, Improved WKB analysis of cosmological
perturbations. Phys. Rev. D 71, 043517 (2005). [arXiv:gr-qc/0410092]
78. U.H. Gerlach, Derivation of the ten Einstein field equations from the semiclassical approxi-
mation to quantum geometrodynamics. Phys. Rev. 177, 1929 (1969)
79. T. Banks, TCP, quantum gravity, the cosmological constant and all that. . . . Nucl. Phys. B 249,
332 (1985)
(1989)
81. C. Kiefer, T.P. Singh, Quantum gravitational corrections to the functional Schrödinger
equation. Phys. Rev. D 44, 1067 (1991)
82. S.P. Kim, New asymptotic expansion method for the Wheeler–DeWitt equation. Phys. Rev.
D 52, 3382 (1995). [arXiv:gr-qc/9511038]
83. J.J. Halliwell, Correlations in the wave function of the universe. Phys. Rev. D 36, 3626 (1987)
84. A. Peres, On Cauchy’s problem in general relativity – II. Nuovo Cim. 26, 53 (1962)
85. H.D. Zeh, Time in quantum gravity. Phys. Lett. A 126, 311 (1988)
86. S.P. Kim, Quantum mechanics of conformally and minimally coupled Friedmann–Robertson–
Walker cosmology. Phys. Rev. D 46, 3403 (1992)
Quantum Grav. 4, 1369 (1987)
88. J.B. Hartle, S.W. Hawking, Wave function of the Universe. Phys. Rev. D 28, 2960 (1983)
89. A. Vilenkin, Creation of universes from nothing. Phys. Lett. B 117, 25 (1982)
90. W.-M. Suen, K. Young, Wave function of the Universe as a leaking system. Phys. Rev. D 39,
2201 (1989)
91. H.D. Conradi, H.D. Zeh, Quantum cosmology as an initial value problem. Phys. Lett. A 154,
321 (1991)
92. H.-D. Conradi, Initial state in quantum cosmology. Phys. Rev. D 46, 612 (1992)
93. S.W. Hawking, The quantum state of the universe. Nucl. Phys. B 239, 257 (1984)
94. D.N. Page, Density matrix of the Universe. Phys. Rev. D 34, 2267 (1986)
95. G.W. Gibbons, L.P. Grishchuk, What is a typical wave function for the universe? Nucl. Phys.
B 313, 736 (1989)
96. J.J. Halliwell, J. Louko, Steepest-descent contours in the path-integral approach to quantum
cosmology. I. The de Sitter minisuperspace model. Phys. Rev. D 39, 2206 (1989)
cosmology. II. Microsuperspace. Phys. Rev. D 40, 1868 (1989)
cosmology. III. A general method with applications to anisotropic minisuperspace models.
Phys. Rev. D 42, 3997 (1990)
99. J.J. Halliwell, J.B. Hartle, Integration contours for the no-boundary wave function of the
universe. Phys. Rev. D 41, 1815 (1990)
100. L.P. Grishchuk, L.V. Rozhansky, Does the Hartle–Hawking wavefunction predict the universe
we live in? Phys. Lett. B 234, 9 (1990)
101. G.W. Lyons, Complex solutions for the scalar field model of the Universe. Phys. Rev. D 46,
1546 (1992)
References 463
102. A. Lukas, The no boundary wave-function and the duration of the inflationary period. Phys.
Lett. B 347, 13 (1995). [arXiv:gr-qc/9409012]
103. J.B. Hartle, S.W. Hawking, T. Hertog, No-boundary measure of the Universe. Phys. Rev. Lett.
100, 201301 (2008). [arXiv:0711.4630]
104. J.B. Hartle, S.W. Hawking, T. Hertog, Classical universes of the no-boundary quantum state.
Phys. Rev. D 77, 123537 (2008). [arXiv:0803.1663]
105. J.B. Hartle, S.W. Hawking, T. Hertog, No-boundary measure in the regime of eternal inflation.
Phys. Rev. D 82, 063510 (2010). [arXiv:1001.0262]
106. A.O. Barvinsky, A.Yu. Kamenshchik, Cosmological landscape from nothing: some like it hot.
JCAP 0609, 014 (2006). [arXiv:hep-th/0605132]
107. A.O. Barvinsky, A.Yu. Kamenshchik, Thermodynamics via creation from nothing:
limiting the cosmological constant landscape. Phys. Rev. D 74, 121502 (2006).
108. A.O. Barvinsky, Why there is something rather than nothing: cosmological constant from
summing over everything in Lorentzian quantum gravity. Phys. Rev. Lett. 99, 071301 (2007).
[arXiv:0704.0083]
109. A. Vilenkin, Quantum creation of universes. Phys. Rev. D 30, 509 (1984)
110. A. Vilenkin, Quantum origin of the universe. Nucl. Phys. B 252, 141 (1985)
111. A. Vilenkin, Boundary conditions in quantum cosmology. Phys. Rev. D 33, 3560 (1986)
112. A. Vilenkin, Quantum cosmology and the initial state of the Universe. Phys. Rev. D 37, 888
(1988)
113. A. Vilenkin, Approaches to quantum cosmology. Phys. Rev. D 50, 2581 (1994).
114. S.W. Hawking, I.G. Moss, Supercooled phase transitions in the very early universe. Phys.
Lett. B 110, 35 (1982)
115. T. Padmanabhan, Cosmological constant: the weight of the vacuum. Phys. Rep. 380, 235
(2003). [arXiv:hep-th/0212290]
116. M. Spradlin, A. Strominger, A. Volovich, de Sitter space, in Unity from Duality: Gravity,
Gauge Theory and Strings, ed. by C. Bachas, A. Bilal, M. Douglas, N. Nekrasov, F. David
(Springer, Berlin, 2002). [arXiv:hep-th/0110007]
117. K. Vandersloot, Hamiltonian constraint of loop quantum cosmology. Phys. Rev. D 71, 103506
(2005). [arXiv:gr-qc/0502082]
118. A. Perez, Regularization ambiguities in loop quantum gravity. Phys. Rev. D 73, 044007
(2006). [arXiv:gr-qc/0509118]
119. T. Thiemann, Anomaly-free formulation of non-perturbative, four-dimensional Lorentzian
quantum gravity. Phys. Lett. B 380, 257 (1996). [arXiv:gr-qc/9606088]
120. T. Thiemann, Quantum spin dynamics (QSD). Class. Quantum Grav. 15, 839 (1998).
121. C. Rovelli, L. Smolin, Knot theory and quantum gravity. Phys. Rev. Lett. 61, 1155 (1988)
122. C. Rovelli, L. Smolin, Loop space representation of quantum general relativity. Nucl. Phys.
B 331, 80 (1990)
124. A. Ashtekar, J. Lewandowski, Background independent quantum gravity: a status report.
Class. Quantum Grav. 21, R53 (2004). [arXiv:gr-qc/0404018]
125. A. Ashtekar, J. Lewandowski, Quantum theory of gravity I: area operators. Class. Quantum
Grav. 14, A55 (1997). [arXiv:gr-qc/9602046]
126. A. Ashtekar, J.C. Baez, A. Corichi, K. Krasnov, Quantum geometry and black hole entropy.
127. A. Ashtekar, J.C. Baez, K. Krasnov, Quantum geometry of isolated horizons and black hole
entropy. Adv. Theor. Math. Phys. 4, 1 (2000). [arXiv:gr-qc/0005126]
128. K.A. Meissner, Black hole entropy in loop quantum gravity. Class. Quantum Grav. 21, 5245
(2004). [arXiv:gr-qc/0407052]
129. D. Oriti, D. Pranzetti, L. Sindoni, Horizon entropy from quantum gravity condensates. Phys.
Rev. Lett. 116, 211301 (2016). [arXiv:1510.06991]
130. M. Dupuis, F. Girelli, Observables in loop quantum gravity with a cosmological constant.
Phys. Rev. D 90, 104037 (2014). [arXiv:1311.6841]
131. V. Bonzom, M. Dupuis, F. Girelli, E.R. Livine, Deformed phase space for 3d loop gravity and
hyperbolic discrete geometries. arXiv:1402.2323
132. R. Gambini, J. Pullin, Does loop quantum gravity imply D 0? Phys. Lett. B 437, 279
(1998). [arXiv:gr-qc/9803097]
133. S.H.S. Alexander, G. Calcagni, Quantum gravity as a Fermi liquid. Found. Phys. 38, 1148
(2008). [arXiv:0807.0225]
134. S.H.S. Alexander, G. Calcagni, Superconducting loop quantum gravity and the cosmological
constant. Phys. Lett. B 672, 386 (2009). [arXiv:0806.4382]
135. S.B. Treiman, R. Jackiw, B. Zumino, E. Witten, Current Algebra and Anomalies (World
Scientific, Singapore, 1985), p. 258
136. H. Kodama, Specialization of Ashtekar’s formalism to Bianchi cosmology. Prog. Theor. Phys.
80, 1024 (1988)
137. H. Kodama, Holomorphic wave function of the Universe. Phys. Rev. D 42, 2548 (1990)
138. R. Gambini, J. Pullin, Loops, Knots, Gauge Theories and Quantum Gravity (Cambridge
139. L. Smolin, Quantum gravity with a positive cosmological constant. arXiv:hep-th/0209079
140. S.S. Chern, J. Simons, Characteristic forms and geometric invariants. Ann. Math. 99, 48
(1974)
141. L. Smolin, C. Soo, The Chern–Simons invariant as the natural time variable for classical and
quantum cosmology. Nucl. Phys. B 449, 289 (1995). [arXiv:gr-qc/9405015]
142. E. Witten, A note on the Chern–Simons and Kodama wave functions. arXiv:gr-qc/0306083
143. C. Soo, Wave function of the universe and Chern-Simons perturbation theory. Class. Quantum
Grav. 19, 1051 (2002). [arXiv:gr-qc/0109046]
144. A. Randono, Generalizing the Kodama state. I. Construction. arXiv:gr-qc/0611073
145. A. Randono, Generalizing the Kodama state. II. Properties and physical interpretation.
arXiv:gr-qc/0611074
146. A. Randono, A mesoscopic quantum gravity effect. Gen. Relat. Grav. 42, 1909 (2010).
[arXiv:0805.2955]
147. S. Weinberg, The Quantum Theory of Fields, vol. II (Cambridge University Press, Cambridge,
1997)
148. C.G. Callan, R.F. Dashen, D.J. Gross, The structure of the gauge theory vacuum. Phys. Lett.
B 63, 334 (1976)
149. R. Jackiw, C. Rebbi, Vacuum periodicity in a Yang–Mills quantum theory. Phys. Rev. Lett.
37, 172 (1976)
150. S. Deser, M.J. Duff, C.J. Isham, Gravitationally induced CP effects. Phys. Lett. B 93, 419
(1980)
151. A. Ashtekar, A.P. Balachandran, S. Jo, The CP problem in quantum gravity. Int. J. Mod.
Phys. A 4, 1493 (1989)
152. R.D. Peccei, H.R. Quinn, CP conservation in the presence of pseudoparticles. Phys. Rev. Lett.
38, 1440 (1977)
153. R.D. Peccei, H.R. Quinn, Constraints imposed by CP conservation in the presence of
pseudoparticles. Phys. Rev. D 16, 1791 (1977)
154. F. Wilczek, Problem of strong P and T invariance in the presence of instantons. Phys. Rev.
Lett. 40, 279 (1978)
155. S. Weinberg, A new light boson? Phys. Rev. Lett. 40, 223 (1978)
156. M. Montesinos, Self-dual gravity with topological terms. Class. Quantum Grav. 18, 1847
(2001). [arXiv:gr-qc/0104068]
157. R. Paternoga, R. Graham, Triad representation of the Chern–Simons state in quantum gravity.
158. N. Weiss, Possible origins of a small, nonzero cosmological constant. Phys. Lett. B 197, 42
(1987)
References 465
159. J.A. Frieman, C.T. Hill, R. Watkins, Late time cosmological phase transitions: particle physics
models and cosmic evolution. Phys. Rev. D 46, 1226 (1992)
160. J.A. Frieman, C.T. Hill, A. Stebbins, I. Waga, Cosmology with ultralight pseudo Nambu–
Goldstone bosons. Phys. Rev. Lett. 75, 2077 (1995). [arXiv:astro-ph/9505060]
161. T. Jacobson, 1 C 1 sector of 3 C 1 gravity. Class. Quantum Grav. 13, L111 (1996);
Erratum-ibid. 13, 3269 (1996). [arXiv:gr-qc/9604003]
162. G.E. Volovik, Superfluid analogies of cosmological phenomena. Phys. Rep. 351, 195 (2001).
163. G.E. Volovik, Field theory in superfluid He-3: what are the lessons for particle physics,
gravity, and high temperature superconductivity? Proc. Natl. Acad. Sci. 96, 6042 (1999).
[arXiv:cond-mat/9812381]
164. G.E. Volovik, The Universe in a Helium Droplet (Clarendon Press, Oxford, 2003)
165. J. Bardeen, L.N. Cooper, J.R. Schrieffer, Microscopic theory of superconductivity. Phys. Rev.
106, 162 (1957)
166. J. Bardeen, L.N. Cooper, J.R. Schrieffer, Theory of superconductivity. Phys. Rev. 108, 1175
(1957)
167. N.N. Bogoliubov, On a new method in the theory of superconductivity. Nuovo Cim. 7, 794
(1958)
168. J. Polchinski, Effective field theory and the Fermi surface, in Recent Directions in Par-
ticle Theory, ed. by J.A. Harvey, J.G. Polchinski (World Scientific, Singapore, 1993).
169. G. Sierra, Conformal field theory and the exact solution of the BCS Hamiltonian. Nucl. Phys.
B 572, 517 (2000). [arXiv:hep-th/9911078]
170. M. Asorey, F. Falceto, G. Sierra, Chern–Simons theory and BCS superconductivity. Nucl.
Phys. B 622, 593 (2002). [arXiv:hep-th/0110266]
171. S.N. Bose, Plancks Gesetz und Lichtquantenhypothese. Z. Phys. A 26, 178 (1924)
172. A. Einstein, Quantentheorie des einatomigen idealen Gases. Sitz.-ber. Kgl. Preuss. Akad.
Wiss. 1924, 261 (1924)
173. A. Einstein, Quantentheorie des einatomigen idealen Gases. Zweite Abhandlung. Sitz.-ber.
Kgl. Preuss. Akad. Wiss. 1925, 3 (1925)
174. F. Dalfovo, S. Giorgini, L.P. Pitaevskii, S. Stringari, Theory of Bose–Einstein condensation
in trapped gases. Rev. Mod. Phys. 71, 463 (1999)
175. A.J. Leggett, Bose–Einstein condensation in the alkali gases: some fundamental concepts.
Rev. Mod. Phys. 73, 307 (2001)
176. C.J. Pethick , H. Smith, Bose–Einstein Condensation in Dilute Gases (Cambridge University
177. L.P. Pitaevskii, S. Stringari, Bose–Einstein Condensation (Clarendon Press, Oxford, 2003)
178. S. Giorgini, L.P. Pitaevskii, S. Stringari, Theory of ultracold atomic Fermi gases. Rev. Mod.
Phys. 80, 1215 (2008). [arXiv:0706.3360]
179. M. Randeria, E. Taylor, BCS-BEC crossover and the unitary Fermi gas. Ann. Rev. Cond.
Matter Phys. 5, 209 (2014). [arXiv:1306.5785]
180. G.A. Mena Marugán, Extent of the Immirzi ambiguity in quantum general relativity. Class.
Quantum Grav. 19, L63 (2002). [arXiv:gr-qc/0203027]
181. C.J. Isham, Some quantum field theory aspects of the superspace quantization of general
relativity. Proc. R. Soc. Lond. A 351, 209 (1976)
182. M. Henneaux, Zero Hamiltonian signature spacetimes. Bull. Soc. Math. Belg. 31, 47 (1979)
183. C. Teitelboim, The Hamiltonian structure of space-time, in General Relativity and Gravita-
tion, ed. by A. Held, vol. 1 (Plenum, New York, 1980)
Chapter 10
Canonical Quantum Cosmology
Not only is the universe stranger than we imagine, it is stranger

than we can imagine.
—Sir Arthur Eddington
Contents
10.1 Mini-superspace. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 468
10.1.1 Classical FLRW Hamiltonian. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 469
10.2 Wheeler–DeWitt Quantum Cosmology. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 471
10.2.1 de Sitter Solutions and Probability of Inflation. . . . . . . . . . . . . . . . . . . . . . . 473
10.2.2 Massless Scalar Field and Group Averaging. . . . . . . . . . . . . . . . . . . . . . . . . . 476
10.2.3 Quantum Singularity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 480
10.2.4 Cosmological Constant and the Multiverse. . . . . . . . . . . . . . . . . . . . . . . . . . . 482
10.2.5 Perturbations and Inflationary Observables. . . . . . . . . . . . . . . . . . . . . . . . . . . 484
10.3 Loop Quantum Cosmology. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 489
10.3.1 Classical FLRW Variables and Constraints. . . . . . . . . . . . . . . . . . . . . . . . . . . 490
10.3.2 Quantization and Inverse-Volume Spectrum. . . . . . . . . . . . . . . . . . . . . . . . . . 493
10.3.3 Mini-superspace Parametrization. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 495
10.3.4 Quantum Hamiltonian Constraint. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 496
10.3.5 Models with Curvature or a Cosmological Constant. . . . . . . . . . . . . . . . . 500
10.3.6 Homogeneous Effective Dynamics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 501
10.3.7 Singularity Resolved?. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 508
10.3.8 Lattice Refinement: Quantum Corrections Revisited. . . . . . . . . . . . . . . . . 510
10.3.9 Perturbations and Inflationary Observables. . . . . . . . . . . . . . . . . . . . . . . . . . . 516
10.3.10 Inflation in Other Approaches. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 525
10.3.11 Is There a Bounce?. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 526
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 531
A complete background-independent quantum theory of gravity is a hard goal to

achieve and the formal discussion of any of the proposals in this direction lies
beyond the scope of this book. However, in order to understand certain techniques it
is often useful to specialize to a simple background and study the quantum system
thereon (Sect. 10.1). Symmetry reduction is done not only for didactic purposes but
also as a (hopefully temporary) necessity in active research. For instance, it is natural

DOI 10.1007/978-3-319-41127-9_10
468 10 Canonical Quantum Cosmology
to consider the big-bang and cosmological constant problems in an expanding

cosmological background. In such a setting, one does not employ the full machinery
of general theories of quantum gravity and a number of simplifications take place.
We apply the procedure of symmetry reduction to canonical gravity and quantum
cosmological scenarios, first in the Wheeler–DeWitt approach in ADM variables
(Sect. 10.2) and then within loop quantum cosmology (Sect. 10.3).
10.1 Mini-superspace
The space S of geometric and matter configurations is rather formal and cannot be
handled easily also due to the fact that it is infinite dimensional. However, if one
restricts the theory to homogeneous geometries, the number of degrees of freedom
becomes finite and one can construct a quantum model on a finite-dimensional
mini-superspace [1, 2]. This reduction has no robust justification a priori and it is
generally agreed that mini-superspace quantization is only a toy model of a quantum
theory of geometry. In this case, the symmetry reduction of the theory is performed
at the kinematical level, that is, before solving the quantum constraints. In fact, we
first restrict the classical theory to a homogeneous background and then quantize:
classical theory ! symmetry reduction ! quantization :
This is in implicit violation of Heisenberg uncertainty principle, since one is

simultaneously freezing inhomogeneous modes out of the wave-function .
On the other hand, a complete quantum canonical theory of gravity should
implement quantization in a background-independent way and then specialize, if
desired, to a homogeneous cosmological setting:
‹
classical theory ! quantization ! symmetry reduction :
The question mark on the last arrow symbolizes the fact that, in many approaches,1
the limit to a continuum (smooth manifold) and the operation of symmetry reduction
are non-trivial and require intermediate steps. In several cases, the formulation of
the theory is incomplete and these steps are still unknown. Therefore, in general,
mini-superspace models can capture only some of the qualitative features of the
cosmology of the full theory. A complete check of that can be done only if
the underlying theory is under control and full solutions can be constructed. In
LQG, this seems to be the case (Sect. 10.3.7). In Sect. 11.5, we will present a
second-quantized extension of the canonical theory where symmetry reduction and
1
For instance, in discrete gravity models geometric variables pick countable labels which character-
ize simplicial complexes. These complexes are mathematical objects quite distinct from manifolds,
but they can approximate smooth manifolds under certain limits.
10.1 Mini-superspace 469
quantization may commute. However, in general they do not [3, 4] and we shall
see a concrete example of this in Sect. 11.3.2. Ignoring for the moment possible
embeddings of canonical quantum gravity into a more complete theory, we can
exploit the knowledge we have of the full canonical theory to get a hint of the level
at which we can trust the mini-superspace results.
Having so cautioned the reader, we proceed to analyze the canonical variables
and first-class constraints in a cosmological setting, beginning with Wheeler–
DeWitt quantization [1, 2]. We limit the discussion to FLRW models in vacuum
or with a minimally coupled scalar field [5], having inflation in mind [6–10].2
We will then discuss loop quantum cosmology and cosmological perturbations.
Mini-superspace FLRW scenarios and their generalization to Bianchi models are
usually sensible and give a very stimulating insight into the problem of the big
bang in a quantum universe, especially in the loop quantization scheme. However,
it has become progressively clear that many of the properties of quantum geometry
become accessible only when inhomogeneous perturbations and, consequently, the
full structure of the constraint algebra, have been included in the picture. In this
case, the properties of the quantum system do change, unless the back-reaction
of anisotropic or inhomogeneous modes (loosely speaking, the “production of
gravitons”) be negligible [3, 4, 24–27].
10.1.1 Classical FLRW Hamiltonian
The symmetry reduction takes place at the level of the classical action. From
this, one can define mini-superspace canonical variables and obtain the first-class
constraints. To get the flavour of what happens, we first perform the symmetry
reduction directly on the classical constraints and then rederive the mini-superspace
constraints from the symmetry-reduced action.
On homogeneous backgrounds, one can always choose the shift vector N ˛ to van-
ish. Then, the extrinsic curvature is (7.132), K˛ˇ D Hh˛ˇ =N, and the spatial Ricci
curvature is .3/R D 6K=a2 . The momentum constraint (9.70) vanishes identically, so
that the only constraint left is the super-Hamiltonian (9.71), which can be written as
" #
3 1 P2 3 H2 K
a C V. / 2 C 2 D 0: (10.1)
2 N2 N2 a
This is nothing but the first Friedmann equation (2.81) with N ¤ 1.

Let us now recover (10.1) from the action. On an ideal FLRW background,
open and flat universes have infinite spatial volume and the super-Hamiltonian
2
Early papers considering a massless scalar field are [11, 12]. The case with non-minimal couplings
has also been studied [6, 7, 13–21]. Adding higher-order curvature terms does not change the results
much [8, 22, 23].
constraint is formally ill-defined because it entails a divergent integration of a

spatially constant quantity over a comoving spatial slice ˙,
Z p
d3 x h D C1 :
˙
To make the integral finite, it is customary to define the constraint on a freely chosen
finite region of size V D a3 V0 , where V0 is the corresponding comoving volume:
Z p Z p
d3 x h ! d3 x h D: a3 V0 < C1 : (10.2)
˙ ˙.V0 /
If the universe is closed, V0 represents its comoving volume and is therefore

physical. In the open and flat cases, the fiducial volume V0 is arbitrary and has no
physical meaning, so that it should not appear in physical observables eventually.
We will face this issue in loop quantum cosmology.
From (9.66),
Z " #
3 3 K H2 1 P2
S D V0 dt Na C V. /
2 a2 N 2 2 N2
Z " !#
3a aP 2 3 1
P2
D V0 dt N K Ca V : (10.3)
2 N2 2 N2
The momenta conjugate to the variables N, a and are 0 D 0 and
ıS 6V0 aPa ıS P
p.a/ :D D 2 ; p :D D V0 a 3 ; (10.4)
ı aP N ıP N
with commutation relations
fa; p.a/g D 1 ; f ;p g D 1: (10.5)
The Dirac Hamiltonian is
HD D 0 NP C p.a/ aP C p P L D 0 NP C NH ;
where, in agreement with (10.1),

2
1 2 2 2
HD a p .a/ C p C U.a; / ; (10.6a)
2V0 a3 6
and

3K
U.a; / :D 2V0 a3 U.a; / ; U.a; / D V0 a a2 V 2 : (10.6b)

10.2 Wheeler–DeWitt Quantum Cosmology 471
Notice that the mini-superspace scalar constraint differs with respect to the Hamil-
tonian constraint of the full theory by the integrated volume factor. To keep the
notation simple, we will use the same symbol H in both cases.
The Hamilton equations are (Problem 10.1)
" ! #
2 2
P2 H2
pP .a/ D f p.a/ ; HD g D 2V0 Na V 3 ; (10.7)
N2 N2
pP D f p ; HD g D NV0 a3 V; : (10.8)
These are, respectively, the second Friedmann equation and the scalar-field equation
of motion: check these expressions for N D 1 with (2.170) and (2.171), noting that
P and pP D V0 a3 . R C 3H P /.
pP .a/ D .6V0 = 2 /a2 .2H 2 C H/
In preparation for a comparison with the results of loop quantum cosmology, we
note that, at the classical level, one can define other canonical coordinates (pairs
of conjugate variables) in phase space and recast the Hamiltonian constraint (10.6)
accordingly. For instance, consider the pair
2 p.a/ 3v0
b :D ; v :D a2.1Cn/ ; (10.9)
6v0 a1C2n .1 C n/ 2
where v0 is a constant. The case n D 1=2 corresponds to the simple scale-factor

variables. From (10.5),
fb; vg D 1 ; (10.10)
while the Hamiltonian constraint (10.6) is

1 v0 2.1Cn/
3
2.1 C n/2 2
HD .bv/2 C p2 C U.v; / ; (10.11a)
2V0 v 3
"
1Cn 1Cn
2 1 #
2 v v 3K
U D 2V0 V. / 2 : (10.11b)
v0 v0
10.2 Wheeler–DeWitt Quantum Cosmology
In the traditional WDW quantization, one promotes b and v to operators and the
Poisson bracket (10.10) to a commutator. This begins the construction of the mini-
superspace quantum theory:
O v
Œb; O D i: (10.12)
Since we are in „ D 1 units,

8 1
2 D 8l2Pl D 2
D 2: (10.13)
mPl MPl
At this point, we can choose two unitarily equivalent representations for the states
of the Hilbert space: one where the volume operator is diagonal or one where bO is
diagonal. We consider first the v representation, where
@
vO :D v ; bO :D i : (10.14)
@v
In the scale-factor representation, pO .a/ D i@a and Œ pO .a/ ; aO D i. In general, the
g
kinematical Hilbert space Hkin of the gravity sector is spanned by the basis fjvig of
eigenstates of the multiplicative operator v,
O which can be rendered orthonormal:
hvjv 0 i D ıvv0 : (10.15)
For the matter sector, one chooses a standard Schrödinger quantization with a
natural representation of the Hilbert space Hkin , the space of square-integrable
functions on R, on which O acts by multiplication and pO by derivation,
O :D @
; pO :D i : (10.16)
@
An orthonormal basis is given by
h j 0 i D 2 ı. 0
/; (10.17)
where the 2 factor is for later convenience (Sect. 11.4). The Hilbert space of the
g
coupled system is then just the tensor product Hkin ˝ Hkin .
Symmetry reduction of the Wheeler–DeWitt equation (9.88) or quantization of
the classical constraint (10.11) with the ordering vO bO vO bO yield the same result:

O Œv; D 1 v0
3
2.1Cn/ 2.1 C n/2 2 @2 @2
H C U Œv; D 0 :
2V0 v 3 .@ ln v/2 @ 2
(10.18)
The mini-superspace Hamiltonian constraint (10.18) is a sort of “Schrödinger
equation without time,” i.e., with no i„@t term. The role of physical time in general
relativity is played by some suitably chosen internal clock. In particular, the WDW
equation is hyperbolic and can be written as a second-order differential equation on
a two-dimensional configuration-space manifold. To get a cleaner expression, we
rewrite the square bracket in terms of the number of e-foldings N D ln a and set
n D 1=2:

@2 6 @2 6U
C 2 ŒN ; D 0 : (10.19)
@N 2 2 @ 2
10.2.1 de Sitter Solutions and Probability of Inflation
The type of information we would like to extract from the solution of the WDW
equation is about the probability in the space of initial conditions that inflation be
realized and whether the big-bang singularity is resolved. To get an idea of how
these solutions behave, we invoke the mini-superspace WKB approximation (9.94)
[13, 24, 28–32], valid only for long-wave-length gravitational modes [33].3 The
further requirement of decoherence excludes significant interference between dif-
ferent WKB solutions [24, 29, 39–41]. In particular, rather than considering a
superposition of expanding and contracting branches of a quantum universe, we
super-select only one of them (the expanding one, which we observe). Finally,
we assume that the wave-function slowly varies with respect to , which is
tantamount to considering a de Sitter background.
In this setting, the configuration space becomes one-dimensional. After rescaling
S ! V0 S in (9.98) and plugging (9.98) into (10.19), (9.99) becomes
s r r
6U 12 2N 3K
@N S ' ˙ D˙ e e2N V 2 ;
V02 2 2
so that
r 3
12 1 3K 2
S'˙ e2N V 2 : (10.20)
2 3V
Using this result and (9.103), we get

1 1 e2N V
0 D @N S@N 0 C 0 @2N S D @N S @N 0 C 1 C 0 ;
2 2 e2N V 3K= 2
giving, via (9.105), the semi-classical limit

" r 3 #
A. / i 12 2 3K 2
Œa; ' exp ˙ aV 2 (10.21)
a.a2 V 3K= 2 /1=4 3V 2
for a2 V 3 K= 2 > 0, while the limit in the classically forbidden region a2 V
3K= 2 < 0 is
" r 32 #
A. / 1 12 3K 2
Œa; ' exp ˙ a V : (10.22)
a.3K= 2 a2 V/1=4 3V 2 2
3
Other related methods were developed in [34–38].
If the scalar potential or cosmological constant is positive, V > 0, then exponential

solutions exist only for closed universes with small enough scale factor. There exist
oscillatory semi-classical solutions (with progressively damped amplitude) for a flat
or open geometry or for a closed geometry with sufficiently large scale factor. In
this case, the Hamilton–Jacobi momenta (9.100) give the classical equations in the
extreme slow-roll approximation.
The signs in (10.20), (10.21) and (10.22) and the coefficients of linear combi-
nations of solutions depend on the choice of boundary conditions. The closed case
K D 1 is of special interest because one can construct solutions obeying the no-
boundary or tunneling conditions discussed in Sect. 9.2.3. Analytic solutions exist
for a pure de Sitter background (2.111), where 2 V D D 3H 2 is the cosmological
constant. In that case, one neglects the field derivative term in (10.19) and the
potential is
2
6U 6V0
D a4 .H 2 a2 1/ : (10.23)
2 2
To solve the WDW equation exactly, one exploits the ambiguity in the ordering
choice of aO and pO .a/ factors and augments the second-order derivative in the scale
factor in (10.19) by a friction term, @2 =@a2 ! @2 =@a2 a1 @=@a. Then, imposing
Œa D 1 D 0, one finds the Hartle–Hawking wave-function [42–44]
AiŒz.a/
HH Œa D A ; (10.24)
AiŒz.0/
where A is a normalization constant, Ai is the Airy function (solution of @2z z D

0) and z.a/ D Œ3V0 =.H 2 2 /2=3 .1 H 2 a2 /. Setting V0 D 1 and using the asymptotic
expansions of Ai,

z!C1 1 1 2 3
Ai.z/ ' p z 4 exp z 2 ;
2 3

z!1 1 14 2 3
Ai.z/ ' p .z/ cos .z/ 2 ;
3 4
one obtains the approximate form of HH in the semi-classical and classically
forbidden ranges [5, 45]:

1 2 2 2 3
HH Œa cos .H a 1/
2 ; H 2 a2 > 1 ;
.H 2 a2 1/1=4 H 2 2 4
(10.25a)

1 2 3
HH Œa 2 2 1=4
exp 2 2 .1 H 2 a2 / 2 ; H 2 a2 < 1 :
.1 H a / H
(10.25b)
The wave-function in the tunneling proposal is complex-valued [42]:
AiŒz.a/ C i BiŒz.a/
V Œa D A ; (10.26)
AiŒz.0/ C i BiŒz.0/
where Bi is the second solution of the Airy equation, with asymptotic limits

z!C1 1 1 2 3
Bi.z/ ' p z exp
4 z 2 ;
3

z!1 1 1 2 3
Bi.z/ ' p .z/ 4 cos .z/ 2 C :
3 4
Thus,

1 2i 2 2 3
V Œa exp 2 2 .H a 1/ 2 ; H 2 a2 > 1 ;
.H 2 a2 1/1=4 H
(10.27a)

1 2 3
V Œa exp .1 H 2 a2 / 2 ; H 2 a2 < 1 :
.1 H 2 a2 /1=4 H2 2
(10.27b)
Figure 9.1 shows the wave-function (10.24) and the real part of the wave-
function (10.26).
We now turn to the problem of whether WDW quantum cosmology predicts
sufficiently long inflation. For the tunneling proposal, it is easy to argue for an
affirmative answer [43]. Let us revert to a scalar-field cosmology in the ESR
approximation, so that H 2 ' 2 V. /=3. The probability density function P. i /
of the initial state of the Universe is the nucleation probability given by the ratio
of the squared wave-function at the classical turning point a D H 1 and at
a D 0, P. i / j ŒH 1 ; = Œ0; i j2 j Œ0; i j2 / expŒ4=.H 2 2 /, where
we used (10.25b) and (10.27b). Thus,

12 12
PHH . i / D exp 4 ; PV . i / D exp 4 : (10.28)
V. i / V. i /
It is worth stressing that the main difference between the probability densities of
the no-boundary and tunneling proposals amounts to just a sign only in the de
Sitter mini-superspace approximation. The Hartle–Hawking wave-function is the
path-integrated exponential of the Euclidean action I < 0 under standard Wick
rotation t ! it, while Vilenkin’s wave-function is a Lorentzian path integral with
Lorentzian action S:
Z Z
HH eI ; V eiS :
For the de Sitter solution, in the tunneling regime these correspond to probability
densities
PHH eI ; PV ejIj :
Both
R CI cases are different from Linde’s t ! Cit prescription [46], leading to L
e and PL eCI PV . The fact that I < 0 in the de Sitter case has sometimes
led to confusion between Linde’s and Vilenkin’s prescriptions [47].
In the tunneling case, PV is maximal but finite at large V, corresponding to
large values of for the monomial potentials of chaotic inflation. Large-field
potentials may then be regarded as typical in realizations of the quantum universe,
although the tunneling wave-function can describe viable small-field inflation
as well (see [48] for the case of natural inflation). The no-boundary proposal
has the opposite behaviour and its probability distribution is strongly peaked at
small potential values, where it diverges (this is not a problem per se, as P is a
probability density). Therefore, PHH is not normalizable unless the scalar field has
a finite range 2 Œ 1 ; 2 . However, maximal probability would correspond to
the absolute minimum of the potential, a condition incompatible with inflationary
dynamics. This result [43] can be refined for a monomial potential (e.g., [49]) but
it holds also for small-field potentials [50, 51]. Later studies, however, reassessed
the way conditional probabilities are weighted in quantum cosmology, apparently
reconciling the Hartle–Hawking no-boundary proposal with sustainable inflation
[52–54]. The same conclusion can be reached by another mechanism, by including
one-loop corrections to the effective action [56–59], also in the case of a non-
minimally coupled scalar field [55, 56, 60, 61]. In general, the modified probability
density function is peaked at configurations favouring eternal inflation in the no-
boundary case, and standard inflation in the tunneling case.4
10.2.2 Massless Scalar Field and Group Averaging
The WKB solutions are useful tools to study the structure of the quantum theory, but
it would be desirable to construct the full Hilbert space of physical states in exactly
solvable models. A case of particular interest in this respect is that of a flat geometry
with a massless scalar field, V. / D 0 D K.
4
The probability issue can be tackled also from other perspectives, some of which are adopted
when facing the cosmological constant problem in the context of eternal inflation (Sects. 5.6.5
and 10.2.1). Gibbons and Turok proposed a classical measure for classical cosmological trajec-
tories [62] which, apparently, disfavours inflation [62]. In fact, in the simplest case of a single
slow-rolling scalar field, this classical probability of having N e-foldings of inflation contains a
damping factor e3N which makes long inflationary periods (N 50 – 60) highly improbable.
There are several caveats and criticisms regarding this measure [63–65], including about the way
initial conditions are imposed and the fact that they are set in regimes where quantum effects should
be taken into account.
For later comparison with loop quantum cosmology, we take the b representation
where (10.14) is replaced by the multiplicative operator bO :D b and
@
vO :D i : (10.29)
@b
Then, keeping the inverse v factor in (10.11) to the left and the ordering .bv/2 !
bO vO bO v,
O the Hamiltonian constraint operator reduces to

O Œ y; :D C @2 Œ y; D 0 ;
C (10.30)
where
s
3 b
:D @2y ; y :D 2 2
ln ; (10.31)
2.1 C n/ b0
and b0 is an integration constant. A first consequence of the absence of a potential

is that the operator annihilating commutes with the parity operator changing the
orientation of the triad, so that both wave-functions encoding either orientation are
solutions. Since physical observables should not depend on the triad orientation, we
can impose to be even under parity in the v representation, which is equivalent
to Œb D Œb in b representation (this can be seen from the Fourier transform
connecting the two representations). Here, however, we have to restrict the sign of b
so that (10.31) be well defined. Therefore, (10.30) implicitly requires a prior fixing
of the frame orientation.
A second consequence is that the scalar field appears only in the second-
derivative term, while the operator is -independent. The WDW equation takes
the form of a massless Klein–Gordon equation where plays the role of time. In
fact, the absence of a potential guarantees that is monotonic throughout the whole
evolution of the wave-function, thus constituting a reliable internal clock.
The operator is self-adjoint and positive definite on the space of square-
integrable functions on the real line; its eigenfunctions are e! . y/ D e˙i!y with
eigenvalues ! 2 and obey the orthonormality relation
Z C1
dy
e . y/ e! 0 . y/ D ı.!; ! 0 / (10.32)
1 2 !
and the completeness relation

Z C1
dy e! . y/ Œ y; D 0 8! , Œ y; D 0 : (10.33)
1
Any solution can be decomposed into a positive-frequency and a negative-

frequency part. The group averaging procedure [66–73] is one way to find this result
[74]. One considers the sub-space S of rapidly decreasing functionals . y; /,
dense in the Hilbert space of square-integrable functionals over the mini-superspace.
These are not physical states unless they satisfy the scalar constraint (10.30).
The latter is implemented by the Schwinger representation (i.e., a one-parameter
O
integral) of the formal operator 2ı.C/:
Z C1
O
Œ y; D Œ D d˛ ei˛C . y; / : (10.34)
1
The first equality stresses that physical states can be regarded as distributions on S:
is the average of over the volume of the one-dimensional group R generated
by the unitary operator ei˛CO , hence the name of “group averaging.” Expanding this
expression in Fourier modes ! and k (eigenvalues of @y and @ ), the integral in ˛
becomes the Dirac delta distribution 2ı.! 2 k2 / D 2Œı.j!j C k/ C ı.j!j
k/=.2jkj/. Indicating as .!; k/ the Fourier transform of . y; / and performing
the integral in k, one obtains
Œ y; D C Œ y; C Œ y; ; (10.35a)
Z C1
d! i!y ˙ij!j
˙ Œ y; :D e e ˙ .!/ ; (10.35b)
1 2
where ˙ .!/ D .!; ˙j!j/=.2j!j/. The auxiliary Hilbert space S is thus

decomposed into positive-, null- and negative-frequency sub-spaces, S D SC ˚S0 ˚
S . In turn, each positive- and negative-frequency wave-function is decomposed
into “left-moving” and “right-moving” components having support on the negative
(respectively, positive) !-axis:
˙ Œ y; D ˙;L Œ y˙ C ˙;R Œ y ; (10.36a)

Z C1 Z 0
d! i!y d! i!y
˙;L Œ y˙ :D e ˙ .!/
˙ .!/ D e ˙
˙ .!/ ; (10.36b)
1 2 1 2
Z C1 Z C1
d! i!y d! i!y
˙;R Œ y :D e .!/ ˙ .!/ D e ˙ .!/ ; (10.36c)
1 2 0 2
where y˙ D y ˙ and is the Heaviside step function: .x/ D 0 for x < 0 and
.x/ D 1 for x > 1. Individually, ˙ obey the “square root” of the WDW equation,
which is a first-order Schrödinger-type equation:
p
i@ ˙ D ˙ : (10.37)
Let D 0 be the initial internal time where the system starts its evolution with
wave-function Œ y; 0 . Then, (10.37) evolves the state into the solution
p
˙ Œ y; D e˙i . 0/
˙ Œ y; 0 : (10.38)
From (10.34), one would be tempted to define the scalar product

Z Z
i˛ 0 CO i˛ CO
.1 ; 2 / d˛ d˛ 0 h 1 je e j 2i ;
where h j i is the scalar product in S, but this quantity is clearly divergent (perform
one integration, or just invoke invariance of the Haar measure over the group
manifold to factorize the infinite group volume). Instead, the physical inner product
in the general group averaging procedure is defined as
Z
i˛ CO
.1 ; 2 / :D d˛h 1 je j 2i : (10.39)
In the case under examination, the Hilbert space of physical states is naturally
equipped with the inner product on constant slices:
Z
ˇ
.1 ; 2 / :D i dy 1 @ 2 2 @ 1 ˇ D 0
: (10.40)
In terms of the cross-products of negative- and positive-frequency parts,

Z C1
d! h 1 j!j˙2 j!j/ 2 j!j˙1 j!j/
i
.1˙ ; 2˙ / D ˙2 ei. ˙1 ei.
1 2

1˙ .!/j!j 2˙ .!/ (10.41)
and one has four terms:

Z C1
d!
.1C ; 2C / D 2 1C .!/j!j 2C .!/ ; (10.42a)
1 2
Z C1
d!
.1 ; 2 / D 2 1 .!/j!j 2 .!/ ; (10.42b)
1 2
Z C1
d!
.1C ; 2 / D 2i sin.2j!j / 1C .!/j!j 2 .!/ ; (10.42c)
1 2
Z C1
d!
.1 ; 2C / D 2i sin.2j!j / 1 .!/j!j 2C .!/ : (10.42d)
1 2
The inner product (10.40) is formally equivalent to the one for a relativistic particle
but other choices are possible [75–77]. For instance, the “non-relativistic” form
Z
.1 ; 2 /nrel :D dy 1 Œ y; 0 2 Œ y; 0 (10.43)
corresponds to defining the functionals ˙ over the auxiliary Hilbert spaces

0
S˙ 3 2j!j ˙ .!/ rather than on S˙ . The two theories equipped with the inner
products (10.40) and (10.43) lead to unitarily equivalent representations of the
algebra of Dirac observables, so that there is no preferred choice from the point
of view of physics.
The wave-function (10.35) is non-normalizable with respect to the scalar prod-
uct (10.40) because the eigenvalue ! D 0 is in the continuous part of the spectrum
of the operator. In fact, the sum of the terms (10.42) is not positive definite.
The problem can be avoided if there exists a super-selection rule allowing us
to consider only the positive- or the negative-frequency sector. This rule can be
realized by identifying a complete set of Dirac observables which preserve the
two sectors separately [12, 74, 78]. In the present case, the self-adjoint operators
realizing these Dirac observables can be chosen to be pO D i@ and the self-adjoint
part of the volume operator vO D i@b [78]. Therefore, we can elect the positive-
frequency wave-functions C to be the actual physical states. Setting D 0,
the inner product (10.40) reduces to (10.42a). Moreover, also the left- and right-
moving sectors are preserved and they can be studied separately. This means that the
expanding and contracting cosmological branches never interfere with each other.
10.2.3 Quantum Singularity
What happens to the big bang singularity? The original hope when the WDW
formalism was proposed was that, in general, the wave-function solving the
Hamiltonian constraint is D 0 at the big bang, so that the probability for the
universe to hit the initial singularity is zero [1]. To solutions with this property,
there would correspond a set of semi-classical effective equations of motion solved
by an effective, non-singular scale factor. Although these solutions exist (e.g.,
[38]), the problem is the same as in classical cosmology: How “typical” are
they? Classical singularity theorems show that, roughly speaking, non-singular
solutions are the exception to the rule. In quantum mini-superspace models, we
do not have such powerful results to achieve the same conclusion, but the fact
that both the no-boundary and tunneling wave-functions discussed in Sect. 10.2.1
are non-vanishing at a D 0 is not encouraging. At small scales, however, the
WKB approximation breaks down and spacetime becomes “fuzzy,” a regime where
singularities could possibly be smeared. This is roughly what we would expect in
a quantum-mechanical universe where Heisenberg uncertainty principle applies to
quanta of matter and spacetime. It is difficult to make these concepts more precise
in second-order formalism.
The issue of the singularity is still unsettled in WDW cosmology. We begin

by reviewing the cons, some general arguments suggesting that the big bang is
not resolved, then moving onto the pros. Consider the massless scalar model of
the previous section and a classical trajectory . cl ; p ;cl ; ycl ; vcl /. When followed
backwards or forwards in time, the classical system falls into a big-bang singularity.
In fact, the classical expanding and contracting solutions in synchronous time are,
up to integration constants,
r r ! 13
2 3 2
cl˙ .t/ D˙ ln t ; acl˙ .t/ D ˙ t : (10.44)
3 2 2
This can be recast as a solution in phase space, N / ln v, where N is a constant.

Also, is monotonic and can be used as an internal clock throughout the whole
evolution.
Take now a Gaussian semi-classical physical state (positive frequency, subscripts
C omitted) peaked at this trajectory at the time D 0 :
Z C1
d! i!y ij!j. .!!cl /2
0 cl /
sc Œ y; 0 D e e sc .!/ ; sc .!/ D e 2 ;
1 2
(10.45)
where the variance is constant and !cl D p ;cl is one of the two branches.
Equation (10.38) determines univocally the state Œ y; , which is simply (10.45)
with 0 replaced by . Therefore, a semi-classical state always remains such all the
way through the classical trajectory and up to the big-bang or big-crunch singularity.
We can conclude that, at least in the massless model, the big-bang problem also
affects the WDW quantum evolution.
The same result holds actually for any physical state, not only semi-classical.
Since the left- and right-moving sectors are super-selected, they must be considered
separately rather than in the joint expression (10.184).5 The expectation value of jvj
O
on (a dense sub-set of) physical states at a time for each sector is (Problem 10.2)
hjvji O C;L / D VL e0 ;

O L D .C;L ; v (10.46)
hjvji O C;R / D VR e0 ;
O R D .C;R ; v (10.47)
where VL;R are two positive constants. Consider first the left sector. The expectation
value of the volume goes from zero at D 1 to infinity at the infinite future;
this corresponds to a quantum big-bang singularity. On the other hand, for the right
sector one starts with an infinitely large universe and eventually encounters a big-
crunch singularity at D C1.
In the presence of a potential, it is possible to obtain quantum wave-functions
that vanish at a D 0 and avoid the big bang [79]. This result depends on the form of
5
The reason is that the frame orientation has been fixed at the very beginning.
the potential V. / near the classical singularity but it is not difficult to find specific
examples.
A final answer may demand further proof beyond the FLRW mini-superspace
isotropic approximation. A natural direction to look at are Bianchi IX wave-
functions and the quantum version of the mixmaster universe [2, 80–82]. For a
small-enough realistic universe, it is not possible to follow a semi-classical evolution
down to the singularity and one does not expect any counterpart to the classical
infinite series of oscillations. Nonetheless, one can study the evolution of wave
packets and the expectation values of anisotropic variables [83–90], especially with
the methods of quantum billiards [86, 88, 89, 91–93]. Quantum theories can be
chaotic in a precise sense [94, 95]. For diagonal Hamiltonians, simple dynamical
variables are represented by pseudo-random matrices with many non-zero entries.
The expectation values of these variables tend to equilibrium values which are
independent of the initial conditions. The quantum mixmaster universe belongs
to this class of models, so that the chaotic nature of the BKL singularity is
thus preserved, mutatis mutandis, at the quantum level. This is not the case in
string theory, where chaos and quantum effects at ultra-microscopic scales could
eventually solve the big-bang problem (Sect. 13.9.3).
10.2.4 Cosmological Constant and the Multiverse
The probability (10.28) to have inflation also gives valuable information about the
cosmological constant problem. Paradoxically, here the tunneling proposal fares
worse than the no-boundary one: what was a virtue of the former and a (solvable)
problem for the latter turns out to be the other way around when applying the
same formulæ to the cosmological constant. Interpreting D = 2 today as the
effective contribution generated by all matter fields near their ground states, we
assume that matter is non-dynamical, such as a scalar field without kinetic term or a
3-form field [96–99] in four dimensions, and replace V in (10.28) with this effective
cosmological constant. The tunneling proposal predicts large values of as the
most probable:

3m2
PV ./ D exp Pl : (10.48)
2
A similar conclusion (probability density peaked away from D 0) is reached in

tunneling models where matter fields compensate a negative cosmological constant
to produce a positive effective cosmological constant [33]. On the other hand, for
the no-boundary proposal one gets [100–104]:

3m2Pl
PHH ./ D exp : (10.49)
2
Thus, the most probable configuration for the Hartle–Hawking wave-function is

D 0, in regions where fields are near their ground state.6 Notice the similarity
with the probability distribution (7.138) obtained in generic holographic theories of
gravity.
However, a -dependent normalization of the wave-function may erase the effect
or, in other words, the latter is only a normalization artifact. One cannot check this
point in canonical theories, which are linear in the wave-function. In some loose
sense, this issue is related to the absence of interactions between wave-functions
describing different universes. Turning on these interactions is tantamount to going
beyond quantum mechanics and promoting wave-function to fields. This is a second-
quantization procedure but, since the degrees of freedom of a single universe
are already fields, eventually to be quantized, such a framework is sometimes
called of “third quantization” [44, 105–124], recently revived also in the context
of group field cosmology [125–131]. While in canonical schemes geometry is
fully dynamical but the topology of the Universe is fixed by construction, in third
quantization the Universe is allowed to change topology and, in particular, to branch
into disconnected components. This interacting multiverse scenario obeys a set of
quantum rules which greatly vary from model to model. Various early realizations
of a branching multiverse seemed to solve or alleviate the cosmological constant
problem: predictions on the most probable value of range from D 0 (with
0 < 1 requiring some amount of fine tuning) [106, 107, 112, 115]7 to small
[105, 111], as in the single-universe probability (10.49) [5, 100]. Most of these
results, however, are obtained in Euclidean gravity [106, 107, 111, 112, 115] and
Lorentzian signature can even spoil the peak at D 0 in the probability density,
making the cosmological constant in third-quantized scenarios a random variable
[116]. Also, as recalled in Sect. 7.2.3, nowadays we would like to obtain a very
small but non-zero cosmological constant, a requirement in tension with models
with D 0.
To circumvent these problems (while creating others), one can exploit the fact
that in eternal inflation there are infinitely many inflating regions, each with a
different value of the true vacuum of the scalar field. Combining this picture with
some anthropic arguments [135–139], one may conceive that it is highly probable
that we live in one such region with a scale reproducing the observed value of the
cosmological constant. As seen in Sect. 7.1.1, the inflationary mechanism turns out
to be too short a blanket to solve both the inflationary and problems at the same
time and cosmic inhomogeneities should then be explained by an alternative to (or
modification of) inflation. Another problem with this model is that the probability
6
The same conclusion might be reached also in scalar-tensor models, both for Vilenkin’s and
Hartle–Hawking’s boundary conditions [16, 19, 20], albeit with some recent reservations [21].
7
This probabilistic enhancement of the D 0 configuration, known as Coleman mechanism, has
been extended also to Einstein–Yang–Mills dynamics [132] and scalar-tensor theories [133, 134].
The latter are not favoured with respect to general relativity.
measure is highly speculative and an endless source of debate (see, e.g., [140–145]
and bibliography in Sect. 5.6.5).
The main message one may get from these efforts is that, in general, quantum
cosmology offers a probabilistic interpretation of the cosmological constant rather
than explaining why takes the value (7.11). In our realization of the universe,
(7.11) holds but could have taken any other value among the most probable ones.
In the best-case scenario, no fine tuning is needed to achieve extremely small values
such as (7.11). Similar considerations hold for quantum models lying outside the
canonical or the third-quantization framework, as in proposals based on decoherence
[146].
10.2.5 Perturbations and Inflationary Observables
Inflationary perturbations and the associated spectra allow us to track down quantum
corrections and confront them with the observed CMB power spectrum. Although
the outcome of this procedure is a constraint on the free parameters of the models
rather than an actual prediction, as a minimal present-day achievement we can at
least state that quantum cosmology WDW models are compatible with observations.
10.2.5.1 Perturbations
When inhomogeneities are switched on, the FLRW mini-superspace framework

breaks down and one should consider the full Dirac Hamiltonian (9.75) for gravity
and a scalar field. Since the super-momentum and super-Hamiltonian constraints
are non-linear in the canonical variables, the problem quickly becomes intractable
unless one resorts to some approximations. To obtain the inflationary spectra, linear
perturbation theory is sufficient. For a closed universe, a full treatment begins with
the expansion of scalar, vector and tensor perturbations in spherical harmonics
over the 3-sphere [147], thus extending mini-superspace to the infinite number of
degrees of freedom of superspace. Eventually, the expansion in spherical harmonics
can be combined with the WKB approximation to find solutions of the WDW and
diffeomorphism equations [28, 39, 43, 147–149]. In this section, however, we ignore
both vector modes and the metric back-reaction ıg , in which case the scalar is
regarded as a test field. In the standard cosmological model, back-reaction does not
affect the power spectrum at lowest order in perturbation theory and in the slow-roll
truncation. This suffices for our purposes also in WDW quantum cosmology. The
scalar perturbation is decomposed into Fourier modes,
X
ı .t; x/ D ı k .t/ eikx ; (10.50)
k
where we assumed spatial slices to be compact ( K D 1) and that each Fourier mode
depends on k D jkj. Replacing .t/ with .t; x/ in the WDW equation (10.19)
for a quadratic potential V. / D 12 m2 2 , the mini-superspace is augmented by

the infinity of modes ı k . The wave-function ŒN ; ; fı k gQ k can be factorized
as a background plus the rest, ŒN ; ; fı k gk D 0 ŒN ; k>0 k ŒN ; ; ı k .
In doing so, one drops self-interaction terms which are consistently negligible in
first-order perturbation theory. Eventually, ignoring the curvature term one obtains
[28, 147]
" #
@2 6 @2 6N 6H 2 6 6N 2 4N 2
2
Ce C 2 e m Ce k ı k k ' 0;
@N 2 2 @ı k2 2
(10.51)
where k ŒN ; ı k D 0 ŒN ; k ŒN ; ; ı k and the dependence is omitted
because we used the slow-roll approximation 6H 2 = 2 ' m2 2 to express the
background potential in terms of the Hubble parameter.
Noting that N and ı k correspond, respectively, to slow- and fast-evolving
variables, at this point one can make a Born–Oppenheimer approximation on the
solution [150, 151]. The latter is written as
k ŒN ; ı k D expŒiS.N ; ı k / (10.52)
and the functional S is expanded in m2Pl D l2Pl

D 8= 2 : S D m2Pl S0 C S1 C m2
Pl
S2 C
: : : . Plugging the Ansatz (10.52) into (10.51) and expanding, the O.m4Pl / and O.m2Pl /
terms imply S0 D ˙e3N H=.4/ (the equivalent of (10.20)), while at the next two
orders one finds
.0/
k ŒN ; ı k :D A.N / eiS1 .N ;ı k/
; (10.53)
.1/ .0/ im2
Pl S2 .N ;ı k/
k ŒN ; ı k :D B.N / k ŒN ; ı k e ; (10.54)
where A and B are chosen to match the amplitudes in the WKB approximation.
10.2.5.2 Observables
.0/ .1/
The wave-functions k and k can be found semi-analytically [152–154]. From
the explicit solutions, one can calculate the two-point correlation function
.n/ .n/ 2 .n/

P .k/ :D h k j jı k j j k i (10.55)
of the scalar perturbation order by order. This quantity is directly related to the
imprint (5.110) of inhomogeneous fluctuations in the cosmic microwave back-
ground, when approximated to the long-wave-length limit k k and evaluated
at k D k . This is the n-th order power spectrum
k3 .n/ ˇ
Ps.n/ .k/ :D 2
P .k k /ˇkDk : (10.56)
2
.0/
The lowest-order result Ps coincides with the classical one (5.152).
The next-to-lowest-order expression is the standard one times a quantum correc-
tion:
Ps .k/ ' Ps.1/ .k/ D Ps.0/ .k/Ck2 : (10.57)
.1/
For k , there are actually two solutions with the same boundary condition, giving
rise to quantum corrections with opposite sign and slightly different size:8
3 2
9:2 40:1
.Ck /2 ' 1 .lPl H/2 1 .lPl H/ 2
.k=k0 /3 .k=k0 /3
4 4
52:6 2 l H
D 1 3
.lPl H/ C O Pl 6 ; (10.58)
.k=k0 / k
3 2
11:5 1:7
.CkC /2 ' 1 .l Pl H/
2
1 C .lPl H/
2
.k=k0 /3 .k=k0 /3
4 4
38:0 l H
D 1C 3
.lPl H/2 C O Pl 6 ; (10.59)
.k=k0 / k
where k0 is some reference scale. We can write the leading Wheeler–DeWitt

quantum correction as
3
k0
.Ck˙ /2 ' 1CıWDW
˙ 2
.k/CO.ıWDW /; ˙
ıWDW .k/ :D ˙50 .lPl H/2 : (10.60)
k
The approximation scheme used to derive (10.58) and (10.59) breaks down in the
limit Ck ! 0 and the critical k at which that happens should not be taken as a
physical threshold. While Ck ! 1 in the small-scale limit (k ! 1), at large scales
(k k ) the quantum-corrected power spectrum acquires a mild scale dependence
which suppresses () or enhances (C) the signal with respect to the standard result.
The imaginary part of the solution is discontinuous; if one demands continuity,
then the prediction of the model is a power enhancement, as in loop quantum
cosmology (Sect. 10.3.9). In the case of suppression of the spectrum, a similar effect
happens also in other models where a bounce is present [155, 156] or geometry is
quantized, such as non-commutative and string inflation [157–159]. It might seem
counter-intuitive that quantum gravity affects large scales more than small scales.
However, large-scale perturbations left the horizon before (and hence re-entered
after) smaller-scale fluctuations and they were longer exposed to quantum effects.
From the power spectrum and using (5.51) and (5.141), we get the scalar spectral
index (4.58) and its running (4.59). Since at horizon crossing d=d ln k ' d=.Hdt/
8 Q Pl2 D
Numerical coefficients differ from those in [154], where the rescaled Planck mass M
.3=2/m2Pl is used instead of (1.4).
and H ' const, we obtain

˙
dıWDW ˙
' 3ıWDW (10.61)
d ln k
and [154, 160]
˙
ns 1 ' 2 4 3ıWDW ; (10.62)
2 2
˙
˛s ' 2 5 4 C 9ıWDW ; (10.63)
where we dropped higher-order terms in the combined ıWDW /slow-roll expansion.

The slow-roll part of these expression reproduces the classical results (5.153)
and (5.155). Using the slow-roll tower (5.55) and relations (5.65), equations (10.62)
and (10.63) are recast as
˙
ns 1 ' 6V C 2V 3ıWDW ; (10.64)
2 2 ˙
˛s ' 24V C 16V V 2V C 9ıWDW : (10.65)
The scalar power spectrum expanded to all orders in the perturbation wave-
number about a pivot scale k0 is the generalization of (4.67),
1
˛s .k0 / 2 X ˛s .k0 / m
.m/
ln Ps .k/ D ln Ps .k0 / C Œns .k0 / 1x C x C x ; (10.66)
2 mD3
mŠ
where x :D ln.k=k0 /. As the order of the observables
dm2 ˛s
˛s.m/ :D ˙
D O. m / C .3/m ıWDW (10.67)
.d ln k/m2
increases, the classical part becomes smaller and smaller but the leading-order
quantum correction survives. At some order m, the quantum correction dominates
over the standard part. Taking (10.67) into account, (10.66) can be recast as
" 3 #
k0
ln Ps .k/ ' ln Ps.0/ .k/ ˙
ıWDW .k0 / 1 : (10.68)
k
10.2.5.3 Experimental Bounds
The typical energy scale during inflation is estimated to be about the grand-
unification scale, H 1015 GeV, corresponding to an energy density infl
H 2 =l2Pl 1068 GeV4 . In contrast, classical gravity is believed to break down at
distances shorter than the Planck length lPl , i.e., at energies above mPl 1019 GeV.
The ratio between the inflationary and Planck energy density is very small,
infl
.lPl H/2 108 ; (10.69)
Pl
and quantum corrections are expected to be of the same order of magnitude or lower,
well below any reasonable experimental sensitivity threshold. WDW quantum
cosmology realizes precisely this type of corrections: their size is set by the energy
scale of inflation.
To make this argument more precise, we choose the reference scale k0 in (10.60).
One possibility is k0 D kmin 1:4 104 Mpc1 , the largest observable scale.
Via (4.38), one can re-express ıWDW in terms of spherical multipoles. The lowest
early-universe contribution to the CMB spectrum is the quadrupole ` D 2, so that
k=kmin D `=`min D `=2. A more generous estimate for the quantum correction is
obtained by replacing kmin by the pivot scale k0 kmin of a CMB experiment, which
we adopt from now on. From the PLANCK bound (5.196) for the Hubble parameter
at k0 D 0:002 Mpc1 , the WDW quantum correction is constrained to be
˙
jıWDW .k0 /j < 3 109 : (10.70)
˙
With kmin instead of k0 , the quantum correction is further suppressed, jıWDW .k0 /j <
13
7 10 . As anticipated, quantum corrections are too small to be detected. Their
dependence on the inflationary energy scale is crucial for this result. Another reason
is that, at large scales, cosmic variance is the leading source of error. Quantum-
gravity effects should be compared with the error bars due to cosmic variance (4.34)
.0/
with respect to the classical spectrum Ps .`/. The latter is determined up to the
normalization Ps .`0 /, so that the region in the .`; Ps .`/=Ps.`0 // plane affected by
cosmic variance is roughly delimited by the two curves
q !
.0/
Ps .`/ ˙ 2 .0/ .`/ r .0/
Ps 2 Ps .`/
.0/
D 1˙ .0/
; (10.71)
Ps .`0 / 2` C 1 Ps .`0 /
where we take the classical spectrum as a reference. The WDW-corrected spectrum

is given by (10.68) with the classical spectrum (4.67) truncated at second order,
since slow-roll parameters are at most of order D O.102 /. To plot the WDW
spectrum, we only need to plug in values for the scalar index and its running. These
are yielded by (10.64) and (10.65) for a given potential.
The quadratic potential V. / / 2 is under strong experimental pressure, but it
illustrates well the effect of quantum-gravity corrections. In that case, from (5.85a)
and (5.85b),
2
V D ; V D V ; V2 D 0 :
2 2
10.3 Loop Quantum Cosmology 489
Fig. 10.1 Log-linear plot of the Wheeler–DeWitt primordial scalar spectrum (10.68) for a
quadratic inflaton potential, with V .k0 / D 0:009 and for the pivot wave-number k0 D
0:002 Mpc1 , corresponding to `0 D 29. The shaded region, delimited by the two curves (10.71),
is affected by cosmic variance. The inset shows the negligible difference between the standard
.0/
“classical” spectrum Ps .`/ (dotted line) and the spectrum Ps .`/ with Wheeler–DeWitt quantum
C
correction ıWDW (solid line, enhanced spectrum) and ıWDW (dashed line, suppressed spectrum), at
˙
2 < ` < 3. Quantum corrections are magnified to jıWDW .k0 /j D 105 in order to show their effect
(Credit: [160])
This allows one to reduce the slow-roll parameters to just one. A realistic theoretical
value for V at the pivot scale is V .k0 / D 0:009. As shown in Fig. 10.1, WDW
quantum corrections are extremely small even in the most conservative estimate and
they are completely drowned by cosmic variance. The loop quantization inspired by
loop quantum gravity will produce corrections potentially much larger that those
of WDW cosmology. The intuitive reason will be given in Sect. 10.3.8, while a
comparison with experiments along the same lines above is in Sect. 10.3.9.
10.3 Loop Quantum Cosmology
Loop quantum gravity aims to quantize the gravitational interaction in a rigorous

and consistent way. At the classical level, it relies on the Hamiltonian formalism
in Ashtekar–Barbero canonical variables and features the real Barbero–Immirzi
constant . While the understanding of physical consequences implied by full LQG
is still an open research topic, insight has been gained by studying the theory in
symmetry-reduced spacetimes. In particular, loop quantum cosmology (LQC) opens
up the possibility of resolving the singularities that plague classical cosmological
spacetimes [161–166]. For a spatially flat FLRW background with a massless
scalar field, the model can be analyzed rigorously in terms of physical observables
[167]. Although the symmetry reduction is performed at the classical level as in
standard mini-superspace quantization, the techniques used in LQC closely follow
those of loop quantum gravity. Consequently, LQC quantization is inequivalent to

the standard Wheeler–DeWitt quantization and, thus, it can lead to significantly
different predictions [166].
In fact, most of the conceptual framework of the previous sections is rather
general and its applications go beyond the scope of the ADM formalism. Having
reviewed some basic aspects of loop quantum gravity in Sect. 9.3, we can proceed
with the LQC construction of kinematical operators and the Hamiltonian constraint
on a spatially-flat, homogeneous and isotropic background. Then, the big-bang
problem is faced at a fourfold level: (a) looking at the spectrum of the kinematical
volume operator, (b) studying how zero-volume physical modes decouple from the
Hamiltonian evolution, (c) computing the expectation value of the volume operator
on physical states and (d) showing that the effective cosmological dynamics displays
a bounce rather than a singularity. All these approaches agree. The interest in
loop quantum cosmology, however, is not exhausted by the big-bang problem and
modifications to the standard inflationary picture may arise. Although the basic
idea of inflation as a dynamical scalar field remains the same, geometric quantum
corrections do change cosmological perturbation theory and the observable spectra.
Again, we will limit the discussion to a minimally coupled scalar field, although
scalar-tensor and f .R/ systems can also be contemplated (Problems 7.1, 7.2, 7.3,
7.4, 7.5 and [168–170]).
10.3.1 Classical FLRW Variables and Constraints
As in the ADM case, the symplectic structure is reduced at the classical level. We
consider only the K D 0 case. We already had occasion to mention that the volume
of a spatially flat universe is infinite and, to render the volume integral (9.112) well
defined, we need to consider a patch of the universe with finite fiducial comoving
volume V0 . To the fiducial patch there correspond a fiducial triad o e˛i and co-triad
e˛ with fixed orientation. The spin connection ˛i vanishes for spatially flat slices,
o i
so that in this background the symmetric Ashtekar–Barbero connection is just the

extrinsic curvature, which is expressed as
1=3
Ai˛ D cQ o ei˛ ; cQ :D cV0 D aP : (10.72)
The relation c / aP is valid only at the classical level and it will be modified later
in the case of the effective quantum dynamics. The symmetry-reduced conjugate
momentum is
2=3
Ei˛ D pQ jo ej o e˛i ; pQ :D pV0 D a2 : (10.73)
1=3 2=3
The definitions of c D cQ V0 and p D pQ V0 are suggested by the holonomy and
flux expressions (9.107) and (9.111) in the full theory.
In view of implementing the various quantization ambiguities and parametriza-
tions, it is convenient to define a new pair of variables as [78, 167]
c
N 6 p
bD ; vD 2
; (10.74)
2 .1 C n/ N
where N is an arbitrary dimensionless function of a. For a power law,

n a
p 2n
N D D ; (10.75)
p a
where n 2 R and p and a are, respectively, constants of dimension Œ p D 2

2.1Cn/=3 n 2=3
and Œa D 0. In particular, v0 D 2V0 p D 2V0 a2n in (10.9) and v is
dimensionless. It is easy to show that b and v are canonically conjugate and (10.10)
holds (notice that fc; pg D 2 =3).
The mini-superspace Hamiltonian constraint in these variables is (10.11). When
n D 0, it is expressed directly in terms of the symmetry-reduced connection and
densitized triad:
3 p 2
HD pc CH :
2 2
(10.76)
The goal of this section is to recast H as a functional of holonomies and positive

powers of the triad.
In the K D 0 FLRW case, comoving spatial slices are flat and the tangent vector
in (9.107) is the same at every point along the edge e. In particular, the holonomy of
oriented length l0 along the i-th direction is
1=3
hi D exp.l0 V0 c i / D: exp.c i / ; (10.77)
where the path ordering P in (9.107) becomes trivial and
l0
:D 1=3
(10.78)
V0
is the ratio between the holonomy length and the size of the comoving volume.
Defining :D =N as the path length in units of , N we can express the
holonomy (10.77) in terms of the new variables as
hi D exp .2b i / D 12 cos b C 2 i sin b ; (10.79)

where we employed the identity in (9.109) valid in the fundamental representation

of the algebra. Using (10.79) and i j i D j =4, one can show that
tr. k hij / D 12 ijk sin2 .2b/ : (10.80)
This equation will be needed to compute the field strength (9.114) as in (9.115) and,
from that, the Hamiltonian.
We have already mentioned the issue of writing inverse powers of the densitized
triad using classical relations. In particular, the formula
fb; v l g D lv l1 (10.81)
is valid for arbitrary values of l. Via (10.79) and (10.81), one can express powers of
v as
1 X
v l1 D tr. i hi fh1
i ; v g/
l
3l i
1
D .cos b fsin b; v l g sin b fcos b; v l g/ ; (10.82)
l
where we explicitly wrote the sum over indices when summation convention is not
clear, used (9.110) and took the trace in the j D 1=2 representation (for which
C2 D j. j C 1/ D 3=4).
For a flat homogeneous background, the scalar constraint (9.60) becomes
(Problem 10.3)
3
12 1 .1 C n/ 2 pn 2.1Cn/ 2.1Cn/
3.1q/
HD lim
3 4 q l0 !0 l3
v sin2 2b
0 3 2
3q 3q
Œsin b fcos b; v 2.1Cn/ g cos b fsin b; v 2.1Cn/ g C H : (10.83)
The gravitational sector is only a function of b and v. The scalar-field part only
contains volume factors,
p2
H D C p3=2 V. / : (10.84)
2p3=2
The other constraints vanish. The constraint (10.83) should be compared with the
WDW Hamiltonian (10.11) in the same variables. These expressions are classical
and, therefore, physically equivalent. However, the following quantization will make
the LQC Hamiltonian crucially deviate from the WDW model.
For later purposes, we note that the full dynamics is encoded in H if we set the
lapse function N D 1. A different gauge choice at the classical levelp can simplify
the Hamiltonian HD D NH significantly. In particular, setting N D j det Ej one
avoids the need to rewrite inverse powers of the volume in terms of Poisson brackets:
1 Ei˛ Ej ij
HD D N 2 2
p k F˛k C NH
2 j det Ej
1 tr. k hij /
lim jej2
ij
D C NH
(9.114)
2 2 l0 !0 k
l20
3 N 2 2 2
D v sin 2b C NH :
(10.80)
lim (10.85)
4 2 2 l0 !0 l20
10.3.2 Quantization and Inverse-Volume Spectrum
In the quantization procedure (9.81) of loop quantum cosmology, the classical

Poisson bracket is replaced by the commutator bracket as in standard quantum
mechanics. As mentioned earlier, the connection operator does not have a well-
defined action on the kinematical Hilbert space of LQG, which suggests not to define
a connection-dependent operator b.O Nonetheless, the other elementary variable v can
be promoted to the self-adjoint operator (10.29), v ! vO :D i@b . It is easy to check
that the functionals
ev .b/ D eivb D: hbjvi (10.86)
are the eigenfunctions in b-representation of v.

O In Dirac notation, we can write down
the action of (10.29) on the states jvi as
vjvi
O D vjvi : (10.87)
These eigenstates form a basis fjvig of the kinematical Hilbert space. On this basis,
the action of a holonomy operator of (dimensionless) edge v 0 can be constructed
b 0
from the basic operator eiv b , which acts simply as a translation:
b 0
eiv b jvi D jv C v 0 i : (10.88)
Comparing with (10.87), one sees that the holonomy operator increases the volume
of the universe by attaching edges to the symmetry-reduced, simplified spin
network. What is the size of an edge? Eigenvalues v D O.1/ correspond to areas
of order of the Planck area 2 D 8l2Pl , so that the holonomy edge is of order of
the Planck length. Note also that, contrary to the full theory, due to the symmetry
reduction the spectrum of flux-area-volume operators is continuous and labelled by
the real parameter v. However, all eigenstates are normalizable with respect to the
inner product (10.15) for the Hilbert space of square-integrable functions on the
quantum configuration space, including the state with zero eigenvalue. In this sense,
the spectrum of volume and dreibein operators is discrete [162, 165, 166, 171–173].
This has an important consequence for the spectrum of inverse-volume operators
because, de facto, v can be regarded as a parameter on natural numbers. Since the
normalization in (10.15) has a Kronecker delta rather than a Dirac distribution, one
is dealing with a non-separable Hilbert space.
We consider the range of l being 0 < l < 1 in (10.82), so that v l1 is an
inverse power of the volume. The ambiguity parameter l determines the initial
slope of the effective geometrical density. To preserve coordinate invariance when
quantizing geometrical densities before symmetry reduction, l must be discrete,
lm D 1 .2m/1 , m 2 N [174, 175]. Hence one can select the bound
1
2 6 l < 1; (10.89)
which is also favoured phenomenologically [175]; a natural choice is l D 3=4.

To quantize the system in terms of elementary operators, we fix the length of
the holonomy to be such that D N and we consider their symmetric ordering so
that (10.82) (with D 1) becomes the self-adjoint operator
1
jvjl1 D
ih b
cos b jvjb b b i h
b
cl cos b D 1 eib jvj
cl sin b sin b jvj cl ebib ebib jvj b i
cl eib ;
l 2l
(10.90)
cl D jvj
where jvj O l . The absolute value of vO is taken in order for the eigenvalues of
1
jvjl1 to be real. The basis states jvi are also the eigenstates of the operator (10.90),
1
jvjl1 jvi D
1
jv C 1jl jv 1jl jvi : (10.91)
2l
Remarkably, the spectrum of this operator is bounded from above. In fact, the
maximum eigenvalue is obtained when the negative term in (10.91) vanishes, i.e.,
for v D 1:
2l1
jvjmax
l1
D : (10.92)
l
Classically, as „ ! 0 and the Planck length goes to zero, the value (10.92) diverges.
In general, however, the inverse volume cannot go arbitrarily to infinity towards
a singular geometric configuration. Here we have the first evidence that, contrary
to ADM quantum cosmology, in LQC the big-bang problem could be avoided.
This happens because we chose the holonomy rather than the connection as the
fundamental variable.
Therefore, the volume spectrum (10.87) does admit the v D 0 eigenvalue and
its inverse fails to be a densely defined operator. In spite of that, the inverse-volume
operator (10.90) is regular (with zero eigenvalue) at v D 0: volume singularities
are healed by purely geometrical effects. However, (10.91) and (10.92) are only
kinematical relations and they do not guarantee that similar properties be realized
at the dynamical level. In particular, nothing has been said about the quantum
dynamical evolution of the system: Does it stop at a big-bang singularity at v D 0?
To check this, we need to consider physical states or, in other words, the action of
the quantum Hamiltonian constraint on a wave-function.
10.3.3 Mini-superspace Parametrization
While in LQG the area spectrum is bounded from below by the minimal
area (9.116), due to symmetry reduction the same property is not shared by loop
quantum cosmology. Nevertheless, one may draw inspiration from the full theory
and assume that the kinematical area of any loop inside the comoving volume V0
is bounded by the area gap for the gauge-invariant states which are likely to be
realized in a homogeneous context. This value is twice the area gap (9.116) [176],
p
Pl :D 2Q Pl D 4 3 l2Pl ; (10.93)
so that
.al0 /2 > Pl : (10.94)
This step is rather speculative inasmuch as it borrows a result of the background-

independent framework and forces it into the symmetry-reduced model. It is
necessary, however, because the quantum scalar constraint in mini-superspace
would be singular if one maintained the limit l0 ! 0.
If the inequality (10.94) is saturated (smallest possible holonomy path), then the
comoving cell area is also the comoving area gap, that is, the smallest non-vanishing
eigenvalue of the area operator measuring comoving surfaces.9 In particular,
1n
2 l20 Pl p
D 2=3
> D N : (10.95)
V0 p Pl
Taking the equality in the second step, one has D N if p D Pl and
s
1 Pl
nD ; D N D ; (10.96)
2 p
9
Big-bang nucleosynthesis can place a bound over the smallest physical area [177].
a choice known as improved quantization scheme or improved dynamics [78, 167,

178]. The set fjvig becomes the eigenstate basis of the volume operator itself, v /
p3=2 D V. From the point of view of spin-network dynamics, the Universe grows by
changing a graph with all spin labels fixed to j D 1=2. As the Universe expands, the
comoving area gap shrinks to zero and the geometry is better and better described by
classical general relativity, while near the putative big bang quantum effects become
important.
Originally, the variables p and c were used instead of v and b, corresponding to
N D 1 (n D 0). In this old quantization scheme, the states jvi D ji coincide with
the basis eigenstates of the momentum operator pO , with eigenvalues v / p [166,
179]. Here the Universe grows by changing the spin label j on a fixed graph. This
case leads to severe restrictions on the matter sector if the wave-functions solving
the Hamiltonian constraint are required to be normalizable and to reproduce the
classical limit at large scales [180]. Also for such reason, the improved quantization
scheme seems to be the most natural and, as we are going to see very soon, the most
reasonable in a purely homogeneous context. However, later motivations lead us to
keep n, p and the other free parameters of the model as general as possible. In this
case, p is some physical squared length determined by the theory which may differ
from the mass gap Pl .
10.3.4 Quantum Hamiltonian Constraint
If we assume that holonomy plaquettes cannot be shrunk indefinitely, then we

1=3
must replace the limit l0 ! 0 in (10.83) with l0 ! V0 . With this substitu-
tion, the quantum Hamiltonian operator corresponding to (10.83) is well-defined.
When D , N
1 1
HO D 4 sin 2b CO sin 2b C HO ; (10.97)
where
CO D
3i
3.1C2n/ 3
.1 C n/ 2 2.1Cn/ pn 2.1Cn/ 4 3.1C2nq/
jvj 2.1Cn/
8qV0 3 4 3 2
b 2b b 2b

3q 3q
cos b jvj 2.1Cn/ sin b sin b jvj 2.1Cn/ cos b
D
3
3.1C2n/ 3
.1 C n/ 2 2.1Cn/ pn 2.1Cn/ 4
3.1C2nq/
jvj 2.1Cn/
16qV0 3 4 3 2

ib
e jvjb2 3q
2.1Cn/ b b
e e jvj
ib ib 2b
3q
2.1Cn/ e ib

; (10.98)
and we used the relation p D Œ.1Cn/ 2 pn v=61=.1Cn/ . The states jvi are eigenstates
O
of C,
O
Cjvi D cv jvi ; (10.99)
with eigenvalues
3.1C2n/ 3
3 .1 C n/ 2 2.1Cn/ pn 2.1Cn/
cv D 3 4
16qV0 3 2
ˇ ˇ 3q ˇ ˇ 3q !
3.1C2n/ ˇ 1 ˇ 2.1Cn/ ˇ 1 ˇ 2.1Cn/
jvj 2.1Cn/ ˇˇ1 C ˇˇ ˇˇ1 ˇˇ : (10.100)
v v
The quantization leading to (10.97) is also known as polymeric and is based on the
replacement (formal in general, rigorous in the case of LQC) of a classically small
canonical variable x with a trigonometric function, x ! sin x. Then,
O
Hjvi D cvC2 jv C 4i .cvC2 C cv2 /jvi C cv2 jv 4i C HO jvi : (10.101)
Taking HO " , one obtains the same expression, thus showing that the Hamiltonian
constraint is real-valued. In fact, CO " D C, O independently of the parametrization
(choice of n and q). It is more delicate to show that the Hamiltonian constraint
is self-adjoint. This property depends on the choice of measure in the physical
inner product (see [181], Sect. IIIB) and it is also important for numerical purposes
(solutions of non-self-adjoint constraints may lead to unstable modes [182]). In
order to have a self-adjoint extension, the constraint must be positive definite and
symmetric under the chosen inner product (which can be achieved also by a state
redefinition) [183]. The details of the proof depend on those of the constraint [184],
but the qualitative features of the effective dynamics are not greatly affected by this
property [185].
A physical state is a linear superposition of the eigenstates jvi over the discrete
variable v (recall (10.15)),10
X
D v jvi ; (10.102)
v
and such that the coefficients v obey the LQC version of the Wheeler–DeWitt
equation:
cvC2 vC4 .cvC2 C cv2 /v C cv2 v4 C hvjHO jviv D 0 : (10.103)
10
A precise description of physical states as bras in a rigged Hilbert space is given in [166].
Upon quantization, the scalar-field Hamiltonian is
1 1 b
HO D p3=2 pb2 C p3=2 V. O / : (10.104)
2
The operator O acts multiplicatively on the states jvi, so that
p2
:D p3=2 hvjHO jvi D CV; (10.105)
2p3
where the correction function is
b1 bb 6
:D hvjp3=2 p3=2 jvi D .hvjv 1l v l1 jvi/ .1l/
" ˇ ˇ ˇ ˇ !# 6
v ˇˇ 1 ˇˇl ˇˇ 1 ˇˇl .1l/
D 1 C ˇ ˇ1 ˇ ; (10.106)
2l ˇ v v
and
D 4.1 C n/ : (10.107)
The first departing point from the Wheeler–DeWitt equation in ADM variables
is that (10.103) is a difference, rather than differential, equation in v. Secondly,
while in ADM quantization the wave-function of the universe vanishes at the big
bang, here it is undetermined. In fact, (10.103) fixes v recursively for all v except
v D 0. In that case, the coefficient hvjHO jvi vanishes on the zero-volume state
if the potential V is non-singular there and so does the coefficient c2 C c2 D 0
(because jvj D j vj). Therefore, 0 decouples from the evolution automatically,
the wave-function at either side of the classical big bang singularity are uniquely
determined and one “jumps across” the singularity, which is effectively removed
from the quantum evolution [186]. Restricting the support of the wave-function onto
a discrete lattice without D 0 picks out a separable sub-space of the originally
non-separable Hilbert space.
In this sense, the big-bang problem receives an answer different from the one
possibly expected by the questioner. In classical cosmology, the issue is whether
the “universe” (matter density, metric) is finite at the beginning of time. In the
ADM quantization, it is whether the probability of having a classical big-bang
configuration is zero or not. Here, the problem is simply cut off from the range
of questions we can ask about the quantum dynamics of the system.
Furthermore, there exists a semi-classical limit where the effective equations of
motion show how the big bang is replaced by a bounce with finite energy density.
Before considering the effective dynamics, however, it is instructive to further
compare (10.103) with the WDW equation (10.18). The states annihilated by the
scalar constraint are in the v-representation and, in fact, v was denoted as Œv;
in the ADM theory. In (10.103), the length of the discrete steps is fixed but in the
limit where it vanishes one can show that the WDW equation is recovered. This
statement is not immediately obvious because, despite the use of the same variables,
HO has been engineered in two quite different ways. The b-representation is more
convenient for the comparison.
To this purpose, we choose the gauge N D a3 [78] and quantize the Hamilto-
nian (10.85). In the absence of a scalar potential and up to an overall factor 1=2, the
operator ordering can be arranged so that

3 2 2
2 2
.sin 2b @b / @ Œb; D 0 : (10.108)
2
Because one has a discrete one-dimensional lattice in v space and the Fourier
transform in b-space has support on the interval b 2 .0; =2/ [78], one can define
r
2 2
:D @2z ; z :D ln tan b ; (10.109)
6
so that we get

O Œz; :D C @2 Œz; D 0 :
C (10.110)
This expression is formally identical to the WDW equation (10.30) and the ensuing
quantization follows step by step that of Sect. 10.2.2. A key difference, however, is
that invariance of the wave-function under parity (frame re-orientation) is not gauge-
fixed ab initio and physical states are required to satisfy C Œz; D C Œz; .
It follows that the left- and right-moving sectors are not super-selected and must be
considered together [78]. In particular, we can write
C Œz; D C;L ŒzC C C;R Œz D .zC / .z / ; (10.111)
where z˙ D z ˙ and is some function.

This fact is crucial for the resolution of the big-bang singularity. The volume
operator (10.29) in the z variable is
r "s #
2 2 2 6
vO D i cosh 2 2
z @z D: iv cosh.0 z/ @z ; (10.112)
3
and one has

Z

.1C ; jvj
O 2C / D 2v dz .@z 1C;L / cosh.0 z/.@z 2C;L /
ˇ

C.@z 1C;R / cosh.0 z/.@z 2C;R / ˇ D 0
: (10.113)
The relative sign of the two terms is positive since we are taking the expectation
value of jvj,
O which coincides with that of vO on the left sector and with that of vO on
the right sector. At any internal time and on any physical state (Problem 10.4),
hjvji
O D .C ; jvj
O C / D V cosh.0 / ; (10.114)
where V > 0 is the minimal volume at the bounce. Equation (10.114) completes
the proof that the big-bang singularity is avoided in mini-superspace loop quantum
cosmology. Further evidence comes from noting that matter energy density has an
absolute upper bound on the whole physical Hilbert space [78]. We can reach the
same quantitative conclusion, albeit not as robustly, when looking at the effective
dynamics on semi-classical states.
10.3.5 Models with Curvature or a Cosmological Constant
In the previous sections, we ignored the contribution both of the intrinsic curvature
˛i D .K=2/ı˛i and of a cosmological constant . Here, we sketch scenarios where
the universe is not flat (K D ˙1) and ¤ 0.
10.3.5.1 Closed Universe
The case of a universe with positive-definite spin connection was studied in

[178, 187–193]. Due to the extra term in the connection, the form of the classical
Hamiltonian constraint (10.76) as a function of c (classically related to metric
variables as c D aP C K with K D 1) is modified by the replacement c2 ! c.c
1/ C .1 C 2 /=4, up to fiducial-volume factors. In the classical Friedmann equation,
this replacement corresponds to H 2 ! H 2 C 1=a2 . The quantum constraint and
the resulting difference equation are modified accordingly. There is no arbitrariness
in the fiducial volume V0 , since it can be identified with the total volume of the
Universe, which is finite and well defined. Then, the choice of elementary holonomy
is more natural than in the flat case and, locally, one can distinguish between the
group structure of SU.2/ and that of SO.3/ [190]. As in the flat case, the constraint
operator is essentially self-adjoint [190] and the singularity at v D 0 is removed
from the quantum evolution [178, 187, 190]. However, instead of a single-bounce
event one now has a cyclic model [178]. This happens because the classical and
quantum scalar constraints have both contracting and expanding branches coexisting
in closed-universe solutions, while these branches correspond to distinct solutions
in the flat case.
10.3.5.2 Open Universe
Loop quantum cosmology of an open universe [181, 193, 194] is slightly more
delicate to deal with. In contrast with the flat and closed cases K D 0; 1, the spin
connection is non-diagonal, so that also the connection is non-diagonal and it has
two (rather than one) dynamical components c.t/ and c2 .t/. The Gauss constraint
eventually fixes c2 .t/ D 1 and one ends up with the same number of degrees of
freedom as usual. The volume of the universe is infinite as in the flat case and a
fiducial volume must be defined. The classical Hamiltonian constraint is (10.76)
2=3
with c2 ! c2 V0 2 . The quantum constraint is constructed after defining a
suitable holonomy loop; the bounce still takes place and the v D 0 big-bang state
factors out of the dynamics.
10.3.5.3 ¤ 0
Another generalization is to add a cosmological constant term, positive [184, 187,

192] or negative [183, 191, 194, 195]. The finite-difference equations of these
models have been studied in relation to the self-adjoint property.
For > 0, below a critical value (of order of the Planck energy) the
Hamiltonian constraint operator admits many self-adjoint extensions, each with a
discrete spectrum. Above , the operator is essentially self-adjoint but there are
no physically interesting states in the Hilbert space of the model [184].
For < 0, the scalar constraint is essentially self-adjoint and its spectrum is
discrete [183] (while for D 0 it is continuous and with support on the positive
real line), also when K D 1 [194]. As in the D 0, K D 1 case, the universe
undergoes cycles of bounces [195].
10.3.6 Homogeneous Effective Dynamics
Effective cosmological equations of motion are derived from the expression of

the Hamiltonian constraint on a semi-classical state jsc i. The latter
P is typically
decomposed into a gravitational and a matter sector, jsc i D A;B jgraviA ˝
jmatiB . In general, geometrical and matter operators do not act separately on
physical states because solutions to the Hamiltonian constraint already incorporate
correlations between the two sectors. Therefore, operators on such states are in
general complicated, entangled observables. However, on a semi-classical state
geometrical and matter operators commute and they can be treated separately.
The semi-classical states jsc i are peaked around some point in the classical
phase space [74, 196] and they can be defined also in the group-averaging formalism
[197]. One then computes the expectation value of the Hamiltonian constraint oper-
ator thereon, using an appropriate inner product. Accordingly, for the gravitational
part of the Hamiltonian operator (10.97) we approximate its expectation value as
c c/
hsc jsin. c c/j
N CO sin. N 2
sc i ' cv sin .c/
N and we may write
O sc i ' 3 p sin2 .c/

N
hsc jHj 2
˛ p 2
C p3=2 D 0 ; (10.115)
N 2
where the correction function ˛ is

ˇ ˇ 6q ˇ ˇ 6q !
4V0 2 2
N 2 ˇ 1 ˇ ˇ 1 ˇ
˛ :D p cv D v ˇˇ1 C ˇˇ ˇˇ1 ˇˇ ; (10.116)
3 p 12q v v
is given in (10.107) and the range of the ambiguity q is
0 < q 6 1: (10.117)
When ˛ D 1 and the matter sector is a massless free scalar field, (10.115) is exact
[198]. In general, however, the evolution of a finitely spread semi-classical state will
produce quantum fluctuations leading to additional corrections to (10.115) [199,
200]. Assuming that the semi-classical wave-packet of the Universe does not spread
appreciably, we can stick with (10.115) also in the presence of a non-trivial scalar
potential. Then, the matter energy density is given by (10.105).
The Hamilton equation of motion pP D f p; hNHig for the densitized triad gives
the Hubble parameter
pP sin.2c/
N
H :D D˛ p : (10.118)
2p 2 N p
In the classical limit, c ! aP and the right-hand side tends to aP =a for small c.
N
Equation (10.115) yields

˛ sin2 .c/
N D ; (10.119)

where
3
:D 2 2
: (10.120)
N 2p
Combining (10.119) and (10.118), one gets the Friedmann equation

2
H2 D ˛ : (10.121)
3
The equation of motion P D f ; hNHig yields (compare with (10.4))
P
p D p3=2 ; (10.122)
N
while pP D f p ; hNHig leads to the effective Klein–Gordon equation

R C 3H P P C V; D 0 : (10.123)

As > 0 has a maximum at v D 1 and then decreases down to unity for large v,
the friction term in (10.123) changes sign during the evolution of the universe, the
first stage being of super-acceleration.
Setting ˛ D 1 D in the equations of motion (10.121) and (10.123), one ignores
inverse-volume corrections. On the other hand, in the limit sin.2c/ N ! 2c N one
neglects holonomy corrections and the second term in (10.121) is dropped.
The left-hand side of (10.121) is positive definite and, if > 0 (˛ > 0 if n > 1),
the energy density is bounded from above:
6 ˛ : (10.124)
When n ¤ 1=2, / a2.2n1/ varies with time and, at some point during the history
of the universe, it can be made arbitrarily small, thus loosing physical meaning as an
absolute lower bound. This is avoided in the improved quantization (10.96), where
the critical density is constant:
3
D 2 2p
: (10.125)

For the particular choice p D Pl , the critical density is less than half the Planck
density,
p
3
D m4 0:41 m4Pl : (10.126)
32 2 3 Pl
The numerical prefactor depends on (10.93) and (10.94) and it could change in a
more complete formulation of the model, but not in a way leading to qualitative
differences. If the ambiguity q is set equal to 1, then ˛ D 1 and the lower
bound (10.124) is the constant (10.126) [167, 201, 202]. Thus, also at the level of
the effective dynamics the big-bang singularity is removed.
The homogeneous effective dynamics describes the evolution of sharply peaked

semi-classical states down to Planck densities. In these regimes, quantum fluctu-
ations of the canonical variables due to their relative uncertainties might become
large enough to spoil the effective equations of motion. However, this is not the
case [203]. Provided the fiducial volume V D a3 V0 is chosen to be much larger
than a Planck volume, states that are initially sharply peaked will remain so during
their evolution, even through the bounce. Conversely, effective equations may be
less reliable for states which are not sharply peaked, or when the fiducial volume is
too small. Barring this possibility, effective homogeneous dynamics is an excellent
approximation of the exact dynamics of homogeneous LQC [167, 197]. Of course,
the system is fully quantum and local quantum fluctuations have not disappeared.
However, canonical variables are average quantities and, in the large-V0 limit, they
are insensitive to local fluctuations. These homogeneous fields behave similarly to
the center of mass of a many-particle body. While individual atoms experience
quantum fluctuations, quantum back-reaction effects are negligible on the center
of mass when the number of atoms composing the body is large.
10.3.6.1 Inverse-Volume Corrections
We discuss now the correction functions ˛ and from the point of view of their
1
asymptotic limits, later stressing an interpretational issue.
On a semi-classical state, the eigenvalues of jvjl1 are approximated by the
classical variable v l1 itself. Consistently, the classical limit corresponds to a large-
volume approximation where v 1, while in the near-Planck regime (“small
volumes”; the reason for quotation marks will be clear soon) v 1. Since
p 4
12 3 p p
vD ; (10.127)
Pl p
“near the Planck scale” (v 1) the correction functions read

6q 6.2l/
q q
˛ ' v 2 D: ˛1 ıinv ˛ ; ' v .1l/ D: 1 ıinv ; (10.128)
where
p !2q˛
3q 12 3 p
q˛ D 1 ; ˛1 D ; (10.129)
Pl
p !2q
3.2 l/ 12 3 p
q D ; 1 D ; (10.130)
.1 l/ Pl
and
2 a
p
ıinv :D D : (10.131)
p a
From the calculation leading to ˛ and , it is clear that the “natural” choice of
the ambiguities l and q can be set at the middle of their range:11
3 1
lD ; qD : (10.132)
4 2
In the mini-superspace parametrization, the old quantization scheme corresponds to
D 4 and
5 15
q˛ D ; ˛1 D 3 8 D O.10/ ; (10.133)
8
15 45
q D ; 1 D 3 4 D O.105 / ; (10.134)
4
while the improved scheme has D 6 and
3
q˛ D ; ˛1 D 23=2 33=4 6 ; (10.135)
4
5
q D ; 1 D 25 35=2 500 : (10.136)
2
In homogeneous models with n D 0, the duration of this regime depends on the spin
representation of the holonomies, small j implying a very short super-inflationary
period and, actually, almost no intermediate stage between the discrete quantum
regime and the continuum classical limit [175]. Since small-j representations are
theoretically favoured, this may constitute a problem. It will be relaxed in a different
parametrization when inhomogeneities are taken into account.
In the quasi-classical limit (large volumes), (10.116) and (10.106) can be
approximated as
˛ ' 1 C ˛0 ıinv ; ' 1 C 0 ıinv ; (10.137)
11
Different parameter choices have been made in the literature. For instance, the same sequence
of steps we reviewed was followed in [172, 173] (arbitrary j, l D 1=2, D 4, ˛ D 1), [204]
(arbitrary j and l, D 4, ˛ D 1), [167] (j D 1=2, l D 1=2, q D 1, D 6), [205] (arbitrary j and
l, q D 1, D 6), and partly [180] (j D 1=2, arbitrary l, q D 1, arbitrary ). All these papers use
the area gap QPl (10.93) rather than Pl .
where
2 2
.3q /.6q / Pl .2 l/ Pl
˛0 D ; 0 D : (10.138)
64 p 63 p
For the natural choice (10.132), the old and the improved quantization schemes in
mini-superspace parametrization correspond, respectively, to
5 5
D 4; ˛0 D 25 34
0:002 ; 0 D 63
0:02 ; (10.139)
and
1 5
D 6; ˛0 D 96
0:01 ; 0 D 144
0:03 : (10.140)
Taking q D 1 instead, one gets a negative ˛0 D 1=648 for D 4 and ˛0 D 0 for

D 6.
Although one can resort to different quantization schemes, equations (10.128)
and (10.137) maintain the same structure. The coefficients , q˛ and q are robust in
the choice of the parameters, inasmuch as their order of magnitude does not change
appreciably [205]. All these parameters can be set to their “natural” values, which
are dictated by the form of the Hamiltonian and other considerations.
Now we examine an interpretation issue related to any parametrization in pure
mini-superspace. For a compact universe (e.g., a sphere or a torus) the fiducial
volume defined in (10.2) is identified with the total physical volume, which is
given by the theory; no issue arises in this case. Otherwise, the fiducial volume in
a non-compact setting is an arbitrary quantity and should not appear in physical
observables. However, both N and other volume-dependent objects do manifest
themselves in a number of cases, from the effective equations of motion to
cosmological spectra. In particular, V appears in the correction function (10.131)
as ıinv a V =3 . To make ıinv dimensionless, one can use the Planck length
lPl to write
3
l3Pl
ıinv a : (10.141)
V0
Physically, the parameter is related to how the number of plaquettes of an

underlying discrete state changes with respect to the volume as the universe expands.
This is a phenomenological prescription for the area of holonomy plaquettes, but
ideally it should be an input from the full theory [206]. For phenomenology at the
current level of precision, the most significant parameter among f˛0 ; 0 ; g is ,
which is not as much affected by different choices of the mini-superspace scheme.
Since ıPl is V0 -dependent, inverse-volume corrections are not strictly meaningful
in a pure mini-superspace treatment. Casting the problem into another way, the
arbitrariness of V0 corresponds to the freedom to p perform a conformal rescaling
of non-closed FLRW metrics. The ratio a =a D p =p appearing in correction
functions, as well as statements such as “the quasi-classical regime takes place at

a < a ,” seem to violate the conformal invariance a ! Aa, since a is supposed
to be some parameter fixed once and for all by the underlying theory. One could
interpret V0 as a regulator and send V0 ! 1 at the end of calculations, so that in the
quasi-classical limit there are no inverse-volume corrections at all. However, the full
theory does contain these corrections and they should appear also in a cosmological
setting. This highlights that something is missing in the theoretical construction
of the homogeneous LQC effective dynamics. To get a clearer picture, we should
include inhomogeneous perturbations. In fact, in the presence of inhomogeneities
there is no conformal freedom in the volume choice, which depends on a local scale
factor. In general, the strength of inverse-volume corrections for a given perturbation
mode is expected to depend on the scale of the problem, which is measured by the
wave-length of a given mode relative to the Planck size [207].
10.3.6.2 Models with K ¤ 0 and ¤ 0
The flat effective dynamical model has been extended to cases with curvature and a
cosmological constant.
For a closed universe, K D 1, as mentioned in Sect. 10.3.5 there is no
fiducial-volume problem and inverse-volume corrections make sense also in a
pure homogeneous and isotropic setting. The cyclic bounces appearing in the
dynamics of the difference evolution equation [178] exist also at the effective level
[189, 191, 192]; in particular, the big crunch of classical closed universes can be
avoided [188]. The bounce persists in an open universe, K D 1 [181]. In general,
all past and future strong curvature singularities are resolved in K D ˙1 isotropic
models; for the closed model, weak singularities in the past evolution may also be
resolved [193].
There is evidence that a cosmological constant, if suitably tuned, does not spoil
the singularity resolution. When > 0 and K D 1, the bounce is preserved if
the cosmological constant is sufficiently small [192]. Above a certain critical value,
however, periodic oscillations take place. When < 0, recollapse of the universe
is possible, even cyclically [191, 195]. Whichever the sign of the cosmological
constant, the effective Friedmann equation is (10.121), with the critical density
shifted by a constant, -dependent term.
10.3.6.3 Probability of Inflation
In Sect. 10.2.1, we asked whether inflation is “probable” in WDW quantization.

The same question has been posed in LQC, relying on the Gibbons–Turok classical
measure. The suppression factor e3N is still present if one ignores holonomy
corrections [208] but, in general, the bounce fixes the ambiguities in the measure
present in the classical case. It turns out that the probability to have observationally
viable inflation is very close to 1 [209]. We do not elaborate further on this point, as
the measure problem in cosmology is still under scrutiny.
10.3.7 Singularity Resolved?
To summarize, the avoidance of the big-bang singularity in homogeneous and

isotropic LQC is confirmed in four ways:
1. At the kinematical level, via the spectrum (10.91) and (10.92) of the inverse
volume operator.
2. By the finite-difference quantum Hamiltonian constraint (10.103).
3. By the expectation value (10.114) of the volume operator on physical states in
the massless case.
4. Through the effective dynamical equations (10.124), (10.125) and (10.126) in the
improved quantization scheme.
The big bang is replaced by a bounce at H D 0, where the energy density is
about half the Planck energy. Relaxing the condition of isotropy does not spoil
the bounce, as shown in Bianchi I and Gowdy models [176, 210–214],12 Bianchi
II [216] and Bianchi IX cosmology [217]. In general, strong singularities are
effectively excluded in purely homogeneous LQC [193, 218]. In accordance with
the disappearance of the singularity, the chaotic behaviour of the mixmaster universe
is drastically changed by LQC quantum effects and the BKL singularity is removed
[219–221]. As the billiard ball representing the dynamics falls down the potential U
from a Kasner regime towards the classical singularity, it starts bouncing off the
walls as in the classical case (6.56). However, at sufficiently small volumes the
BKL sequence of Kasner eras comes to an end because quantum gravity becomes
important and the shape of the potential is modified. The walls are now of finite
height, beyond which U falls off smoothly to negative values approaching zero
(where ai D 0) from below. Thus, the zero-volume limit corresponds to a local
maximum of the potential, not to a global minimum. Also spin-foam cosmology,
the mini-superspace path-integral version of LQC, agrees that the singularity a D 0
disappears from the quantum dynamics (Sect. 11.4).
These results are encouraging but insufficient to establish a definitive solution of
the big-bang issue. The mini-superspace quantization is characterized by a number
of ambiguities encoding our ignorance about the details of the full theory and of
the underlying state [200]. Even choosing the improved quantization scheme, the
critical density ˛ may be non-constant if q ¤ 1, in which case one would lose the
neat bounce interpretation.
These ambiguities cannot easily be dismissed as artifacts of the mini-superspace
quantization. It is possible to obtain cosmology directly from the full quantum
12
Gowdy spacetime is globally hyperbolic and has the spatial topology of a 3-torus [215].
theory by reducing its degrees of freedom at the level of kinematical quantum states.
One fixes a gauge in the kinematical Hilbert space of LQG where the triad (hence
the metric) is diagonal. Spin networks with SU.2/ labels are therefore reduced to
a three-dimensional cubic lattice whose links are labelled by elements of a sub-
group U.1/ of SU.2/. Each 3-valent node is therefore characterized by the elements
of a sub-group U.1/ ˝ U.1/ ˝ U.1/ of SU.2/ ˝ SU.2/ ˝ SU.2/. Since one
has a grid with different intertwiners at each point, these reduced states represent
inhomogeneous geometries; and since one has different spin labels on the links
associated with the three directions at each node, the states are also anisotropic. The
homogeneous kinematical sector is thus obtained when all the intertwiners are equal,
while the anisotropic sector is recovered when all spins are equal. This procedure
is known as quantum reduced loop gravity [222–225]. The effective Friedmann
equation is obtained by picking a semi-classical reduced state, peaked at large spin
quantum numbers and calculating the expectation value of the Hamiltonian. The
spread of the state gives rise to quantum corrections. These results depend on three
regulators ˛ , representing the length of the links along the ˛-th direction; in the
isotropic case, the regulators ˛ D are all equal. Sending ˛ ! 0 produces the
correct classical limit of Bianchi I cosmology, plus corrections. On the other hand,
fixing D N one obtains the dynamics (10.115) of homogeneous loop quantum
cosmology [225]. Clearly, the quantization ambiguities of canonical formalism as
well as the uncertainty on the function .a/
N still remain.
Therefore, from the point of view of the full theory the mini-superspace
loop quantization of gravity is less of a toy model than expected (an intriguing
confirmation of this result will be discussed in Sect. 11.5.2). But as far as empirical
cosmology is concerned, it is not enough. On one hand, loop quantum gravity could
resolve the big-bang singularity but, on the other hand, we are far from a complete
control of the full theory. An important step forward consists in extending the mini-
superspace (regardless of whether the symmetry reduction is done before or after
quantization) to include inhomogeneities. In this case, although the notion of a fixed
background is still present and one must assume a number of approximations, one
can better appreciate the interplay of the degrees of freedom of the full theory and
check whether the bounce is an artifact of homogeneity or is a robust feature of the
theory. Moreover, the study of inhomogeneous perturbations is essential to extract
early-universe cosmological observables.
This is the subject of the remainder of the chapter. As a spoiler for the reader,
we anticipate that no clear-cut answer is yet available in the state of the art. In the
so-called deformed-algebra approach, the bounce picture is even put on trial by the
appearance of something interpreted as an effective signature change of spacetime,
suggesting that the universe “dissolves” at Planckian scales rather than bouncing off
a minimal size. However, there is no general consensus on this issue. The deformed-
algebra method is still under inspection and other formalisms such as the dressed-
metric approach and hybrid quantization do not see any signature change.
10.3.8 Lattice Refinement: Quantum Corrections Revisited
Until now, we have not given any motivation for taking D N / pn . This is the
next subject and it resides in a framework which does not enjoy all the symmetries
of a purely FLRW background.
In loop quantum gravity, the classical continuum of general relativity is replaced
by the appearance of discrete spatial structures. It is often expected that the discrete-
ness scale is determined by the Planck length lPl but, if discreteness is fundamental,
its scale must be set by the dynamical parameters of some underlying spin-network
state. The scaling of the plaquette in the area law (9.117) is determined by the Planck
length for dimensional reasons, but the actual size is given by the spin quantum
number. Its values in a specific physical situation have to be derived from the LQG
dynamical equations, a task which remains extremely difficult to date. However,
given the form in which je appears in the dynamical equations, its implications for
physics can be understood in certain phenomenological situations, such as in cos-
mology. Then, instead of using the spin labels je , it is useful to refer to an elementary
quantum-gravity length scale L, which needs not be exactly the Planck length.
The scale L naturally arises if translation invariance is broken, e.g., by the
presence of clustering matter or inhomogeneous perturbations. The comoving
volume V0 of the system can be discretized as a lattice whose N cells or patches
are nearly isotropic, have characteristic comoving volume l30 and correspond to the
vertices of the spin network associated with V0 . The proper size of a cell is
V
L3 :D a3 l30 D : (10.142)
N
To calculate the curvature at the lattice sites within V0 , we need to specify closed
holonomy paths around such points. As we have seen in Sect. 9.3, a generic
holonomy plaquette is given by the composition of elementary holonomies over
individual plaquettes. Therefore, we can set the length of the elementary holonomy
to be that of the characteristic lattice cell. In other words, the elementary loops of
comoving size l0 we have talked about until now define the cells’ walls, while in
a pure FLRW background there is only one cell of volume V0 (the number N is
arbitrary). We naturally identify the previously ad hoc function . N p/ as the ratio of
the cell-to-lattice size, under the requirement that the lattice be refined in time:
1
N D D N 3 : (10.143)
The patch size l30 is independent of the size of the fiducial region, since both V0 and
N scale in the same way when the size of the region is changed. Physical predictions
should not feature the region one chooses unless one is specifically asking region-
dependent questions (such as: What is the number of vertices in a given volume?).
Lattice refinement addresses the fiducial-volume issue mentioned in Sects. 10.1.1
and 10.3.6. Fluxes EŒ˙ are determined by the inhomogeneous spin-network
quantum state of the full theory associated with a given patch [204]. This implies
that the number of vertices of the underlying physical state would change when
changing the fiducial volume a3 V0 . Therefore, there is no scaling ambiguity in the
equations of motion [204, 226], although the physical observables will depend on
the choice of spin-network state.
The spin-network state described by the lattice can be (and usually is) excited by
the action of the Hamiltonian operator on the spin vertices, increasing their number
and changing their edge labels [227, 228]. This process has not yet been established
univocally in the full theory, so that it is convenient to parametrize the number
of vertices as in (10.142) [206], where the length L.t/ is state dependent and, by
assumption, coordinate independent; its time dependence is inherited from the state
itself. Since the kinematical Hilbert space is usually factorized into gravitational and
matter sectors, the problem arises of how to define a natural clock when matter does
not enter the definition of a (purely geometrical) spin network. This issue requires
a much deeper understanding of the theory. Thus, as unsatisfactory as the free
scale (10.142) may be, we take it as a phenomenological ingredient in the present
formulation of inhomogeneous LQC.
The general form (10.75) of .N p/ is obtained if L.t/ scales as
L / a3.12n/ : (10.144)
Homogeneous models adopting (10.95) feature holonomies which depend on triad

variables; in other words, curvature components are constrained by the area operator
although this does not appear in the full constraint. This does not seem justified
in a purely homogeneous setting. On the other hand, in inhomogeneous models
the dependence of the parameter N on p is implemented at the state (rather than
operatorial) level, in closer conformity with the full theory [206]. This partly
explains why inverse-volume corrections should depend only on the densitized triad.
A better understanding of this fact is gained in a more complete derivation of these
corrections within lattice refinement, showing that they naturally arise in a moment
expansion of flux operators [229].
10.3.8.1 Lattice Refinement and Inhomogeneous Perturbations
The patches of volume L3 find a most natural representation in inhomogeneous

cosmologies within the separate universe picture. Regions of size L3 are generated
by an underlying discrete state and correspond to quantum degrees of freedom
absent classically. The discrete nature of the state implies that inhomogeneities
are unavoidable and no perfectly homogeneous geometry can exist. Given these
inhomogeneities and their scale provided by the state, one can reinterpret them in a
classical context, making use of the separate universe approach. There, the volume
V can be regarded as a region of the universe where inhomogeneities are non-zero
but small. This region is coarse-grained into smaller regions of volume L3 , each
centered at some point x, wherein the universe is FLRW and described by a “local”
scale factor a.t; x/ D ax .t/. The difference between scale factors separated by the
typical perturbation wave-length jx0 xj V 1=3 defines a spatial gradient
interpreted as a metric perturbation. In a perfectly homogeneous context, L3 V
and there is no sensible notion of dividing V into cells; this is tantamount to stating
that only the fiducial volume enters the quantum corrections and the observables.
On the other hand, in an inhomogeneous universe the quantity L3 carries a time
dependence which, in turn, translates into a momentum dependence. The details of
the cell sub-division (number of cells per unit volume) are intimately related to the
structure of small-scale perturbations and their spectrum. Thus, lattice refinement
naturally fits into the cosmological perturbation analysis. As long as perturbations
are linear and almost scale invariant, the size of the volume within which the study
is conducted is totally irrelevant.
10.3.8.2 Critical Density and Quantum Corrections
Under the replacement V0 ! l30 everywhere, also the definition of the classical
canonical pair .c; p/ changes: c D l0 aP , p D l20 a2 . From (10.120), (10.143)
and (10.142), the critical density is
2=3
3 N 3
D 2 2
D 2 2 L2
: (10.145)
V
In all quantization schemes but the improved one (n D 1=2), the patch size
L is dynamical and is not constant. In any case, the critical density is a
number density which depends neither on the size of the fiducial volume nor on
coordinates explicitly, so that it is physically well-defined even outside the improved
quantization scheme. Holonomy corrections are defined as

ıhol :D ; (10.146)

and the Friedmann equation (10.121) can be written as H 2 D . 2 =3/.˛ ıhol /,

where the Hubble parameter (10.118) is
sin.2 LHcl / aP
HD˛ ; Hcl :D : (10.147)
2 L a
The maximum density (size of H) is obtained when sin.2 LHcl / D 1, i.e.,

for LHcl D =.4 / 3:3 In terms of the classical energy density D
3Hcl2 = 2 , (10.146) can be quantified as
ıhol . Hcl L/2 : (10.148)

Intuitively, holonomy corrections become large when the Hubble scale Hcl1 D
a=Pa L is of the size of the discreteness scale, certainly an extreme regime in
cosmology. In fact, the relative size of the quantum effect at maximal density is
very large, since for ˛ D 1
Hcl H sin.2 LHcl / 2

D1 D 1 30 % : (10.149)
Hcl 2 LHcl
Similar modifications occur for the quantum correction ıinv . In inverse-volume

as well as in holonomy corrections, one refers to elementary building blocks of
a discrete state, respectively, the plaquette areas and the edge lengths. A pure
mini-superspace quantization makes use of macroscopic parameters, such as the
volume of some fiducial region, and fluxes are calculated on comoving areas
2=3
V0 . In a purely homogeneous universe, the only way to write down (10.131)
is ıinv / .lPl =V 1=3 / , which is volume dependent. On the other hand, in the lattice
interpretation of loop quantum cosmology one uses the microscopic volume of a
cell and both holonomy plaquettes and fluxes are defined on comoving areas l20 .
In particular, the individual cell area L2 is nothing but the expectation value of the
O
flux operator EŒ˙ through a surface ˙ on a semi-classical state. Inverse-volume
corrections are determined by the inhomogeneous state through the patch size L:
Q
lPl
ıinv D ; (10.150)
L
with some phenomenological parameter Q > 0 [229]. The latter is not the parameter
determined by (10.107); n D 1=2 will not imply Q D 6. The inverse-volume
correction (10.150) does not depend on holonomies due to the use of the Thiemann
identity [229]. We saw that inverse powers of L cannot be quantized to a densely
defined operator because the spectrum of the volume contains 0. Inverse volumes
appear in the dynamics via the Hamiltonian constraint (of both gravity and matter, as
in kinetic matter terms) and are an unavoidable consequence of spatial discreteness
in loop quantum gravity. This requires to recast their classical expressions via
Poisson brackets, which in turn feature derivatives by L. Quantum discreteness then
replaces classical p
continuous derivatives
p by finite-difference
p quotients.
p For example,
the expression .2 L/1 Dp@ L=@L would become . L C lPl L lPl /=.2lPl /,
strongly differing from .2 L/1 when L lPl . For larger L, corrections are
perturbative and of order lPl =L, so that in general this type of inverse-volume
quantum corrections are encoded by the ratio (10.150). In terms of energy densities,
2
lPl
. 1: (10.151)
Pl L
In practice, the actual size of LQC effects on the inflationary spectrum is well
below the over-optimistic upper bound (10.151) but above the naive estimate (10.69)
and the WDW effect (10.70). The non-local nature of LQG effects prevents the
formation of singularities one would typically find classically. The physical inter-
pretation of inverse-volume corrections stems exactly from the same mechanism:
classically divergent quantities such as inverse powers of volumes remain finite due
to intrinsically quantum effects. Loosely speaking, quanta of geometry cannot be
compressed too densely and they determine the onset of a repulsive force at Planck
scale [230, 231] which is responsible for the various corrections to the dynamics.
There are indications that holonomy corrections are not significant in the energy
regime of inflation, but only at near-Planckian densities [167]. This is suggested
by effective equations for certain matter contents with a dominant kinetic energy
[198, 232]. In order to compare inverse-volume and holonomy corrections, we
notice from (10.150) and (10.148) that
Q
1=2
ıinv lPl Hıhol : (10.152)
For a universe of causal size H 1 lPl , inverse-volume corrections are considerable

and behave very differently from what is normally expected for quantum gravity.
Holonomy corrections are small for small densities, but inverse-volume corrections
may still be large because they are magnified by an inverse power of ıhol . As
the energy density decreases in an expanding universe, holonomy corrections fall
to small values, while inverse-volume corrections increase. For instance, in an
10
inflationary regime with a typical energy
9
p scale of 10 Pl , we can use (10.152)
with Q D 4 to write ıhol 10 = ıinv . Small holonomy corrections of size
ıhol < 106 then require inverse-volume correction larger than ıinv > 106 . This
interplay of holonomy and inverse-volume corrections can make loop quantum
cosmology testable because it leaves only a finite window for consistent parameter
values, rather than just providing Planckian upper bounds. It also shows that inverse-
volume corrections become dominant for sufficiently small densities, although they
are eventually suppressed as the density decreases.
10.3.8.3 Lattice-Refinement Parametrization
In the lattice refinement picture, (10.150) replaces the total lattice fiducial volume V
with the “patch” (i.e., cell) volume L3 [233]. This means that one makes the formal
replacement V ! V=N everywhere in mini-superspace expressions, which can also
be justified as follows.
While writing down the semi-classical Hamiltonian with inverse-volume and
holonomy corrections, one is at a non-dynamical quantum-geometric level. At this
kinematical level, internal time is taken at a fixed value but the geometry still varies
on the whole phase space. In this setting, we must keep N fixed to some constant N0
while formulating the constraint as a composite operator. Since the vertex density
does not depend on the choice of fiducial volume, it is physically reasonable to

p
expect the N0 factor to be hidden in the kinematical quantity a D p . The net
result is the Hamiltonian constraint operator of the old formulation of loop quantum
cosmology [162, 179] not taking into account any refinement, corresponding to
n D 0 and Q D .
However, when one solves the constraint or uses it for effective equations, one
has to bring in the dynamical nature of N from an underlying full state. This is the
motivation for promoting N to a time-dependent quantity, a step which captures
operator as well as state properties of the effective dynamics. For some stretches of
time, one can choose to use the scale factor a as the time variable and to represent
N .a/ as a power law,
N D N0 a6n : (10.153)
Then, from (10.143) one gets (10.75). Overall, quantum corrections are of the
form (10.150),
Q Q
N 3 N0 3 .2n1/Q
ıinv D l3Pl D l3Pl a ; (10.154)
V V0
where Q > 0. This equation cannot be obtained in a pure mini-superspace setting.

The parameter a plays two roles, one as a dynamical geometric quantity and the
other as internal time. The parametrization (10.153) as a power law of the scale
factor is simply a way to encode the qualitative (yet robust) phenomenology of the
theory. The general viewpoint is similar to mean-field approximations which model
effects of underlying degrees of freedom by a single, physically motivated function.
Comparing with the earlier mini-superspace parametrization, (10.154) gives D
.1 2n/.Q Since @N =@V > 0, one has n > 0: the number of vertices N must not
decrease with the volume and it is constant for n D 0. Also, l0 / a12n is the
geometry as determined by the state; in a discrete geometrical setting, this has a
lower non-zero bound which requires n 6 1=2. In particular, for n D 1=2 we have
a constant patch volume as in the improved mini-superspace quantization scheme
[167]. In contrast with the mini-superspace parametrization (10.107), in the effective
parametrization of (10.154) we have D 0 for the improved quantization scheme
n D 1=2. The range of n is then
1
0<n6 2
: (10.155)
The critical density (10.145), / a6.2n1/ , is constant for n D 1=2.

The exponent Q in (10.153) can be taken as a positive integer. In fact, the
correction function ıinv depends on flux values, corresponding to p for the isotropic
background. Since p changes sign under orientation reversal but the operators are
parity invariant, only even powers of p can appear, giving Q D 4 as the smallest
value. Therefore we set Q > 4.
To summarize, is a time-independent parameter given by the quasi-classical

theory and with range
> 0: (10.156)
The lattice parametrization replaces the one for homogeneous LQC. In fact, strictly
speaking, the use of one parametrization instead of the other is not a matter
of choice. A perfectly homogeneous FLRW background is an idealization of
reality which, in some applications, may turn out to be untenable. The study of
cosmological perturbations is an example in this respect. In that case, the lattice
refinement parametrization is not only useful, but also required for consistency when
inverse-volume corrections are considered.
In the near-Planckian regime, the small-j problem in the homogeneous
parametrization is reinterpreted and relaxed in terms of the lattice embedding.
The volume spectrum
p depends on the quadratic Casimir in j representation:
N n L2 C2 . j/ j. A higher-j effect can be obtained as a refinement
of the lattice (smaller )N [234], thus allowing for long enough super-acceleration.
A change in . N p/ can be achieved by varying the comoving volume. This is an
arbitrary operation in pure FLRW, while in inhomogeneous models N is a physical
quantity related to the number of vertices of the underlying reduced spin-network
state. As long as we lack a calculation of this effect in the full theory, we will not
be able to predict the duration of the small-volume regime.
In the quasi-classical regime in the lattice parametrization (10.156), may be
different in ˛ and for an inhomogeneous model but we assume that the background
equations (10.137) are valid also in the perturbed case. The coefficients ˛0 and
0 become arbitrary but positive parameters. In fact, from the explicit calculations
of inverse-volume operators and their spectra in exactly isotropic models and for
regular lattice states in the presence of inhomogeneities [165, 172, 205], correction
functions implementing inverse-volume corrections approach the classical value
always from above:
˛0 > 0 ; 0 > 0 : (10.157)
10.3.9 Perturbations and Inflationary Observables
The original formulation of LQC mainly dealt with the quantization of homoge-
neous spaces but efforts have been made to incorporate inhomogeneities at the
quantum level as well as within an effective dynamics. The goal is to identify
characteristic observational signatures allowing one to place bounds on the model.
Currently, there are three main approaches to the problem of LQC perturbations.
In the deformed-algebra (or effective-constraints) approach [226, 235–237], one
encodes quantum corrections in effective constraints. In hybrid quantization [238–
243], it is assumed that the main effects of quantum geometry are encoded in the
homogeneous background, which is quantized polymerically as in Sects. 10.3.4
and 10.3.6, while inhomogeneities are Fock quantized in a standard way [212].
This gauge-invariant formalism aims to capture the quantum dynamics in a regime
between full quantum gravity and quantum field theory on curved spacetimes and it
served as a basis where to deal with inhomogeneous Gowdy cosmology [244–247].
In the dressed metric approach, classical constraints are solved for gauge-invariant
modes before quantizing. The quantization is of hybrid type and in the geometry
sector a “dressed” metric encodes the quantum corrections [248–251]. At the time
of writing, most of the LQC approaches to inhomogeneities are still under active
inspection and there is ongoing debate concerning both their mathematical aspects
and, whenever available, the differences in their physical predictions [237]. The
reader is invited to look at the literature and thread carefully on this ground, which is
still at the research frontier. It is with this mind attitude that, to obtain the dynamics
of inhomogeneities, we briefly report on the deformed-algebra scheme.
10.3.9.1 Perturbations with Deformed Algebra
In standard cosmology, it is common to perturb the action or the Einstein equations

(Chap. 3). However, it is also possible to apply perturbation theory to the constraints
in Hamiltonian formalism; the two methods agree [252]. In canonical quantum
cosmology, the second option is forced upon us by the structure of the theory and
constitutes the first step of the following strategy. In the classical theory with a set
of constraints C˛ , closure of the constraint algebra fC˛ ; Cˇ g D f˛ˇ C is guaranteed
by general covariance. For instance, the perturbed smeared constraints in ADM
variables automatically obey the canonical algebra (9.166); in Ashtekar–Barbero
variables, we have seven first-class constraints (9.52): the super-Hamiltonian, the
three components of the diffeomorphism constraint and the three components of
the Gauss constraint generating infinitesimal su.2/ gauge transformations in the
internal space. However, when quantum corrections are included the constraint
algebra can present anomalies, terms which do not let the algebra close. In order
to guarantee closure of the algebra, one must include counter-terms order by order
in perturbation theory that cancel the anomalies. LQG effects can be captured
by effective constraints, where quantum corrections are inserted by hand in all
the places and shapes expected from the full theory. Since one is solving the
quantum-corrected diffeomorphism and Gauss constraints in parallel with the super-
Hamiltonian, the theory is fully gauge and diffeomorphism invariant, but not with
respect to the standard classical transformations. The structure of spacetime itself is
deformed by quantum effects, via the effective constraints. Gauge transformations
belonging to a deformed algebra no longer correspond to ordinary coordinate
transformations on a manifold. Thus, in order to take the new gauge structure into
account, one should rely only on gauge-invariant perturbations. This procedure (first
quantize the classical system, then cast it in gauge-invariant variables) is the core
of the effective-dynamics approach and it gives rise to the somewhat surprising
possibility that LQC quantum corrections be large even during inflation.
One may wonder whether one would get the same results by fixing the gauge
before quantizing (examples of this possibility are the partial gauge-fixing at the
classical level of [207, 253, 254] and early papers on hybrid quantization). In
principle, the answer is No. Gauge fixing and quantization do not commute because
quantization deeply affects the very notion of gauge invariance. Moreover, whenever
gauge-ready variables are not constructed after quantizing, one will generally
obtain a physically different quantum system. This is one of the reasons why
the effective-dynamics approach (which defines gauge invariance from deformed
effective constraints, i.e., supposedly at the quantum level) differs from other
proposals.13
In the case of inhomogeneous LQC, the seven first-class constraints (9.52)
display three types of quantum corrections. These are not limited to inverse-volume
and holonomy corrections but can include higher-moment terms generating higher
derivatives in the effective constraints [256–260]. The setting is an inflationary
era driven by a slowly rolling scalar field, the only difference with respect to the
standard case being the presence of the quantum corrections. Ignoring moment
corrections, the background equations of motion are (10.121) (with ! 1 when
holonomy corrections are dropped) and (10.123). As in standard general relativity,
linear perturbations can be decomposed into scalar, vector and tensor modes which
can be studied independently. Triad and connection components are separated into
a flat FLRW background and an inhomogeneous perturbation,

Ei˛ D a2 ıi˛ C ıEi˛ ; Ai˛ D cı˛i C ı˛i C ıK˛i ; (10.158)
where the curvature and triad perturbations are canonically conjugate:
ˇ
fıK˛i .x/; ıEj .x0 /g D 2 ı˛ˇ ıji ı.x; x0 / : (10.159)
The form of the perturbations depends on the sector one considers. In turn, perturbed
effective dynamics changes according to whether one includes only holonomy cor-
rections, inverse-volume corrections, moments, or all. After some early works based
on toy models [173, 175, 261–266], the constraint algebra has been closed explicitly
for holonomy corrections only (for scalar [255, 267, 268], vector [235, 269] and
tensor modes [226]), inverse-volume corrections only (for scalar [236, 270], vector
[235] and tensor modes [226]), and both at the same time [226, 235, 271]. Contrary
13
Eventually, the discrepancy might be not so severe. The dynamics of [207, 253] is the same found
in [255] once recast in the longitudinal gauge. This match is not sufficient to have a complete
proof of equivalence because the longitudinal gauge is legal only in the absence of inverse-volume
corrections.
to the WDW case, ignoring back-reaction of the metric and considering just a
perturbed test scalar is undesirable, because back-reaction contributes to the actual
form of quantum gauge transformations and hence of the gauge-invariant variables.
This can lead to an incomplete treatment in partial disagreement with the full gauge-
invariant equations. The inflationary tensor spectrum and its index are known for
holonomy corrections only [272–279], inverse-volume corrections only [229, 280–
283] and both corrections simultaneously [284], while the scalar spectrum and
index have been computed in the presence of inverse-volume [229, 282, 283] and
small holonomy corrections [285] separately. We concentrate on the inverse-volume
case, since one needs the full set of first-order plus non-Gaussianity inflationary
observables in order to make a stronger comparison with observations. At the end
of the chapter, we will sketch the status of LQC spectra with holonomy corrections.
Since the near-Planckian regime is intrinsically non-perturbative, one can safely
trust linear perturbation theory only in the quasi-classical limit, where a consistent
closure of the effective constraint algebra has been verified explicitly. One defines
the smeared effective Hamiltonian constraint with large-volume inverse-volume
corrections as
Z
HŒN D d3 x NŒ˛.E/Hg C .E/H˘ C .E/Hr C HV ; (10.160)
where ˛, and are arbitrary functions of the densitized triad and different
contributions Hg;˘;r;V pertain to the gravitational sector and to the scalar field
kinetic, gradient and potential terms, respectively. Similarly, one considers the
perturbed Gauss and diffeomorphism constraints. The resulting perturbed equations
contain counter-terms which fix the functions ˛, and and guarantee anomaly
cancellation [235, 236]. These counter-terms have been computed in the quasi-
classical limit where (10.137) holds and they depend only on the three parameters
˛0 , 0 and . A consistency condition further reduces the parameter space to two
dimensions:

˛0 1 0 C1 1 D 0: (10.161)
6 6 3
Notably, when D 6 this equation is satisfied only if 0 D 0, which is not
possible in a pure mini-superspace calculation (see (10.138)). Further tension with
the mini-superspace parametrization with D 6 arises from inflationary and de
Sitter background solutions, which exist for 0 . . 3 [282].
Once a closed constraint algebra and the full set of linearized perturbed equations
are obtained, one can study the dynamics of the perturbations. Inverse-volume (and
holonomy) corrections suppress vector perturbations even faster than in classical
cosmology [235] and we can ignore them as usual. Scalar fluctuations and ı in
the metric and in the scalar field generate the gauge-invariant curvature perturbation
on comoving hypersurfaces,
H 0
RDC 0
1 ıinv ı :
6
At large scales, this quantity is conserved thanks to a delicate cancellation of

counter-terms [282]:
h ˛ i 2H
0
R0 D 1 C C 20 ıinv 2 02 r 2 :
2
Because of this property, one can argue (and also rigorously show) that the
Mukhanov–Sasaki scalar variable u D zinv R, where
a 0h ˛
0
i
zinv :D 1C 0 ıinv ; (10.162)
H 2
obeys the simple dynamical equation in momentum space

z00
u00k C s2inv k2 inv uk D 0 ; (10.163)
zinv
where
0 ˛0
s2inv D 1 C ıinv ; :D C1 C 5 ; (10.164)
3 6 2 3
is the (squared) propagation speed of the perturbation.
Tensor observables can be calculated analogously and display the same type of
corrections. In terms of the transverse traceless part of the perturbed 3-metric, the
triad and curvature perturbations are

ˇ 1 ˇ 1 1 ˇ0 c ˇ
ıEi D a2 hi ; ıKˇi D h C hi : (10.165)
2 2 ˛ i
Defining hk as the Fourier transform of the tensor polarization mode h and

˛0
wk :D ainv hk ; ainv :D a 1 ıinv ; (10.166)
2
the Mukhanov–Sasaki equation for the individual tensor mode hk is [226, 281, 282]

a00inv
w00k 2 2
C ˛ k wk D 0 : (10.167)
ainv
This is formally identical to the scalar Mukhanov–Sasaki equation and the analysis
is exactly the same up to the substitutions zinv ! ainv , ! 2˛0 .
10.3.9.2 Observables
According to the inflationary paradigm, observables are expanded in terms of the

slow-roll parameters. Asymptotic solutions to the Mukhanov–Sasaki equation at
large scales, evaluated at horizon crossing, yield the scalar spectrum, the scalar index
and its running to lowest order in the SR parameters:
k3 ˝ ˛ˇ
2 ˇ 2 H2
Ps D 2
ju kH j ˇ D .1 C s ıinv / ; (10.168)
2 2 zinv kj jD1 8 2 a2
ns 1 D 2 4 C ns ıinv ; (10.169)
2 2
˛s D 2.5 4 / C .4Q ns /ıinv ; (10.170)
where
˛0
s :D 0 C1 C ;
6 2 C1

ns :D ˛0 20 C ;
C1

Q :D ˛0 C 2 C 0 1 :
2 6
One can notice a large-scale enhancement of power via the term
ıinv a j j k : (10.171)
If large enough, quantum corrections dominate and ˛s ' fs ıinv to lowest SR order,
where
Œ3˛0 .13 3/ C 0 .6 C 11/

fs :D :
18. C 1/
Bounds on the scalar running turn out to be the main constraint on the parameters.
Due to cosmic variance, there is an intrinsic uncertainty in the determination of the
power spectrum at large scales (small multipoles `), which should be compared with
the strength of the typical signal from quantum corrections.
From (10.167), one derives the tensor spectrum and its index:
4 2 k3 ˝ ˛ˇ
2 ˇ 2 2 H2
Pt D 2
jwkH j jD1
D .1 C t ıinv / ; (10.172)
2 ainv kj 2 a2
1
nt D 2 t ıinv ; t :D ˛0 : (10.173)
C1
Finally, a modified consistency relation between scalar and tensor perturbations

holds:
r D 8fnt C Œnt . t s/ C t ıinv g : (10.174)
10.3.9.3 Experimental Bounds
As in Sect. 10.2.5, one rewrites the observables in terms of the potential-dependent

slow-roll parameters (5.54); the resulting lengthy expressions can be found in [229].
.m/
Since ˛s .k0 / ' .1/m m1 fs ıinv .k0 /, the scalar spectrum (10.66) becomes

˛s .k0 / 2
Ps .k/ ' Ps .k0 / exp Œns .k0 / 1x C x
2

1 1
Cfs ıinv .k0 / x 1 x C .e x 1/ ; (10.175)
2
where x D ln.k=k0 /. This is the expression to be used in numerical analyses

and when comparing the LQC signal with cosmic variance. Before doing so,
we notice the existence of a theoretical upper bound on the quantum correction
ıLQC :D ˛0 ıinv . (We recall that 0 is not independent and it can be removed from
parameter space via (10.161), except in the case D 3 which can be treated
separately.) For the validity of the linear expansion of the perturbation formulæ
where the O.ıinv / truncation has been systematically implemented, we require that
ıLQC .k/ D ıLQC .k0 /.k0 =k/ D ıLQC .k0 /.`0 =`/ < 1 for all wave-numbers relevant
to the CMB anisotropies. For the pivot scale `0 D 29, the quadrupole ` D 2 gives
the bound
ıLQC .k0 / < ıLQC

max
D 14:5 : (10.176)
One can choose an inflationary potential and, for any given choice of , find an
upper bound for ıLQC . For instance, for a quadratic potential and using the 7-year
WMAP data combined with large-scale structure, the Hubble constant measurement
from the Hubble Space Telescope, Supernovae type Ia and BBN, one finds an
experimental upper bound that depends on the value of [229, 283]. The 95%-
confidence-level upper limits of ı constrained by observations for the quadratic
potential with k0 D 0:002 Mpc1 are
ıLQC
max
D 0:27; 3:5102 ; 6:8105 ; 4:3107 ; D 0:5; 1; 2; 3 : (10.177)
For D 3, the parameter ı D 0 ıinv is used instead. For . 1, the signal can be
greater than cosmic variance at large scales (Fig. 10.2). All the cases in (10.177) are
compatible with the theoretical upper limit (10.176) but ıLQC .k0 / ıLQC max
for D
0:5. More recent constraints on other potentials are about 2 3 orders of magnitude
Fig. 10.2 Log-linear plot of the LQC primordial scalar spectrum (10.175) with inverse-volume
quantum corrections for a quadratic inflaton potential, with V .k0 / D 0:009 and for the pivot
wave-number k0 D 0:002 Mpc1 , corresponding to `0 D 29. The classical case is represented by
the dotted line, while solid curves correspond to D 1; 1:5; 2 (decreasing thickness). The shaded
region is affected by cosmic variance (Credit: [160])
stronger [286]. For instance, given the same pivot scale, ıLQC max
103 105 for
D 0:5.
We conclude by reviewing the path which led to the above observational
implications of quantum gravity. Corrections to the general relativistic dynamics are
expected to arise in different ways. For instance, loop corrections are always present
in perturbative graviton field theory, which can be captured in effective actions with
higher-curvature corrections to the Einstein–Hilbert action. The additional terms
change the Newtonian potential as well as the cosmological dynamics. However,
in currently observable regimes the curvature scale is very small and one expects
only tiny corrections of (dimensionless) size at most lPl H. In such cases, exemplified
by (10.69), tests of quantum gravity are possible at best indirectly, for instance if it
provides concrete and sufficiently constrained models for inflation. So far, however,
models do not appear tight enough. The same type of modifications arises in WDW
quantum cosmology, predicting corrections of the size of (10.70), jıWDW j < 3109 .
The LQC quantum corrections are not governed by the energy scale of inflation
but by some quantum-state scale. In background-independent frameworks such
as loop quantum gravity, stronger modifications of the theory are possible since
the usual covariant continuum dynamics is generalized, and entirely new effects
may be contemplated. In LQG, gauge transformations as well as the dynamics
are generated by constraint equations. Since the latter are modified with respect
to the classical constraints, gauge transformations change and a new spacetime
structure becomes apparent. This global deformation of the classical geometry is
the ultimate responsible for possibly sizable inverse-volume quantum effects of the
form (10.151). The stark contrast between WDW and LQC quantum corrections
arising in these scenarios highlights how sensitive the physics is to the quantization
scheme and to the choice of canonical variables.
10.3.9.4 Holonomy Corrections
Another type of quantum effect in the dynamics, holonomy corrections, is realized

in a highly non-linear fashion, from the exponentiation he of curvature components.
Just as in the case of inverse-volume corrections, the constraint algebra is deformed
by quantum effects and gauge transformations do not correspond to standard
diffeomorphisms. This has important effects not only for the inflationary spectra,
but also for the fate of the homogeneous bounce.
The Mukhanov–Sasaki equations for scalar and tensor modes are [226, 255, 268]

z00 a00
u00 s2hol r 2 C hol u D 0 ; w00 s2hol r 2 C hol w D 0 ; (10.178)
zhol ahol
where the effective propagation speeds and background functions zhol and ahol read
0
a
s2hol :D cos.2 LHcl / D 1 2 ; zhol :D ; ahol :D ; (10.179)
H jshol j
and H is given by (10.147) with ˛ D 1. These expressions should be com-

pared with their inverse-volume counterparts (10.162), (10.163), (10.164), (10.166)
and (10.167).
Although the full set of equations of motions for perturbations with holonomy
corrections is known, cosmological observational signatures of holonomy effects
have been studied for the tensor sector alone. The phenomenology of the scalar
sector has not been developed in sufficient detail to compare with experiments as
in the WDW and inverse-volume LQC cases. With respect to the inverse-volume
case, the analysis of the spectra is complicated by the analytic form of holonomy
corrections. In general, tensor modes are amplified during the bounce. However,
after the bounce these modes are enhanced by inflationary expansion later than in
the classical case and the spectrum is thus suppressed at low multipoles, as
Pt k 2 ;
k!0
(10.180)
on a de Sitter background. It also shows an oscillatory pattern, progressively damped

towards small scales. The gravitational spectrum is notoriously difficult to detect by
itself and information from the scalar spectrum (which, from (10.178), is expected
to behave similarly to the tensor one) is needed, also to determine whether the
large-scale suppression is observable past the cosmic-variance noise. Recent results
invoking the PLANCK upper bounds on B-modes show that inflation with holonomy
corrections might be excluded observationally [279]. This does not rule out the
deformed-algebra approach because the interplay with inverse-volume corrections
should also be taken into account when comparing the model with experiments.
10.3.9.5 Non-Gaussianity
The effect of quantum corrections goes beyond linear perturbation theory and
higher-order observables can be calculated. As the perturbative level increases,
the statistics of inhomogeneous fluctuations deviates from the Gaussian one and
odd-order correlation functions acquire non-vanishing values. In particular, the
bispectrum (three-point correlation function of the curvature perturbation) can be
constrained by observations.
Both WDW and LQC quantum cosmology with inverse-volume corrections
predict a small scalar index ns 1, given by (10.62) and (10.169). In Sect. 4.6.3,
we saw that any inflationary model with such a prediction implies a negligible
non-Gaussianity in the squeezed limit (4.108) [287]. One can therefore conclude
that no appreciable WDW or inverse-volume-corrected LQC non-Gaussian sig-
nal of local form can be detected [160]. A detailed calculation of the LQC
momentum-dependent bispectrum for inverse-volume corrections reaches the same
conclusion [288].
10.3.10 Inflation in Other Approaches
Compared with the effective constraints, the hybrid quantization approach [240–
243] is more closely related to the dressed-metric framework [249–251]. In the
former case, the gauge-invariant Mukhanov–Sasaki variables are determined at the
classical level; perturbations are afterwards quantized on a Fock space. Making a
Born–Oppenheimer Ansatz for physical states (valid, for instance, if the latter are
semi-classical with a sharp peak), the Mukhanov–Sasaki equation for, say, the scalar
perturbation uk is

hOz00 i
u00k C k2 uk D 0 ; (10.181)
hOzi
where hzi is the expectation value on the chosen vacuum of a suitable background
operator. The Mukhanov equation in the dressed-metric framework is formally the
same as (10.181) but with a different operator zO (and possibly different choices
of vacuum). In this case, zO contains the scale-factor operator aO whose expectation
value hOai D a is solution to the modified Friedmann equations. hOzi D z is
approximated with the expectation value at which zO is peaked. The two approaches
yield qualitatively the same spectra, which show a suppression of power at large
scales [242, 251].
The inflationary spectra have been computed also in the partial gauge-fixing
scenario of [253, 254] and it is possible to suppress the tensor-to-scalar ratio with
respect to the standard result. Although, as discussed in Sect. 10.3.9.1, the quantum
theory after gauge fixing is different from the other approaches, it is a full loop
quantization of both the background and the perturbations, in the sense that it makes
use of the well-known standard LQC quantization in each patch in the separate-
universe approach.
At intermediate and large scales, there is agreement between the tensor spectrum
in the deformed-algebra approach in the absence of inverse-volume corrections (or
when holonomy corrections
p dominate) and the dressed-metric approach [289]. For
wave-numbers k < 2 D O.l1 Pl
/, the form of all these spectra is quantitatively
the same and is independent of unknown quantum-geometry parameters (there
is a dependence on the inflaton initial conditions). Therefore, at large scales the
effective-dynamics and dressed-metric approaches predict suppressed power spectra
and are in mutual agreement.14 However, in the deep ultraviolet (i.e., at small scales,
large wave-number k) the effective-dynamics spectra have a different asymptotic
behaviour with respect to the other three formalisms. This is expected from the
different way the classical system is quantized and raises an interpretational issue
near Planckian scales, which we now examine.
10.3.11 Is There a Bounce?
Let us go back to the issue of singularity resolution. In homogeneous LQC, the

big bang is replaced by a bounce; this result can be found in so many different
ways that its robustness seems beyond question. Is this feature preserved in the
presence of inhomogeneities? In approaches with gauge fixing or where quantum
gauge invariance stems directly from classical gauge invariance, the answer is in
the affirmative. However, in the deformed-algebra framework gauge invariance is
implemented at the quantum level, which can result in a dramatic rewriting of the
behaviour of the Universe near Planckian densities.
The propagation speed of perturbations is never super-luminal (js2hol j 6 1) but
it does change sign near the bouncing point [255, 271, 278]. This marks a possible
instability and a change of effective spacetime signature at near-Planckian scales
[290, 291], in a super-inflationary early era. This happens because s2hol appears in
the right-hand side of the deformed commutator fH; Hg, (9.166c). In general, this
constraint is deformed as
fHŒN; HŒMg D dŒˇh˛ .N@ M M@ N/ ; (10.182)
where ˇ is a function of the phase-space variables. When ˇ D C1, one recovers the
constraint (9.166c) of general relativity in Lorentzian signature, while when ˇ D
1, one has the algebra of Riemannian gravity (spacetime with Euclidean signature)
[292]. The deformation of LQC with inverse-volume and holonomy corrections is,
respectively, ˇ D s2inv and ˇ D s2hol ; only in the latter case ˇ changes sign [268].
14
It is not clear whether the gauge-fixed quantization via the separate-universe approach [253, 254]
makes the same prediction of a suppression of power at large scales.
The same structure (10.182) appears also in the loop quantization of spherically
symmetric spacetimes [237, 291, 293–296] and in .2C1/-dimensional loop quantum
gravity [297, 298]. In the case of LQC holonomy corrections, at the critical energy
density D one has ˇ D s2hol 1, while at low energies ˇ D s2hol C1.
Going backwards in time from today, ˇ changes sign at some point before reaching
the bounce density, which poses a conceptual problem: Is the homogeneous bounce
reached at all or does spacetime geometry change so much as to invalidate the mini-
superspace approximation?
A change of sign in front of the Laplacian of the Mukhanov–Sasaki equations
is not a novelty in cosmology. The same effect occurs, for instance, in higher-order
gravity where the Gauss–Bonnet curvature invariant is non-minimally coupled with
a scalar field (see [299] and references therein). In that case, this change of sign
is simply interpreted as a classical instability of the perturbations on the FLRW
background affecting cosmological spacetime scales. Although ghost instabilities
and super-luminal propagation are problematic and can be avoided by a restriction
of the parameter space, the nature of the spacetime wherein perturbation modes
propagate remains purely classical and Lorentzian. The higher-order terms of
Gauss–Bonnet theory, in fact, do not lead to deformations of the constraint algebra.
In LQC, however, the change in the perturbation equations is a direct conse-
quence of the deformation (10.182) of the constraint algebra of gravity and, hence,
of a deformation of the classical spacetime structure. The type of field equations
changes from hyperbolic to elliptic for all modes simultaneously. Moreover, the
manifold on which physical fields are defined has no causal structure at high
curvature. These are the main reasons why, in the present context, such an effect
is interpreted as a signature change of spacetime rather than a simple perturbative
instability. This effect is not a transition to classical Euclidean space, since ˇ D 1
is realized only on one hypersurface. Rather, it is a change in the type of partial
differential equations.
Within the effective-dynamics approach, there are choices of counter-terms
which avoid the signature change but still imply the existence of a space-like
surface where the propagation speed of perturbations vanishes and initial conditions
must be set [271]. There is also the possibility that inverse-volume and moment
terms may compensate these deformations so that no signature change takes
place. However, these corrections would not cancel holonomy effects exactly and
are presently unknown. Overall, although the deformation (10.182) happens on
different backgrounds in the deformed-algebra approach, the resolution of classical
gravitational singularities via a Lorentzian mechanism is not as evident as in
homogeneous cosmology. This preliminary conclusion is reinforced in full .3 C 1/-
dimensional LQG, where inverse-volume operators are well-defined but unbounded
from above on zero-volume eigenstates (including the big bang) [300, 301].
The Laplacian modification of the Mukhanov–Sasaki equation does not take
place in the hybrid quantization and dressed-metric approaches. Since gauge invari-
ants are defined with respect to the classical gauge transformations, no deformation
of the constraint algebra (10.182) arises which could give rise to a signature change.
In fact, the propagation speed of the perturbations in (10.181) is always equal to 1
[240–243]. As we already mentioned, both the hybrid quantization and the dressed-
metric frameworks are different ways of quantizing the same classical system and, as
long as they are self-consistent, they should be regarded as physically inequivalent
but equally valid alternatives to the effective-dynamics setting. In the gauge-fixed
approach [207, 253, 254], the factor in front of the kinetic term does change
sign. However, for sub-Planckian modes the derivative term in the Mukhanov–
Sasaki equation is always sub-leading with respect to the effective mass. Since
this framework cannot be applied to trans-Planckian modes, where the kinetic
term would be important, it does not give any conclusive evidence about signature
change.
At present, the signature change remains a possibility, although its appearance,
not just in the effective-dynamics approach but also from operator computations
of off-shell constraint algebras, may be very generic [237, 302, 303]. If it was
confirmed that spacetime changes from Lorentzian to Euclidean when reaching the
critical density from below, this would not mean that singularities are not resolved
in loop quantum cosmology. One could no longer talk of a bounce in the sense of
Lorentzian cosmology, but there might be some tunneling or topology change as in
WDW quantum cosmology, to be translated into some general mechanism in the
full theory more subtle than the requirement of having well-defined inverse-volume
operators. It may turn out that there is a mechanism within loop quantum gravity
by which signature change could be avoided after all. Even if this were the case,
however, one could not show it within mini-superspace models because one must
have access to temporal and spatial variations. This is the reason why such models
do not seem to be completely understood at high density.
10.1 Classical FLRW Hamilton equations. Derive (10.7).

" #
3p2 2 3 2 p2.a/
pP .a/ D f p.a/; HD g D N 3V0 a V C 2 V0 K
2V0 a4 12V0 a2
" # " ! #
3 P 2
3 K 3 H 2 P 2
H 2
D V0 Na2 3V C 2 2 2 2 D2V0 Na2 2 V 3 2 :
2N 2 a N N2 N
10.2 Volume expectation value. Noting that @ ˙;L D ˙@y ˙;L and
@ ˙;R D @y ˙;R for any state, rewrite the inner product (10.40). From
there, derive (10.46) and (10.47).
Solution After integration by parts, (10.40) becomes

Z
ˇ
.1C ; 2C / D 2i
dy 2C;L @y 1C;L
2C;R @y 1C;R ˇ : (10.183)
D 0
Writing the volume operator (10.29) as

s q
3 2 2
vO D i e 3 .1Cn/ y
@y D: iv e0 y @y ;
2.1 C n/2 2 b20
one has
Z

.1C ; v
O 2C / D 2v dy .@y 1C;L /e0 y .@y 2C;L /
ˇ

.@y 1C;R /e0 y .@y 2C;R / ˇ D 0
D .1C;L ; jvj
O 2C;L / .1C;R ; jvj
O 2C;R / : (10.184)
The relative sign comes from the fact that vO leaves the left sector invariant and the
right sector anti-invariant. Physically, what matters is the expectation value of jvj.O
Finally, we have
Z
hjvji
O L D .C;L ; v
O C;L / D 2v dyj@y C;L Œ yC j2 e0 y
Z
D 2v dyC j@yC C;L Œ yC j2 e0 . yC /
D: VL e0 ;
Z
hjvji
O R D .C;R ; v
O C;R / D 2v dyj@y C;R Œ y j2 e0 y
Z
D 2v dy j@y C;R Œ y j2 e0 . y C /
D: VR e0 :
10.3 Classical super-Hamiltonian in LQC. Derive (10.83).
1 Ei˛ Ej ij
H D p Fk C H
2 2 2 j det Ej k ˛
4 V 1q X 0
D tr. k hij / iji tr. k hi0 fh1
i0 ; V g/ C H
(9.115) q
lim
3 4 l0 !0 ql3
0 i;j;k;i0
4 V 1q X
D sin2 2b tr. k hk fh1
k ; V g/ C H
(10.80) q
lim
3 4 l0 !0 ql3
0 k
3
12 1 .1 C n/ 2 pn 2.1Cn/ 2.1Cn/
3.1q/
D v sin2 2b
(10.82)
lim
3 4 q l0 !0 l3
0 3 2
3q 3q
Œsin b fcos b; v 2.1Cn/ g cos b fsin b; v 2.1Cn/ g C H :
10.4 Expectation value of the volume operator in LQC. Derive (10.114).
O D .C ; jvj
hjvji O C/
Z

D 2v dz j@z C;L ŒzC j2 C j@z C;R Œz j2 cosh.0 z/
Z
D 2v dzC j@zC j2 coshŒ0 .zC /
Z
C2v dz j@z j2 coshŒ0 .z C /
Z
2
D 4v dzj@z j cosh.0 z/ cosh.0 /
D: V cosh.0 / :
References 531
References
2. C.W. Misner, Quantum cosmology. I. Phys. Rev. 186, 1319 (1969)
3. K.V. Kuchař, M.P. Ryan, Can mini superspace quantization be justified? in Gravitational
Collapse and Relativity, ed. by H. Sato, T. Nakamura (World Scientific, Singapore, 1986)
4. K.V. Kuchař, M.P. Ryan, Is minisuperspace quantization valid?: Taub in mixmaster. Phys.
Rev. D 40, 3982 (1989)
5. S.W. Hawking, The quantum state of the universe. Nucl. Phys. B 239, 257 (1984)
6. S.P. Kim, Quantum mechanics of conformally and minimally coupled Friedmann–Robertson–
Walker cosmology. Phys. Rev. D 46, 3403 (1992)
7. I.G. Moss, W.A. Wright, Wave function of the inflationary universe. Phys. Rev. D 29, 1067
(1984)
8. S.W. Hawking, Z.C. Wu, Numerical calculations of minisuperspace cosmological models.
Phys. Lett. B 151, 15 (1985)
9. U. Carow, S. Watamura, Quantum cosmological model of the inflationary universe. Phys. Rev.
D 32, 1290 (1985)
10. C. Kiefer, Wave packets in minisuperspace. Phys. Rev. D 38, 1761 (1988)
11. D.J. Kaup, A.P. Vitello, Solvable quantum cosmological model and the importance of
quantizing in a special canonical frame. Phys. Rev. D 9, 1648 (1974)
12. W.F. Blyth, C.J. Isham, Quantization of a Friedmann universe filled with a scalar field. Phys.
Rev. D 11, 768 (1975)
13. R. Brout, G. Horwitz, D. Weil, On the onset of time and temperature in cosmology. Phys.
Lett. B 192, 318 (1987)
14. L. Liu, C.-G. Huang, The quantum cosmology in the Brans–Dicke theory. Gen. Relat. Grav.
20, 583 (1988)
15. D.N. Page, Minisuperspaces with conformally and minimally coupled scalar fields. J. Math.
Phys. 32, 3427 (1991)
16. Z.H. Zhu, Boundary conditions in quantum cosmology in the Brans–Dicke theory. Chin. Phys.
Lett. 9, 273 (1992)
17. C. Kiefer, E.A. Martínez, On time and the quantum to classical transition in Jordan–Brans–
Dicke quantum gravity. Class. Quantum Grav. 10, 2511 (1993). [arXiv:gr-qc/9306029]
18. J.E. Lidsey, Scale factor duality and hidden supersymmetry in scalar-tensor cosmology. Phys.
Rev. D 52, 5407 (1995). [arXiv:gr-qc/9510017]
19. Z.-H. Zhu, Y.-Z. Zhang, X.-P. Wu, On the cosmological constant in quantum cosmology of
the Brans–Dicke theory. Mod. Phys. Lett. A 13, 1333 (1998)
20. Z.-H. Zhu, Cosmic wave functions with the Brans–Dicke theory. Chin. Phys. Lett. 17, 856
(2000)
21. D.-i. Hwang, H. Sahlmann, D.-h. Yeom, The no-boundary measure in scalar-tensor gravity.
22. S.W. Hawking, J.C. Luttrell, Higher derivatives in quantum cosmology: (I). The isotropic
case. Nucl. Phys. B 247, 250 (1984)
23. P.F. González-Díaz, On the wave function of the universe. Phys. Lett. B 159, 19 (1985)
(1989)
25. S. Sinha, B.L. Hu, Validity of the minisuperspace approximation: an example from interacting
quantum field theory. Phys. Rev. D 44, 1028 (1991)
26. F.D. Mazzitelli, Midisuperspace-induced corrections to the Wheeler–DeWitt equation. Phys.
Rev. D 46, 4758 (1992). [arXiv:hep-th/9203072]
27. A. Ishikawa, T. Isse, The stability of the minisuperspace. Mod. Phys. Lett. A 08, 3413 (1993).
Quantum Grav. 4, 1369 (1987)
29. J.J. Halliwell, Correlations in the wave function of the universe. Phys. Rev. D 36, 3626 (1987)
30. R. Brout, On the concept of time and the origin of the cosmological temperature. Found.
Phys. 17, 603 (1987)
31. R. Brout, G. Venturi, Time in semiclassical gravity. Phys. Rev. D 39, 2436 (1989)
32. D.P. Datta, Geometric phase in vacuum instability: applications in quantum cosmology. Phys.
Rev. D 48, 5746 (1993). [arXiv:gr-qc/9306028]
33. T. Banks, TCP, quantum gravity, the cosmological constant and all that. . . . Nucl. Phys. B 249,
332 (1985)
34. S.P. Kim, New asymptotic expansion method for the Wheeler–DeWitt equation. Phys. Rev.
D 52, 3382 (1995). [arXiv:gr-qc/9511038]
35. S.P. Kim, Classical spacetime from quantum gravity. Class. Quantum Grav. 13, 1377 (1996).
36. C. Bertoni, F. Finelli, G. Venturi, The Born–Oppenheimer approach to the matter-gravity
system and unitarity. Class. Quantum Grav. 13, 2375 (1996). [arXiv:gr-qc/9604011]
37. S.P. Kim, Problem of unitarity and quantum corrections in semiclassical quantum gravity.
38. S.P. Kim, Quantum potential and cosmological singularities. Phys. Lett. A 236, 11 (1997).
39. J.J. Halliwell, Decoherence in quantum cosmology. Phys. Rev. D 39, 2912 (1989)
40. C. Kiefer, Decoherence in quantum electrodynamics and quantum gravity. Phys. Rev. D 46,
1658 (1992)
42. A. Vilenkin, Boundary conditions in quantum cosmology. Phys. Rev. D 33, 3560 (1986)
43. A. Vilenkin, Quantum cosmology and the initial state of the Universe. Phys. Rev. D 37, 888
(1988)
44. A. Vilenkin, Approaches to quantum cosmology. Phys. Rev. D 50, 2581 (1994).
45. J.B. Hartle, S.W. Hawking, Wave function of the Universe. Phys. Rev. D 28, 2960 (1983)
46. A.D. Linde, Quantum creation of an inflationary universe. Zh. Eksp. Teor. Fiz. 87, 369 (1984)
[Sov. Phys. JETP 60, 211 (1984)]; Quantum creation of the inflationary universe. Lett. Nuovo
Cim. 39, 401 (1984)
47. A. Vilenkin, Wave function discord. Phys. Rev. D 58, 067301 (1998). [arXiv:gr-qc/9804051]
48. G. Calcagni, C. Kiefer, C.F. Steinwachs, Quantum cosmological consistency condition for
1996). [arXiv:gr-qc/0101003]
50. L.P. Grishchuk, L.V. Rozhansky, Does the Hartle–Hawking wavefunction predict the universe
we live in? Phys. Lett. B 234, 9 (1990)
51. A. Lukas, The no boundary wave-function and the duration of the inflationary period. Phys.
Lett. B 347, 13 (1995). [arXiv:gr-qc/9409012]
52. J.B. Hartle, S.W. Hawking, T. Hertog, No-boundary measure of the Universe. Phys. Rev. Lett.
100, 201301 (2008). [arXiv:0711.4630]
53. J.B. Hartle, S.W. Hawking, T. Hertog, Classical universes of the no-boundary quantum state.
Phys. Rev. D 77, 123537 (2008). [arXiv:0803.1663]
54. J.B. Hartle, S.W. Hawking, T. Hertog, No-boundary measure in the regime of eternal inflation.
Phys. Rev. D 82, 063510 (2010). [arXiv:1001.0262]
55. A.O. Barvinsky, A.Yu. Kamenshchik, C. Kiefer, C.F. Steinwachs, Tunneling cosmological
state revisited: origin of inflation with a non-minimally coupled standard model Higgs
inflaton. Phys. Rev. D 81, 043530 (2010). [arXiv:0911.1408]
56. A.O. Barvinsky, A.Yu. Kamenshchik, 1-loop quantum cosmology: the Normalizability of the
Hartle-Hawking wave function and the probability of inflation. Class. Quantum Grav. 7, L181
(1990)
References 533
57. A.O. Barvinsky, Unitarity approach to quantum cosmology. Phys. Rep. 230, 237 (1993)
58. A.O. Barvinsky, Reduction methods for functional determinants in quantum gravity and
59. A.O. Barvinsky, A.Yu. Kamenshchik, C. Kiefer, Effective action and decoherence by fermions
in quantum cosmology. Nucl. Phys. B 552, 420 (1999). [arXiv:gr-qc/9901055]
60. A.O. Barvinsky, A.Yu. Kamenshchik, Quantum scale of inflation and particle physics of the
early universe. Phys. Lett. B 332, 270 (1994). [arXiv:gr-qc/9404062]
61. A.O. Barvinsky, A.Yu. Kamenshchik, Effective equations of motion and initial conditions for
inflation in quantum cosmology. Nucl. Phys. B 532, 339 (1998). [arXiv:hep-th/9803052]
62. G.W. Gibbons, N. Turok, The measure problem in cosmology. Phys. Rev. D 77, 063516
(2008). [arXiv:hep-th/0609095]
63. A.D. Linde, Inflationary cosmology. Lect. Notes Phys. 738, 1 (2008). [arXiv:0705.0164]
64. J.S. Schiffrin, R.M. Wald, Measure and probability in cosmology. Phys. Rev. D 86, 023521
(2012). [arXiv:1202.1818]
65. A. Kaya, Comments on the canonical measure in cosmology. Phys. Lett. B 713, 1 (2012).
[arXiv:1203.2807]
66. A. Higuchi, Quantum linearization instabilities of de Sitter space-time. II. Class. Quantum
Grav. 8, 1983 (1991)
67. A. Higuchi, Linearized quantum gravity in flat space with toroidal topology. Class. Quantum
Grav. 8, 2023 (1991)
68. N.P. Landsman, Rieffel induction as generalized quantum Marsden–Weinstein reduction. J.
Geom. Phys. 15, 285 (1995). [arXiv:hep-th/9305088]
69. D. Marolf, The spectral analysis inner product for quantum gravity, in Proceedings of the
Seventh Marcel Grossman Meeting on General Relativity, ed. by R. Ruffini, M. Keiser (World
Scientific, Singapore, 1994). [arXiv:gr-qc/9409036]
70. A. Ashtekar, J. Lewandowski, D. Marolf, J. Mourão, T. Thiemann, Quantization of diffeo-
morphism invariant theories of connections with local degrees of freedom. J. Math. Phys. 36,
6456 (1995). [arXiv:gr-qc/9504018]
71. D. Marolf, Refined algebraic quantization: systems with a single constraint.
arXiv:gr-qc/9508015
72. J.B. Hartle, D. Marolf, Comparing formulations of generalized quantum mechanics for
reparametrization-invariant systems. Phys. Rev. D 56, 6247 (1997). [arXiv:gr-qc/9703021]
73. A. Ashtekar, L. Bombelli, A. Corichi, Semiclassical states for constrained systems. Phys. Rev.
D 72, 025008 (2005). [arXiv:gr-qc/0504052]
74. A. Ashtekar, T. Pawłowski, P. Singh, Quantum nature of the big bang: an analytical and
numerical investigation. I. Phys. Rev. D 73, 124038 (2006). [arXiv:gr-qc/0604013]
75. J.J. Halliwell, M.E. Ortiz, Sum-over-histories origin of the composition laws of rel-
ativistic quantum mechanics and quantum cosmology. Phys. Rev. D 48, 748 (1993).
76. J.J. Halliwell, J. Thorwart, Decoherent histories analysis of the relativistic particle. Phys. Rev.
D 64, 124018 (2001). [arXiv:gr-qc/0106095]
77. G. Calcagni, S. Gielen, D. Oriti, Two-point functions in (loop) quantum cosmology. Class.
78. A. Ashtekar, A. Corichi, P. Singh, Robustness of key features of loop quantum cosmology.
Phys. Rev. D 77, 024046 (2008). [arXiv:0710.3565]
79. A. Kamenshchik, C. Kiefer, B. Sandhöfer, Quantum cosmology with big-brake singularity.
Phys. Rev. D 76, 064032 (2007). [arXiv:0705.1688]
80. S.W. Hawking, J.C. Luttrell, The isotropy of the universe. Phys. Lett. B 143, 83 (1984)
81. W.A. Wright, I.G. Moss, The anisotropy of the universe. Phys. Lett. B 154, 115 (1985)
82. P. Amsterdamski, Wave function of an anisotropic universe. Phys. Rev. D 31, 3073 (1985)
83. T. Furusawa, Quantum chaos of mixmaster universe. Prog. Theor. Phys. 75, 59 (1986)
84. T. Furusawa, Quantum chaos of mixmaster universe. II. Prog. Theor. Phys. 76, 67 (1986)
85. B.K. Berger, Quantum chaos in the mixmaster universe. Phys. Rev. D 39, 2426 (1989)
86. R. Graham, P. Szépfalusy, Quantum creation of a generic universe. Phys. Rev. D 42, 2483
(1990)
87. B.K. Berger, Application of Monte Carlo simulation methods to quantum cosmology. Phys.
Rev. D 48, 513 (1993)
88. R. Graham, Chaos and quantum chaos in cosmological models. Chaos Solitons Fractals 5,
1103 (1995). [arXiv:gr-qc/9403030]
89. R. Benini, G. Montani, Inhomogeneous quantum mixmaster: from classical toward quantum
mechanics. Class. Quantum Grav. 24, 387 (2007). [arXiv:gr-qc/0612095]
90. E. Calzetta, Chaos, decoherence and quantum cosmology. Class. Quantum Grav. 29, 143001
(2012). [arXiv:1205.1841]
92. A. Csordás, R. Graham, P. Szépfalusy, Level statistics of a noncompact cosmological billiard.
Phys. Rev. A 44, 1491 (1991)
93. R. Graham, R. Hübner, P. Szépfalusy, G. Vattay, Level statistics of a noncompact integrable
billiard. Phys. Rev. A 44, 7002 (1991)
94. A. Peres, Ergodicity and mixing in quantum theory. I. Phys. Rev. A 30, 504 (1984)
95. M. Feingold, N. Moiseyev, A. Peres, Ergodicity and mixing in quantum theory. II. Phys. Rev.
A 30, 509 (1984)
97. M.J. Duff, P. van Nieuwenhuizen, Quantum inequivalence of different field representations.
Phys. Lett. B 94, 179 (1980)
98. A. Aurilia, H. Nicolai, P.K. Townsend, Hidden constants: the parameter of QCD and the
cosmological constant of N D 8 supergravity. Nucl. Phys. B 176, 509 (1980)
99. M. Henneaux, C. Teitelboim, The cosmological constant as a canonical variable. Phys. Lett.
B 143, 415 (1984)
100. E. Baum, Zero cosmological constant from minimum action. Phys. Lett. B 133, 185 (1983)
101. S.W. Hawking, The cosmological constant is probably zero. Phys. Lett. B 134, 403 (1984)
102. M.J. Duff, The cosmological constant is possibly zero, but the proof is probably wrong. Phys.
Lett. B 226, 36 (1989)
103. M.J. Duncan, L.G. Jensen, Four-forms and the vanishing of the cosmological constant. Nucl.
Phys. B 336, 100 (1990)
104. Z.C. Wu, The cosmological constant is probably zero, and a proof is possibly right. Phys.
Lett. B 659, 891 (2008). [arXiv:0709.3314]
105. T. Banks, Prolegomena to a theory of bifurcating universes: a nonlocal solution to the
cosmological constant problem or little lambda goes back to the future. Nucl. Phys. B 309,
493 (1988)
106. S.R. Coleman, Why there is nothing rather than something: a theory of the cosmological
constant. Nucl. Phys. B 310, 643 (1988)
107. S.B. Giddings, A. Strominger, Baby universes, third quantization and the cosmological
constant. Nucl. Phys. B 321, 481 (1989)
108. M. McGuigan, Third quantization and the Wheeler–DeWitt equation. Phys. Rev. D 38, 3031
(1988)
109. A. Hosoya, M. Morikawa, Quantum field theory of the Universe. Phys. Rev. D 39, 1123
(1989)
110. V.A. Rubakov, P.G. Tinyakov, Gravitational instantons and creation of expanding universes.
Phys. Lett. B 214, 334 (1988)
111. V.A. Rubakov, On third quantization and the cosmological constant. Phys. Lett. B 214, 503
(1988)
112. I.R. Klebanov, L. Susskind, T. Banks, Wormholes and the cosmological constant. Nucl. Phys.
B 317, 665 (1989)
113. W. Fischler, L. Susskind, A wormhole catastrophe. Phys. Lett. B 217, 48 (1989)
114. M. McGuigan, Universe creation from the third-quantized vacuum. Phys. Rev. D 39, 2229
(1989)
References 535
115. J. Preskill, Wormholes in spacetime and the constants of nature. Nucl. Phys. B 323, 141 (1989)
116. W. Fischler, I.R. Klebanov, J. Polchinski, L. Susskind, Quantum mechanics of the googol-
plexus. Nucl. Phys. B 327, 157 (1989)
117. Y.-M. Xiang, L. Liu, Third quantization of a solvable model in quantum cosmology in Brans–
Dicke theory. Chin. Phys. Lett. 8, 52 (1991)
118. H.J. Pohle, Coherent states and Heisenberg uncertainty relation in a third-quantized minisu-
perspace. Phys. Lett. B 261, 257 (1991)
119. K. Kuchař, Time and interpretations of quantum gravity, in Proceedings of the Fourth Cana-
dian Conference on General Relativity and Relativistic Astrophysics, ed. by G. Kunstatter et
al. (World Scientific, Singapore, 1992) [Int. J. Mod. Phys. Proc. Suppl. D 20, 3 (2011)]
120. C. Isham, Canonical quantum gravity and the problem of time, in Integrable Systems,
Quantum Groups, and Quantum Field Theories, ed. by L.A. Ibort, M.A. Rodríguez (Kluwer,
Dordrecht, 1993). [arXiv:gr-qc/9210011]
121. S. Abe, Fluctuations around the Wheeler–DeWitt trajectories in third-quantized cosmology.
Phys. Rev. D 47, 718 (1993)
122. T. Horiguchi, Uncertainty relation in a third-quantized universe. Phys. Rev. D 48, 5764 (1993)
123. M.A. Castagnino, A. Gangui, F.D. Mazzitelli, I.I. Tkachev, Third quantization, decoherence
and the interpretation of quantum gravity in minisuperspace. Class. Quantum Grav. 10, 2495
(1993)
124. L.O. Pimentel, C. Mora, Third quantization of Brans–Dicke cosmology. Phys. Lett. A 280,
191 (2001). [arXiv:gr-qc/0009026]
125. S. Gielen, D. Oriti, Discrete and continuum third quantization of gravity, in Quantum Field
Theory and Gravity, ed. by F. Finster et al. (Springer, Basel, 2012). [arXiv:1102.2226]
126. G. Calcagni, S. Gielen, D. Oriti, Group field cosmology: a cosmological field theory of
quantum geometry. Class. Quantum Grav. 29, 105005 (2012). [arXiv:1201.4151]
127. Y. Ohkuwa, Y. Ezawa, Third quantization of f .R/-type gravity. Class. Quantum Grav. 29,
215004 (2012). [arXiv:1203.1361]
128. M. Faizal, Super-group field cosmology. Class. Quantum Grav. 29, 215009 (2012).
[arXiv:1209.2346]
129. Y. Ohkuwa, Y. Ezawa, Third quantization of f .R/-type gravity II: general f .R/ case. Class.
130. M. Faizal, Absence of black holes information paradox in group field cosmology. Int. J.
Geom. Methods Mod. Phys. 11, 1450010 (2014). [arXiv:1301.0224]
131. S. Gielen, D. Oriti, L. Sindoni, Cosmology from group field theory formalism for quantum
gravity. Phys. Rev. Lett. 111, 031301 (2013). [arXiv:1303.3576]
132. A. Hosoya, W. Ogura, Wormhole instanton solution in the Einstein–Yang–Mills system. Phys.
Lett. B 225, 117 (1989)
133. F.S. Accetta, A. Chodos, B. Shao, Wormholes and baby universes in scalar-tensor gravity.
Nucl. Phys. B 333, 221 (1990)
134. L.J. Garay, J. García-Bellido, Jordan–Brans–Dicke quantum wormholes and Coleman’s
mechanism. Nucl. Phys. B 400, 416 (1993). [arXiv:gr-qc/9209015]
135. J. García-Bellido, A.D. Linde, D.A. Linde, Fluctuations of the gravitational constant in the
inflationary Brans–Dicke cosmology. Phys. Rev. D 50, 730 (1994). [arXiv:astro-ph/9312039]
136. A. Vilenkin, Predictions from quantum cosmology. Phys. Rev. Lett. 74, 846 (1995).
137. J. García-Bellido, A.D. Linde, Stationarity of inflation and predictions of quantum cosmology.
138. A. Vilenkin, Making predictions in eternally inflating universe. Phys. Rev. D 52, 3365 (1995).
139. H. Martel, P.R. Shapiro, S. Weinberg, Likely values of the cosmological constant. Astrophys.
J. 492, 29 (1998). [arXiv:astro-ph/9701099]
140. J. Garriga, A. Vilenkin, On likely values of the cosmological constant. Phys. Rev. D 61,
083502 (2000). [arXiv:astro-ph/9908115]
141. S. Weinberg, A priori probability distribution of the cosmological constant. Phys. Rev. D 61,
103505 (2000). [arXiv:astro-ph/0002387]
142. S. Weinberg, The cosmological constant problems. arXiv:astro-ph/0005265
143. M.L. Graesser, S.D.H. Hsu, A. Jenkins, M.B. Wise, Anthropic distribution for cosmo-
logical constant and primordial density perturbations. Phys. Lett. B 600, 15 (2004).
144. B. Feldstein, L.J. Hall, T. Watari, Density perturbations and the cosmological constant from
inflationary landscapes. Phys. Rev. D 72, 123506 (2005). [arXiv:hep-th/0506235]
145. J. Garriga, A. Vilenkin, Anthropic prediction for and the Q catastrophe. Prog. Theor. Phys.
Suppl. 163, 245 (2006). [arXiv:hep-th/0508005]
146. C. Kiefer, F. Queisser, A.A. Starobinsky, Cosmological constant from decoherence. Class.
147. J.J. Halliwell, S.W. Hawking, Origin of structure in the Universe. Phys. Rev. D 31, 1777
(1985)
148. S. Wada, Quantum cosmological perturbations in pure gravity. Nucl. Phys. B 276, 729 (1986);
Erratum-ibid. B 284, 747 (1987)
149. T. Vachaspati, A. Vilenkin, Uniqueness of the tunneling wave function of the Universe. Phys.
Rev. D 37, 898 (1988)
150. C. Kiefer, Quantum Gravity (Oxford University Press, Oxford, 2012)
151. C. Kiefer, T.P. Singh, Quantum gravitational corrections to the functional Schrödinger
equation. Phys. Rev. D 44, 1067 (1991)
152. C. Kiefer, M. Krämer, Quantum gravitational contributions to the CMB anisotropy spectrum.
Phys. Rev. Lett. 108, 021301 (2012). [arXiv:1103.4967]
153. C. Kiefer, M. Krämer, Can effects of quantum gravity be observed in the cosmic microwave
background? Int. J. Mod. Phys. D 21, 1241001 (2012). [arXiv:1205.5161]
154. D. Bini, G. Esposito, C. Kiefer, M. Krämer, F. Pessina, On the modification of the cosmic
microwave background anisotropy spectrum from canonical quantum gravity. Phys. Rev. D
87, 104008 (2013). [arXiv:1303.0531]
155. Y.-S. Piao, B. Feng, X. Zhang, Suppressing CMB quadrupole with a bounce from contracting
phase to inflation. Phys. Rev. D 69, 103520 (2004). [arXiv:hep-th/0310206]
156. Z.-G. Liu, Z.-K. Guo, Y.-S. Piao, Obtaining the CMB anomalies with a bounce from the
contracting phase to inflation. Phys. Rev. D 88, 063539 (2013). [arXiv:1304.6527]
158. S. Tsujikawa, R. Maartens, R. Brandenberger, Non-commutative inflation and the CMB.
159. Y.-S. Piao, S. Tsujikawa, X. Zhang, Inflation in string inspired cosmology and suppression of
CMB low multipoles. Class. Quantum Grav. 21, 4455 (2004). [arXiv:hep-th/0312139]
160. G. Calcagni, Observational effects from quantum cosmology. Ann. Phys. (Berlin) 525, 323
(2013); Erratum-ibid. 525, A165 (2013). [arXiv:1209.0473]
161. M. Bojowald, Loop quantum cosmology I: kinematics. Class. Quantum Grav. 17, 1489
(2000). [arXiv:gr-qc/9910103]
162. M. Bojowald, Loop quantum cosmology. II: volume operators. Class. Quantum Grav. 17,
1509 (2000). [arXiv:gr-qc/9910104]
163. M. Bojowald, Loop quantum cosmology III: Wheeler–DeWitt operators. Class. Quantum
Grav. 18, 1055 (2001). [arXiv:gr-qc/0008052]
164. M. Bojowald, Loop quantum cosmology IV: discrete time evolution. Class. Quantum Grav.
18, 1071 (2001). [arXiv:gr-qc/0008053]
165. M. Bojowald, Inverse scale factor in isotropic quantum geometry. Phys. Rev. D 64, 084018
(2001). [arXiv:gr-qc/0105067]
166. A. Ashtekar, M. Bojowald, J. Lewandowski, Mathematical structure of loop quantum
cosmology. Adv. Theor. Math. Phys. 7, 233 (2003). [arXiv:gr-qc/0304074]
167. A. Ashtekar, T. Pawłowski, P. Singh, Quantum nature of the big bang: improved dynamics.
References 537
168. X. Zhang, Y. Ma, Extension of loop quantum gravity to f .R/ theories. Phys. Rev. Lett. 106,
171301 (2011). [arXiv:1101.1752]
169. X. Zhang, Y. Ma, Loop quantum f .R/ theories. Phys. Rev. D 84, 064040 (2011).
[arXiv:1107.4921]
170. X. Zhang, Y. Ma, Nonperturbative loop quantization of scalar-tensor theories of gravity. Phys.
Rev. D 84, 104045 (2011). [arXiv:1107.5157]
171. M. Bojowald, Loop quantum cosmology. Living Rev. Relat. 11, 4 (2008)
172. M. Bojowald, Quantization ambiguities in isotropic quantum geometry. Class. Quantum Grav.
19, 5113 (2002). [arXiv:gr-qc/0206053]
173. M. Bojowald, Inflation from quantum geometry. Phys. Rev. Lett. 89, 261301 (2002).
174. T. Thiemann, Quantum spin dynamics (QSD): V. Quantum gravity as the natural regulator of
the Hamiltonian constraint of matter quantum field theories. Class. Quantum Grav. 15, 1281
(1998). [arXiv:gr-qc/9705019]
175. M. Bojowald, J.E. Lidsey, D.J. Mulryne, P. Singh, R. Tavakol, Inflationary cosmology and
quantization ambiguities in semiclassical loop quantum gravity. Phys. Rev. D 70, 043530
(2004). [arXiv:gr-qc/0403106]
176. A. Ashtekar, E. Wilson-Ewing, Loop quantum cosmology of Bianchi I models. Phys. Rev. D
79, 083535 (2009). [arXiv:0903.3397]
177. M. Bojowald, R. Das, R.J. Scherrer, Dirac fields in loop quantum gravity and big bang
nucleosynthesis. Phys. Rev. D 77, 084003 (2008). [arXiv:0710.5734]
178. A. Ashtekar, T. Pawłowski, P. Singh, K. Vandersloot, Loop quantum cosmology of k D 1
FRW models. Phys. Rev. D 75, 024035 (2007). [arXiv:gr-qc/0612104]
179. M. Bojowald, Isotropic loop quantum cosmology. Class. Quantum Grav. 19, 2717 (2002).
180. W. Nelson, M. Sakellariadou, Lattice refining LQC and the matter Hamiltonian. Phys. Rev. D
76, 104003 (2007). [arXiv:0707.0588]
181. K. Vandersloot, Loop quantum cosmology and the k D 1 Robertson–Walker model. Phys.
Rev. D 75, 023523 (2007). [arXiv:gr-qc/0612070]
182. M. Bojowald, P. Singh, A. Skirzewski, Coordinate time dependence in quantum gravity. Phys.
Rev. D 70, 124022 (2004). [arXiv:gr-qc/0408094]
183. W. Kamiński, J. Lewandowski, The flat FRW model in LQC: the self-adjointness. Class.
184. W. Kamiński, T. Pawłowski, The LQC evolution operator of FRW universe with positive
cosmological constant. Phys. Rev. D 81, 024014 (2010). [arXiv:0912.0162]
185. K. Vandersloot, Hamiltonian constraint of loop quantum cosmology. Phys. Rev. D 71, 103506
(2005). [arXiv:gr-qc/0502082]
186. M. Bojowald, Absence of singularity in loop quantum cosmology. Phys. Rev. Lett. 86, 5227
(2001). [arXiv:gr-qc/0102069]
187. M. Bojowald, K. Vandersloot, Loop quantum cosmology, boundary proposals, and inflation.
188. P. Singh, A. Toporensky, Big crunch avoidance in k D 1 semiclassical loop quantum
189. J.E. Lidsey, D.J. Mulryne, N.J. Nunes, R. Tavakol, Oscillatory universes in loop quan-
tum cosmology and initial conditions for inflation. Phys. Rev. D 70, 063521 (2004).
190. Ł. Szulc, W. Kamiński, J. Lewandowski, Closed FRW model in loop quantum cosmology.
191. M. Bojowald, R. Tavakol, Recollapsing quantum cosmologies and the question of entropy.
Phys. Rev. D 78, 023515 (2008). [arXiv:0803.4484]
192. J. Mielczarek , O. Hrycyna, M. Szydłowski, Effective dynamics of the closed loop quantum
cosmology. JCAP 0911, 014 (2009). [arXiv:0906.2503]
193. P. Singh, F. Vidotto, Exotic singularities and spatially curved loop quantum cosmology. Phys.
Rev. D 83, 064027 (2011). [arXiv:1012.1307]
194. Ł. Szulc, Open FRW model in loop quantum cosmology. Class. Quantum Grav. 24, 6191
(2007). [arXiv:0707.1816]
195. E. Bentivegna, T. Pawłowski, Anti-de Sitter universe dynamics in LQC. Phys. Rev. D 77,
124025 (2008). [arXiv:0803.4446]
196. P. Singh, K. Vandersloot, Semiclassical states, effective dynamics and classical emergence in
loop quantum cosmology. Phys. Rev. D 72, 084004 (2005). [arXiv:gr-qc/0507029]
197. A. Corichi, E. Montoya, Coherent semiclassical states for loop quantum cosmology. Phys.
Rev. D 84, 044021 (2011). [arXiv:1105.5081]
198. M. Bojowald, Large scale effective theory for cosmological bounces. Phys. Rev. D 75,
081301(R) (2007). [arXiv:gr-qc/0608100]
199. M. Bojowald, H.H. Hernández, A. Skirzewski, Effective equations for isotropic quantum
cosmology including matter. Phys. Rev. D 76, 063511 (2007). [arXiv:0706.1057]
200. M. Bojowald, Quantum nature of cosmological bounces. Gen. Relat. Grav. 40, 2659 (2008).
[arXiv:0801.4001]
201. P. Singh, Loop cosmological dynamics and dualities with Randall–Sundrum braneworlds.
202. P. Singh, K. Vandersloot, G.V. Vereshchagin, Nonsingular bouncing universes in loop
quantum cosmology. Phys. Rev. D 74, 043510 (2006). [arXiv:gr-qc/0606032]
203. C. Rovelli, E. Wilson-Ewing, Why are the effective equations of loop quantum cosmology so
accurate? Phys. Rev. D 90, 023538 (2014). [arXiv:1310.8654]
204. M. Bojowald, Loop quantum cosmology: recent progress. Pramana 63, 765 (2004).
205. M. Bojowald, H.H. Hernández, M. Kagan, A. Skirzewski, Effective constraints of loop
quantum gravity. Phys. Rev. D 75, 064022 (2007). [arXiv:gr-qc/0611112]
206. M. Bojowald, Loop quantum cosmology and inhomogeneities. Gen. Relat. Grav. 38, 1771
(2006). [arXiv:gr-qc/0609034]
207. E. Wilson-Ewing, Holonomy corrections in the effective equations for scalar mode
perturbations in loop quantum cosmology. Class. Quantum Grav. 29, 085005 (2012).
[arXiv:1108.6265]
208. C. Germani, W. Nelson, M. Sakellariadou, On the onset of inflation in loop quantum
209. A. Ashtekar, D. Sloan, Probability of inflation in loop quantum cosmology. Gen. Relat. Grav.
43, 3619 (2011). [arXiv:1103.2475]
210. D.-W. Chiou, Loop quantum cosmology in Bianchi type I models: analytical investigation.
211. D.-W. Chiou, K. Vandersloot, Behavior of nonlinear anisotropies in bouncing Bianchi I
models of loop quantum cosmology. Phys. Rev. D 76, 084015 (2007). [arXiv:0707.2548]
212. M. Martín-Benito, L.J. Garay, G.A. Mena Marugán, Hybrid quantum Gowdy cosmology:
combining loop and Fock quantizations. Phys. Rev. D 78, 083516 (2008). [arXiv:0804.1098]
213. G.A. Mena Marugán, M. Martín-Benito, Hybrid quantum cosmology: combining loop and
Fock quantizations. Int. J. Mod. Phys. A 24, 2820 (2009). [arXiv:0907.3797]
214. M. Martín-Benito, G.A. Mena Marugán, E. Wilson-Ewing, Hybrid quantization: from Bianchi
I to the Gowdy model. Phys. Rev. D 82, 084012 (2010). [arXiv:1006.2369]
215. R.H. Gowdy, Vacuum space-times with two parameter spacelike isometry groups and compact
invariant hypersurfaces: topologies and boundary conditions. Ann. Phys. (N.Y.) 83, 203
(1974)
216. A. Ashtekar, E. Wilson-Ewing, Loop quantum cosmology of Bianchi type II models. Phys.
Rev. D 80, 123532 (2009). [arXiv:0910.1278]
217. E. Wilson-Ewing, Loop quantum cosmology of Bianchi type IX models. Phys. Rev. D 82,
043508 (2010). [arXiv:1005.5565]
218. P. Singh, Are loop quantum cosmos never singular? Class. Quantum Grav. 26, 125005 (2009).
[arXiv:0901.2750]
219. M. Bojowald, G. Date, Quantum suppression of the generic chaotic behavior close to
cosmological singularities. Phys. Rev. Lett. 92, 071302 (2004). [arXiv:gr-qc/0311003]
References 539
220. M. Bojowald, G. Date, K. Vandersloot, Homogeneous loop quantum cosmology: the role of
the spin connection. Class. Quantum Grav. 21, 1253 (2004). [arXiv:gr-qc/0311004]
221. M. Bojowald, G. Date, G.M. Hossain, The Bianchi IX model in loop quantum cosmology.
222. E. Alesci, F. Cianfrani, A new perspective on cosmology in loop quantum gravity. Europhys.
Lett. 104, 10001 (2013). [arXiv:1210.4504]
223. E. Alesci, F. Cianfrani, Quantum-reduced loop gravity: cosmology. Phys. Rev. D 87, 083521
(2013). [arXiv:1301.2245]
224. E. Alesci, F. Cianfrani, C. Rovelli, Quantum-reduced loop gravity: relation with the full
theory. Phys. Rev. D 88, 104001 (2013). [arXiv:1309.6304]
225. E. Alesci, F. Cianfrani, Quantum reduced loop gravity: semiclassical limit. Phys. Rev. D 90,
024006 (2014). [arXiv:1402.3155]
226. M. Bojowald, G.M. Hossain, Loop quantum gravity corrections to gravitational wave
dispersion. Phys. Rev. D 77, 023508 (2008). [arXiv:0709.2365]
227. T. Thiemann, Quantum spin dynamics (QSD). Class. Quantum Grav. 15, 839 (1998).
228. C. Rovelli, L. Smolin, The physical Hamiltonian in nonperturbative quantum gravity. Phys.
229. M. Bojowald, G. Calcagni, S. Tsujikawa, Observational test of inflation in loop quantum
cosmology. JCAP 1111, 046 (2011). [arXiv:1107.1540]
230. M. Bojowald, Quantum geometry and its implications for black holes. Int. J. Mod. Phys. D
15, 1545 (2006). [arXiv:gr-qc/0607130]
231. A. Ashtekar, Quantum space-times. Fund. Theories Phys. 165, 163 (2010). [arXiv:0810.0514]
232. M. Bojowald, W. Nelson, D. Mulryne, R. Tavakol, The high-density regime of kinetic-
dominated loop quantum cosmology. Phys. Rev. D 82, 124055 (2010). [arXiv:1004.3979]
233. M. Bojowald, Consistent loop quantum cosmology. Class. Quantum Grav. 26, 075020 (2009).
[arXiv:0811.4129]
234. M. Bojowald, D. Cartin, G. Khanna, Lattice refining loop quantum cosmology, anisotropic
models and stability. Phys. Rev. D 76, 064018 (2007). [arXiv:0704.1137]
235. M. Bojowald, G.M. Hossain, Cosmological vector modes and quantum gravity effects. Class.
236. M. Bojowald, G.M. Hossain, M. Kagan, S. Shankaranarayanan, Anomaly freedom in
perturbative loop quantum gravity. Phys. Rev. D 78, 063547 (2008). [arXiv:0806.3929]
237. A. Barrau, M. Bojowald, G. Calcagni, J. Grain, M. Kagan, Anomaly-free cosmological per-
turbations in effective canonical quantum gravity. JCAP 1505, 051 (2015). [arXiv:1404.1018]
238. M. Fernández-Méndez, G.A. Mena Marugán, J. Olmedo, Hybrid quantization of an inflation-
ary universe. Phys. Rev. D 86, 024003 (2012). [arXiv:1205.1917]
239. M. Fernández-Méndez, G.A. Mena Marugán, J. Olmedo, Hybrid quantization of an inflation-
ary model. The flat case. Phys. Rev. D 88, 044013 (2013). [arXiv:1307.5222]
240. L. Castelló Gomar, M. Fernández-Méndez, G.A. Mena Marugán, J. Olmedo, Cosmological
perturbations in hybrid loop quantum cosmology: Mukhanov–Sasaki variables. Phys. Rev. D
90, 064015 (2014). [arXiv:1407.0998]
241. L. Castelló Gomar, M. Martín-Benito, G.A. Mena Marugán, Gauge-invariant perturbations in
hybrid quantum cosmology. JCAP 1506, 045 (2015). [arXiv:1503.03907]
242. D. Martín de Blas, J. Olmedo, Primordial power spectra for scalar perturbations in loop
quantum cosmology. JCAP 1606, 029 (2016). [arXiv:1601.01716]
243. L. Castelló Gomar, M. Martín-Benito, G.A. Mena Marugán, Quantum corrections to the
Mukhanov–Sasaki equations. Phys. Rev. D 93, 104025 (2016). [arXiv:1603.08448]
244. L.J. Garay, M. Martín-Benito, G.A. Mena Marugán, Inhomogeneous loop quantum cos-
mology: hybrid quantization of the Gowdy model. Phys. Rev. D 82, 044048 (2010).
[arXiv:1005.5654]
245. D. Brizuela, G.A. Mena Marugán, T. Pawłowski, Big bounce and inhomogeneities. Class.
246. M. Martín-Benito, D. Martín de Blas, G.A. Mena Marugán, Matter in inhomogeneous

loop quantum cosmology: the Gowdy T 3 model. Phys. Rev. D 83, 084050 (2011).
[arXiv:1012.2324]
247. M. Martín-Benito, D. Martín de Blas, G.A. Mena Marugán, Approximation methods in loop
quantum cosmology: from Gowdy cosmologies to inhomogeneous models in Friedmann–
Robertson–Walker geometries. Class. Quantum Grav. 31, 075022 (2014). [arXiv:1307.1420]
248. I. Agulló, A. Ashtekar, W. Nelson, Quantum gravity extension of the inflationary scenario.
249. I. Agulló, A. Ashtekar, W. Nelson, Extension of the quantum theory of cosmological
perturbations to the Planck era. Phys. Rev. D 87, 043507 (2013). [arXiv:1211.1354]
250. I. Agulló, A. Ashtekar, W. Nelson, The pre-inflationary dynamics of loop quantum cos-
mology: confronting quantum gravity with observations. Class. Quantum Grav. 30, 085014
(2013). [arXiv:1302.0254]
251. I. Agulló, N.A. Morris, Detailed analysis of the predictions of loop quantum cosmology for
the primordial power spectra. Phys. Rev. D 92, 124040 (2015). [arXiv:1509.05693]
252. D. Langlois, Hamiltonian formalism and gauge invariance for linear perturbations in inflation.
253. E. Wilson-Ewing, Lattice loop quantum cosmology: scalar perturbations. Class. Quantum
Grav. 29, 215013 (2012). [arXiv:1205.3370]
254. E. Wilson-Ewing, Separate universes in loop quantum cosmology: framework and applica-
tions. Int. J. Mod. Phys. D 25, 1642002 (2016). [arXiv:1512.05743]
255. T. Cailleteau, J. Mielczarek, A. Barrau, J. Grain, Anomaly-free scalar perturbations with
holonomy corrections in loop quantum cosmology. Class. Quantum Grav. 29, 095010 (2012).
[arXiv:1111.3535]
256. M. Bojowald, A. Skirzewski, Effective equations of motion for quantum systems. Rev. Math.
Phys. 18, 713 (2006). [arXiv:math-ph/0511043]
257. M. Bojowald, A. Skirzewski, Quantum gravity and higher curvature actions. Int. J. Geom.
Methods Mod. Phys. 4, 25 (2007). [arXiv:hep-th/0606232]
258. M. Bojowald, S. Brahma, E. Nelson, Higher time derivatives in effective equations of
canonical quantum systems. Phys. Rev. D 86, 105004 (2012). [arXiv:1208.1242]
259. M. Bojowald, B. Sandhöfer, A. Skirzewski, A. Tsobanjan, Effective constraints for quantum
systems. Rev. Math. Phys. 21, 111 (2009). [arXiv:0804.3365]
260. M. Bojowald, A. Tsobanjan, Effective constraints for relativistic quantum systems. Phys. Rev.
D 80, 125008 (2009). [arXiv:0906.1772]
261. S. Tsujikawa, P. Singh, R. Maartens, Loop quantum gravity effects on inflation and the CMB.
Class. Quantum Grav. 21, 5767 (2004). [arXiv:astro-ph/0311015]
262. G.M. Hossain, Primordial density perturbation in effective loop quantum cosmology. Class.
263. G. Calcagni, M. Cortês, Inflationary scalar spectrum in loop quantum cosmology. Class.
264. E.J. Copeland, D.J. Mulryne, N.J. Nunes, M. Shaeri, Super-inflation in loop quantum
265. M. Artymowski, Z. Lalak, Ł. Szulc, Loop quantum cosmology: holonomy corrections to
inflationary models. JCAP 0901, 004 (2009). [arXiv:0807.0160]
266. M. Shimano, T. Harada, Observational constraints on a power spectrum from super-inflation
in loop quantum cosmology. Phys. Rev. D 80, 063538 (2009). [arXiv:0909.0334]
267. T. Cailleteau, A. Barrau, Gauge invariance in loop quantum cosmology: Hamilton–Jacobi
and Mukhanov-Sasaki equations for scalar perturbations. Phys. Rev. D 85, 123534 (2012).
[arXiv:1111.7192]
268. T. Cailleteau, A. Barrau, J. Grain, F. Vidotto, Consistency of holonomy-corrected scalar,
vector and tensor perturbations in loop quantum cosmology. Phys. Rev. D 86, 087301 (2012).
[arXiv:1206.6736]
269. J. Mielczarek, T. Cailleteau, A. Barrau, J. Grain, Anomaly-free vector perturbations with
[arXiv:1106.3744]
References 541
270. M. Bojowald, G.M. Hossain, M. Kagan, S. Shankaranarayanan, Gauge invariant cosmological

perturbation equations with corrections from loop quantum gravity. Phys. Rev. D 79, 043505
(2009); Erratum-ibid. D 82, 109903(E) (2010). [arXiv:0811.1572]
271. T. Cailleteau, L. Linsefors, A. Barrau, Anomaly-free perturbations with inverse-volume and
[arXiv:1307.5238]
272. A. Barrau, J. Grain, Holonomy corrections to the cosmological primordial tensor power
spectrum. arXiv:0805.0356
273. J. Mielczarek, Gravitational waves from the big bounce. JCAP 0811, 011 (2008).
[arXiv:0807.0712]
274. J. Grain, A. Barrau, Cosmological footprints of loop quantum gravity. Phys. Rev. Lett. 102,
081301 (2009). [arXiv:0902.0145]
275. J. Mielczarek, Tensor power spectrum with holonomy corrections in loop quantum cosmol-
ogy. Phys. Rev. D 79, 123520 (2009). [arXiv:0902.2490]
276. J. Mielczarek, T. Cailleteau, J. Grain, A. Barrau, Inflation in loop quantum cosmol-
ogy: dynamics and spectrum of gravitational waves. Phys. Rev. D 81, 104049 (2010).
[arXiv:1003.4660]
277. J. Grain, A. Barrau, T. Cailleteau, J. Mielczarek, Observing the big bounce with tensor
modes in the cosmic microwave background: phenomenology and fundamental loop quantum
cosmology parameters. Phys. Rev. D 82, 123520 (2010). [arXiv:1011.1811]
278. L. Linsefors, T. Cailleteau, A. Barrau, J. Grain, Primordial tensor power spectrum in
holonomy corrected ˝ loop quantum cosmology. Phys. Rev. D 87, 107503 (2013).
[arXiv:1212.2852]
279. B. Bolliet, A. Barrau, J. Grain, S. Schander, Observational exclusion of a consistent quantum
cosmology scenario. Phys. Rev. D 93, 124011 (2016). [arXiv:1510.08766]
280. E.J. Copeland, D.J. Mulryne, N.J. Nunes, M. Shaeri, The gravitational wave background
from super-inflation in loop quantum cosmology. Phys. Rev. D 79, 023508 (2009).
[arXiv:0810.0104]
281. G. Calcagni, G.M. Hossain, Loop quantum cosmology and tensor perturbations in the early
universe. Adv. Sci. Lett. 2, 184 (2009). [arXiv:0810.4330]
282. M. Bojowald, G. Calcagni, Inflationary observables in loop quantum cosmology. JCAP 1103,
032 (2011). [arXiv:1011.2779]
283. M. Bojowald, G. Calcagni, S. Tsujikawa, Observational constraints on loop quantum
cosmology. Phys. Rev. Lett. 107, 211302 (2011). [arXiv:1101.5391]
284. J. Grain, T. Cailleteau, A. Barrau, A. Gorecki, Fully loop-quantum-cosmology-corrected
propagation of gravitational waves during slow-roll inflation. Phys. Rev. D 81, 024040 (2010).
[arXiv:0910.2892]
285. J. Mielczarek, Inflationary power spectra with quantum holonomy corrections. JCAP 1403,
048 (2014). [arXiv:1311.1344]
286. T. Zhu, A. Wang, K. Kirsten, G. Cleaver, Q. Sheng, Q. Wu, Inflationary spectra with inverse-
volume corrections in loop quantum cosmology and their observational constraints from
Planck 2015 data. JCAP 1603, 046 (2016). [arXiv:1510.03855]
0410, 006 (2004). [arXiv:astro-ph/0407059]
288. L.-F. Li, R.-G. Cai, Z.-K. Guo, B. Hu, Non-Gaussian features from the inverse volume
corrections in loop quantum cosmology. Phys. Rev. D 86, 044020 (2012). [arXiv:1112.2785]
289. B. Bolliet, J. Grain, C. Stahl, L. Linsefors, A. Barrau, Comparison of primordial tensor
power spectra from the deformed algebra and dressed metric approaches in loop quantum
290. M. Bojowald, G.M. Paily, Deformed general relativity and effective actions from loop
quantum gravity. Phys. Rev. D 86, 104018 (2012). [arXiv:1112.1899]
291. M. Bojowald, G.M. Paily, A no-singularity scenario in loop quantum gravity. Class. Quantum
Grav. 29, 242002 (2012). [arXiv:1206.5765]
292. S.A. Hojman, K. Kuchař, C. Teitelboim, Geometrodynamics regained. Ann. Phys. (N.Y.) 96,
88 (1976)
293. M. Bojowald, J.D. Reyes, R. Tibrewala, Nonmarginal Lemaitre–Tolman–Bondi-like models
with inverse triad corrections from loop quantum gravity. Phys. Rev. D 80, 084002 (2009).
[arXiv:0906.4767]
294. J.D. Reyes, Spherically Symmetric Loop Quantum Gravity: Connections to 2-Dimensional
Models and Applications to Gravitational Collapse. Ph.D. thesis, Pennsylvania State Univer-
sity, University Park (2009)
295. M. Bojowald, G.M. Paily, J.D. Reyes, Discreteness corrections and higher spatial derivatives
in effective canonical quantum gravity. Phys. Rev. D 90, 025025 (2014). [arXiv:1402.5130]
296. M. Bojowald, S. Brahma, Signature change in 2-dimensional black-hole models of loop
quantum gravity. arXiv:1610.08850
297. A. Perez, D. Pranzetti, On the regularization of the constraints algebra of quantum gravity
in 2 C 1 dimensions with non-vanishing cosmological constant. Class. Quantum Grav. 27,
145009 (2010). [arXiv:1001.3292]
298. A. Henderson, A. Laddha, C. Tomlin, Constraint algebra in loop quantum gravity reloaded.
I. Toy model of a U.1/3 gauge theory. Phys. Rev. D 88, 044028 (2013). [arXiv:1204.0211]
300. J. Brunnemann, T. Thiemann, On (cosmological) singularity avoidance in loop quantum
gravity. Class. Quantum Grav. 23, 1395 (2006). [arXiv:gr-qc/0505032]
301. J. Brunnemann, T. Thiemann, Unboundedness of triad-like operators in loop quantum gravity.
302. M. Bojowald, S. Brahma, U. Büyükçam, F. D’Ambrosio, Hypersurface-deformation alge-
broids and effective space-time models. arXiv:1610.08355
303. M. Bojowald, S. Brahma, Signature change in loop quantum gravity: general midisuperspace
models and dilaton gravity. arXiv:1610.08840
Chapter 11
Cosmology of Quantum Gravities
Magna et spatiosa res est sapientia; vacuo illi loco opus est; de
divinis humanisque discendum est, de praeteritis de futuris, de
caducis de aeternis, de tempore. De quo uno vide quam multa
quaerantur: primum an per se sit aliquid; deinde an aliquid
ante tempus sit sine tempore; cum mundo coeperit an etiam ante
mundum quia fuerit aliquid, fuerit et tempus. [: : :]
Quamcumque partem rerum humanarum divinarumque
conprenderis, ingenti copia quaerendorum ac discendorum
fatigaberis. Haec tam multa, tam magna ut habere possint
liberum hospitium, supervacua ex animo tollenda sunt. Non
dabit se in has angustias virtus; laxum spatium res magna
desiderat. Expellantur omnia, totum pectus illi vacet.
— Seneca, Ad Lucilium Epistularum Moralium, XI, 88, 33–35
Wisdom is a great and spacious thing; it needs plenty of free

space. It teaches us about the divine and the human, the past
and the future, the transient and the eternal, and about time. See
how many issues arise just about the latter: First, whether time
is anything by itself; then, if anything existed prior to time and
without time; if time began with the world or, since something
must have existed before the world, if also time existed before
the world. [: : :] Whatever part you embrace of human and
divine sciences, you will have to make a great effort in studying
a vast number of things. These are so many and so important
that, in order for them to have free shelter in your soul, you will
have to remove all superfluous things. Virtue does not confine
itself to narrow quarters; great things wish large space. Let us
expel everything else from our breast and make it empty to make
room for virtue.

DOI 10.1007/978-3-319-41127-9_11
544 11 Cosmology of Quantum Gravities
Contents
11.1 Hausdorff and spectral dimension. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 545
11.2 Asymptotic Safety. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 547
11.2.1 Framework. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 548
11.2.2 Cosmology. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 551
11.3 Causal Dynamical Triangulations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 555
11.3.1 Framework. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 555
11.3.2 Cosmology. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 561
11.4 Spin Foams. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 563
11.4.1 Framework. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 563
11.4.2 Cosmology. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 564
11.5 Group Field Theory. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 569
11.5.1 Framework. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 569
11.5.2 Cosmology. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 572
11.6 Causal Sets. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 584
11.6.1 Framework. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 584
11.6.2 Cosmology. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 588
11.7 Non-commutative Spacetimes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 591
11.7.1 Framework. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 591
11.7.2 Cosmology. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 593
11.8 Non-local Gravity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 596
11.8.1 Non-locality. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 596
11.8.2 Framework. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 599
11.8.3 Cosmology. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 603
11.9 Comparison of Quantum-Gravity Models. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 604
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 607
A theory of everything explaining all the cosmological puzzles is an attractive

utopia. It should explain how an expanding universe can arise from the chaotic
foam of quantum geometry; whether the big-bang singularity of classical general
relativity is replaced by some configuration without infinities; what is the fate of
other gravitational singularities such as those met in black holes; what caused the
early inflationary expansion, and what the late one; how the degrees of freedom of
the Standard Model of particles emerged; what is the cosmological constant and
why it acquired such a small value as observed today; and so on.
In the lack of a concrete realization of a theory of everything, one should be able
to ask the same questions in non-unified quantum-gravity scenarios while keeping
exploring all the possibilities with an open mind. To this end, it is useful to compare
the approaches to quantum gravity listed in Sect. 8.3 and see what they can say
about cosmology and its problems. We will do so in this chapter, summarizing here
the salient points of each framework:
• Asymptotic safety [1–45] (Sect. 11.2). Reviews are [46–51]. Various types of
f .R/ actions are naturally produced. The big-bang problem is not solved. The
cosmological constant problem is reformulated but not quite solved in the
functional renormalization approach, while in the resummed-quantum-gravity
approach to asymptotic safety an actual prediction for is given and is in
11.1 Hausdorff and spectral dimension 545
agreement with the observed value, although the observed equation of state is
not reproduced [52].
• Causal dynamical triangulations [53–75] (Sect. 11.3). Reviews are [76–79]. A
de Sitter universe naturally emerges as the semi-classical limit of full non-
perturbative, background-independent quantum gravity. Neither the nor the
big-bang problem are addressed.
• Spin foams [80–95] (reviews are [96–99]) and their symmetry-reduced cos-
mological version [100–110] (Sect. 11.4). The big bang is removed in an
LQC-related mini-superspace context. The problem is not solved.
• Group field theory [111–141] (Sect. 11.5). Reviews are [142–148]. As in causal
dynamical triangulation, a cosmological limit is obtained from quantum grav-
ity under certain approximations. Semi-classical cosmological equations of
motions arise from the full, background-independent quantum theory via a
non-perturbative condensation mechanism. Interestingly, the homogeneous and
isotropic dynamics can match with that of LQC in the improved quantization
scheme of the canonical theory.
• Causal sets [149–180] (Sect. 11.6). Reviews are [181–186]. A prediction for the
cosmological constant is found which is in agreement with the observed value
(2.118); however, its robustness against inhomogeneities is not clear and there
might be some residual fine tuning. The big-bang problem seems to be removed.
There arises the possibility to have an early de Sitter stage of purely geometric
origin. So far, none of these features have been embedded in a fully realistic
cosmological history, which is difficult to extract since the main building blocks
of the theory are pre-geometric and the notion of a spacetime continuum is not
fundamental but, rather, emergent from a discrete ordered structure.
• In Sects. 11.7, 11.8 and 11.9, we will overview other approaches: non-
commutative spacetimes, non-local gravity and models with dimensional flow.
11.1 Hausdorff and spectral dimension
In preparation for a comparison among these theories (Sect. 11.9), we introduce two
notions of spacetime dimension. We recall that the Hausdorff dimension dH of a set
is the scaling of volumes defined thereon. One can proceed by covering the set with
a minimum number of balls of different size and then send their radii to zero. For
a smooth manifold of topological dimension D, this reduces to a local, operational
definition of the Hausdorff dimension dH as the scaling law for the volume V .D/ .R/
of a D-ball of radius R:
d ln V .D/ .R/
dH :D : (11.1)
d ln R
For Euclidean space, V .D/ / RD and the Hausdorff and topological dimension
coincide, dH D D. For a Riemannian manifold, the same result holds only in
the limit R ! 0, lest curvature or topological effects become important. For
a Lorentzian manifold, one computes the Hausdorff dimension either of spatial
hypersurfaces or of the whole spacetime with imaginary time; the result, in the
limit of vanishing radius, is still dH D D. For a generic spacetime coming from
some effective limit of a quantum-gravity model, the Hausdorff dimension can vary
depending on the scale (in which case one must not take the limit R ! 0 in (11.1))
and takes non-integer values.
A different geometric indicator is the spectral dimension dS . Its most common
interpretation is based upon the diffusion equation method. One places a classical
test particle on the geometry one wants to probe, at some initial point x0 (suitably
defined also in discrete geometries) and let it diffuse pushed around by geometry
itself. Then one asks what the probability P is to find the particle at point x after
some diffusion time T. However, in covariant settings it is difficult to interpret T
as an actual time variable. The time coordinate t is either omitted (when diffusing
on spatial hypersurfaces) or put on equal footing as spatial coordinates (when
considering the Euclideanized version of the geometry under inspection). Therefore,
it is advisable to modify the interpretation of the diffusing process and regard the
latter as a probing of the geometry with a resolution 1=T ! 1=`. The length scale `
represents the minimal detectable separation between points. By definition, the heat
kernel P.x; x0 I `/ is the solution of the running (or diffusion) equation

@ 0
K x P.x; x I `/ D 0 ; P.x; x0 I 0/ D f .x/ ı.x x0 / ; (11.2)
@`2
where Kx is the kinetic operator characteristic of the geometry and acting on the
x dependence of P. For a smooth manifold, Kx is the Euclidean version of the
covariant Laplace–Beltrami operator , i.e., the curved Laplacian r 2 . Effective
continuous quantum geometries can induce strong deviations from (11.2) via
changes in the initial condition, in the diffusion operator @=@`2 , in Kx (which can be
a higher-order or even a non-local derivative operator on a continuum, or a finite-
difference operator if the setting is discrete) and by the presence of source terms
p
[41, 187, 188]. The initial condition at ` D 0 (where the weight factor is f D 1= jgj
for a Riemannian geometry) expresses the fact that, at infinite resolution, the probe
0
is a point-wise particle exactly
R D localized at x .
The trace P.`/ :D d x P.x; xI `/ is called return probability. The spectral
dimension of spacetime is then defined as the scaling of P with respect to `,1
d ln P.`/
dS .`/ :D : (11.3)
d ln `
1
In order to make sense of dS as an indicator of the physical geometry, one should ensure that the
diffusion process or resolution-dependent probing be well defined. If the solution P of (11.2) was
regarded as a probability density function, as in transport theory, then it should be positive semi-
definite and normalized to 1. However, in many approaches to quantum gravity P becomes negative
for certain values of ` and x, in which case the interpretation of (11.2) becomes problematic,
even if the return probability P is positive-definite. In fact, if there were no operationally sensible
procedure by which the test particle could be found “somewhere” with a certain probability, then
no clear geometric and physical meaning could be attached to (11.3). The negative probabilities
problem can be fixed either by modifying (11.2) [41] or by adopting a quantum-field-theory
viewpoint [189], where P is an amplitude rather than a probability.
11.2 Asymptotic Safety 547
As for the Hausdorff dimension, to ensure that large-scale curvature and topo-
logical effects do not vitiate the result,2 one should limit the attention to a
local definition of dS , in the limit ` ! 0. However, in any situation where
we expect a running of the spectral dimension one should not take this limit
and, instead, ignore both curvature and topological effects whenever possible. For
instance, on a Riemannian manifold one ignores the curvature and just considers
D-dimensional Euclidean space. The solution of (11.2) is the Gaussian P.x; x0 I `/ D
.4`2 /D=2 expŒjx x0 j2 =.4`2 /, the return probability is a simple power law
P.`/ / `D and the spectral dimension is dS D D D dH .
11.2 Asymptotic Safety
Although graviton perturbation theory suffers from divergences, the gravitational

interaction can be quantized as a field theory non-perturbatively via the functional
renormalization approach. This is the aim of asymptotically safe gravity. Let i be
the essential couplings of a field theory, i.e., those combinations of the parameters
in the action which are invariant under field redefinitions. Asymptotic safety is a
condition imposed on i such that they approach a fixed point as the momentum-
energy scale k above which all quantum fluctuations are integrated out goes to
infinity [1].3 As in a perturbatively renormalizable theory, there is a finite number
of essential couplings and divergences in the UV are supposedly removed.
If the i ’s have energy dimension di , then N i .k/ D kdi i .k/ are dimensionless
and remain so under an infinitesimal change in the energy scale, so that their
variation can depend only on a combination ˇi of the other couplings:
dN i
D ˇi fN j g : (11.4)
d ln k
For given initial conditions, the solution of the Gell-Mann–Low equation (11.4)
describes the trajectory N i .k/ in the space of couplings. Different initial conditions
correspond to different theories. An ultraviolet non-Gaussian fixed point (NGFP) in
the trajectory is defined by
lim N i .k/ D N i ¤ 0 ; ˇi .fN j g/ D 0 ; (11.5)

k!1
2
For instance, in the diffusion interpretation on a sphere the particle can come back to x0 more easily
than on a plane and, if we wait too long (T ! C1), the return probability tends to a constant.
In the resolution interpretation, waiting too long means taking too low a resolution (1=` ! 0), so
that the sphere cannot be distinguished from a point. In both cases, dS ! 0 instead of dS ! D.
3
In this section, we reserve the symbol k for this cut-off.
where N i are finite constants not necessarily small (hence, perturbation theory may
not apply). In general, any theory possesses the Gaussian fixed point N i D 0,
corresponding to the free limit; the existence of a non-Gaussian fixed point is much
less trivial. The main assumption of asymptotic safety is that there exist initial
conditions such that the fixed point (11.5) lies in the trajectory solving (11.4). If
it does not, then in general N i ! 1 at high energies and the theory may develop a
singularity in physical observables.
11.2.1 Framework
For a given k, effective asymptotic-safety spacetime and dynamics are described

by an effective action k Œg . The metric is subject to quantum fluctuations and,
depending on the resolution 1=`.k/ of the microscope probing the geometry, one
observes different properties of spacetime. The change of the microscope amounts
to a coarse-graining procedure. Denote as hg ik the metric averaged over Wick-
rotated spacetime volumes of linear size `.k/. Then, hg ik is solution of
ık
hg ik D 0 : (11.6)
ıg
Letting k take any non-negative value, one ends up with a continuous family of
actions and of field equations (11.6), all valid simultaneously. From the scale
dependence of k , one can extract the solution hg ik at any scale from the UV to the
IR. Although hg ik is typically a smooth classical metric, its change with the scale,
regulated by the renormalization group (RG) flow, makes effective asymptotic-
safety spacetime a non-smooth multi-scale (in particular, fractal [34]) object with
possibly very “irregular” geometry.
In the absence of matter and in the so-called Einstein–Hilbert truncation, the
effective action is the usual one but with scale-dependent couplings:
Z
1 p
k D dD x g .R 2k / : (11.7)
16Gk
A non-trivial non-Gaussian fixed point .G N ; N / exists in the plane of the dimen-

sionless couplings G N k D kD2 Gk and N k D k2 k [2, 16]. In the IR perturbative
regime near the Gaussian fixed point, Gk D G0 C O.k2 / and k D 0 C
G0 k4 C O.k6 /, where D O.1/ is a constant. In terms of the combined quantities
N T :D .40 G0 /1=2 , kT :D Œ0 =.G0 /1=4 and G N T :D T =.2/, in four dimensions
2 " 2 2 #
Nk D G
NT k 1 kT k
G 4
C O.k /; N k D N T C C O.k4 / :
kT 2 k kT
(11.8)
Fig. 11.1 Renormalization-group flow on the plane of the dimensionless couplings g D G N k and
D N k described in the text, for asymptotically safe quantum gravity with Einstein–Hilbert
effective action. Other classes of trajectories are possible but this is the only one realized by Nature
(positive Newton’s and cosmological constant in the IR). Classical general relativity corresponds
to the segment between P1 and P2 . The question mark indicates a singularity point where the
Einstein–Hilbert approximation breaks down. The separatrix from other trajectories is also shown
(Source: [18])
Here kT represents the scale at which the turning point T in Fig. 11.1 is passed.
Overall, the gravitational coupling decreases from the UV to the IR along the RG
flow, while the cosmological constant (if positive) increases. Both acquire very small
N /
values near the Gaussian fixed point .G; N D .0; 0/ at the origin of the plot [18].
The scale dependence of the average metric is related to the one of the running
cosmological constant k . In fact, the field equations read R Œhgik D Œ2=.2
D/k hg ik . Let k0 be an arbitrary reference scale (typically in the infrared,
k0 =k 1) and assume that the cosmological constant scales according to a
function F,
k D F.k2 /k0 ; (11.9)
where F depends on k2 by the requirement of Lorentz invariance. We also

assume that F stays positive throughout the flow between k and k0 . Then, from

F 1 R Œhgik D Œ2=.2 D/k0 ı and the scaling property R Œc g D c1 R Œg
of the Riemann tensor (c > 0 is a constant), one gets
1
hg ik D hg ik0 ; hg ik D F.k2 /hg ik0 : (11.10)
F.k2 /
The function F acquires different asymptotic forms depending on the regime. In the
far IR, F ' 1 by definition (k ! k0 ! C1). In an intermediate semi-classical
regime near the Gaussian fixed point, F ' k4 . In the deep UV, at the NGFP one has
k ' k2 N and asymptotic safety requires that F ' k2 [21]. All in all, we can write
F ' jkjı (11.11)

in asymptotic regimes, with ı D 0; 4; 2 in the IR, semi-classical limit and UV,

respectively. The semi-classical limit may change depending on the truncation and
on the presence of matter [37], but the main scaling argument is unaltered.
In classical geometries, k can be qualitatively interpreted as encoding dynamical
variables which are coarse-grained on regions of spacetime of size L [190]. In this
context, in the simplest classical non-compact case ` k1 and the resolution is
directly identified with k. This is the sense in which k acts as a “microscope”
with resolving power `.k/. In quantum gravity not only is the functional form
`.k/ more complicated a priori, but the relation between resolving power, proper
distances and IR cut-off k is much subtler. In a general quantum manifold with
fractal properties, the k-dependence of lengths is an example of the typical effect
of momentum dependence of measurements on multi-fractals [191]. There appear
situations where ` can decrease arbitrarily even when there exists a non-vanishing
minimum resolution: the limit k ! C1 is no longer sufficient to probe arbitrarily
small proper distances because the effective quantum geometry “shrinks” faster in
the UV limit. This is a direct consequence of the scaling (11.10) of the average
metric and, more generally, of the running of the couplings in the effective action.
Lengths can also be measured by a macroscopic observer with average metric
hg ik0 , who does see a finite minimal length corresponding to the minimal
resolution [22, 24].
The cut-off scale k is often identified with the physical momentum p, although
sometimes other identifications are more convenient (see below and [41]). With the
identification k D p and ignoring intermediate regimes, one can interpolate the IR
and UV regions via the graviton effective dispersion relation p2 C p4 = 2 D 0 and
the effective Newton’s constant Geff . p/ D G0 =.1 C p2 = 2 /. For small momenta,
Geff . p/ D G0 C O. p2 /, while for large momenta (i.e., near the NGFP) Geff . p/ '
.G0 2 /p2 . The numerical coefficient is such that G0 2 D G N . Remarkably, this is
precisely the behaviour of (8.16) and (8.18) found in Sect. 8.2.3. A direct calculation
yields G N 0:0442 [192].
In Sect. 7.1.2, we have seen that the zero-point energy density of each degree of
freedom of the i-th matter field yields the contribution (7.12) to the cosmological
constant, which stems from the four-dimensional integral
Z
.1/
C1
d4 p p20
vac;i / : (11.12)
1 .2/4 p2 C m2i
In the resummed quantum-gravity framework, the propagator is replaced by its

dressed version (8.15). Summing over all matter particles with Ni degrees of
freedom (negative for fermions), up to an overall constant now one has [52]
X Z
m4Pl X
00
.1/
C1
d4 p p20 eBg . p/ Ni
vac / Ni 2
' : (11.13)
i 1
4
.2/ p C mi
2 64 i Œln.m2Pl =m2i /2
To encode the running of Newton’s constant in (11.13), one should replace the
Planck mass with the one coming from (8.18),

p2
m2Pl . p/ D 1 C 2 m2Pl ; (11.14)
where 1= 2 22:635=m2Pl. This replacement does not have an important effect on

the logarithm in (11.13) and can be done just in the numerator m4Pl :
m4Pl . p/ X Ni
. p/ ' : (11.15)
64 i
Œln.m2Pl =m2i /2
P
For the particle content of the Standard Model, the sum value is i Ni =
Œln.m2Pl =m2i /2 9:194 103 .
Thus, . p/ D 8 . p/=m2Pl . p/ ' 0 C N p2 , where N 0:0817 [52]. This
estimate changes with the number and species of particles and, in turn, can provide
a constraint on the admissible supersymmetric extensions of the Standard Model.
Therefore, the existence of a non-Gaussian fixed point is confirmed indepen-
dently by the functional renormalization approach and by resummed quantum
gravity. It is more difficult to compare the actual numerical values of G N and N in
the two frameworks. In the case of resummed gravity, .G N ; N / .0:0442; 0:0817/
[52], but in the functional renormalization approach the values depend on the
assumptions on the truncation scheme and on the matter content and G N ; jN j D
O.0:1/ – O.1/ [5, 37, 39, 43, 45].
Despite the pending issue of the NGFP value, a posteriori we recognize
resummed quantum gravity as another explicit realization of asymptotic safety
because .G N ; N / ¤ .0; 0/ [25]. This formalism has also the advantage of being
gauge invariant, thus avoiding the issues of gauge dependence that inevitably arise
in the truncation scheme of the functional renormalization approach.
11.2.2 Cosmology
The scale dependence of the couplings in the average effective action k is inherited
by the Einstein equations of this theory. For a FLRW background, the cut-off k D
k.t/ can only depend on time and, dimensionally, the simplest identification is with
the Hubble horizon, k / H.t/ [26]. Then, the RG-improved cosmological dynamics
is encoded into two equation. One is the usual first Friedmann equation except for
the replacements ! .t/ and G ! G.t/, where the precise time dependence is
determined by integrating the RG equations numerically. In four dimensions,
8G.t/ K .t/ .t/

H2 D 2 C ; D : (11.16)
3 a 3 8G.t/
The Bianchi identity is not automatically solved when the energy-momentum tensor
is covariantly conserved:
P C 8G
P
P C 3H. C P/ D ; (11.17)
8G
where and P are the energy density and pressure of a perfect fluid. Energy
is exchanged between matter and geometry. If the energy-momentum tensor is
conserved, then (11.17) splits into P C 3H. C P/ D 0 and P C 8G P D 0,
giving the general solution
! 3.1Cw/
1
1 P GP
D ; a/ ; (11.18)
8 GP P
valid for a constant barotropic index w D P=. Such a dynamics with varying G
and varying is not just a phenomenological model with ad hoc functions G.t/ and
.t/ [193], since the latter are determined by the RG evolution.
In particular, one has a concrete quantum-gravity example of an alternative
acceleration mechanism without scalar fields, solving the horizon and the entropy
problem [6, 26, 51]. Near the non-Gaussian fixed point (11.20), the equations of
motion with flat background (K D 0) admit the power-law asymptotic solution
a / t˛ , / t4 , where
2
˛D (11.19)
3.1 C w/.1 ˝ /
and ˝ D

=crit is evaluated at the fixed point. Therefore, near the NGFP k /
1
H / t and
.t/ / t2 ; G.t/ / t2 (11.20)
in four dimensions. From (11.17), one can find that P = D 4=.3˛/ 1: the
running cosmological constant (11.20) obeys the usual equation of state P '
only for ˛ 1.
The UV asymptotic behaviour (11.20) is readily recovered in resummed quantum
gravity, by identifying p 1=t in the energy density (11.15),
( )
1 2 4 1 X Ni
.t/ ' 1 C 2 2 mPl ; (11.21)
t 64 i Œln.m2Pl =m2i /2
and taking the limit t / t=tPl 1.

The deep UV regime takes place at energy scales k / H mPl , corresponding
to times prior to t ˛tPl . Starting with radiation (w D 1=3), for 0 < t < t
the universe accelerates (˛ > 1) provided 1=2 < ˝ . 1. At times t > t ,
inflation stops and there follows a classical evolution with a / t1=2 . The primordial
power spectrum is generated by matter fluctuations, which are almost scale invariant
near the NGFP. In fact, at the UV fixed point the D D 4 graviton propagator
in momentum space scales as G. Q p2 / ' 1=p4 for p2 m2Pl , amounting to
0 0 2
G.x x / ln jx x j in position space [7]. In particular, the graviton two-
point correlation function on a cosmological spatial slice is (up to tensor indices)
hh.t; x/h.t; x0 /i ln jx x0 j2 , so that hıR.t; x/ıR.t; x0 /i jx x0 j4 for the
curvature fluctuation ıR @2 h. Back to momentum space, this corresponds to a
scale-invariant power spectrum [6].
Unfortunately, the amplitude of tensor modes produced during this stage is too
large, since it is governed by the dimensionless combination GH 2 / G D
O.103 / at the NGFP [36]. To fill the gap of seven orders of magnitude between
this value and the present constraints on the tensor amplitude (from (4.66) and
(4.69), At 0:1As 1010 ), one must abandon the Einstein–Hilbert truncation and
consider higher-order curvature terms, for instance of the Starobinsky type (7.88)
[42] or encoded in a generic functional f .R/ [32]. The Taylor coefficients of f are
determined at the NGFP for any given truncation order.
An f .R/ effective action with limit (7.88) can also arise from the Einstein–Hilbert
truncation when making the scale identification [36]
k2 / R : (11.22)
Near the NGFP, Gk ' G k2 / R1 , so that R=Gk R2 . In the semi-
classical perturbative regime near the Gaussian fixed point, the intermediate limit
f .R/ ' R C bR2 is obtained. This may provide a theoretical justification to
Starobinsky inflation. A more precise determination of the effective action, still by
the identification (11.22), comes from the analytic expressions of Gk and k solving
the linearized flow equations around the NGFP [35]:
ˇ
R R
L D R2 C bR2 cos ! ln ; (11.23)
R0 R0
where b > 0 and ˇ < 0. The logarithmic oscillations arise from the interplay of the
pair of complex-conjugate critical exponents 1 D 2 D ˇ C 2ib governing the
spiral approach to the fixed point (Fig. 11.1). The simplest cosmological solution
with acceleration is de Sitter.
In general, inflation can be sustained by higher-order curvature terms one can
add to the Einstein–Hilbert truncation [31]. It is also possible to introduce matter in
the form of a real scalar field, in which case one obtains an ordinary inflationary
scenario where the RG quantum corrections modify the constraints on viable
potentials [33, 38, 40].
The big-bang problem still persists, as there is a singularity at t D 0. In the
functional renormalization approach, the cosmological constant problems are not
solved either, since today’s value (2.118) corresponds to the IR semi-classical limit
of the running ,N which is tuned by extremely small values of G N T N T 1060 .

However, once the emergence of classical spacetime is achieved in a robust way, the
smallness of is a natural consequence of the fact that the RG trajectory realized in
Nature spends several orders of magnitude of RG time ln.k=kT / near the Gaussian
N /
fixed point .G; N D .0; 0/ [20]. Thus, the old problem is reformulated by asking
why the perturbative semi-classical regime of such a trajectory is fine tuned to this
degree [18]. Another interesting trajectory runs along the N D 0 line from a non-
Gaussian fixed point .G;N /N D .G N ¤ 0; 0/ directly to the Gaussian fixed point
[44]. In this case, the cosmological constant is zero at all scales and, as in any
“overshooting” approach to the old problem, the small value of the dark-energy
density must be explained with something more than the vacuum energy.
In the resummed-propagator approach to asymptotic safety, it is possible to make
a more precise prediction for the value of the effective cosmological constant today.
This estimate relies on a transition between the deep UV regime where (11.21)
holds and the classical regime where the standard cosmological dynamics applies.
The transition is marked by the critical time t D ˛tPl . Numerical integration of the
RG-improved cosmological equations of the functional renormalization approach
points towards a value ˛ 25 [26]. The same value is also obtained in resummed
quantum gravity with the help of Heisenberg uncertainty principle [194]. In the
classical regime, the vacuum equation of state is evaluated at energies smaller than
the Planck mass. One takes the view that the energy density and pressure of the
cosmological constant and matter fields are not separately conserved. Instead, one
should consider the full continuity equation
P C P C 3HŒ. C / C .P C P / D 0 : (11.24)
To maintain the usual equation of state w D 1 that is observed by experiments,

C P D 0 and .t/ is determined dynamically by Einstein’s equations. If the
cosmological constant is small enough during cosmic evolution, then with good
approximation P C 3H. C P/ 0 for matter and radiation separately. At times
t < t < teq , the universe is dominated by radiation and, by virtue of the standard
Friedmann equation, a / t1=2 (equation (2.78)). For t > teq , dust matter dominates
and a / t2=3 . In both eras, D C 3H 2 = 2 / t2 , so that integrating (11.24)
between t and t0 (today), we have
2
t
.t0 / D .t / : (11.25)
t0
The age of the universe is given by (2.14), t0 13:8 Gyr, so that t =t0
3:097 1060 . The key information provided by resummed quantum gravity is
the value of .t /, which is determined by (11.21) at the transition point t :
.t / 1:54 104 m4Pl . Combining all this information into (11.25), one finally
11.3 Causal Dynamical Triangulations 555
obtains the attractive result
.t0 / 1:5 10123 m4Pl .2:4 103 eV/4 ; (11.26)
quite close to the observed value (2.118). A posteriori, one can check that the contri-
bution of is small enough from the time of the big-bang nucleosynthesis onwards
up to the recent acceleration phase [52]. Therefore, the above approximation on the
equations of state is consistent. Since w D 1, (11.26) solves the old cosmological
constant problem without disrupting the evolution of the universe in an unwanted
way.
11.3 Causal Dynamical Triangulations
In analogy with quantum mechanics and quantum field theory, quantization of

geometry can be defined as the sum over histories
Z
ZD ŒDg eiSŒg ; (11.27)
where we set „ D 1, ŒDg is the functional integration measure (performed with

certain boundary conditions) of the space of equivalence classes of Lorentzian D-
dimensional metrics g under diffeomorphisms and S is the gravitational action.
We saw examples of path integrals for gravity in (7.103) and (9.91). For simplicity,
we have factored out and ignored the contribution of matter fields. The path integral
Z depends on the coupling constants in the Lorentzian action S. If S is chosen to be
the Einstein–Hilbert action with cosmological term, the space of couplings is two-
dimensional and spanned by all possible values of G and . The presence of all
allowed geometries in the path creates quantum interference in the evolution of the
metric g .x/ via the dynamics of S.
We had already occasion to mention that the path-integral approach is technically
difficult due to the divergences hidden in (11.27). To overcome this problem, one
can discretize the geometries and thus regularize the functional integration, in a way
preserving the local causal structure of inertial frames. Such is the goal of causal
dynamical triangulations (CDT).
11.3.1 Framework
The most obvious cause of concern regarding (11.27) is the oscillatory integrand
eiS . Going to imaginary time t ! it produces a formally convergent factor eSE ,
where SE is the Euclideanized classical action; the Euclideanized path integral is
then called partition function. The Euclidean analytic continuation of gravity can
Fig. 11.2 Triangulation of a Riemannian disk with positive curvature (Reprinted figure with
permission from [70]. ©2013 by the American Physical Society)
give a wide overview of possible quantum phenomena, as seen in Chaps. 7 and 9,

but its viability as a physical approach has never been established conclusively. One
can appreciate this in Euclidean dynamical triangulations (EDT) [195–203], which
is also helpful to introduce some of the building blocks of CDT.
The triangulation of a given smooth manifold with metric g is a collection of
D-simplices, pieces of a D-dimensional flat space. A 2-simplex is a triangle, a 3-
simplex a tetrahedron. Denoting with ì the length of the edges of a D-simplex, the
continuum limit is defined by ì ! 0 for all edges of all simplices in a triangulation.
In the case of EDT, the continuum limit is a Riemannian manifold, the metric has
positive signature and the D-simplices are chunks of Euclidean space. Simplices are
usually assumed to be equilateral, ì D ` for all i. The meeting point of different
edges is called a vertex. Curvature in a triangulation is introduced by gluing D-
simplices together according to different angles. Take the D D 2 example of a
disk, which is triangulated by six equilateral triangles (Fig. 11.2). Removing one
triangle and gluing (i.e., identifying) the free edges, one realizes a disk with positive
curvature, characterized by a deficit angle (in this case, D =3). Generalizing
to D dimensions and many simplices, one gets triangulations whose geometry is
intrinsically expressed in terms of the edge length ` and deficit angles i , with no
reference to an external embedding space. The picture of a triangulation embedded
in a Riemannian manifold is therefore unnecessary and one can also construct
arbitrary triangulations not associated with a smooth manifold in the continuum
limit. By virtue of a set of manifold conditions, however, one can guarantee that the
triangulation looks like a D-dimensional space everywhere.
A triangulation is said to be dynamical when a classical action for it is considered.
The classical dynamics is described by Regge calculus [204–208], a discrete
approximation of general relativity. When the quantum theory is considered, the
path integral (11.27) over continuous geometries becomes a sum over triangulations.
The action S is discretized according to Regge calculus and the parameter space
can be studied numerically via Monte Carlo simulations. Operationally, different
triangulations with the same topology are realized by creating and destroying
vertices with a set of combinatorial moves [55]. It is found that D D 4 EDT possess
neither a large-scale nor a continuum limit reproducing general relativity.
Contrary to EDT, causal dynamical triangulations do not lose the information
on the causal structure of the original continuous system. At a first step, geometry
is Lorentzian and D-simplices are pieces of Minkowski spacetime. This requires

that the squared lengths `2i of their edges can also take null or negative values,
and that the concurrence of time- and space-like edges at each vertex respect the
local light-cone structure of inertial frames with the correct Lorentzian signature
.; C; ; C/. In the simplest formulation of CDT, there is also a global notion
of causality thanks to the introduction of a preferred foliation in each path-integral
history, thus ensuring the presence of a global, discrete proper time. This foliation
into D D 2 strips (or D D 3 slabs, and so on) is the piecewise discrete realization
of global hyperbolicity. There is good evidence, anyway, that the main features of
CDT (in particular, the de Sitter phase we shall describe below) do not depend on
the existence of a preferred foliation in the triangulation [69, 70].
Each strip (or slab, and so on) of fixed width t D 1 is constituted by an arbitrary
sequence of triangles (tetrahedra) pointing up or down. The requirements of local
and global causality lead, however, to restrictions on the type of simplices allowed.
These are classified according to the number (i,j) of vertices in one spatial slice and
the next (or the previous: in the following we omit to mention the time-reversed
simplices (j,i)). For CDT with foliation, in D D 1 C 1 dimensions triangles can
have only one space-like edge and are therefore all of type (1,2), while in D D
2 C 1 dimensions tetrahedra can have either two non-contiguous space-like edges
(type (2,2) tetrahedra) or one space-like face (type (3,1) tetrahedra), while no edge
can be light-like (Fig. 11.3). In D D 3 C 1, four-simplices are only of type (4,1)
(one space-like tetrahedral face, i.e., six out of ten edges are space-like) and (3,2)
(one space-like triangle and one edge, while the other six edges are time-like); see
Fig. 11.4. Without foliation, there are more classes of simplices allowed.
Therefore, one can parametrize equilateral CDT with one space-like squared
length `2s > 0 and a time-like one `2t D ˛`2s < 0, where ˛ is a dimensionless
constant called asymmetry parameter. In foliated CDT, ˛ > 0 corresponds to
Lorentzian signature and ˛ < 7=12 to the Euclidean one. In numerical codes,
the edge length is fixed to `s D 1.
For actual numerical calculations, one must Wick rotate the system, but this does
not lead to EDT because the causal structure is preserved through the length labels
of the edges. The discrete partition function of foliated CDT is thus
X 1 Regge
ZCDT D eSE .T/ ; (11.28)
T
Aut.T/
the sum of causal triangulations T (piecewise globally hyperbolic geometries)

whose spatial sections are further restricted to have fixed .D 1/-sphere SD1
compact topology. The factor 1=Aut.T/ is the analogue of the path-integral measure
Regge
ŒDg and Aut is the order of the automorphism group of T.4 The action SE is the
Wick-rotated Regge action, which depends on the couplings G and , on ˛ and on
4
An automorphism of a graph is an edge-preserving permutation of vertices. The set of automor-
phisms of a graph forms a group called automorphism group.
a b
(3,1) (3,1)
(2,2)
c
Fig. 11.3 (a) Two-dimensional EDT, consisting of equilateral triangles with Euclidean signature.
(b) Two-dimensional foliated CDT, consisting of equilateral triangles with one space-like (blue,
thick lines) and two time-like (red, thin lines) edges. In both cases, curvature is present at all interior
vertices where the number of concurrent edges is not six. Two-dimensional graphs constitute a
flattening out of a curved surface and, therefore, cannot respect the equilateral length assignment.
(c) A piece of foliated D D 2C1 triangulation (Reprinted figure with permission from [70]. ©2013
by the American Physical Society)
t+1
(4,1) (3,2)
Fig. 11.4 Four-simplices of type (4,1) and (3,2), described in the text (Reprinted figure with
permission from [59]. ©2005 by the American Physical Society)
the weighted volumes of vertices and allowed (i,j) simplices of each triangulation.
Volumes are nothing but the counting of the number of each type of simplex and are
combined together according to the combinatorial structure of T. Euclideanization
is achieved by rotating the asymmetry parameter ˛ ! ˛ by 180 degrees in the
complex plane .Re ˛; Im ˛/, under certain restrictions on the range of ˛.
The form of the action in D D 2; 3; 4 can be found in [53, 55], to which we refer
for the details. In four-dimensional foliated CDT, it reads
Regge
SE .T/ D c0 .G; ˛/ N0 .T/ C c4 .G; ; ˛/ N4 .T/
C.˛/ ŒN41 .T/ 6N0 .T/ ; (11.29)
where N0 is the number of vertices of the triangulation T and N4 D N41 C N32

is the total number of 4-simplices, in turn split into those of type (4,1) and (3,2).
The constant c0 is proportional to 1=G with ˛-dependent coefficient, c4 is an ˛-
dependent linear combination of and 1=G and the parameter depends on ˛;
their explicit form can be found in [79]. Since D 0 when ˛E D ˛ D 1, it
encodes the asymmetry between the lengths of spatial and time-like edges.
We now briefly describe the setting and main results of numerical simulations.
The cosmological constant is usually kept fixed, which corresponds to fixing
the four-volume hV4 i 1 averaged over paths in the path integral R (thatp is,
4
over different computer-generated triangulations). The four-volume p is d x jgj
4
in
p the continuum limit, while in simplicial gravity it is V 4 D ` s . 8˛ C 3 N 41 C
12˛ C 7 N32 /. Fixing the volume is therefore equivalent to constrain the number
of 4-simplices of type (4,1) and (3,2) but, since N41 / N32 / N4 for a given ˛, this
is tantamount to fixing the total number of 4-simplices.
The overall picture that emerges from the simulations is that there are different
phases of geometry depending on the values of the parameters [59]. Although
each individual triangulation in the partition function obeys the manifold conditions
(corresponding, in the continuum limit, to summing over smooth manifolds with
metric g ), the geometry coming out of the sum over histories can be highly non-
classical. Such is the case in phases A and B of the phase-space diagram of the
theory (Fig. 11.5).
The branched-polymer phase (phase A) dominates at sufficiently large c0 (small
Newton’s constant G) above some critical value cN 0 . The time evolution of the three-
volume is characterized by an irregular sequence of maxima of variable size and
minima of the order of the smallest three-volume `3s (i.e., one single vertex). The
lifetime of these spacetime configurations with macroscopic spatial extension is
very short, t 3. These mutually disconnected “lumps” of space are realizations
of branched polymers, which dominate the path for this choice of parameters.
Branched polymers [209] belong to a class of random fractals which include tree
graphs and random combs. These objects have dH D 2, while their spectral
dimension is bounded both from above and from below, according to the relation
2dH =.dH C 1/ 6 dS 6 dH [210–214]. Hence, the geometry of phase A is far from
being Riemannian.
The crumpled phase (phase B) occurs for small c0 < cN 0 and small asymmetry
parameter 0 6 < . N Here, spacetime has a vanishing temporal extension since
it is limited to a large-volume configuration at a thin slice t D 2, where all 4-
simplices are concentrated. Everywhere else along the time evolution, the three-
volume stays close to its minimum `3s (just one vertex), which in the continuum
0.8
0.6
0.4 C
D A
0.2
Bifurcation
0
Quadruple point
B
–0.2
0 1 2 3 4 5
c0
a b c
Fig. 11.5 Upper plot: Phase diagram of four-dimensional causal dynamical triangulations in the
parameter space spanned by the gravitational coupling c0 / 1=G and the asymmetry parameter .
“Bifurcation” is phase D. Thick points mark where numerical simulations have been conducted at
the phase transitions, while the dashed lines converging to the quadruple point and extending to
the outer regions are an extrapolation. The jump in the order parameter N41 6N0 (conjugate to
according to (11.29)) decreases for decreasing c0 , until the B $ D (bifurcation) phase transition
has an end-point moving towards the left of the phase diagram. Lower graphics: snapshots of
typical spacetimes in phases A (N4 D 45;500, c0 D 5:6, D 0, ttot D 20), B (N4 D 22;250,
c0 D 1:6, ttot D 20) and C (N4 D 91;100, c0 D 2:2, D 0:6, ttot D 40). At each snapshot, taken
at intervals t D 1, the circumference is proportional to the spatial three-volume V3 .t/, while the
surface is an interpolation between adjacent spatial volumes, to give the reader the feeling of a
continuous evolution (Source: adaptation from [59, 73])
limit means that the universe has zero spatial extension. The spacetime at the thin
slice is not just a short-lived three-dimensional universe, since it has large-volume
but almost no extension. In fact, there the number of vertices is very small while
the number of simplices concurring at each vertex is extremely high. In particular,

one can jump from one tetrahedron to any other in just a few steps. The geometry is
highly correlated and displays a very high, or possibly infinite, Hausdorff dimension
dH , and does not correspond to any classical smooth configuration.
Within the bifurcation phase (phase D or Cb ) that separates phase B from
C, periodic clusters of volume form around singular vertices and the metric of
spacetime spontaneously changes signature from Lorentzian (in phase C, where an
effective metric is meaningful) to Euclidean [71, 73–75].
Both phase A and B have an analogue in Euclidean dynamical triangulations
[195, 196, 199, 200, 203] but phase C and the bifurcation phase are typical only
of CDT. The intuitive reason is that in EDT all edges have the same length and
D 0, but on the D 0 line one can only access to phases A and B with
numerical methods. The phase transition A $ B is first order: the number of
vertices N0 abruptly jumps from one type of geometry to the other, with no smooth
interpolating configurations in between [201, 202]. This explains why EDT fails to
have a continuum limit, for which existence of a second-order phase transition is
required. On the other hand, in CDT it is still not clear how the continuum limit is
approached. While the A $ C transition is first order [65, 68] and the B $ D
transition is second order [67, 68, 75],5 the order of the transition between the
bifurcation phase and phase C is under study. A first- or second-order transition
would imply that the change of spacetime signature is, respectively, discontinuous
or continuous. Recent results give preliminary indications that the B $ C transition
may be second or higher order [74].
11.3.2 Cosmology
In phase C (or CdS ), a semi-classical limit is reached non-perturbatively. Here the

geometry is macroscopically four-dimensional [55–57, 59] and can be identified, at
large scales, with a de Sitter universe [61, 63, 66, 70]. The typical result is shown
in the bottom right of Fig. 11.5. The number of simplices is fixed to some N4 D
O.104 /; the vertical axis is the time span t 2 Œ0; ttot .
To check the geometric properties of this spacetime, one can study its dimen-
sionality and the time evolution of spatial hypersurfaces. Let dH be the Hausdorff
dimension of the average geometry. For a classical spacetime, time intervals are
1=d 1=d .d 1/=dH .d 1/=dH
expected to scale as N4 H / V4 H and spatial volumes as N4 H / V4 H .
For the output geometry to correspond to a macroscopic semi-classical spacetime,
it should be dH 4 D D with good approximation. One check consists in setting
dH D 4 from the start, rescaling spatial volumes and time as the dimensionless
5
In [67, 68], phase D had not been discovered yet and it was indicated as “phase C.”
quantities
V3 t
V3 ! 3=4
; t! 1=4
; (11.30)
V4 V4
and comparing the rescaled spatial slices for simulations with different N4 ,
for P instance by measuring the volume-volume correlator hV3 .0/ V3 .Nt/i :D
2 ttot
N
ttot jD1 V3 . j/ V3 . j C t/. It turns out that the profiles of the volume-volume
correlator as a function of Nt match. Alternatively, one can let dH free and fit it
to match the volume correlators; indeed, the best fit for dH is 4 within the given
accuracy. As said above, the rescaling (11.30) is typical of a classical spacetime,
which means that it is appropriate to call the geometry under study macroscopic.
A further suggestive result is that the average geodesic distance hri between points
1=3
in the spatial volume V3 .t/ scales, for any given t, as hri V3 . This scaling
and the strong matching of the rescaling (11.30) indicate that what is produced
by the simulations are genuine, macroscopic, four-dimensional spacetimes whose
spatial volumes are three-dimensional. Another geometric indicator is the spectral
dimension. At large scales, dS ' 4, while at small scales there is uncertain numerical
evidence on the values dS ' 2 [58, 59] and dS ' 3=2 [72].
This is not sufficient, by itself, to conclude that we are observing the emergence
of a smooth manifold, since there exist very “irregular” geometries with integer
Hausdorff and spectral dimension. However, at large scales the geometry matches
de Sitter spacetime. First, the Euclidean effective action for a large number N3 of
tetrahedra (i.e., large spatial volumes) is [57, 59, 61]
ttot
X
ŒN3 .t C 1/ N3 .t/2 1=3
SEeff / b1 C b2 N3 .t/ N3 .t/ ; (11.31)
tD0
N3 .t/
P tot
where b1;2 are constants and the parameter is determined so that ttD0 N3 .t/ D
N4 .t/. Calling the three-volume N3 .t/ / V3 .t/ D: a3 .t/ and taking arbitrarily small
time steps, up to an overall minus sign we obtain the action of Euclidean closed
FLRW cosmology in synchronous gauge with scale factor a and cosmological
constant, evaluated within a finite time interval:
Z
ttot
3 3 aP 2 1
SEeff ' SEFLRW / dt a C 2 : (11.32)
0 2 a2 a
Thus, at large scales and in this region of parameter space, one does get cosmology
from background-independent quantum gravity. Furthermore, hV3 .t/i is very well
fitted by the classical three-volume of Euclidean de Sitter space V3 D a3 .t/ with
a.t/ / cos.bt/, for some constant b related to b1 . The linear size of the universe is
O.10/lPl [63]. Small quantum fluctuations accompany this configuration.
The embedding of the CDT semi-classical limit in a realistic history of the uni-
verse is still missing. The cosmological constant is fixed throughout the numerical
11.4 Spin Foams 563
simulations, which therefore cannot say anything about the problems. The typical
de Sitter universe is about 10 times larger than a Planck volume, implying that such
configuration may be of interest only for primordial cosmology, for instance as the
birth of an inflationary universe from quantum physics. How such birth and the
UV properties of this geometry can accommodate the big-bang problem remains
to be seen. Also, the CDT de Sitter universe has a finite life span and one should
study how to extend its evolution to the future, possibly with the inclusion of matter
degrees of freedom. Despite all these open issues, CDT gives encouraging hints that,
indeed, cosmology can naturally come from quantum gravity.
To summarize, in CDT one starts from a fully background-independent, non-
perturbative formulation of quantum gravity and obtains various geometric phases.
In general, the geometry of quantum spacetime does not correspond to any continu-
ous manifold. Under such premises, it is highly non-trivial to show the existence of a
semi-classical limit. However, the theory does possess a phase where at large scales
the geometry naturally tends to de Sitter spacetime, a stable, classical background of
Planckian size. Therefore, accelerating cosmology can emerge at early times from
quantum gravity non-perturbatively. At microscopic scales in the same phase, the
spectral dimension of spacetime is appreciably lower than 4, signalling a highly non-
classical UV geometric regime. This also gives an explicit illustration of the fact,
mentioned in Sect. 10.1, that quantization and mini-superspace symmetry reduction
do not commute: if they did, (11.32) would be valid at all scales, not only in the
large-volume limit. On the contrary, at small scales geometry is radically different
and the semi-classical limit (11.32) breaks down.
11.4 Spin Foams
A disadvantage of canonical quantization is the loss of general covariance. A

complete covariant definition of the quantum dynamics may be achieved by the
path-integral version of LQG going under the name of spin foams.
11.4.1 Framework
In Sect. 9.2.2, we briefly introduced the concept of path integral. In quantum

mechanics and quantum field theory, this is related to the probability for a state
specified at some initial configuration ˙i (initial time ti or field value i ) to evolve
into a state with configuration ˙f (final time tf or field value f ). A notion of global
time is employed in foliated CDT but it is absent in a general-covariant, background
independent theory of gravitation such as the one symbolically represented by
the path integral (9.91). In particular, in canonical formalism the dynamics is
not encoded in an evolution variable but, rather, in the solutions to the quantum
constraints. This can be achieved either via the group averaging detailed in
Sect. 10.2.2 or by a relational time variable that can be found from the available
degrees of freedom of gravity and matter, a procedure known as deparametrization.
The interpretation of the “initial” and “final” states is modified accordingly.
Just as LQG states are graphs labelled by group elements or group represen-
tations, spin-foam histories are 2-complexes labelled by the same type of data.
A 2-complex is a collection of faces, edges and vertices, triangulating a four-
dimensional geometry bounded by an “initial” and a “final” spin-network state j˙i i
and j˙f i. These boundary states are solutions to the quantum first-class constraints.
One usually takes the dual of the triangulation, constructed with the surfaces
orthogonal to the faces of the 4-simplices. Then, the labels of edges and vertices
of a spin network (respectively, half-integer spin numbers and intertwiners) are now
attached to the faces and edges of the dual triangulation (a “foam” of spin labels); a
dual vertex corresponds to a 4-simplex. In the continuum classical limit, the overall
configuration can be heuristically regarded as a four-dimensional manifold with
boundary, describing the “evolution” of the geometry ˙i into ˙f . At the quantum
level, the transition amplitude is the physical inner product Z D h˙f j˙i i between
the two states, in the case of LQG obtained by summing over all possible spin labels
and dual triangulations with the same boundary. The sum over dual triangulations is
called vertex expansion, since the n-th term of the series contains n vertices (i.e., n
4-simplices in the original triangulation). A simplex is a chunk of flat Euclideanized
spacetime, so that the vertex expansion might be roughly interpreted (but see
Sect. 11.5.1) as an approximation around flat spacetime. It well captures both the
classical and quantum features of the continuum limit [94, 95]. The spin-foam
models currently under the most intense scrutiny are the Engle–Pereira–Rovelli–
Livine (EPRL) amplitude [85, 86, 89, 90] and its variant the Freidel–Krasnov (FK)
amplitude [84, 87, 88, 91]. Asymptotically for large spin labels in the unitary
representations of the gauge group, both EPRL and FK models reproduce the path
integral Z ' exp.iSRegge / for the Regge action, of which we saw a foliated example
in (11.29) [92, 93]. Therefore, spin foams possess not only a continuum limit, but
also the correct semi-classical one of simplicial gravity. In this sense, CDT can be
regarded as an asymptotic approximated limit of spin foams.
11.4.2 Cosmology
A path-integral formulation of quantum cosmology also exists, both for the old
Wheeler–DeWitt quantization [215–217] and for LQC [100–110]. The dynamics
of homogeneous and isotropic LQC is recast in a form that formally resembles the
spin-foam formulation of the full theory; in particular, it is expressed in terms of
a vertex expansion. This cosmological model can exemplify issues and conceptual
points that one would eventually have to address in the spin-foam formulation of
quantum gravity and can suggest solutions to the same by providing a testing ground
for novel techniques.
11.4 Spin Foams 565
The covariant construction of two-point functions is particularly simple because

of the close analogy between flat cosmology with a massless scalar field and the
relativistic point particle in quantum mechanics [216, 218]. In fact, not only do both
systems have the same number of degrees of freedom, but they are also formally
identical. To see this, we recall that the parametrized form of the particle action is
Z 00
dx
SD d p NH0 ; H0 D p p C m2 ; (11.33)
0 d
which is invariant under reparametrizations ! f . / of the world-line. H0 is the

Hamiltonian, which for a massless particle in 1 C 1 dimensions reads H0 D p2x p2t
and can be quantized as HO 0 D .@2x @2t /. On the other hand, the space of physical
states in canonical quantum cosmology is obtained by solving the Hamiltonian
constraint

CO Œv; D pO 2 Œv; D 0 ; (11.34)
corresponding to (10.30) for Wheeler–DeWitt cosmology and (10.110) for LQC.

The two expressions only differ in the explicit form of the operator , which we
do not need to use here. In Sects. 10.2.2 and 10.3.4, the space of solution
p states
was reduced in terms of two sectors satisfying, respectively, .Op /˙ D 0,
resulting from the fact that pO is a Dirac observable for the system governed by the
above dynamics. This allowed us to perform an explicit splitting of the space of
physical states into positive- and negative-frequency states.
Now, the form of the constraint (11.34) is that of a relativistic particle in 1+1
dimensions (with a discretized spatial coordinate in the case of LQC), where plays
the role of a second-order spatial derivative (up to a mass terms) and the scalar field
of the time coordinate t. The volume eigenstates jvi with normalization (10.15)
are the counterpart of the position states jxi. The conjugate momentum p is the
analogue of the energy E. Defining
Z
d q
jp i :D 2jp j eip j i; (11.35)
2
one verifies
pO jp i D p jp i ; p 2 R; (11.36)
as well as the positive-definite normalization
hp jp0 i D 2jp j ı. p p0 / : (11.37)

Alternatively, one may remove the absolute value in the definition (11.35) to obtain
the indefinite normalization
hp jp0 i D 2p ı. p p0 / : (11.38)
Just as in the particle case, the definition of different two-point functions is fully
characterized by three choices: (i) the canonical inner product, via the choice of
canonical representation of the physical quantum states in the space, depending
on which normalization between (11.37) and (11.38) is selected; (ii) whether one
restricts the attention to either positive- or negative-frequency sectors of the Hilbert
space, or works in the full Hilbert space (super-selection); (iii) the integration
range in proper time (lapse) in the group-averaging representation of the two-point
function, or, equivalently, the class of histories summed over in the sum-over-
histories (i.e., path-integral) formulation of the same. Integrals over the full real line
give, in general, solutions of the constraint equation and, thus, true inner products for
the canonical theory. Integrals of the positive semi-axis only give Green’s functions
(propagators) for the same constraint equation, i.e., solutions of the constraint
equation in the presence of a delta source.
After making the choices (i)–(iii), one can proceed to define various two-point
functions for quantum cosmology and to give their spin-foam representation in
terms of a sum over histories. As an example, consider a “relativistic” representation
in terms of states
jv; i :D jv; I Ci C jv; I i ; (11.39)
where the kets

Z p
dp
jv; I ˙i D e˙ip ı. p / jv; p i (11.40)
2jp j
are not orthogonal at fixed . Choosing the positive-definite normalization (11.37)

and the corresponding resolution of the identity in momentum space, these states
satisfy the completeness relation
X $ $

1 D i jv; I Ci @ hv; I Cj jv; I i @ hv; I j ; (11.41)
v
$ !
where @ D @ @ . The sum becomes an integral in the WDW case. The inner
product is then positive definite. Super-selecting the positive-frequency sector, only
the first term in the resolution of the identity (11.41) would survive.
11.4 Spin Foams 567
The inner product between two states ji ; i i and jf ; f i can be defined
rigorously in the group averaging procedure (Sect. 10.2.2). In the relativistic
representation (11.39) with both frequency sectors, the inner product is [105, 219]6
Z C1
i˛ CO
GH .vf ; f I vi ; i / :D d˛ hvf ; f je jvi ; i i : (11.42)
1
This two-point function governs the evolution of the universe from one volume-
matter configuration to another. For the relativistic particle, the analogue of (11.42)
is the Hadamard function, hence the subscript H. The Hadamard function has a sum-
over-histories representation obtained by fixing the reparametrization invariance of
the action (11.33) in proper-time gauge dN=d D 0:
Z C1
GH .x00 ; t00 I x0 ; t0 / D d g.x00 ; t00 I jx0 ; t0 I 0/ ; (11.43)
1
where g.x00 ; t00 I jx00 ; t0 I 0/ is a non-relativistic transition amplitude for the Hamil-
tonian H0 , in (proper) time . Using a different (“non-relativistic”) canonical
representation, one ends up with another type of two-point function, named
Newton–Wigner, that can be described also via deparametrization [100, 102, 105].
Group averaging and the deparametrized framework lead to distinct vertex expan-
sions, which correspond to different perturbative expressions of the same two-point
function.
The Hadamard function (11.42) can be recast in the vertex expansion. One
observes that CO consists of two pieces that act on the kinematical Hilbert spaces
g
Hkin and Hkin , respectively, so that for (11.42) the group-averaged inner product
takes the form
Z C1
GH .vf ; f I vi ; i/ D: d˛ AH . I ˛/A .vf ; vi I ˛/ ;
1
where D f i and
Z C1
i˛ dp i˛p2 ip
A .vf ; vi I ˛/ :D hvf je jvi i ; AH . I ˛/ :D e e :
1 2
6
Restriction of the integration interval to the positive semi-axis ˛ 2 Œ0; C1/ leads to the definition
of the Feynman Green’s function GF . While two-point functions such as GH define canonical inner
products, Green’s functions such as GF are not solutions of the Hamiltonian constraint equation but
they are true transition amplitudes (propagators), in the sense that they take into account the relative
ordering in the “time” variable labelling the states (and thus defining a background-independent
notion of “in” and “out”) [105]. In other words, true transition amplitudes propagate solutions of
the constraint equation into other solutions.
A is recast in the manner of Feynman by splitting up the “time” interval ˛ into N

parts of equal length D ˛=N and introducing decompositions of the identity on
g
Hkin :
X
A .vf ; vi I ˛/ D hvf jei jvN N1 ihvN N1 jei jvN N2 i : : : hvN 1 jei jvi i :
vNN1 ;:::;vN1
One then reorganizes the sum by characterizing each possible history .vi ; vN 1 ;
: : : ; vN N2 ; vN N1 ; vf / by the number of volume transitions M that occur. In the limit
N ! 1 and performing the integration over ˛ and, then, the one over p , the final
result has the cosine typical of the Hadamard two-point function [105]:
C1
X X Y
p
1
GH .vf ; f I vi ; i / D vf vM1 : : : v2 v1 v1 vi
MD0 vM1 ;:::;v1 kD1
.n k 1/Š
vm ¤vmC1
nk 1 X
p p
@ 1 cos. wm wm /
p Qp ;
@wk wk mD1
wm wm jD1 .wm wm wj wj /
j¤m
(11.44a)
where
vk vl :D hvk jjvl i (11.44b)
are the matrix elements of and the p distinct values appearing in .vi ; v1 ; : : : ;
vM1 ; vf / are denoted by w1 ; : : : ; wp with multiplicities ni , so that n1 C : : : C np D
M C 1. This last step can only be done formally if the variable v has a continuous
range. The expression (11.44a), which is rigorous for LQC, is understood to hold
for WDW theory in this formal sense.
The purpose of the example (11.44a) is to illustrate how the form of the path
integral depends on the various steps (i)–(iii); other possibilities are available in the
literature. Here we are mainly interested in quoting the cosmological applications
of the theory. To this purpose, we revert to a generic setting where a spin-foam sum
has been properly derived.
To extract phenomenology, one must make a concrete choice for the boundary
states j˙i i D jvi ; i i and j˙f i D jvf ; f i. Assuming j˙i i and j˙f i to be coherent
states and taking the leading term in the vertex expansion, the classical dynamics
is recovered in the large-volume limit [103, 106, 110]. This is a weaker result than
that in CDT, since here the path integral is calculated in the symmetry reduction
of mini-superspace. However, one should appreciate that, away from the classical
limit, the support of the LQC transition amplitude does not include a D 0 [110]. The
canonical and path-integral approaches to FLRW loop quantum cosmology agree on
the removal of the big-bang singularity.
11.5 Group Field Theory 569
11.5 Group Field Theory
In any canonical scheme such as LQG, while geometry is fully dynamical the
topology of the universe is fixed by construction. In general, however, one may
ask whether it is possible to build a quantum theory inclusive of topology change or,
in other words, if one can envisage an interacting multiverse scenario obeying a set
of quantum rules. In its general lines, this is the field (or, more improperly, “third”)
quantization approach mentioned in Sect. 10.2.4.
Beside the issue of topology change, the main difficulty faced by LQG is the
complete definition of the quantum dynamics and the proof that the resulting
theory contains Einstein’s gravity in an appropriate limit. A tentative but complete
definition of the quantum dynamics of spin-network states is obtained, via spin-
foam models, by embedding LQG states into the larger framework of group field
theory (GFT), in turn strictly related to tensor models [220, 221].
11.5.1 Framework
GFTs are quantum field theories on group manifolds. Instead of Lorentzian tensor
fields ::: .x/ on a D-dimensional spacetime manifold, one has a complex-valued
object '.g/ :D '.g1 ; : : : ; gD / dependent on D elements gI (I D 1; : : : ; D) of a
given Lie group G, the local gauge group of gravity. Gauge invariance of vertices is
expressed by the property
'.g1 ; : : : ; gD / D '.g1 h; : : : ; gD h/ 8h 2 G ; (11.45)
i.e., the field ' W GD ! C is invariant under right multiplication of all its arguments
gI with the same group element h. The classical dynamics is governed by the action
Z Z
SGFT Œ'; ' D dD g dD g0 ' .g/K.g; g0 / '.g0 / C VŒ'.g/; ' .g/ ;
G G
(11.46)
where the kinetic operator K is a non-local operator on G ˝ G and the potential V
is a non-linear interaction of the fields; choices of K and V fix the model. Variation
of (11.46) with respect to ' yields the classical equation of motion (group domain
in the integral omitted)
Z
ıV
dD g0 K.g; g0 /'.g0 / C D 0; (11.47)
ı' .g/
while varying with respect to ' gives the conjugate of (11.47).

The classical field '.g/ is interpreted as the D-valent vertex of a spin network,
with group labels g1 ; : : : ; gD attached to the D links. Each gI is then the holonomy
(parallel transport) of the connection along the I-th link, (11.45) being the gauge
transformations acting on the vertex. To each vertex in a spin network there
corresponds a .D 1/-simplex (in D D 4, a tetrahedron) in the dual simplicial
complex. In this representation, '.g/ is a .D 1/-simplex whose D .D 2/-faces
are labelled by the gI ’s and (11.45) is called closure constraint, since it can be shown
to be equivalent to the requirement that the faces (four triangles, for a tetrahedron)
close to form the simplex. The interaction term V in the action describes how
.D 1/-simplices are glued together along their faces to form a D-simplex.
GFTs are the direct second quantization of spin networks [146]. The quantum
scalar field 'O can be expanded in terms of creation and annihilation operators on a
Fock space. A choice of operator ordering is necessary in the interaction term, since
'O and 'O " do not commute and obey the algebra
'.g/;
O 'O " .g0 / D 1G .g; g0 / ; '.g/;
O O 0 / D 0 D 'O " .g/; 'O " .g0 / ;
'.g
(11.48)
where 1G is the identity operator on G compatible with (11.45). For a compact
R Q 0 1
group, 1G .g; g0 / :D G dh D ID1 ı gI hgI , where dh is the Haar measure of the
R
group such that the group volume is normalized to 1, G dh D 1.
The Fock states of GFT closely resemble spin networks or their dual and they
represent certain geometries according to the choice of the group. In the continuum
limit of differential manifolds, for Riemannian four-dimensional geometries one
has G D SO.4/, while for covariant Lorentzian four-dimensional models G D
SL.2; C/. When GFT is constructed as the generalization of loop quantum gravity,
the group is G D SU.2/ and the geometry described by the states is three-
dimensional and spatial. In this case, the connection is the Ashtekar–Barbero
connection and the elements of the group are the holonomies (9.107) along the edges
of the tetrahedra.
The GFT Hilbert space is very similar to the standard Hilbert space of loop
quantum gravity, although there are some technical differences.7 These are not
very important in what follows and we will stick with the LQG terminology in
D D 4. The Fock vacuum j;i is, by definition, annihilated by ', O 'j;i
O D 0, and
corresponds to a “no-spacetime” configuration where no quantum-geometry degree
of freedom is present and all area and volume operators have vanishing expectation
value. It is normalized to 1 by convention, h;j;i D 1. The one-particle GFT state
jgi :D 'O " .g/j;i is interpreted as the creation of a four-valent spin-network vertex or
of its dual tetrahedron with labels g1 ; : : : ; g4 . The labels of all the vertices of a spin
network with N vertices are specified by constructing an N -particle state.
7
Contrary to LQG spin networks, the combinatorial structure of GFT states is not embedded in
an abstract space; no cylindrical equivalence conditions are imposed; states associated to different
graphs have a different scalar product.
Just like in LQG, one can define area and volume operators on the kinematical
Hilbert space with different spectra. For instance, the square of the total area hAO I i2
associated with all I-th faces in a given state will differ, in general, from the sum
hAb2I i of squared areas associated with each I-th face, where angular brackets denote
the expectation value on the state of the operators
Z s Z
AO I :D 2 d4 g 'O " .g/ d4 g0 K.gI ; g0 / '.g
O 0/ ; (11.49a)
Z Z
Ab2I :D . 2 /2 d4 g 'O " .g/ d4 g0 K.gI ; g0 / '.g
O 0/ ; (11.49b)
where is the Barbero–Immirzi parameter. The minus signs are such that expec-
tation values are positive definite in the case of gravity in three spatial dimensions.
Contrary to LQG, the number of spin-network vertices N is an observable in GFT,
given by the expectation value of the operator
Z
NO :D d4 g 'O " .g/ '.g/
O ; hNO i D: N : (11.50)
The elements BI 2 g of the Lie algebra associated with the group G are the
coordinates of the GFT momentum space, related to group space via the momentum
transform [122, 123, 129, 222, 223]
Z 4
Y
Q 1 ; : : : ; B4 / :D
'.B d4 g egI .BI / '.g1 ; : : : ; g4 / ; (11.51)
ID1
where feg .B/g is a basis of “plane waves” on G. Since G is non-Abelian and

gh ¤ hg for any two group elements, the product of two elements of the basis is
non-commutative: eg .B/ ? eh .B/ D egh .B/ ¤ ehg .B/ D eh .B/ ? eg .B/. The bivectors
.BI /˛ˇ can be directly related to the flux variables of Sect. 9.3 and represent the
oriented areas, spanned by the triads e˛i , of the face I of the dual tetrahedron labelled
by the gI ’s. Therefore, the one-particle momentum-space Fock state 'OQ " .B/j;i
fully determines the metric of one tetrahedron. In quantum mechanics, it would
correspond to the momentum particle state jpi with p being a vector multiplet with
components pI .
The quantum dynamics of the theory can be studied with the various methods
available in quantum field theory, adapted to group manifolds. The GFT path
integral or partition function can be written perturbatively
P as a series of Feynman
diagrams. Decomposing the interaction V D n n Vn as a sum of polynomials of
order n, one has
Z Q
X n n
N . /
SGFT Œ';'
ZGFT :D ŒD'ŒD' e D n
A ; (11.52)

Aut./
where are cellular complexes dual to the GFT Feynman diagrams, Nn ./ is the
number of vertices with n legs in the Feynman diagram dual to , Aut./ is the order
of the automorphisms of and A is the Feynman amplitude assigned to . This
expression can be compared with (11.28). The dynamics of GFT is defined by the
superposition of interaction processes (creation and annihilation) of spin-network
vertices, forming complexes of arbitrary topology; thus, topology is naturally made
dynamical. The Feynman amplitudes A of the theory can be represented as spin-
foam models and, therefore, encode four-dimensional spacetime geometries. The
spin-foam vertex expansion is nothing but the GFT perturbative expansion in powers
of the GFT coupling constant in V; in this case, however, its interpretation as an
expansion around flat spacetime is uncertain, since the GFT vacuum is highly non-
geometric.
General relativity is not a sub-sector of the theory but a regime obtained wherever
a continuum and classical limit of the pre-geometric fundamental construct exists.
Gravity is therefore an emergent phenomenon within group field theory, as we are
going to see now.
Matter fields can be added to the picture, either as emergent degrees of freedom
of the theory arising as perturbations around background dynamical solutions [117–
120] or by hand as new coordinates in an extension of the group manifold GD D
G ˝ ˝ G [116, 121]. In the second case and for a real scalar field, the GFT
field becomes '.g/ ! '.g; /, where the dependence is rendered dimensionless.
The generalization of the action (11.46), of the right-hand side of the commutation
relations (11.48) and of the rest of the theory is straightforward.
11.5.2 Cosmology
We had several occasions to remark that loop quantum cosmology is not the cosmo-
logical limit of full LQG but, rather, a mini-superspace model (plus perturbations)
employing LQG techniques. One of the advantages of GFT over LQG is the
possibility to extract cosmology directly from the full quantum theory [130, 132–
135, 139–141, 148]. The ensuing model warrants a thorough comparison with the
findings of homogeneous LQC and has offered important consistency checks that
the known features of LQC do belong to the complete theory and are not artifacts of
the symmetry reduction before quantization. As a companion to CDT, GFT allows
one to face the problem of how to obtain cosmological dynamical equations from a
background-independent, non-perturbative formulation of quantum gravity.
A problem in GFT, not present in CDT, in obtaining the double limit of
semi-classicality and the continuum is that the fundamental pre-geometric discrete
structure contains different information with respect to gravity on a fixed topology.
The corresponding method for defining this double limit is therefore more involved
that those described in the preceding sections, albeit not overly so.
The physical interpretation of the continuum limit is somewhat delicate. We want

to translate statements regarding differential manifolds (homogeneity in D 1 D
3, for instance) into the language of simplicial complexes, but to do so it should
be made possible to embed any such complex into a smooth continuous geometry.
If a limit where the complex describes a differentiable spatial hypersurface exists,
each tetrahedron is nearly flat. This flatness condition consists in having the triads
e˛i , associated with the bivectors BI labelling the faces of each
P tetrahedron, almost
constant. Since the four BI ’s must close into a tetrahedron ( I BI D 0), for each
tetrahedron one can reparametrize this information by three linearly independent
bivectors defined by the simplicity constraints
ˇ
.Bi /˛ˇ D i e˛j ek :
jk
(11.53)
Consequently, one can approximate the integral of the triad along the edges of
the tetrahedron T by its value e˛i .xv.T/ / at the point xv.T/ where the dual spin-
network vertex v.T/ associated with T is located in the abstract embedding. Then,
for each tetrahedron one has triad data that can be converted into information on the
j
metric g˛ˇ .xv.T/ / D ıij ei˛ .xv.T/ / eˇ .xv.T/ / at each point, and a description in terms
of continuous fields is made possible. Retroactively, this allows one to compute
the curvature of the geometry as a whole and compare it with the linear size of
each tetrahedron. The flatness condition is thus nothing but the requirement that
the tetrahedra be much smaller than the overall curvature radius of the embedding
geometry. It is not an assumption imposed on the model a priori but, rather, a self-
consistency check to be done at the end of the calculation. If this check failed, then
the interpretation of the resulting quantum state in terms of continuum geometry
would lose robustness or even break down.
With this caveat on board, let us construct GFT quantum states in the Fock space
capable of describing homogeneous semi-classical geometries with a continuum
limit. We begin with the vacuum theory and the double requirement of continuity
and semi-classicality.
Consider an n-particle complex given by the superposition of different discrete
geometries:
Y
n Z
j˝n i D Oj j;i ; Oj ´ d4 g j .g/ 'O " .g/ ;
jD0
where the operators Oj are composed of creation operators summed over all
possible group configurations. The group weights j are left-invariant under gauge
transformations, j .g1 ; : : : ; g4 / D j .hg1 ; : : : ; hg4 / for all h 2 G. Intuitively, a
continuum limit is approximated by states with an infinite number of particles:
N
X
jcont i D lim bn j˝n i :
N !C1
nD0
This most generic state corresponds to an arbitrarily non-classical geometry, which

may or may not be well defined depending on whether the formal limit N ! C1
can actually be performed. To achieve semi-classicality, one should also be able
to replace operators 'O in the quantum dynamics with a classical group field .g/
representing their expectation value on the state. Note that the constituents of the
GFT field remain quantum even if the field itself is taken to a classical limit. This
means that, at small scales, the system is in a quantum regime where individual
“molecules” or “atoms” interact quantum mechanically, while at large scales it will
admit a classical description in terms of collective or “hydrodynamical” degrees
of freedom. In the LQG interpretation, the scale separating the microscopic and
macroscopic regimes is the Planck scale.
In fact, a configuration with (i) an infinite number of particles with a concrete
N ! C1 limit and (ii) realizing semi-classicality but with fully quantum elements
is nothing but a condensate, of which we saw examples in Sects. 7.6.4 and 9.4.2.
GFT condensates are known for quantum cosmology [130, 132–137] and for
spherically-symmetric quantum geometries with an horizon [138]. Here we consider
cosmology and the case where the condensate is of elementary building blocks, i.e.,
tetrahedra. Let
Z
O WD d4 g .g/ 'O ".g/ (11.54)
be a single-tetrahedron operator with group weight (called in the original

literature) left-invariant under gauge transformations. By changing (11.54), one can
obtain other types of states where the building blocks are composite (“molecules” or
bound states such as pairs of tetrahedra). According to the lore of condensed matter,
a condensate of tetrahedra is represented by the gauge-invariant kinematical state
O
ji WD A e j;i ; (11.55)
where A is a normalization constant. If, as in realistic situations, the state lies in

the Fock space so that it does not define an inequivalent representation of the same
quantum system, one can choose A so that hji D 1:
Z
1 4 2
A D exp d g j.g/j : (11.56)
2
An easy calculation shows that ji is a coherent state, that is, an eigenstate of the
annihilation operator 'O with eigenvalue ,
'.g/ji
O D .g/ji : (11.57)
The metric can be reconstructed from when working in the space of bivectors via
the momentum transform (11.51).
Expression (11.55) defines a non-perturbative vacuum on the kinematical Fock

space, on which the GFT field acquires a non-trivial expectation value: hj'ji
O D
¤ 0. In condensed-matter physics, expanding the field ' D C ı' around its
non-zero vacuum expectation value and truncating the equations of motion up to
some order in the fluctuations ı' is called mean-field approximation. To spell this
approximation in precise terms, we consider the single-particle condensate (11.55).
Full quantum dynamics is given by the infinite tower of constraints hjOO Cji
O D 0,
where
Z
ı VO
CO :D d4 g0 K.g; g0 /'.g
O 0/ C (11.58)
ı 'O " .g/
is the quantum version of the classical equation of motion (11.47) and OŒ O ';O 'O "
is an arbitrary operator of the field and its Hermitian conjugate. Exact solutions
to the quantum dynamics are able to solve all these conditions simultaneously.
Approximated solutions can be found by imposing only the first of such constraints,
with O D 1. The expectation value of the quantum equation of motion (11.58) on
ji is the analogue of the Gross–Pitaevskii equation for Bose–Einstein condensation
[224–226]. Taking a normal ordering in VO such that all 'O " are to the left of all the ',O
from (11.57) one has
Z ˇ
ˇ
O D d4 g0 K.g; g0 /.g0 / C ıV ˇ
0 D hjCji : (11.59)
ı' .g/ ˇ'D

Solutions .g/ of this equation give, when plugged into (11.55), approximate
physical states.
The scalar weight jj is interpreted as a probability distribution on the space of
homogeneous geometries. It is not a wave-function of the quantum geometry in the
canonical sense, since (11.59) is non-linear in general (just like the Gross–Pitaevskii
equation, which is a “non-linear Schrödinger equation” for the condensate wave-
function). In WDW and loop quantum cosmology, a wave-function describes
a single quantum Universe with fixed topology. In mini-superspace spin-foam
cosmology, the quantum Universe is represented by a path integral analogous to that
for an individual quantum particle. In GFT cosmology (or GFC in short), the scalar
is a highly quantum object, the interpretation of a continuum geometry and the
semi-classical limit being recovered only by the macroscopic, large-scale collective
behaviour of this many-particle ensemble. In the case of molecular condensates,
also encodes the correlation between different quanta.
In a homogeneous manifold, all points of space carry the same information on the
metric or the connection. In a classical dual simplicial complex where the flatness
condition holds for each individual tetrahedron, the equivalent of points of space
are tetrahedra and their metric information is carried by their group or algebra
labels. The redundancy required by homogeneity is thus achieved by asking that
all the building blocks of the combinatorial structure be in the same microscopic
configuration. This does not mean that the Hilbert space has been reduced to states
with certain symmetry requirements, which would correspond to an ad hoc mini-
superspace symmetry reduction. Rather, we have considered special states in the full
Hilbert space and we have imposed that the quantum distribution .g/ of geometric
data g is identical for all the quanta. Accordingly, and in this precise sense, the GFT
condensate (11.55), composed by an infinite number of tetrahedra all with the same
wave-function, represents a quantum continuous geometry which is homogeneous
at the scale of the quanta (the Planck scale). At ultra-macroscopic scales where
only global properties are apparent, by construction the geometry is classical and
homogeneous.
Note that the definition of semi-classicality employed here differs from the one
from the point of view of spacetime geometry. In canonical quantum gravity, semi-
classicality means peakedness of the wave-function of the Universe around some
classical phase-space point. In GFT, this characteristic is additional to all the rest of
the construction and should be imposed, if desired, after reaching the “classical”
(in the sense of the mean-field approximation) hydrodynamical description in
terms of a distribution on mini-supespace. This distribution, which enters final
hydrodynamical equations, is formally equal to the quantum weight but, at this
effective level, it encodes information on the collective behaviour of the system
rather than on the individual tetrahedra.
Inhomogeneities could be included in different ways, most of which are still
under exploration. In general, at the microscopic level, inhomogeneities would
correspond to deviations from the simple condensate states. Some possibilities
include (i) quasi-particle fluctuations over the condensate, (ii) states in which a non-
negligible number of fundamental GFT quanta do not aggregate to the condensate
and (iii) multi-condensate states, i.e., states in which several large sub-sets of the
GFT atoms condense in different states, each characterized by a different wave-
function.
Another scenario, which amounts to an effective description of inhomogeneities,
is the following. Once the hydrodynamical description of the quantum system
is given, one can also interpret it as the purely statistical approximation of the
dynamics of classical building blocks; the weight is then regarded as a statistical
probability distribution. These constituents might be reconstructed from the classi-
cal fluid by regarding any hydrodynamical observable (for instance, the total area
associated with the state) as the composition of elementary observables (the area of
each element). The statistically reconstructed constituents are not the fundamental
GFT quanta. In fact, condensed quanta yield exactly the same distribution of
geometries, but the reconstructed effective elements may be defined to differ in
the value of their geometric observables (the value of their area) even if they are
homogeneous individually.8 The resulting effective geometry captures the properties
8
Starting from the fundamental level, this can be realized by grouping a different number N of
fundamental quanta in the effective building blocks.
of the universe at mesoscopic scales intermediate between the fundamental one and
the global picture and it is interpreted as an inhomogeneous universe [137].
This picture is strikingly coincident with the separate-universe approach: the
global geometry is described as a collection of patches of continuous and homo-
geneous space and inhomogeneities are the correlations among these patches. The
effective constituents are the equivalent of the homogeneous patches and, in order
for the mesoscopic regime to take place, their size turns out to be a few orders
of magnitude larger than the Planck scale. The level of inhomogeneity depends,
among other details, on the shape of the mean field .g/. If the latter is sharply
peaked around one particular set of group labels, then inhomogeneities are expected
to be small, since the excursion of the geometric data between different patches is
considerably limited. This aspect of the theory is still under development.
To summarize, the GFC condensate (11.55) is characterized by three scales. At
the Planck scale, the geometry is quantum but continuous and homogeneous. At
macroscopic scales many orders of magnitude larger than the Planck scale or than
the present Hubble horizon, the geometry is classical (in the sense of the mean-
field approximation), continuous and homogeneous. Some extra ingredient might
be necessary to extend this notion of classicality with the one of canonical gravity
(there, a classical spacetime is represented by a point in the classical phase space).
In the effective approach to inhomogeneities, the transition between the micro- and
macroscopic regimes is set by a mesoscopic scale a few orders of magnitude larger
than the Planck scale, at which inhomogeneities are incorporated as an effect of
statistical fine graining of the hydrodynamical description.9
To study a concrete model of quantum dynamics, one must make a choice of
operators in (11.59). Renormalization analyses indicate that finiteness of the theory
requires the kinetic operator K to be the Laplacian g on the group manifold [124,
127, 128, 131]. At first, we will ignore the matter part and assume that non-linear
interactions are negligible, V D M 2 j'.g/j2 , where M is a dimensionless constant.
The dynamical equation to solve is thus
4
!
X
2
gI C M .g/ D 0 : (11.60)
ID1
We now use the representation of SU.2/p in a neighborhood of the identity in terms of

the generators (9.109), where gI .I / D 1 E I E I 12 C2E E I is a 22 matrix and
the four three-vectors E I are elements of the Lie algebra su.2/ such that jE I j 6 1.
The Laplacian with these coordinates reads (index I omitted everywhere)
g Œ.g/ D .ı ij i j /@i @j ./ 3 i @i ./ : (11.61)
9
The reader should consider this scale hierarchy cum grano salis, since the techniques and
interpretations in GFC are still under intense study at the time of writing.
Thanks to gauge invariance and the closure condition of the group elements on the
tetrahedron, one can manipulate (11.60) with the replacement (11.61) to express
it only in terms of the elements E I of the first three links (dual faces). The E I ’s
can be combined into the matrix invariants IJ :D .g E I g1 E J g1
4 / .g 4 /, where
I; J D 1; 2; 3, jIJ j 6 1 and II > 0. For isotropic states in which depends
only on the diagonal components II , one gets
X3
@2 @
2II .1 II / 2 C .3 4II /
ID1
@II @II
3
Xp p 2
@ 2
C 1 II 1 J J IJ 5C M D 0: (11.62)
@II @J J 4
J ¤I
Note that the assumption of isotropy D .11 ; 22 ; 33 / does not amount to a
classical symmetry reduction of the theory as in Chap. 10, since it is performed
after quantization. Therefore, homogeneity is an ingredient recovered after taking
the continuum limit of a special but fully quantum state, while isotropy is imposed
only as a useful requirement to find analytic solutions of the GFT condensate giving
rise to a FLRW cosmological background.
Compatibility with isotropyP requires the last term in square brackets in (11.62) to
vanish, implying that D I I .II / for some functions I . For simplicity, one
can further assume that the diagonal components are all equal, II D for all I,
so that (11.62) is finally recast as [133]
d2 ./ d./ M2
2.1 / 2
C .3 4/ C m ./ D 0 ; m :D ; (11.63)
d d 12
where 0 6 6 1. To give an interpretation to E and , we notice that, assuming

that the connection remains approximately constant along a dual link with length
l0 , the holonomy thereon is g ' exp.l0 !E E /, where ! i D e˛ Ai˛ . Expanding as in
(10.79), one has g D cos.l0 j!j=2/
E 12 C 2E .!=j E !j/
E sin.l0 j!j=2/,
E leading to the
identification E D .!=j
E !j/
E sin.l0 j!j=2/
E and of as the square of the sine of the
connection:

l0 j!j
E c
N
D sin2 D: sin2 : (11.64)
2 2
In the second step, we used a notation reminiscent of the cosmological setting of

(10.72) and (10.74), where we encoded the information on the holonomy length
and on the macroscopic fiducial volume into a parameter N D .a/, N whose
time evolution is parametrized by the scale factor. In accordance with the lattice-
refinement interpretation, it is natural to regard N as the number (10.143) of
1=3
elementary cells per fiducial volume D l0 V0 D N 1=3 . Near the identity,
p
E ' !=2
E and ' c=2 N is proportional to the connection c at low curvature. At
the classical level for K D 0, c / aP , so that the low-curvature classical limit is
N 2 1:
/ .aH/ (11.65)
The general solution of (11.63) is a linear combination of the associated Legendre

functions of the first and second kind (here Pm ` D .1/ P`;m ),
m
./ D m ./

1 1=4 1 1
:D A1 P 21 .p1C2m1/ .2 1/ C A2 Q 21 .p1C2m1/ .2 1/ ;
2 2
(11.66)
which can be compared with (10.25) and (10.27) of WDW quantum cosmology.
The solution (11.66), where A1;2 are constants, is always normalizable with respect
to the group measure. Its behaviour depends on the sign and value of m but, in the
most general case,
1 1
./ p : (11.67)

The probability density ./ diverges when the connection variable tends to
zero. Therefore, the general isotropic vacuum solution is infinitely peaked at small
curvature, meeting the expectation that, in the continuum limit, tetrahedra of a
classical geometry are nearly flat (spatially constant triad and connection). This is
an important self-consistency check of the theory.
The exact vacuum solutions of (11.63) are well defined also in high-curvature
regimes where 1 and the flatness condition fails; away from 0, they
can even have finite oscillatory maxima. Moreover, a special class of solutions
diverges also at D 1. These regimes and special solutions are not unphysical
but, according to the above discussion on the embedding picture, they do not admit
a simple geometric interpretation in the language of continuous smooth manifolds.
This situation is strongly remindful of what happens in LQC, where a non-classical
dynamics is effectively described by equations on a continuum even if there is
no underlying smooth manifold structure. It is in this sense that the Universe
described by ./, although semi-classical, retains many quantum features, contrary
to the WKB wave-functions of canonical quantum cosmology which represent
conventional semi-classical geometries for all values of their arguments (scale factor
a and matter field ).
Including a matter scalar field, the kinetic operator reads
4
!
X
1 0 0 2 2
K D ı.g g /ı. / gI C 12E @ C M ; (11.68)
ID1
where E 2 is some constant whose sign will be chosen later in relation with the
classical equations of motion. The analytic extension of all the following equations
to E 2 < 0 is straightforward. Equation (11.60) becomes
4
!
X
2 2 2
gI C 12E @ C M .g; / D 0 : (11.69)
ID1
Under the isotropy condition, a one-parameter family of exact solutions in factorized

form is
p
i˝
.1 ˇ/m
.; / D ˇm ./ BC e i˝
C B e ; ˝ :D ;
E
where B˙ are constant coefficients and ˇ is real. When ˇ D 1 and ˇ D 0, one

p the special solutions .; / D m ./ .B1 C B2 / and .; / D .C1 C
gets
C2 1= 1/.BC ei˝ C B ei˝ /, respectively (B1;2 and C1;2 are constants).
Another and no less important check is the recovery of the continuum classical
equations of motion, but it may be possible to get even more: the modified
Friedmann equation of homogeneous and isotropic LQC. At first, we take a WKB
Ansatz of the form
/I.; /=l2Pl
WKB .; / D A.; / eŒiS.; : (11.70)
A purely semi-classical approximation with A D const, S D S./ and I D 0 is

incompatible with the exact solutions [133], which do not oscillate rapidly in the
semi-classical region 1. In general, the functions A, S and I must be tuned so
that the continuum and classical geometric interpretation with 1=2 be given
in vacuum by the simultaneous limits lPl ! 0 and ! 0. Incidentally, the problem
of the normalization of WKB states discussed in Sect. 9.2.2 is solved in GFT, since
the general exact solution of (11.69) is known.
From now on, we look for solutions of the form S D S.; / and I D 0; damping
terms can be included in A. Plugging (11.70) into (11.63), expanding the constant
m as m D m4 l4Pl
C m2 l2
Pl
C m0 and separating order by order in lPl , one obtains the
WKB equations
0 D 2.1 /.S; /2 E 2 .S; /2 C m4 ; (11.71a)

A;
0 D 2.1 /S; C 4.1 /S; C .3 4/S;
A

A;
CE 2 S; C 2S; im2 ; (11.71b)
A
A; A; A;
0 D 2.1 / C .3 4/ C E2 C m0 : (11.71c)
A A A
Setting m4 D 0 in the first equation, we get

2
2 S;
2.1 / D E : (11.72)
S;
Ignoring matter one has 2.1 / 0 which, consistently with the above analysis,
admits solutions both in the continuum geometric limit 0 and away from it
where 1.
As in the usual Hamilton–Jacobi formalism, we identify @ S / p and @ S / p
with the semi-classical momenta. Classically, from (10.3) in N D 1 gauge they
would correspond to
a
p pN 2 aP 2 ; p a3 P : (11.73)
N 2 H
Two main results stem from (11.72): (A) the purely classical limit fixes the behavior
of N and (B) the limit of LQC effective dynamics is also recovered and confirms
(A) [134].
(A) Plugging (11.65) into (11.71a), one has H 2 / E 2 P 2 m4 a6 C .a/ N 2 ,
2
which is, assuming E > 0, the standard Friedmann equation for a massless scalar
field and two extra contributions. One is a stiff matter term which can be removed
by setting m4 D 0. The other is a curvature term K D 1 if N D 1 or a cosmological
constant if N / a1 . The first possibility is excluded because a curvature term could
only come from the classical connection c D aP C 1 and also because, if we want
to embed LQC in group field cosmology and to identify the GFC function N with
the LQC function (10.75), restrictions on the LQC matter content forbid a constant
N The other choice is more interesting, but we will see that E 2 > 0 does not lead
.
to Lorentzian LQC. Also, Wick-rotating the above equation to compensate for a
positive E 2 (H 2 ! H 2 , P 2 ! P 2 ) would give a negative cosmological constant.
We therefore turn to another derivation of the classical equation of motion.
Taking the extreme regime 0, we now make the expansion .1 / '
in (11.72) and get
N 2 / E 2 P 2 :
.a/ (11.74)
If we take E 2 < 0, the right-hand side is the scalar field energy density plus a
cosmological constant. The left-hand side is H 2 only if N / 1=Pa. For the inverse
power law (10.75) and an expanding universe, this condition is verified if a /
t1=.12n/ for n < 1=2, or if a / eHt when H is constant in the improved quantization
scheme n D 1=2. Although both cases rely on a specific form of the scale factor,
the second is more realistic in the presence of a cosmological constant , which
is bound to dominate over matter asymptotically (de Sitter attractor, Sect. 5.10.1).
Preliminary calculations confirm that only interactions, which we ignored here,
could generate a term when the scalar is in slow rolling. Remarkably, the choice
1
N / (11.75)
a
is the one of the improved quantization scheme of Sect. 10.3.3. In the canonical
theory, we saw how other quantization choices are compatible with the classical
limit. If we want to embed LQC in group field cosmology and identify the GFC
function N with the LQC function (10.75), and if we demand the classical limit
of the GFC dynamics to be the Einstein-gravity Friedmann equation in de Sitter
approximation, then one can conclude that the quantization ambiguity of the canon-
ical theory is partially removed in this model of GFT. Of the functions N decreasing
with the scale factor, only (11.75) gives rise to the classical Friedmann equation. At
the level of homogeneous and isotropic cosmology, any other quantization choice
of the canonical theory would make the embedding of LQC into GFC inconsistent
with the classical limit of general relativity.
(B) One can also obtain the Friedmann equation (10.119) (equivalent to (10.121))
of loop quantum cosmology for general . Observing that 4.1 / D sin2 .c/, N
(11.72) and (10.119) agree provided E 2 < 0 and the Hamilton–Jacobi momentum
p be
r
˛ a2
p / ; (11.76)
N
where ˛ and are the inverse-volume LQC corrections of the gravity and matter
sectors. Equation (11.76) and the functions ˛ and have not been derived from first
principles, but the characteristic structure of LQC dynamics is indeed reproduced.
The classical limit ˛; ! 1 agrees with (11.73) only if N / 1=Pa, consistently with
(11.74).
The WKB approximation may be too limited, or even inadequate [133], to
describe GFT condensates, but there are other ways to recover the improved
dynamics of LQC. One of them makes the same assumptions on the kinetic term
of GFT (the group Laplacian up to a constant) and on the potential (quadratic if the
constant
R in the Laplacian is zero; vanishing if such constant is non-zero) [137]. With
OO D d4 g 'O " .g/ and recalling the operators (11.49) and (11.50), the constraint
hjOO Cji
O D 0 can be written as
X
hAb2I i . 2 M/2 hNO i D 0 ; (11.77)
I
which is an alternative form of (11.59) and (11.60). The first expectation value is
2=3
proportional to the square of the fiducial area a2 V0 but the proportionality coeffi-
cient is also important: for consistency, in the low-curvature limit one should recover
the Poisson brackets (classical algebra) between area and holonomy R variables. In
GFT, these variables are given by the expectation values i 2 h d4 g 'O " @i 'i
O / a2
R
and h d4 g i 'O " 'i=N
O / sin.N 1=3 c=2/, where we omitted numerical coefficients.
In particular, the second expectation value is an extensive quantity and, therefore,
scales as N ; the overall scaling of the right-hand side is thus independent of N .
Therefore, the only way to compensate the N dependence in the holonomy variable
in the low-curvature limit (and hence to have a Poisson bracket fc; a2 g ' const) is to
attach a factor N 1=3 to the area 2
q variable a . In the homogeneous limit, this implies
' N hAb2 i. Assuming all I contributions in (11.77) are
2=3
that hAO i ' N 1=3 a2 V
I 0 I
4=3
equal, the latter reads 4N 1=3 a4 V0 . 2 M/2 N D 0, implying that
N / a3 :
This is nothing but the volume–number-of-vertices scaling in the improved quanti-

zation scheme of LQC.
We have thus found that the effective dynamics of the isotropic condensate state
of group field theory with Laplacian kinetic operator may be equivalent to that
of homogeneous and isotropic LQC in the improved quantization scheme (11.75)
where the area of elementary holonomies is constant. This constitutes a somewhat
surprising example of a cosmological model of quantum gravity where the opera-
tions of mini-superspace symmetry reduction and quantization possibly commute
(see the diagrams at the beginning of Sect. 10.1). Group field theory can have the
potential of deriving homogeneous and isotropic LQC from a fundamental quantum
setting. In doing so, at least part of the quantization ambiguities unavoidable in the
Hamiltonian formalism are removed for a given GFC model. The lattice-refinement
interpretation of Sect. 10.3.8 is also naturally embedded in GFT: the number of
vertices N corresponds to the expectation value of the number operator NO , the
number of particles of the condensate for a given macroscopic volume.
Evidence has been piling up that LQC at large can be fully derived from GFT
but these results need some extensions. The Laplacian g and the mass term used
above do not correspond to GFT models based on popular spin-foam amplitudes
such as the EPRL model, and one can conceive more involved kinetic operators K
and interactions V. Regarding the kinetic term, a generalization of the EPRL model
has indeed been considered in [139, 140], where LQC-like Friedmann equations
with a bounce were derived from a condensate. Non-trivial interactions V have
just begun to be implemented in models with the g [141], while there are still
no studies for the EPRL Laplacian. Non-linear interactions are carriers of topology
change and can produce an effective cosmological constant term as well as non-
minimally coupled effective models when matter is included [227]. Moreover, the
mean-field approximation should be controlled by computing fluctuations around
and checking that they remain small. If they did, then the quantum solutions
above would remain peaked at small curvature. This check would also be crucial
for the study of cosmological inhomogeneities and their comparison with LQC
perturbations. Cosmological perturbations, in fact, would be naturally interpreted
as phonons, fluctuations propagating in the condensate.
The GFC approach will likely have something to say about the cosmological
constant problem. The peak (11.67) in the probability density can be translated
into one for , since in the classical limit (where the standard Friedmann equation
holds), in the improved-dynamics prescription (11.75) and for negligible or nearly
constant matter energy density one has ' H 2 / and
1
./ p : (11.78)

This peak is less pronounced than the exponential probabilities found in WDW
quantum cosmology, but it is perhaps better motivated, as it does not rely on a mini-
superspace quantization.
The resonance between the ideas exposed here and in Sect. 9.4.2 might not
be accidental, either. The equation of motion (9.131) and the corresponding
Hamiltonian constraint have interaction terms that could find a justification in the
second quantization of GFT. The issues with the Chern–Simons state disappear
in GFT, where we deal with normalizable states of the full theory. It is quite
conceivable that the condensate of the Ashtekar connection of the rudimentary
model of Sect. 9.4.2 might be the approximation of a GFT condensate on the Fock
space in group variables. To date, this possibility has not been investigated yet.
11.6 Causal Sets
A spacetime with a discrete texture may avoid the infinities plaguing general
relativity and quantum field theory. Loop quantum gravity is an example where such
type of geometry apparently removes gravitational singularities. Causal sets [149]
is another theory with fundamental discreteness devised for the same motivation.
Its development is still at an early stage but one of its most intriguing features is an
actual prediction for the value (2.118) of the cosmological constant.
11.6.1 Framework
The type of discreteness considered in causal sets is such that any bounded region
of spacetime is made of a finite (rather than just countable) number of points. This
means that measuring any D-dimensional spacetime volume V is tantamount to
counting the N points within V. To match units, we assume that an average number
˛ 2 of points occupy a volume of Planck size tPl lD1
Pl
D mD
Pl
:
N ˛ 2 mD
Pl
V; (11.79)
where ˛ > 0 is a free real parameter and the symbol “” will be clarified shortly.
If ˛ D O.1/, then the Planck scale lPl is also the discreteness scale l characterizing
11.6 Causal Sets 585
the theory, and Planck volumes are elementary. If ˛ 1, then l lPl and the scale
at which discreteness effects become apparent is larger than the Planck size.
A second assumption is that this set of points has an ordering, which reproduces
the causal structure of continuous Lorentzian geometries whenever these can be
recovered after some coarse-graining or large-scale approximation on the set. The
most general set will not possess a macroscopic light-cone continuum structure,
which will therefore be a byproduct of the fundamental ordering. The appealing
economy of the recipe “order and number” has the potential of unifying topology,
geometry and causal structure.
11.6.1.1 Kinematics and Poisson Statistics
These properties can be made rigorous and embodied, with no reference to any
embedding, in a causal set (or causet) C, defined by a partial-order relation and
the following rules:
1. Reflexivity: for all x 2 C, x x.
2. Transitivity: for all x; y; z 2 C, x y z ) x z.
3. Anti-symmetry (or non-circularity, or acyclicity): for all x; y 2 C, x y x )
x D y. In the continuum approximation, this property forbids closed time-like
curves.
4. Local finiteness: for all x; y 2 C, Card.Œx; y/ < 1, where Card is the cardinality
(the number of points) of the “interval” set Œx; y :D fz 2 C j x z yg, i.e., all
the ordered points comprised between x and y.
In particular, whenever a Lorentzian-manifold limit exists, the relation x y
corresponds to the continuum statement x 2 J .y/, i.e., the point x belongs to the
causal past of y (Sect. 6.1.1). The partial order is therefore the discrete equivalent
of the time ordering. The order is said to be partial because not all points are related
by . In the continuum, two such unrelated points are separated by a space-like
vector.
Before considering any dynamics, one should understand the kinematical prop-
erties of causets, namely, the circumstances under which concepts such as length,
geodesic, spacetime dimensionality and continuum can emerge. The latter is often
studied by sprinkling (i.e., randomly generating) a finite set of points in the portion
of a Lorentzian manifold M and then imposing an order relation reproducing the
light-cone structure of the given continuum geometry. One then attempts to classify
the type of causal sets compatible with this embedding. The intuitive conjecture,
by now extensively verified, underlying this procedure is that a causet C admits a
continuum approximation M if it can be generated by sprinkling on M.
In order to preserve both Lorentz invariance and the point-volume relation
(11.79), the random sprinkling must be done via a Poisson process. The probability
that the random variable N take the value n is given by the Poisson distribution of n
points into a spacetime volume V:
1
P.N D n/ D .V/n eV ; (11.80)
nŠ
where is the sprinkling density. By this definition, the expected value of N is
thus N D V and its variance is N2 D hN 2 i D V D N. Consequently, any
determination of N will carry a statistical uncertainty N given by the standard
deviation
p
p V
N :D N D N ; VD D=2
: (11.81)
˛mPl
Such fluctuations, often called Poisson noise, are purely kinematical and do not
entail any dynamics. Thus, the relation (11.79) is not an exact equality but a
statistical one, valid up to fluctuations of the size (11.81).
The Poisson distribution (11.80) is manifestly covariant since it depends only on
the D-volume element V and on the sprinkling density . This does not ensure that
the resulting causet be covariant but, for a Minkowski embedding, it can be proven
that individual realizations of the Poisson process are indeed Lorentz invariant
[164]. A Lorentz transformation on a causet would change the relative position of
the elements but not the properties of their statistical distribution. Also, no finite-
valency graph (i.e., such that each node has a finite number of neighbors) can
be associated with a sprinkling of Minkowski spacetime consistent with Lorentz
invariance. A regular lattice or the graphs used in spin foams and CDT, for instance,
do not satisfy this property. The intuitive reason is that a regular distribution of
points can be boosted to a distribution of over-dense and under-dense regions,
thus changing the statistics (e.g., [161]). Therefore, the double requirement of
discreteness and Lorentz invariance seems to be inextricably related to a random
structure at the fundamental level.
The calculation of geodesic distances [152] and of other geometric indicators
such as the spacetime dimension [151, 178, 181], the Riemann curvature [173,
174, 177] and topology [163, 170] clarifies the relations between discrete ordered
structures and macroscopic faithful embeddings. In general, these quantities depend
on the number N of points in a given volume V and on the number R of relations
within the interval Œx; y between any two points in the set.
11.6.1.2 Dynamics
The dynamics of causets is under construction through different approaches. In one

of them (classical sequential growth), classical dynamics is regarded as a process of
stochastic growth of elements of the causet from existing elements [154, 162]. There
is no actual direction of growth apart from that determined by the order relation ;
however, one can take N as a monotonically growing “time” variable along which
the birth process unfolds. The details of the process determine the probability of
each transition from a set of N elements to one with NC1 elements and, in particular,
the probability of forming specific causal sets for a given N. Each new element can
either be in the future of an existing point or be space-like, that is, not related by
; no elements can be added to the past. Matter modes can emerge in the effective
action without introducing them by hand.
The choice of action S is related to the recovery of the gravity+matter continuum
limit SŒC ! Seff Œg ; . Call link the relation y x without other elements
in between. For a Minkowski embedding, the probability to find n points y in
a region A with volume V sharing a link with a given point x near A is finite
due to discreteness. By Lorentz invariance, the same statement holds for any
other region A0 (disjoint from A) obtained by a sufficiently large boost on A, so
that x actually has an infinite number of links, many of which at large spatial
distances. The theory is therefore non-local at the kinematical level [150, 151]. If
a continuum limit exists, then higher-order derivative terms in the dynamics are
suppressed by powers of the discreteness fundamental length scale, L D a0 l4 C
a1 l2 2 2
RCa2 R CO.l / (the ai ’s are dimensionless constants), and the effective action
Seff is approximately local. General covariance then guarantees that the Einstein–
Hilbert action is recovered [173, 181]. Evidence has been collected that kinematical
Laplace–Beltrami operators can be constructed so that the effective dynamics is
approximately local at sufficiently large scales, where the transition scale between
the non-local and local regimes can be much larger than l [173, 174, 177]. These
operators constitute the local limit of a family of highly non-local operators, which
can capture salient features of causet dynamics in the continuum approximation but
in regimes well below the non-locality length scale [179, 180].
At the quantum level, transition probabilities are replaced by transition ampli-
tudes. The models of classical sequential growth are a somewhat intermediate step
between classical and quantum dynamics, since their intrinsic stochastic character is
akin to a quantum process; their relation to quantum causal sets is similar to the one
between Brownian motion and quantum mechanics [186]. Since time (and emergent
Lorentz invariance) is discrete and inevitably included in the picture via the ordered
causal structure, the canonical formalism is perhaps less convenient than the study
of the local properties of amplitudes in a sum-over-histories approach. In this case,
the dynamics is governed by the path integral
X
ZŒN D eiSŒC ; (11.82)
C2˝
where ˝ is a sample of causets suitably chosen [154, 159, 175] and SŒC is an action
dependent on the details of the causet C, in particular N and R. In the classical
growth model of [154], N acts as a stochastic time parameter, so that one does not
expect to extend the sum in (11.82) to causets with different N. This is the reason
why we made the N dependence explicit in the left-hand side.
11.6.2 Cosmology
Without entering into the details of the dynamics, classical sequential growth shows
evidence of some cosmology, where a large homogeneous and isotropic universe
emerges after a period of cycles of expansion and contraction with increasing
maximal spatial volumes [155–157, 172]. The initial configuration of each cycle
is a single causet element (called “post” in combinatorial language). It corresponds
to what we would call, in a continuum geometry, a big bang or big crunch, since
the whole universe is constituted by just one spacetime point. Each post is followed
by an exponential volume expansion resembling a de Sitter phase. After each cycle,
the dynamics repeats itself but with different values of the fundamental constants.
Since there is no infinity associated with a post, the big-bang problem somehow
disappears altogether. Such results have not been developed to the stage of a full,
realistic cosmological scenario. Here we concentrate on two much more generic
predictions of causets about, respectively, the production of B-modes in the CMB
polarization [161, 169, 228] and the smallness of [153, 158, 176, 181].
11.6.2.1 Generation of B-Modes
Since Lorentz invariance is preserved in the Minkowski embedding, the effects of

microscopic discreteness and stochastic fluctuations can be encoded in Lorentz-
invariant dynamical equations. In particular, one can study the diffusion of point
particles, for instance the Brownian motion in a medium or the propagation of pho-
tons across cosmological scales. Consider the latter. The small random fluctuations
in the momentum of the particle are reflected, in the continuum approximation, into
a drift term in the diffusion equation of the particle in momentum space [161, 169].
For a polarized photon on a FLRW background, this drift or friction term depends
on a coefficient d and on the first derivative of the azimuthal angle parametrizing
the polarization space. It violates parity explicitly and causes a rotation of the
polarization vector during the propagation of the photon by an angle D 2td=,
where is the photon frequency. Therefore, B-modes are generated in the CMB
from gradient E-modes and non-vanishing cross-spectra, even in the absence of
primordial gravitational waves [228]:
C`BB / CÈE sin2 .2/ ; C`TB / C`TE sin.2/ ; CÈB / CÈE sin.4/ : (11.83)
When d D 0, one has C`TB D CÈB D C`BB D 0. The model of diffusion discussed
here has a wider application than causal sets and holds in any Lorentz-invariant
theory with an uncertainty in the spacetime structure at small scales. Observations
on the CMB polarization spectra can in principle constrain the parameter space of
causets in the continuum approximation. In Sect. 5.9.2, we saw other examples of
Lagrangian models where non-vanishing TB and EB spectra can be originated [229].
11.6.2.2 Cosmological Constant
To address the problem in causet theory, one first argues that an O.1/ cosmolog-
ical constant in the effective dynamics is compensated by some mechanism, so that
D 0 in average. Since the observable patch of the universe is made of a random
distribution of points n with mean N, this compensation does not take place exactly
but within some statistical accuracy. The observed cosmological constant is given
by this effect.
To begin with, one notes that the kinematical predetermination of N in the causet
approach is closely akin to the volume fixing of unimodular gravity discussed in
Sect. 7.6.3. In fact, in the quantum theory (11.82) the sum is taken over a family
of causets with fixed N which, in the continuum or macroscopic approximation,
corresponds to a partition function over D-dimensional geometries with a given
volume V. Classically and in the continuum approximation, it is clear that V
and are conjugate quantities:
R they appear in the effective action via the usual
p
combination V= 2 D dD x g.= 2 /, where 2 / m2D Pl
. Since V is fixed,
the parameter in front of it is a Lagrange multiplier which can be chosen arbitrarily
to compensate a bare cosmological constant. However, we also have to take into
account the statistical fluctuations of these variables. Just as time and energy are
conjugate quantities in ordinary quantum mechanics, so are the volume and the
cosmological constant in a statistical sense and their fluctuations obey an uncertainty
principle:
V & 2 : (11.84)
If these were just ordinary quantum fluctuations, we would be unable to determine

either their magnitude or the mean hi. However, they are governed by the Poisson
distribution (11.80), from which we know both the mean and the variance for V.
From the combination V in the action, one can interpret as the action of ˛ 2
fundamental elements; supposing each contribution is independent and fluctuates in
sign by ˙˛ 2 , then hi is expected to be zero or very close to zero [158]:
hi D 0 : (11.85)
A further suggestion in favour of (11.85) is that only D 0 configurations are

stable against fluctuations of the causal set not corresponding to manifolds [153].
To summarize: (i) discreteness and Lorentz invariance require a Poisson sprin-
kling; (ii) this is assumed also in curved scenarios; (iii) the statistical uncertainty
and the volume fixing imply the uncertainty relation (11.84), so that large Poisson
fluctuations of the volume are compensated by small fluctuations of the cosmolog-
ical constant and vice versa; (iv) from (11.81) and (11.84) and assuming (11.85),
when the uncertainty principle is minimized one has the estimate
˛ 2 mD
D=2
1 ˛mPl
D D p D p Pl : (11.86)
2 V V N
In general, one can choose V as the volume of the past light cone of any
representative point on the hypersurface for which one wants to calculate the value
of V. In a cosmological context, the total D-volume of the observable universe
is, at any given time t since the big bang, V tH .D1/ H D , so that
˛.mPl H/D=2 . Only in D D 4 dimensions is, at any given time, of the
same order of magnitude of the critical density,
˛m2Pl H 2 : (11.87)
To explain dark energy, ˛ should be in the range ˛ D O.102 / – O.1/. In particular,

according to (2.11) and (2.13), if ˛ D O.1/ we get the very large number today
(t D t0 )
N0 m4Pl V0 10244 : (11.88)
From (11.86), one thus obtains the same order of magnitude as in (2.118),
10122 m4Pl .
Like asymptotic safety (Sect. 11.2.2), this scenario resembles phenomenological
models where .t/ is time dependent but without kinetic term. Equation (11.87) is
the basis for the solution of the cosmological constant problems in causet theory.
The old problem is solved because acquires the observed value today, while
the coincidence problem can be potentially addressed because (11.87) embodies an
ever-present component with tracking behaviour, as numerical solutions indicate
[158, 176]. Note also that (11.87) is valid only in four dimensions; the resolution
of the cosmological constant problem is therefore tightly related to the number
of topological dimensions of the embedding. This is one of the many instances
in theoretical physics where one question (Why is the cosmological constant so
small?) is answered by replacing it by yet another question of different type (Why
four dimensions?).
However, some fine tuning lurks behind the causet solution of the problem.
Fluctuations of the ever-present cosmological constant cannot be as large as ˛ D
O.102 / – O.1/, lest they shift the main milestone epochs of cosmic evolution
(matter-radiation equality, decoupling, and so on) to unacceptable values. Such
shifts would displace the peak positions of the CMB and galaxy-clustering power
spectra away from their observed values. In particular, the large-scale CMB
spectrum constrains the parameter ˛ to be very small, ˛ < 107 – 105 depending
on the assumptions [166, 167]. One can get a rough idea of such a bound by noting
that the inhomogeneous fluctuation of Newton’s potential ˚ is proportional to the
one of at some cosmological scale k via the Poisson equation ˚ k2 ı .
Taking into account that these fluctuations are of order ˛ times the critical density,
the temperature fluctuation at last scattering is of order
T ˚ ı
˛; (11.89)
T ˚ tot
11.7 Non-commutative Spacetimes 591
which implies that ˛ . 105 [166]. This range is insufficient to explain late-time
acceleration.
In order to make the considerations in favour of a viable solution of the
problem more robust and to better understand those against it, it should be made
possible to follow the cosmic evolution through all its main stages. The argument
leading to (11.87) should be embedded in a more complete framework where
dynamics were under control. The mechanism by which D 0 has to be worked out
in detail; the quantum partition function (7.104) of unimodular gravity gives some
support to a zero average cosmological constant, but results obtained in Euclidean
signature are notoriously delicate. The role of matter should also be taken into
account. It is not yet known whether emergent matter degrees of freedom [154] from
pure “order and number” are viable from the point of view of both cosmological and
particle physics. Furthermore, the stochastic process modeled by a constant ˛ could
be modified by dynamics. An effectively time-varying ˛.t/ could accommodate
both the need of small fluctuations at early times and of larger ones at late times.
This resolution of the fine-tuning problem would not be dissimilar from the lattice-
refinement picture in loop quantum cosmology, where the number of elementary
nodes per unit volume changes dynamically. Its phenomenological origin might as
well be the same, namely, the role of inhomogeneities in the full dynamics. All these
elements may influence the results (11.86), (11.87) and (11.88) in ways which are
presently difficult to assess.
11.7 Non-commutative Spacetimes
11.7.1 Framework
Divergences in ordinary quantum field theory may be cured if the structure of

spacetime undergoes suitable modifications below some physical scale l? often
assumed to be the Planck length lPl . On a smooth continuous manifold, one
can identify classical events by points, which means that the uncertainty in the
determination of time and space coordinates is exactly zero, x x D 0. In
quantum mechanics this uncertainty does not vanish and acquires a lower bound,
given by Heisenberg’s principle. The idea of non-commutativity is to promote this
feature of quantum mechanics to an intrinsic characteristic of spacetime, where
the coordinates X are the elements of a Lie algebra g and quantum fields can be
defined to live thereon [230–235]. The uncertainty X X > 0 is expressed by
the commutation relation
ŒX ; X D i
.X/ ; (11.90)
where is an anti-symmetric tensor of dimension (length)2 . A special case is

that of a linear algebra ŒX ; X D C X , for some structure constants C . In
particular, -Minkowski spacetime is defined by the algebra ŒX 0 ; X ˛ D il2Pl X ˛ ,

ŒX ˛ ; X ˇ D 0, for ˛; ˇ D 1; 2; 3.
Momentum space is made of the elements of the Lie group G dual to g. Since
the measure on G will be in general non-trivial, non-commutative spacetimes are
tightly related to a momentum space endowed with curvature and a non-standard
dispersion relation.
The most conservative scenario where this structure makes its appearance is
classical .2 C 1/-dimensional Einstein gravity coupled to point particles [236, 237].
In three dimensions, gravity is topological and particles are introduced as topolog-
ical defects on the spacetime manifold [238]. The particle phase space is deeply
modified by the curved background and it turns out that momenta are group valued
and the particle spacetime coordinates have non-zero Poisson bracket.
The momentum space of GFT is also of this sort (see (11.51)) and, in fact, a
non-commutative theory of matter fields emerges from the GFT action [117–120].
This model is obtained by expanding '.g/ D ' .0/ .g/ C .g/ in terms of a classical
solution ' .0/ describing flat building blocks and a perturbation . Plugging this
decomposition into the GFT action, the resulting effective action for is a non-
commutative field theory invariant under a deformation of the Poincaré group.
It is no surprise, then, that a class of spin foams reproduces the Feynman-graph
amplitudes of the same non-commutative field theory [222, 239, 240].
Non-commutativity shows up also in a variety of other contexts, from low-energy
phenomenology of quantum gravity [241, 242] and heuristic approaches to quantum
spacetime [243, 244] to deformations of relativistic symmetries [245, 246] and
multi-scale spacetimes [247, 248]. In string theory, in the presence of a background
Neveu–Schwarz field the end-points of open strings obey the non-commutative
algebra (11.90) [249–252], where is assumed to be constant in flat spacetime
and covariantly constant in the presence of gravity. Since the closed-string sector and
hence gravity is unaffected, the low-energy effective action is non-commutative only
in matter fields. Independent field-theory models of a non-commutative gravitational
action based on (11.90) can be constructed [253–256]. The same type of algebra
(11.90) arises in compactifications of M-theory [257].
The treatment of field theories on non-commutative spacetimes is greatly
simplified by the introduction of a Weyl map [258–260] putting in correspondence
functions of the non-commuting variables X with functions of the commuting
“classical” coordinates x . The map is invertible (i.e., the correspondence is one-to-
one) provided the space of classical functions is equipped with a non-commutative
?-product. It is customary to define the Weyl map on plane waves, the basic building
blocks of a field theory, and then to construct the functions via Fourier transform.
The ?-product can also be defined explicitly: on a curved background [261],
C1
X .i=2/n 1 1 n n
. f ? g/.x/ :D .x/ : : : .x/.r1 : : : rn f /.r1 : : : rn g/ ;
nD0
nŠ
(11.91)
for any f and g in the extended space of classical functions. When is constant and
spacetime is flat, (11.91) is the associative Moyal product [262, 263]

i @ @ ˇ
.f g/.x/ :D exp
Œf .x/g.y/ˇyDx : (11.92)
2 @x @y
In general, however, the product (11.91) is non-associative, . f ? g/ ? h ¤ f ? .g ?

h/. It is also an example of operator with an infinite number of derivatives; non-
commutativity and non-locality are closely connected. When is constant, the
algebra (11.90) of non-commutative coordinates translates into
Œx ; x ? :D x ? x x ? x D i
;
D const ; (11.93)
corresponding to the uncertainty relation
x x > 12 j
j: (11.94)
The action of a scalar field theory on a curved background and with a monomial
potential is
Z
4 p 1
SD d x g .@ / ? .@ / C ? ? ; (11.95)
2
where an operator ordering has been chosen such that there is no ?-product between
p
the measure weight g and the Lagrangian density.
11.7.2 Cosmology
In cosmological applications of the theory, new effects show only in inhomoge-

neous configurations, since the homogeneous background dynamics is unchanged.
Therefore, non-commutativity can affect inflationary perturbations [264–266] and
astrophysics [267].10
10
For WDW quantum cosmology with non-commutative mini-superspace coordinates or non-
commutative phase-space variables, see [268] and [269], respectively. In the second case, where
not only mini-superspace coordinates but also their conjugate momenta obey a non-commutative
algebra, also black-hole backgrounds have been studied and the wave-function found to vanish at
the central singularity [270–272].
11.7.2.1 Non-commutative Inflation Without Inflaton
In one particular non-commutative model, early-time acceleration without a slow-

rolling scalar field is obtained from semi-classical Einstein equations where the
expectation value of the matter energy-momentum tensor is calculated on coherent
states [273, 274].
Another, more developed scenario is based on a modified dispersion relation
.k0 /2 C f .k0 /jkj2 D 0 inspired by non-commutative geometry [275–278], (5.199)
with
f .k0 / D Œ1 C .˛k0 / 2 ; (11.96)
where ˛ and are real parameters. Here, the content of the universe is pure radiation
and cosmological perturbations are generated by thermal rather than quantum
fluctuations. This mechanism of seeding of cosmic structures has been considered
also in standard cosmology [279–281], but in that case it is difficult to produce
inflation in the absence of a slow-rolling field or of quantum-gravity modifications
to the dynamics [282]. On the other hand, radiation in non-commutative spacetime
can sustain an accelerated expansion thanks to the deformed dispersion relation.
The spectrum thus produced is almost scale invariant with a slight red tilt and can
be compatible with observations.
11.7.2.2 Non-commutative Inflation with Inflaton
Another class of models has a more direct contact with non-commutative field
theories. Here the inflationary mechanism is fairly standard, as it is driven by an
ad hoc scalar in slow roll (scalars may appear in quite specific non-commutative
geometries [283]). Expanding (11.91) and (11.95) to leading order in , one obtains
effective models with small non-commutativity where standard slow-roll inflation
takes place [284–286]. When non-commutative effects become appreciable, (11.93)
may lead to large anisotropies (the anti-symmetric spatial tensor ˛ˇ selects
preferred directions) and non-Gaussianities [261, 287].
To make the problem tractable, the components of are usually assumed to be
constant in a particular coordinate frame. Two common choices are the comoving
frame (Œx ; x ? D i D const, with x0 D t) [287–290] and the physical or proper

frame (Œxp ; xp ? D i p D const, with x0p D t and x˛p D a.t/ x˛ ) [261, 287]. The
relation between the matrix elements of the two frames is dictated by the relative
˛ˇ
scaling of proper and comoving spatial coordinates: p0˛ D a 0˛ , p D a2 ˛ˇ .
Taking (11.93) to hold for comoving coordinates, the scalar spectrum reads [288–
290]
Ps. / .k/ D Ps.0/ Œˇ cosh.H 0˛

k˛ / C i.1 ˇ/ sinh.H 0˛
k˛ / ; (11.97)
.0/
where Ps is the standard commutative spectrum and 0 < ˇ 6 1 is a free parameter.
The non-linear parameter fNL is modified by similar effects as well as by the
contribution of thepspace-space part ˛ˇ . CMB data constrain the non-commutative
length scale to be < 1019 m, corresponding to a lower bound on the energy of
about 10 TeV [291].
Another possibility is to regard the coordinates in (11.94)p to be the proper
ones and consider a simplified model where ˛ˇ D 0 and 0˛ D l . Then,
?
the spacetime uncertainty relation reduces to t xp D Q x > l2? =2 for all
spatial directions, where d Q :D a.t/dt and dxp D a.t/dx. The algebra of classical
coordinates (11.93) is then [292]
Œ Q ; x? D il2? : (11.98)
In this setting (Brandenberger–Ho model), inflation driven by a scalar field

has been considered [292–306]. The observed spectra can be produced in two
different regimes. If the cosmological energy scale at which a perturbation with
physical wave-number p D k=a crosses the horizon is much smaller than the non-
commutativity scale E? D 1=l? , l? p D l? H 1, then non-commutative effects
are soft. During inflation, Q ' a=H, so that at horizon crossing l? k D l? a H '
.H l? /2 Q =l? Q =l? . These modes are contained in a UV region characterized by
perturbations generated inside the Hubble horizon.
If the observed power spectra lie in this regime, one gets corrections to the
standard observables suppressed as .l? H/4 . Omitting scalar and tensor subscripts,
the power spectrum, spectral index and running read
P . / D Ps.0/ ˙ 2 ; n. / D n.0/ C ; ˛. /
' ˛ .0/ C O. 2 / ; (11.99)
with
˙ 2 ' 1 bıUV ; ' 4b ıUV ; ıUV :D .l? H/4 ; (11.100)
where the numerical value of the coefficient b D O.1/ > 0 depends on the operator
ordering in the action [292, 301]. These corrections are similar to those of WDW
quantum cosmology (Sect. 10.2.5), but more strongly suppressed. Moreover, their
sign is fixed: power spectra are suppressed at large scales and the spectral index is
blue tilted.
On the other hand, in the IR regime l? H 1 the wave modes are generated
outside the horizon. Since they are frozen until they enter the horizon for the first
time, their magnitude depends on the conformal time 0 when they were generated.
This corresponds to the time when the spacetime uncertainty relation is saturated
and quantum fluctuations start out with their vacuum amplitude. Then, one can show
that the IR region is characterized by l? k ' Q0 =l? at horizon crossing. Expanding
the exact spectrum in the small parameter .l? H/2 , the observables (11.99) now
have
1=2
˙ 2 ' ıIR
3
or ' ıIR ; D O.1/ ; ıIR :D ıUV D .l? H/2 :
(11.101)
This regime is somewhat speculative since the Hubble radius is smaller than l? .
Yet, it brings forward some non-trivial imprint of non-commutativity. While the UV
corrections are akin to those in WDW quantum cosmology, where the corrections
are of the “naive” kind .H=E? /n and suppressed by the ratio of the Hubble
parameter during inflation with respect to a fundamental energy scale E? D 1=l? ,
the IR corrections are just the opposite. This situation is loosely remindful of
the relative status of inverse-volume and holonomy corrections, expressed by the
heuristic relation (10.152).
At intermediate scales, the two typespof spectra can be interpolated
p and they are
of the form (11.99) with ˙ 2 ' 1 c ıUV and ' 2c ıUV , for some model-
dependent c D O.1/ > 0 [293, 301].
The characteristic scale signaling the transition between the UV and the IR
regime is not fixed by the model, which is rather phenomenological. Therefore, for
the purpose of placing observational constraints one has the freedom to tune the free
parameters so that to describe the whole power spectrum by only one of the three
regimes (UV, intermediate or IR). Both in the UV [259, 294, 297, 299, 300, 304, 305]
and intermediate [293] regime, the effect of new physics is typically small and
can be rendered compatible with observations for realistic values of the non-
commutativity scale l? . Assuming instead the strongly non-commutative IR regime,
the blue tilt in the scalar index is large and such models are much more severely
constrained, to the point of being almost ruled out for the simplest choices of scalar-
field potential [303, 306].
11.8 Non-local Gravity
11.8.1 Non-locality
We saw in Sect. 7.5.1 that the presence of derivative operators of order higher than
two in a Lagrangian often leads to ghosts at the perturbative level. Consider the
instance of a massless scalar field in Minkowski spacetime with action
Z
1
Sloc D D
d x .1 M / V. / ; M WD ; (11.102)
2 M2
where M 2 > 0 is a mass scale. The classical equation of motion is .1 M /

V; D 0. The propagator of the free theory (V D 0) obeys the Green equation
. 2 =M 2 /G.x x0 / D ı.x x0 / which, transformed to momentum space, gives
11.8 Non-local Gravity 597
Q p2 / D 1, i.e.,
.p2 p4 =M 2 /G.
Q p2 / D 1 C
G.
1
: (11.103)
p2 p2 C M 2
Thus, the action (11.102) actually encodes two degrees of freedom, a massless
ordinary scalar (residue 1) and a massive ghost (residue C1) with mass M. Higher-
order derivatives do not inevitably give rise to ghosts: their presence can be easily
avoided in Horndeski theory (Sect. 7.5.1) as well as in higher-spin models [307].
However, outside these special classes of dynamics one can reasonably expect to
encounter instabilities whenever one has derivatives higher than second order.
In quantum gravity, the problem depends on the background one chooses to
expand the metric in graviton modes. In a cosmological setting, there exist non-
trivial regions in the parameter space where ghosts disappear. However, since
ghosts persist on the Minkowski background where particle field theory is usually
developed, it is important and perhaps more elegant to find a general mechanism
preserving unitarity. The situation can radically change when non-perturbative
effects are taken into account. For example, the unitarity problem in higher-order
gravity (in particular, Stelle’s theory) is resolved in a suitable strong-coupling limit
in the presence of fermionic matter [308–311].
Another possibility is the following. Introducing operators n in the example
(11.102), one soon realizes that the higher the order of derivatives, the larger the
number of degrees of freedom [312–314]. At least one ghost mode, sometimes
called Ostrogradski instability [312, 315], is ever persistent. But what happens when
the number of derivatives is infinite? These theories are called non-local or non-
polynomial.
In systems defined in continuous spacetimes, non-locality is the appearance of
an infinite number of derivatives
R in the kinetic terms f .@/ of fields or, which is
equivalent, by interactions dx0 .x/F.x x0 / .x0 / dependent on non-coincident
points. In fact,
Z Z
0 0 0
dx dx .x/F.x x / .x / D dx dz .x/F.z/ .x C z/
Z Z
D dx dz .x/F.z/ dp eip.xCz/ Q . p/
Z Z
D dx dz .x/F.z/ ez@x dp eipx Q . p/
Z Z
D dx .x/ dz F.z/ e z@x
.x/
Z
D: dx .x/f .@x / .x/ : (11.104)
Historically, non-locality was first found in the interactions of quantum field theory
[316–335].
Consider now the action
Z
1
Snon-loc D dD x Œ We .M / V. / ; (11.105)
2
where
We .M / :D eM (11.106)
is called form factor and V is a local interaction. A field redefinition
Q :D eM =2 (11.107)
cannot reduce the action to a local one unless V D 0. The free propagator of is
now
p2 =M2
Q p2 / D e
G. : (11.108)
p2
Adopting the series definition of the operator eM ,
C1
X .M /n
eM :D ; (11.109)
nD0
nŠ
one can truncate up to order n D 1 and find out that, in this limit, (11.102) is an
approximation of (11.105), Snon-loc D Sloc C O.3 /. However, higher-order and
non-local theories are physically quite distinct. The main difference with respect to
the higher-order propagator (11.103) is that the exponential in (11.108) is an entire
function which does not introduce extra poles. The spectrum of the model (11.105)
is the same as in the local field theory with M 2 D 0 (this is obvious from the field
redefinition (11.107), which trivializes the model in the free limit V ! 0) but, if
M is real, the propagator (11.108) vanishes much more rapidly in the UV in the
Euclidean limit p2 ! p2E ! C1. The problem, in general, is that solutions to
non-local equations of motion cannot be expressed via series representations such
as (11.109), unless certain restrictive conditions (for instance, slow variation of the
fields or convergence of perturbative series on the functional space of solutions)
are satisfied. It is therefore clear that, to solve this type of non-local models in
the presence of non-trivial interactions, a perturbative truncation of the operators
is inadequate and one should employ non-perturbative techniques.
Generally speaking, non-local operators are troublesome also for other reasons.
On one hand, a particle interpretation of a quantum field theory may even be absent,
due to the replacement of poles in the propagators with branch cuts for certain
operators such as ˛ , where ˛ is non-integer [336, 337]. Moreover, causality may

be violated at microscopic scales [319]. On the other hand, the Cauchy problem
can be ill defined or highly non-standard [319, 321]. It entails an infinite number of
initial conditions .0/, P .0/, R .0/, . . . , representing an infinite number of degrees
of freedom. Since the Taylor expansion of .t/ around t D 0 is given by the full
set of initial conditions, specifying the Cauchy problem would be tantamount to
knowing the solution itself, if analytic [338]. This makes it very difficult to find
analytic solutions to the equations of motion, even on Minkowski spacetime.
Fortunately, the exponential operator (11.106) and its higher-derivative gener-
alization We Œ.M /l with l 2 NC are under much greater control than other
non-local operators. In these cases, the diffusion-equation method is available to
find analytic solutions [339–346] which are well defined when the perturbative
expansion (11.109) is not [340]. The Cauchy problem can be rendered meaningful,
both in the free theory [347] and in the presence of interactions, by transferring
the infinite number of initial conditions .0/, P .0/, R .0/, . . . into a finite number
of initial conditions for an auxiliary scalar field living in an extended spacetime
with a fictitious extra coordinate [341]. Causality, too, is respected in theories with
exponential operators, both at the micro- and the macroscopic level [333, 348].
Exponential non-locality arises in string field theory, as we shall discuss in
Sect. 13.7.4.
The non-locality (11.106) is also responsible for a feature of interest for gravity:
asymptotic freedom, namely the fact that the theory becomes weakly coupled at
short distances. If the toy model (11.105) schematically represented the action for
the linearized graviton modes, the Fourier transform of the spatial counterpart of
(11.108) would give the Newtonian potential ˚.r/. While at large distances ˚.r/
1=r in D 1 D 3 spatial dimensions, in the limit p ! 1 the potential does not
diverge but reaches a constant value determined by the UV cutoff r0 1=M, ˚.r/
1=r0 [349]. Notice that ˚.r/ 0 in asymptotic safety when r ! 0, as we saw in
(8.17): asymptotic safety is therefore a stronger condition than asymptotic freedom.
11.8.2 Framework
The strong damping in the UV of the propagator (11.108) after Wick rotation
suggests, together with other cumulative evidence, that theories with exponential
non-locality have interesting renormalization properties. After early studies of
quantum scalar field theories [327, 331–333, 350] (see also [351]) and gauge and
gravitational theories [352–359], in recent years there has been a resurgence of
interest in non-local classical and quantum gravity [348, 349, 360–369]. A non-
local theory of gravity aims at fulfilling a synthesis of minimal requirements: (i)
spacetime is a continuum where Lorentz invariance is preserved at all scales; (ii)
classical local (super)gravity should be a good approximation at low energy; (iii) the
theory has to be perturbatively super-renormalizable or finite at the quantum level;
(iv) the model has to be unitary and ghost free, without extra degrees of freedom in
addition to those present in the classical theory; (v) typical classical solutions must
be singularity-free.
To meet these demands, one considers a general action combining curvature
tensors, covariant derivatives of the curvature tensors and non-polynomial terms
[363, 370]:
X
NC2 Z
p
Sg D ˛2n M D2n dD x g O2n .@ g / C SNP
nD0
Z (
1 p X
N
D 2 dD x g R 2 an R.M /n R C bn R .M /n R
2 nD0

R F2 .M / R RF0 .M /R C O.R3 / : (11.110)
Here, M is some characteristic mass-energy scale and O2n .@ g / denotes general-
covariant operators containing 2n derivatives of the metric g , while SNP is a non-
polynomial action defined in terms of two entire functions F2;0 which we will fix
shortly [348]. The couplings and the non-local functions of the theory have the
following dimensions in mass units: Œan D Œbn D 2, Œ 2 D 2 D, ŒF2 D
ŒF0 D 2.
The maximal number of derivatives in the local part of the action is 2N C 4.
From the discussion of the local theory (7.79) in Sect. 8.2, renormalizability sets
2N C4 > D. To avoid fractional powers of the Laplace–Beltrami operator (which
do not even admit a series representation), we take 2N C 4 D D in even dimensions
and 2N C 4 D D C 1 in odd dimensions. For N > 0 and n > 2, only the operators
R n2 R , Rn2 R and R n2 R contribute to the graviton propagator
(with 2n derivatives in total) [370], but using the Bianchi and Ricci identities the
third operator can be eliminated. The last line of (11.110) includes only these
leading-order terms, with the redefinitions D ˛0 2 M D O0 , 1=.2 2/ D ˛2 M D2 ,
and so on.
The entire functions F0;2 .z/ in (11.110) are (here z :D M )
PN
W.z/1 1 M 2 z bQ n zn
F2 .z/ D nD0
; (11.111)
M2 z
PN
W.z/1 1 C 2M 2 z aQ n zn
F0 .z/ D nD0
; (11.112)
2M 2 z
for general parameters aQ n and bQ n , where W.z/ is a non-polynomial entire function

without zeros in the whole complex plane.
The Lagrangian can be expanded at second order in the graviton fluctuation
(7.101) around the Minkowski background. Adding a gauge-fixing term LGF due
to a local gauge symmetry under infinitesimal coordinate transformations [371], the
linearized gauge-fixed Lagrangian reads Llin C LGF D h O h =2. Inverting
the operator O [372], one finds the following two-point function in the harmonic
gauge (3.18) and in momentum space:
P.2/ P.0/
O1 . p/ D C ; (11.113a)
p2 FN 2 . p2 / .D 2/p2 FN 0 . p2 /
where
FN 2 . p2 / D 1 C p2 ˇ. p2 / ; (11.113b)
XN 2 n 2
2 p p
ˇ. p / D bn 2
C F2 ; (11.113c)
nD0
M M2
Dˇ. p2 / C 4.D 1/˛. p2 /

FN 0 . p2 / D 1 p2 ; (11.113d)
D2
XN 2 n 2
2 p p
˛. p / D an 2
C F0 : (11.113e)
nD0
M M2
In (11.113a), we omitted the tensorial indices of the operator O1 and of the
.2/
projectors P.0/ and P.2/ , defined as [372, 373] P . p/ D . C /=2
.0/ 2
=.D 1/, P . p/ D =.D 1/, D p p =p .
We now assume that the theory is renormalized at some scale 0 . Setting aQ n D
an .0 / and bQ n D bn .0 /, the bare propagator only possesses the gauge-invariant,
physical massless spin-2 graviton pole, and FN2 D FN 0 D W 1 . Choosing another
renormalization scale, the bare propagator acquires poles which cancel with a shift
in the self-energy in the dressed propagator. Thus, (11.113a) reads

1 W. p2 =M 2 / .2/ P.0/
O . p/ D P : (11.114)
p2 D2
A non-local field theory is unitary and micro-causal provided the following

properties are satisfied by W.z/ [333]: (i) W.z/ is an entire analytic function in
the complex z-plane and it has a finite order of growth 1=2 6 < C1, i.e.,

9 b > 0; c > 0 such that jW.z/j 6 c eb jzj ; (ii) when Re.z/ ! C1 (p2 ! C1 or
2
pE ! C1), W.z/ decreases quite rapidly: limRe.z/!C1 jzjN jW.z/j D 0, 8 N > 0;
(iii) ŒW.z/ D W.z / and W.0/ D 1; (iv) the function W 1 .z/ is real and positive
on the real axis and it has no zeros on the whole complex plane jzj < C1. This
requirement implies that there are no gauge-invariant poles other than the transverse
massless physical graviton pole.
11.8.2.1 Super-Renormalizable and Finite Quantum Gravities?
We now specialize to a form factor satisfying the above properties [348]:
W.z/ D eH.z/ ; (11.115)
where H is a polynomial. In particular, we consider W.z/ D exp.zl /, where l 2

NC ; the case l D 1 corresponds to the exponential form factor (11.106). The high-
energy propagator takes the form O1 . p/ D expŒ. p2 =M 2 /l =p2 . The n-graviton
interaction has the same scaling, since it can be written schematically as L.n/
hn h expŒ. /l =M 2l h C : : : , where D @ @ and “: : : ” are other sub-
leading interaction terms. Placing an upper bound to the amplitude with L loops,
one finds heuristically
Z h iI h 2 2 l iV
2 2 l
A.L/ .dD p/L e. p =M / p2 e. p =M / p2
Z h iL1
2 2 l
D .dp/DL e. p =M / p2 : (11.116)
In the last step, we used the topological identity (8.10). The L-loops amplitude
is ultraviolet finite for L > 1 and it diverges at most as pD for L D 1. Only
one-loop divergences survive and, therefore, this theory is power-counting super-
renormalizable and unitary, as well as micro-causal [331–333, 348, 350, 352, 363,
369, 374]. A more rigorous power counting than the one of (11.116) can be found
in [369].
All of this applies to spacetimes of any dimension D > 3. In odd dimensions,
there are no counter-terms for pure gravity at the one-loop level in dimensional
regularization and the theory is finite; it stays so also when matter is added to fill up
the supergravity multiplet [375, 376]. One may then infer that the amplitudes with
an arbitrary number of loops are finite and all the beta functions vanish. In particular,
we can fix all the coefficients of the higher-curvature terms in (11.110) to zero, while
the couplings an ./ and bn ./ do not run with the energy: an ./ D aQ n D const,
bn ./ D bQ n D const. Using (11.111) and (11.112) for the exponential form factor
(11.106), the gravitational action (11.110) simplifies to
Z
p R 2
Sg D d x g
D
R F2 .M / R RF0 .M /R (11.117)
2 2
Z
1 p
D 2 dD x g R 2 C G ./ R ; (11.118)
2
where
eM 1
./ :D : (11.119)

The equations of motion are eM G C O.R2 / D 2 T . For generic form factors,
they can be found in [349] when F2 D 0 and in [377, 378] for F0 ¤ 0 ¤ F2
(following the calculation for higher-order actions of [379]).
A non-perturbative analysis of the renormalization group flow of the theory has
not been performed yet but the damping of the form factor already points to the fact
that this class of non-local gravitational theories is asymptotically free and the only
fixed point in the UV is Gaussian (relevant couplings go to zero in the ultraviolet).
This should be contrasted with asymptotically-safe gravity, where the couplings do
not vanish in the UV. The effective action of the two theories is quite different (the
latter, in particular, is local) and, therefore, it does not come as a surprise that their
UV behaviour and the way they resolve the infinities of perturbative Einstein gravity,
do not match.
11.8.3 Cosmology
In the simplest classical cosmological applications of the operator (11.106), gravity

is local and the only non-local content is a scalar field [340, 342, 345, 380–401].
When also the gravitational dynamics is dominated by exponential operators, the
big-bang [349, 361, 362, 377, 402–405] and black-hole [406] singularities are
apparently resolved, thanks to asymptotic freedom.11
The simplest means to find cosmological solutions is to require that curvature
terms are eigenstates of the form factor. If, for instance, the Ricci tensor R is
an eigenfunction of the Laplace–Beltrami operator, so will the Ricci scalar R in any
theory where the metric is covariantly constant. One is therefore interested in finding
profiles a.t/ such that
R D 1 R C 2 ; (11.120)
where 1;2 are constants. Two such typical profiles in D D 4 are

r
!
a.t/ D a cosh t ; (11.121a)
2

H1 2
a.t/ D a exp t ; (11.121b)
2
where ! and H1 are constants. From

p (2.72) and for zero intrinsic curvature K, R D
6.2H 2 C H/P D 3!f1 C Œtanh. !=2t/2 g, 1 D ! and 2 D 6! 2 in the first
case, while R D 6H1 .1 C 2H1 t2 /, 1 D 6H1 and 2 D 12H12 in the second. The
scale factors (11.121) are non-singular at t D 0, hence the big bang is replaced by
11
For other types of non-locality which include infrared modifications to gravity, see [407–429].
a classical bounce where a.0/ D a ¤ 0. The relation between ! or H1 and the

other parameters of the model (matter energy density and cosmological constant)
is determined by the equations of motion. The solutions (11.121) are, in general,
exact in vacuum and for certain classes of form factors F0;2 in (11.117),12 while
they can be regarded as a good approximation near the bounce in the presence of
matter fields.
In the specific example (11.118), one can find a profile aaf .t/ by non-perturbative
methods exploiting asymptotic freedom [405]. Near the bounce, this scale factor
takes the asymptotic form (11.121b), while at late times it matches the usual power
law aaf .t/ ' tp (2.93). The shape (11.121b) is rather insensitive to the type of matter
dominating at late times, which only determines the numerical value of H1 / p.
One can fit the solution aaf .t/ with the effective Friedmann equation
" ˇ #
22 2
H D eff :D 1 ; (11.122)
3 3
where D .a / is the critical energy density at which the bounce occurs and
ˇ D O.1/ is a real parameter. The exponent ˇ is determined by plugging the
profile aaf .t/ into (11.122) for a given energy density profile .aaf /. The effective
equation (11.122) is similar to (10.121) for homogeneous LQC in the absence of
inverse-volume corrections.
To solve the big-bang problem convincingly, one should show by a stability
analysis that bouncing solutions such as (11.121) do not rely on special initial
conditions; infinitely many solutions with a big bang (for instance, a power-law
scale factor) are, after all, known for certain non-local actions [361, 377]. This task
is still in progress but points towards a reassuring answer.
11.9 Comparison of Quantum-Gravity Models
The variety of approaches to quantum gravity and cosmology may confuse the
reader accustomed to the clear-cut answers of traditionally accepted physics. This
state of affairs is due partly to the lack of any recognized imprint of quantum gravity
in the observable world and partly to the difficulty in developing all these lines of
research to an adequate level of rigorousness and contact with phenomenology. It
is therefore helpful to ask whether some of these theories are physically equivalent
and, if not, in what they differ.
Just like LQG, the frameworks of asymptotic safety and causal dynamical
triangulations are quantizations of standard general relativity, with no exotic
modification to the classical dynamics. A difference among these approaches is
12
Notice that F2 D 0 in [349, 362, 403] and F0 ¤ 0 ¤ F2 in [377, 405].
11.9 Comparison of Quantum-Gravity Models 605
in the way quantization is carried out: via the Hamiltonian formalism in LQG,
through the covariant functional renormalization approach in asymptotic safety, and
via a discretization of the covariant path integral in causal dynamical triangulations.
The choice of the gravitational degrees of freedom deeply affects the quantization
procedure. In LQG, Ashtekar–Barbero variables and spin networks are used; in
asymptotic safety, the covariant metric is quantized; in causal dynamical triangu-
lations, discrete triangulations regularize the divergences of the continuum.
On the other hand, all the other approaches introduce some fundamental
ingredients foreign to both classical general relativity and standard quantization
frameworks. In spin foams, one reformulates gravity as a constrained topological
theory and builds a combinatorial and group structure encoded in labelled simplicial
complexes. Group field theory shares the same type of ingredients as in spin
foams but with the addition of a group-manifold structure and, in principle, of
non-gravitational degrees of freedom. In causal sets, spacetimes descend from a
discrete ordered structure. In non-commutative geometry, spacetime has a non-
trivial algebraic structure. In non-local gravity, operators with infinitely many
derivatives are included in the Lagrangian.
All these methods except those based on non-commutativity are background-
independent, although a 3 C 1 splitting is necessarily assumed in LQG as well as in
the best studied versions of group field theory and causal dynamical triangulations.
Except non-commutative field theories and non-local gravity, they are also non-
perturbative in the sense that no graviton expansion around a fixed background
is used, although spin foams and group field theories can be treated via several
perturbative techniques according to the needs.
Of these approaches, asymptotic safety, causal dynamical triangulations and
non-local gravity essentially describe continuous spacetimes. In contrast, LQG is
characterized by a discrete quantum geometry, which is a consequence of the choice
of canonical quantum variables rather than an assumption; spin foams and group
field theories inherit the same property. In non-commutative spacetimes, below the
fundamental scale appearing in the coordinates algebra it is not possible to maintain
the continuum picture.
In general, these theories are not physically equivalent despite some similarities.
Non-commutativity and non-locality arise, in one form of another, as effective or
intrinsic properties of the other theories, but they can be conceived as stand-alone,
independent proposals as defined in Sects. 11.7 and 11.8. Asymptotic safety, CDT
and LQG are all based on the same classical theory (general relativity), but the
different quantization procedures lead to inequivalent quantum models. Both LQG
and spin foams are apparently recovered in GFT under certain approximations, but
GFT has, by construction, more degrees of freedom than any of these.
Nevertheless, almost all these theories share a rather typical characteristic worth
mentioning, since it can have cosmological applications. The dimension of the
effective spacetimes emerging as a suitable limit of quantum geometry (to be defined
case by case) changes with the probed scale and its value in the far ultraviolet
is often smaller than 4 [430–433]. This phenomenon, called dimensional flow,13

has been detected in all the theories considered in this chapter: asymptotically-
safe gravity (Hausdorff dimension dH D D always, while the spectral dimension
UV
is dS ' D=2 in D topological dimensions at the non-Gaussian fixed point;
analytic results) [21, 41]; causal dynamical triangulations (for phase-C geometries,
UV UV
dS ' D=2 [58, 59, 434, 435] or, more recently, dS ' 3=2 [72]; numerical
results) and the related models of random combs [213, 436] and random multi-
graphs [437, 438]; spin foams [439, 440]; causal sets [178]; non-commutative
geometry [237, 441, 442] and -Minkowski spacetime [247, 443–446]; non-local
super-renormalizable quantum gravity (dS < 1 in the UV in D D 4) [363]. In
LQG, while there is no conclusive evidence of variations of the spectral dimension
for individual quantum-geometry states based on given graphs or complexes [447],
genuine dimensional flow has been encountered in non-trivial superpositions of
spin-network states [448], as an effect of quantum discreteness of geometry. Since
these states appear also in GFT, the latter inherits the same feature.
Other examples, all based on analytic results, are Hořava–Lifshitz gravity
UV
(dS ' 2 for any D) [41, 435, 449], spacetimes with black holes [450–452], fuzzy
spacetimes [453], multi-fractal and multi-fractional spacetimes (variable model-
dependent dH and dS ) [187, 188, 432, 454–456] and string theory [457]. When the
spectral dimension dS decreases to values lower than D at small scales, the UV
properties of the theory usually improve; counter-examples, however, exist [458].
Some properties of dimensional flow are universal simply because one always
reaches the IR in a smooth asymptotic way, a fact that has repercussions in the profile
of the flow also at mesoscopic and small scales [456]. However, geometries with
the same dimensionality profiles are not necessarily isomorphic. The same value of
the spectral dimension may appear in widely different theories without implying a
physical duality among them. This is due to the fact that the spectral dimension is
obtained from a diffusion equation but, as is well known in transport theory, different
diffusion equations can accidentally give rise to the same correlation properties at
small scales [41, 187]. Therefore, there is no precise link between renormalizability
or UV finiteness on one hand, and dimensional flow on the other hand.
In models with dimensional flow, however, the change of spacetime dimension-
ality can give rise to almost scale-invariant primordial spectra without invoking
a slow-rolling scalar field. In one such case, one considers a phenomenological
dispersion relation (5.199) as it could stem from quantum gravity. The correction
function f is higher order in spatial momenta, as in Hořava–Lifshitz gravity. In
particular, for
f .p/ D 1 C ˇjpj4 (11.123)
13
The alternative name of “dimensional reduction” is often employed in the quantum-gravity
literature, despite the fact that it is already in use in Kaluza–Klein and string scenarios, where
spacetime has D > 4 topological dimensions and compactification to four observable dimensions
is performed. For this reason, and to include also all scenarios where the dimension in the UV is
not smaller than in the IR, we prefer the naming “flow.”
References 607
the leading UV term in the kinetic operator K in (11.2) is .r 2 /3 and the spectral
dimension in the UV is dS ' 2. Cosmological perturbations generated by the same
dispersion relation obey the Mukhanov–Sasaki equation (3.32) with k2 ! k2 f . k/,
and they are scale invariant both outside and inside the Hubble horizon [459–461].
Deviations from scale invariance can be obtained easily by a modification of the
dispersion relation; the resulting spacetime has a spectral dimension close but not
exactly equal to 2 in the UV.
Another instance is provided by multi-fractional spacetimes [462]. There, the
ordinary integration measures of position and momentum space are replaced by
some measures with non-trivial weights, dD x ! dD x v.x/, dD p ! dD p w. p/. These
weights contain a hierarchy of fundamental time and length scales such that both the
Hausdorff and spectral dimension change continuously. The form of v is dictated by
multi-fractal geometry and by the universal IR behaviour of dimensional flow [456],
while that of w follows from the requirement of an invertible momentum transform.
Kinetic operators are also modified according to the symmetries imposed on the
Lagrangian. If the spectral dimension is sufficiently small in the UV, power spectra
are almost scale invariant [463]. Other features of the multi-fractal measures can also
help to reformulate the big-bang and cosmological constant problem and trigger a
phase of cyclic evolution in the early universe.
References
1. S. Weinberg, Ultraviolet divergences in quantum gravity, in General Relativity: An Einstein

Centenary Survey, ed. by S.W. Hawking, W. Israel (Cambridge University Press, Cambridge,
1979)
2. M. Reuter, Nonperturbative evolution equation for quantum gravity. Phys. Rev. D 57, 971
(1998). [arXiv:hep-th/9605030]
3. D. Dou, R. Percacci, The running gravitational couplings. Class. Quantum Grav. 15, 3449
(1998). [arXiv:hep-th/9707239]
4. W. Souma, Non-trivial ultraviolet fixed point in quantum gravity. Prog. Theor. Phys. 102, 181
(1999). [arXiv:hep-th/9907027]
5. A. Bonanno, M. Reuter, Renormalization group improved black hole spacetimes. Phys. Rev.
D 62, 043008 (2000). [arXiv:hep-th/0002196]
6. A. Bonanno, M. Reuter, Cosmology of the Planck era from a renormalization group for
quantum gravity. Phys. Rev. D 65, 043508 (2002). [arXiv:hep-th/0106133]
7. O. Lauscher, M. Reuter, Ultraviolet fixed point and generalized flow equation of quantum
gravity. Phys. Rev. D 65, 025013 (2002). [arXiv:hep-th/0108040]
8. O. Lauscher, M. Reuter, Is quantum Einstein gravity nonperturbatively renormalizable? Class.
Quantum Grav. 19, 483 (2002). [arXiv:hep-th/0110021]
9. M. Reuter, F. Saueressig, Renormalization group flow of quantum gravity in the Einstein–
Hilbert truncation. Phys. Rev. D 65, 065016 (2002). [arXiv:hep-th/0110054]
10. O. Lauscher, M. Reuter, Towards nonperturbative renormalizability of quantum Einstein
gravity. Int. J. Mod. Phys. A 17, 993 (2002). [arXiv:hep-th/0112089]
11. O. Lauscher, M. Reuter, Flow equation of quantum Einstein gravity in a higher-derivative
truncation. Phys. Rev. D 66, 025026 (2002). [arXiv:hep-th/0205062]
12. R. Percacci, D. Perini, Constraints on matter from asymptotic safety. Phys. Rev. D 67, 081503
(2003). [arXiv:hep-th/0207033]
13. R. Percacci, D. Perini, Asymptotic safety of gravity coupled to matter. Phys. Rev. D 68,
044018 (2003). [arXiv:hep-th/0304222]
14. D. Perini, Gravity and matter with asymptotic safety. Nucl. Phys. Proc. Suppl. 127, 185
(2004). [arXiv:hep-th/0305053]
15. M. Reuter, F. Saueressig, Nonlocal quantum gravity and the size of the universe. Fortsch.
Phys. 52, 650 (2004). [arXiv:hep-th/0311056]
16. D.F. Litim, Fixed points of quantum gravity. Phys. Rev. Lett. 92, 201301 (2004).
17. R. Percacci, D. Perini, Should we expect a fixed point for Newton’s constant? Class. Quantum
Grav. 21, 5035 (2004). [arXiv:hep-th/0401071]
18. M. Reuter, H. Weyer, Quantum gravity at astrophysical distances? JCAP 0412, 001 (2004).
19. A. Bonanno, M. Reuter, Proper time flow equation for gravity. JHEP 0502, 035 (2005).
20. M. Reuter, F. Saueressig, From big bang to asymptotic de Sitter: complete cosmologies in a
quantum gravity framework. JCAP 0509, 012 (2005). [arXiv:hep-th/0507167]
21. O. Lauscher, M. Reuter, Fractal spacetime structure in asymptotically safe gravity. JHEP
0510, 050 (2005). [arXiv:hep-th/0508202]
22. M. Reuter, J.-M. Schwindt, A minimal length from the cutoff modes in asymptotically safe
quantum gravity. JHEP 0601, 070 (2006). [arXiv:hep-th/0511021]
23. P. Fischer, D.F. Litim, Fixed points of quantum gravity in extra dimensions. Phys. Lett. B
638, 497 (2006). [arXiv:hep-th/0602203]
24. M. Reuter, J.-M. Schwindt, Scale-dependent metric and causal structures in Quantum Einstein
Gravity. JHEP 0701, 049 (2007). [arXiv:hep-th/0611294]
25. B.F.L. Ward, Massive elementary particles and black holes. JCAP 0402, 011 (2004).
26. A. Bonanno, M. Reuter, Entropy signature of the running cosmological constant. JCAP 0708,
024 (2007). [arXiv:0706.0174]
27. M. Reuter, H. Weyer, Background independence and asymptotic safety in conformally
reduced gravity. Phys. Rev. D 79, 105005 (2009). [arXiv:0801.3287]
28. D. Benedetti, P.F. Machado, F. Saueressig, Taming perturbative divergences in asymptotically
safe gravity. Nucl. Phys. B 824, 168 (2010). [arXiv:0902.4630]
29. E. Manrique, M. Reuter, Bimetric truncations for quantum Einstein gravity and asymptotic
safety. Ann. Phys. (N.Y.) 325, 785 (2010). [arXiv:0907.2617]
30. J.E. Daum, U. Harst, M. Reuter, Running gauge coupling in asymptotically safe quantum
gravity. JHEP 1001, 084 (2010). [arXiv:0910.4938]
31. S. Weinberg, Asymptotically safe inflation. Phys. Rev. D 81, 083535 (2010).
[arXiv:0911.3165]
32. A. Bonanno, A. Contillo, R. Percacci, Inflationary solutions in asymptotically safe f .R/
theories. Class. Quantum Grav. 28, 145026 (2011). [arXiv:1006.0192]
33. A. Contillo, M. Hindmarsh, C. Rahmede, Renormalisation group improvement of scalar field
34. M. Reuter, F. Saueressig, Fractal space-times under the microscope: a renormalization group
view on Monte Carlo data. JHEP 1112, 012 (2011). [arXiv:1110.5224]
35. A. Bonanno, An effective action for asymptotically safe gravity. Phys. Rev. D 85, 081503
(2012). [arXiv:1203.1962]
36. M. Hindmarsh, I.D. Saltas, f .R/ gravity from the renormalisation group. Phys. Rev. D 86,
064029 (2012). [arXiv:1203.3957]
37. S. Rechenberger, F. Saueressig, R2 phase-diagram of QEG and its spectral dimension. Phys.
Rev. D 86, 024018 (2012). [arXiv:1206.0657]
38. A. Kaya, Exact renormalization group flow in an expanding Universe and screening of the
cosmological constant. Phys. Rev. D 87, 123501 (2013). [arXiv:1303.5459]
39. A. Codello, G. D’Odorico, C. Pagani, Consistent closure of renormalization group flow
equations in quantum gravity. Phys. Rev. D 89, 081701 (2014). [arXiv:1304.4777]
References 609
40. Y.-F. Cai, Y.-C. Chang, P. Chen, D.A. Easson, T. Qiu, Planck constraints on Higgs modulated
reheating of renormalization group improved inflation. Phys. Rev. D 88, 083508 (2013).
[arXiv:1304.6938]
41. G. Calcagni, A. Eichhorn, F. Saueressig, Probing the quantum nature of spacetime by
diffusion. Phys. Rev. D 87, 124028 (2013). [arXiv:1304.7247]
42. E.J. Copeland, C. Rahmede, I.D. Saltas, Asymptotically safe Starobinsky inflation. Phys.
Rev. D 91, 103530 (2015). [arXiv:1311.0881]
43. P. Donà, A. Eichhorn, R. Percacci, Matter matters in asymptotically safe quantum gravity.
Phys. Rev. D 89, 084035 (2014). [arXiv:1311.2898]
44. K. Falls, Asymptotic safety and the cosmological constant. JHEP 1601, 069 (2016).
[arXiv:1408.0276]
45. P. Donà, A. Eichhorn, R. Percacci, Consistency of matter models with asymptotically safe
quantum gravity. Can. J. Phys. 93, 988 (2015). [arXiv:1410.4411]
46. M. Niedermaier, The asymptotic safety scenario in quantum gravity: an introduction. Class.
47. M. Niedermaier, M. Reuter, The asymptotic safety scenario in quantum gravity. Living Rev.
Relat. 9, 5 (2006)
48. M. Reuter, F. Saueressig, Functional renormalization group equations, asymptotic safety and
quantum Einstein gravity. arXiv:0708.1317
49. A. Codello, R. Percacci, C. Rahmede, Investigating the ultraviolet properties of gravity
with a Wilsonian renormalization group equation. Ann. Phys. (N.Y.) 324, 414 (2009).
[arXiv:0805.2909]
50. D.F. Litim, Renormalisation group and the Planck scale. Philos. Trans. R. Soc. Lond. A 369,
2759 (2011). [arXiv:1102.4624]
51. M. Reuter, F. Saueressig, Asymptotic safety, fractals, and cosmology. Lect. Notes Phys. 863,
185 (2013). [arXiv:1205.5431]
52. B.F.L. Ward, An estimate of in resummed quantum gravity in the context of asymptotic
safety. Phys. Dark Univ. 2, 97 (2013)
53. J. Ambjørn, R. Loll, Non-perturbative Lorentzian quantum gravity, causality and topology
change. Nucl. Phys. B 536, 407 (1998). [arXiv:hep-th/9805108]
54. J. Ambjørn, J. Jurkiewicz, R. Loll, A non-perturbative Lorentzian path integral for gravity.
55. J. Ambjørn, J. Jurkiewicz, R. Loll, Dynamically triangulating Lorentzian quantum gravity.
56. J. Ambjørn, J. Jurkiewicz, R. Loll, Emergence of a 4D world from causal quantum gravity.
57. J. Ambjørn, J. Jurkiewicz, R. Loll, Semiclassical universe from first principles. Phys. Lett. B
607, 205 (2005). [arXiv:hep-th/0411152]
58. J. Ambjørn, J. Jurkiewicz, R. Loll, Spectral dimension of the universe. Phys. Rev. Lett. 95,
171301 (2005). [arXiv:hep-th/0505113]
59. J. Ambjørn, J. Jurkiewicz, R. Loll, Reconstructing the universe. Phys. Rev. D 72, 064014
(2005). [arXiv:hep-th/0505154]
60. J. Ambjørn, J. Jurkiewicz, R. Loll, The universe from scratch. Contemp. Phys. 47, 103 (2006).
61. J. Ambjørn, A. Görlich, J. Jurkiewicz, R. Loll, Planckian birth of the quantum de Sitter
universe. Phys. Rev. Lett. 100, 091304 (2008). [arXiv:0712.2485]
62. J. Ambjørn, J. Jurkiewicz, R. Loll, The self-organized de Sitter universe. Int. J. Mod. Phys.
D 17, 2515 (2009). [arXiv:0806.0397]
63. J. Ambjørn, A. Görlich, J. Jurkiewicz, R. Loll, The nonperturbative quantum de Sitter
universe. Phys. Rev. D 78, 063544 (2008). [arXiv:0807.4481]
64. J. Ambjørn, A. Görlich, J. Jurkiewicz, R. Loll, Geometry of the quantum universe. Phys.
Lett. B 690, 420 (2010). [arXiv:1001.4581]
65. J. Ambjørn, A. Görlich, S. Jordan, J. Jurkiewicz, R. Loll, CDT meets Hořava–Lifshitz gravity.
Phys. Lett. B 690, 413 (2010). [arXiv:1002.3298]
66. J. Ambjørn, A. Görlich, J. Jurkiewicz, R. Loll, J. Gizbert-Studnicki, T. Trześniewski, The

semiclassical limit of causal dynamical triangulations. Nucl. Phys. B 849, 144 (2011).
[arXiv:1102.3929]
67. J. Ambjørn, S. Jordan, J. Jurkiewicz, R. Loll, A second-order phase transition in causal
dynamical triangulations. Phys. Rev. Lett. 107, 211303 (2011). [arXiv:1108.3932]
68. J. Ambjørn, S. Jordan, J. Jurkiewicz, R. Loll, Second- and first-order phase transitions in
causal dynamical triangulations. Phys. Rev. D 85, 124044 (2012). [arXiv:1205.1229]
69. S. Jordan, R. Loll, Causal dynamical triangulations without preferred foliation. Phys. Lett. B
724, 155 (2013). [arXiv:1305.4582]
70. S. Jordan, R. Loll, De Sitter universe from causal dynamical triangulations without preferred
foliation. Phys. Rev. D 88, 044055 (2013). [arXiv:1307.5469]
71. J. Ambjørn, J. Gizbert-Studnicki, A. Görlich, J. Jurkiewicz, The effective action in 4-dim
CDT. The transfer matrix approach. JHEP 1406, 034 (2014). [arXiv:1403.5940]
72. D.N. Coumbe, J. Jurkiewicz, Evidence for asymptotic safety from dimensional reduction in
causal dynamical triangulations. JHEP 1503, 151 (2015). [arXiv:1411.7712]
73. J. Ambjørn, D.N. Coumbe, J. Gizbert-Studnicki, J. Jurkiewicz, Signature change of the metric
in CDT quantum gravity? JHEP 1508, 033 (2015). [arXiv:1503.08580]
74. D.N. Coumbe, J. Gizbert-Studnicki, J. Jurkiewicz, Exploring the new phase transition of CDT.
JHEP 1602, 144 (2016). [arXiv:1510.08672]
75. J. Ambjørn, J. Gizbert-Studnicki, A. Görlich, J. Jurkiewicz, N. Klitgaard, R. Loll, Character-
istics of the new phase in CDT. arXiv:1610.05245
76. R. Loll, The emergence of spacetime, or, quantum gravity on your desktop. Class. Quantum
Grav. 25, 114006 (2008). [arXiv:0711.0273]
77. J. Ambjørn, J. Jurkiewicz, R. Loll, Causal dynamical triangulations and the quest for quantum
gravity, in [78]. [arXiv:1004.0352]
78. G.F.R. Ellis, J. Murugan, A. Weltman (eds.), Foundations of Space and Time (Cambridge
79. J. Ambjørn, A. Görlich, J. Jurkiewicz, R. Loll, Nonperturbative quantum gravity. Phys. Rep.
519, 127 (2012). [arXiv:1203.3591]
80. M.P. Reisenberger, C. Rovelli, “Sum over surfaces” form of loop quantum gravity. Phys. Rev.
D 56, 3490 (1997). [arXiv:gr-qc/9612035]
81. J.W. Barrett, L. Crane, Relativistic spin networks and quantum gravity. J. Math. Phys. 39,
3296 (1998). [arXiv:gr-qc/9709028]
82. J.W. Barrett, L. Crane, A Lorentzian signature model for quantum general relativity. Class.
83. A. Perez, C. Rovelli, Spin foam model for Lorentzian general relativity. Phys. Rev. D 63,
041501 (2001). [arXiv:gr-qc/0009021]
84. E.R. Livine, S. Speziale, New spinfoam vertex for quantum gravity. Phys. Rev. D 76, 084028
(2007). [arXiv:0705.0674]
85. J. Engle, R. Pereira, C. Rovelli, Loop-quantum-gravity vertex amplitude. Phys. Rev. Lett. 99,
161301 (2007). [arXiv:0705.2388]
86. J. Engle, R. Pereira, C. Rovelli, Flipped spinfoam vertex and loop gravity. Nucl. Phys. B 798,
251 (2008). [arXiv:0708.1236]
87. L. Freidel, K. Krasnov, A new spin foam model for 4D gravity. Class. Quantum Grav. 25,
125018 (2008). [arXiv:0708.1595]
88. E.R. Livine, S. Speziale, Consistently solving the simplicity constraints for spinfoam quantum
gravity. Europhys. Lett. 81, 50004 (2008). [arXiv:0708.1915]
89. R. Pereira, Lorentzian LQG vertex amplitude. Class. Quantum Grav. 25, 085013 (2008).
[arXiv:0710.5043]
90. J. Engle, E. Livine, R. Pereira, C. Rovelli, LQG vertex with finite Immirzi parameter. Nucl.
Phys. B 799, 136 (2008). [arXiv:0711.0146]
91. F. Conrady, L. Freidel, Path integral representation of spin foam models of 4D gravity. Class.
References 611
92. F. Conrady, L. Freidel, Semiclassical limit of 4-dimensional spin foam models. Phys. Rev. D
78, 104023 (2008). [arXiv:0809.2280]
93. J.W. Barrett, R.J. Dowdall, W.J. Fairbairn, F. Hellmann, R. Pereira, Lorentzian spin foam
amplitudes: graphical calculus and asymptotics. Class. Quantum Grav. 27, 165009 (2010).
[arXiv:0907.2440]
94. C. Rovelli, Discretizing parametrized systems: the magic of Ditt-invariance. arXiv:1107.2310
95. C. Rovelli, On the structure of a background independent quantum theory: Hamilton function,
transition amplitudes, classical limit and continuous limit. arXiv:1108.0832
96. D. Oriti, Spacetime geometry from algebra: spin foam models for non-perturbative quantum
gravity. Rep. Prog. Phys. 64, 1489 (2001). [arXiv:gr-qc/0106091]
97. A. Perez, Spin foam models for quantum gravity. Class. Quantum Grav. 20, R43 (2003).
98. C. Rovelli, A new look at loop quantum gravity. Class. Quantum Grav. 28, 114005 (2011).
[arXiv:1004.1780]
99. A. Perez, The spin-foam approach to quantum gravity. Living Rev. Relat. 16, 3 (2013).
100. A. Ashtekar, M. Campiglia, A. Henderson, Loop quantum cosmology and spin foams. Phys.
Lett. B 681, 347 (2009). [arXiv:0909.4221]
101. C. Rovelli, F. Vidotto, On the spinfoam expansion in cosmology. Class. Quantum Grav. 27,
145005 (2010). [arXiv:0911.3097]
102. A. Ashtekar, M. Campiglia, A. Henderson, Casting loop quantum cosmology in the spin foam
paradigm. Class. Quantum Grav. 27, 135020 (2010). [arXiv:1001.5147]
103. E. Bianchi, C. Rovelli, F. Vidotto, Towards spinfoam cosmology. Phys. Rev. D 82, 084035
(2010). [arXiv:1003.3483]
104. A. Henderson, C. Rovelli, F. Vidotto, E. Wilson-Ewing, Local spinfoam expansion in loop
quantum cosmology. Class. Quantum Grav. 28, 025003 (2011). [arXiv:1010.0502]
105. G. Calcagni, S. Gielen, D. Oriti, Two-point functions in (loop) quantum cosmology. Class.
106. E. Bianchi, T. Krajewski, C. Rovelli, F. Vidotto, Cosmological constant in spinfoam cosmol-
ogy. Phys. Rev. D 83, 104015 (2011). [arXiv:1101.4049]
107. H. Huang, Y. Ma, L. Qin, Path integral and effective Hamiltonian in loop quantum cosmology.
Gen. Relat. Grav. 45, 1191 (2013). [arXiv:1102.4755]
108. F. Hellmann, Expansions in spin foam cosmology. Phys. Rev. D 84, 103516 (2011).
[arXiv:1105.1334]
109. L. Qin, G. Deng, Y.-G. Ma, Path integrals and alternative effective dynamics in loop quantum
cosmology. Commun. Theor. Phys. 57, 326 (2012). [arXiv:1206.1131]
110. J. Rennert, D. Sloan, A homogeneous model of spinfoam cosmology. Class. Quantum Grav.
30, 235019 (2013). [arXiv:1304.6688]
111. D.V. Boulatov, A model of three-dimensional lattice gravity. Mod. Phys. Lett. A 07, 1629
(1992). [arXiv:hep-th/9202074]
112. H. Ooguri, Topological lattice models in four dimensions. Mod. Phys. Lett. A 07, 2799
(1992). [arXiv:hep-th/9205090]
113. R. De Pietri, L. Freidel, K. Krasnov, C. Rovelli, Barrett–Crane model from a Boulatov–
Ooguri field theory over a homogeneous space. Nucl. Phys. B 574, 785 (2000).
114. M.P. Reisenberger, C. Rovelli, Space-time as a Feynman diagram: the connection formulation.
115. A.R. Miković, Quantum field theory of spin networks. Class. Quantum Grav. 18, 2827 (2001).
116. D. Oriti, J. Ryan, Group field theory formulation of 3D quantum gravity coupled to matter
fields. Class. Quantum Grav. 23, 6543 (2006). [arXiv:gr-qc/0602010]
117. W.J. Fairbairn, E.R. Livine, 3D spinfoam quantum gravity: matter as a phase of the group
field theory. Class. Quantum Grav. 24, 5277 (2007). [arXiv:gr-qc/0702125]
118. E.R. Livine, Matrix models as non-commutative field theories on R 3 . Class. Quantum Grav.
26, 195014 (2009). [arXiv:0811.1462]
119. F. Girelli, E.R. Livine, D. Oriti, Four-dimensional deformed special relativity from group field
120. D. Oriti, Emergent non-commutative matter fields from group field theory models of quantum
spacetime. J. Phys. Conf. Ser. 174, 012047 (2009). [arXiv:0903.3970]
121. R.J. Dowdall, Wilson loops, geometric operators and fermions in 3d group field theory. Centr.
Eur. J. Phys. 9, 1043 (2011). [arXiv:0911.2391]
122. A. Baratin, D. Oriti, Group field theory with noncommutative metric variables. Phys. Rev.
Lett. 105, 221302 (2010). [arXiv:1002.4723]
123. A. Baratin, B. Dittrich, D. Oriti, J. Tambornino, Non-commutative flux representation for loop
quantum gravity. Class. Quantum Grav. 28, 175011 (2011). [arXiv:1004.3450]
124. J. Ben Geloun, V. Bonzom, Radiative corrections in the Boulatov–Ooguri tensor model: the
2-point function. Int. J. Theor. Phys. 50, 2819 (2011). [arXiv:1101.4294]
125. A. Baratin, D. Oriti, Quantum simplicial geometry in the group field theory formalism:
reconsidering the Barrett–Crane model. New J. Phys. 13, 125011 (2011). [arXiv:1108.1178]
126. A. Baratin, D. Oriti, Group field theory and simplicial gravity path integrals: a model for
Holst–Plebanski gravity. Phys. Rev. D 85, 044003 (2012). [arXiv:1111.5842]
127. J. Ben Geloun, Two- and four-loop ˇ-functions of rank-4 renormalizable tensor field theories.
128. S. Carrozza, D. Oriti, V. Rivasseau, Renormalization of tensorial group field theories: Abelian
U.1/ models in four dimensions. Commun. Math. Phys. 327, 603 (2014). [arXiv:1207.6734]
129. C. Guedes, D. Oriti, M. Raasakka, Quantization maps, algebra representation and non-
commutative Fourier transform for Lie groups. J. Math. Phys. 54, 083508 (2013).
[arXiv:1301.7750]
130. S. Gielen, D. Oriti, L. Sindoni, Cosmology from group field theory formalism for quantum
gravity. Phys. Rev. Lett. 111, 031301 (2013). [arXiv:1303.3576]
131. S. Carrozza, D. Oriti, V. Rivasseau, Renormalization of an SU.2/ tensorial group field theory
in three dimensions. Commun. Math. Phys. 330, 581 (2014). [arXiv:1303.6772]
132. S. Gielen, D. Oriti, L. Sindoni, Homogeneous cosmologies as group field theory condensates.
JHEP 1406, 013 (2014). [arXiv:1311.1238]
133. S. Gielen, Quantum cosmology of (loop) quantum gravity condensates: an example. Class.
134. G. Calcagni, Loop quantum cosmology from group field theory. Phys. Rev. D 90, 064047
(2014). [arXiv:1407.8166]
135. S. Gielen, D. Oriti, Quantum cosmology from quantum gravity condensates: cosmological
variables and lattice-refined dynamics. New J. Phys. 16, 123004 (2014). [arXiv:1407.8167]
136. D. Oriti, D. Pranzetti, J.P. Ryan, L. Sindoni, Generalized quantum gravity condensates
for homogeneous geometries and cosmology. Class. Quantum Grav. 32, 235016 (2015).
[arXiv:1501.00936]
137. S. Gielen, Identifying cosmological perturbations in group field theory condensates. JHEP
1508, 010 (2015). [arXiv:1505.07479]
138. D. Oriti, D. Pranzetti, L. Sindoni, Horizon entropy from quantum gravity condensates. Phys.
Rev. Lett. 116, 211301 (2016). [arXiv:1510.06991]
139. D. Oriti, L. Sindoni, E. Wilson-Ewing, Bouncing cosmologies from quantum gravity conden-
sates. arXiv:1602.08271
140. D. Oriti, L. Sindoni, E. Wilson-Ewing, Emergent Friedmann dynamics with a quantum
bounce from quantum gravity condensates. Class. Quantum Grav. 33, 224001 (2016).
[arXiv:1602.05881]
141. A.G.A. Pithis, M. Sakellariadou, P. Tomov, Impact of nonlinear effective interactions
on group field theory quantum gravity condensates. Phys. Rev. D 94, 064056 (2016).
[arXiv:1607.06662]
142. L. Freidel, Group field theory: an overview. Int. J. Theor. Phys. 44, 1769 (2005).
143. D. Oriti, The group field theory approach to quantum gravity, in [144]. [arXiv:gr-qc/0607032]
144. D. Oriti (ed.), Approaches to Quantum Gravity (Cambridge University Press, Cambridge,
2009)
References 613
145. D. Oriti, The microscopic dynamics of quantum space as a group field theory, in [78].
[arXiv:1110.5606]
146. D. Oriti, Group field theory as the second quantization of loop quantum gravity. Class.
147. A. Baratin, D. Oriti, Ten questions on group field theory (and their tentative answers). J. Phys.
Conf. Ser. 360, 012002 (2012). [arXiv:1112.3270]
148. S. Gielen, L. Sindoni, Quantum cosmology from group field theory condensates: a review.
SIGMA 12, 082 (2016). [arXiv:1602.08104]
149. L. Bombelli, J. Lee, D. Meyer, R. Sorkin, Space-time as a causal set. Phys. Rev. Lett. 59, 521
(1987)
150. C. Moore, Comment on “Space-time as a causal set”. Phys. Rev. Lett. 60, 655 (1988)
151. L. Bombelli, J. Lee, D. Meyer, R.D. Sorkin, Bombelli et al. reply. Phys. Rev. Lett. 60, 656
(1988)
152. G. Brightwell, R. Gregory, Structure of random discrete spacetime. Phys. Rev. Lett. 66, 260
(1991)
153. R.D. Sorkin, Forks in the road, on the way to quantum gravity. Int. J. Theor. Phys. 36, 2759
(1997). [arXiv:gr-qc/9706002]
154. D.P. Rideout, R.D. Sorkin, Classical sequential growth dynamics for causal sets. Phys. Rev.
D 61, 024002 (2000). [arXiv:gr-qc/9904062]
155. R.D. Sorkin, Indications of causal set cosmology. Int. J. Theor. Phys. 39, 1731 (2000).
156. X. Martín, D. O’Connor, D.P. Rideout, R.D. Sorkin, “Renormalization” transformations
induced by cycles of expansion and contraction in causal set cosmology. Phys. Rev. D 63,
084026 (2001). [arXiv:gr-qc/0009063]
157. A. Ash, P. McDonald, Moment problems and the causal set approach to quantum gravity. J.
Math. Phys. 44, 1666 (2003). [arXiv:gr-qc/0209020]
158. M. Ahmed, S. Dodelson, P.B. Greene, R. Sorkin, Everpresent . Phys. Rev. D 69, 103523
(2004). [arXiv:astro-ph/0209274]
159. G. Brightwell, H.F. Dowker, R.S. García, J. Henson, R.D. Sorkin, “Observables” in causal set
160. D. Rideout, Dynamics of Causal Sets. Ph.D. thesis, Syracuse University, Syracuse (2001).
161. F. Dowker, J. Henson, R.D. Sorkin, Quantum gravity phenomenology, Lorentz invariance and
discreteness. Mod. Phys. Lett. A 19, 1829 (2004). [arXiv:gr-qc/0311055]
162. S. Major, D. Rideout, S. Surya, Spatial hypersurfaces in causal set cosmology. Class.
163. S. Major, D. Rideout, S. Surya, On recovering continuum topology from a causal set. J. Math.
Phys. 48, 032501 (2007). [arXiv:gr-qc/0604124]
164. L. Bombelli, J. Henson, R.D. Sorkin, Discreteness without symmetry breaking: a theorem.
Mod. Phys. Lett. A 24, 2579 (2009). [arXiv:gr-qc/0605006]
165. D. Rideout, S. Zohren, Evidence for an entropy bound from fundamentally discrete gravity.
166. J.D. Barrow, Strong constraint on ever-present . Phys. Rev. D 75, 067301 (2007).
167. J.A. Zuntz, The cosmic microwave background in a causal set universe. Phys. Rev. D 77,
043002 (2008). [arXiv:0711.2904]
168. S. Johnston, Particle propagators on discrete spacetime. Class. Quantum Grav. 25, 202001
(2008). [arXiv:0806.3083]
169. L. Philpott, F. Dowker, R.D. Sorkin, Energy-momentum diffusion from spacetime discrete-
ness. Phys. Rev. D 79, 124047 (2009). [arXiv:0810.5591]
170. S. Major, D. Rideout, S. Surya, Stable homology as an indicator of manifoldlikeness in causal
set theory. Class. Quantum Grav. 26, 175008 (2009). [arXiv:0902.0434]
171. S. Johnston, Feynman propagator for a free scalar field on a causal set. Phys. Rev. Lett. 103,
180401 (2009). [arXiv:0909.0944]
172. M. Ahmed, D. Rideout, Indications of de Sitter spacetime from classical sequential growth
dynamics of causal sets. Phys. Rev. D 81, 083528 (2010). [arXiv:0909.4771]
173. D.M.T. Benincasa, F. Dowker, Scalar curvature of a causal set. Phys. Rev. Lett. 104, 181301
(2010). [arXiv:1001.2725]
174. D.M.T. Benincasa, F. Dowker, B. Schmitzer, The random discrete action for two-dimensional
spacetime. Class. Quantum Grav. 28, 105018 (2011). [arXiv:1011.5191]
175. S. Surya, Evidence for the continuum in 2D causal set quantum gravity. Class. Quantum Grav.
29, 132001 (2012). [arXiv:1110.6244]
176. M. Ahmed, R. Sorkin, Everpresent . II. Structural stability. Phys. Rev. D 87, 063515 (2013).
[arXiv:1210.2589]
177. F. Dowker, L. Glaser, Causal set d’Alembertians for various dimensions. Class. Quantum
Grav. 30, 195016 (2013). [arXiv:1305.2588]
178. A. Eichhorn, S. Mizera, Spectral dimension in causal set quantum gravity. Class. Quantum
Grav. 31, 125007 (2014). [arXiv:1311.2530]
179. S. Aslanbeigi, M. Saravani, R.D. Sorkin, Generalized causal set d’Alembertians. JHEP 1406,
024 (2014). [arXiv:1403.1622]
180. S. Johnston, Correction terms for propagators and d’Alembertians due to spacetime discrete-
ness. Class. Quantum Grav. 32, 195020 (2015). [arXiv:1411.2614]
181. R.D. Sorkin, Spacetime and causal sets, in Relativity and Gravitation: Classical and
Quantum, ed. by J.C. D’Olivo, E. Nahmad-Achar, M. Rosenbaum, M.P. Ryan, L.F. Urrutia,
F. Zertuche (World Scientific, Singapore, 1991)
182. D.D. Reid, Introduction to causal sets: an alternate view of spacetime structure. Can. J. Phys.
79, 1 (2001). [arXiv:gr-qc/9909075]
183. J. Henson, The causal set approach to quantum gravity, in [144]. [arXiv:gr-qc/0601121]
184. J. Henson, Discovering the discrete universe. arXiv:1003.5890
185. S. Surya, Directions in causal set quantum gravity, in Recent Research in Quantum Gravity,
ed. by A. Dasgupta (Nova Science, Hauppauge, 2011). [arXiv:1103.6272]
186. F. Dowker, Introduction to causal sets and their phenomenology. Gen. Relat. Grav. 45, 1651
(2013)
187. G. Calcagni, Diffusion in multiscale spacetimes. Phys. Rev. E 87, 012123 (2013).
[arXiv:1205.5046]
188. G. Calcagni, G. Nardelli, Spectral dimension and diffusion in multiscale spacetimes. Phys.
Rev. D 88, 124025 (2013). [arXiv:1304.2709]
189. G. Calcagni, L. Modesto, G. Nardelli, Quantum spectral dimension in quantum field theory.
Int. J. Mod. Phys. D 25, 1650058 (2016). [arXiv:1408.0199]
190. J. Berges, N. Tetradis, C. Wetterich, Non-perturbative renormalization flow in quantum field
theory and statistical physics. Phys. Rep. 363, 223 (2002). [arXiv:hep-ph/0005122]
191. G. Calcagni, Multifractional spacetimes, asymptotic safety and Hořava–Lifshitz gravity. Int.
J. Mod. Phys. A 28, 1350092 (2013). [arXiv:1209.4376]
192. B.F.L. Ward, Planck scale cosmology in resummed quantum gravity. Mod. Phys. Lett. A 23,
3299 (2008). [arXiv:0808.3124]
193. D. Kalligas, P.S. Wesson, C.W.F. Everitt, Bianchi type I cosmological models with variable
G and : a comment. Gen. Relat. Grav. 27, 645 (1995)
194. B.F.L. Ward, Einstein–Heisenberg consistency condition interplay with cosmological con-
stant prediction in resummed quantum gravity. Mod. Phys. Lett. A 30, 1550206 (2015).
[arXiv:1507.00661]
195. J. Ambjørn, S. Varsted, Three-dimensional simplicial quantum gravity. Nucl. Phys. B 373,
557 (1992)
196. J. Ambjørn, D.V. Boulatov, A. Krzywicki, S. Varsted, The vacuum in three-dimensional
simplicial quantum gravity. Phys. Lett. B 276, 432 (1992)
197. J. Ambjørn, J. Jurkiewicz, Four-dimensional simplicial quantum gravity. Phys. Lett. B 278,
42 (1992)
198. M.E. Agishtein, A.A. Migdal, Critical behavior of dynamically triangulated quantum gravity
in four dimensions. Nucl. Phys. B 385, 395 (1992). [arXiv:hep-lat/9204004]
References 615
199. J. Ambjørn, S. Jain, J. Jurkiewicz, C.F. Kristjansen, Observing 4d baby universes in quantum
gravity. Phys. Lett. B 305, 208 (1993). [arXiv:hep-th/9303041]
200. S. Catterall, J.B. Kogut, R. Renken, Phase structure of four-dimensional simplicial quantum
gravity. Phys. Lett. B 328, 277 (1994). [arXiv:hep-lat/9401026]
201. P. Bialas, Z. Burda, A. Krzywicki, B. Petersson, Focusing on the fixed point of 4D simplicial
gravity. Nucl. Phys. B 472, 293 (1996). [arXiv:hep-lat/9601024]
202. B.V. de Bakker, Further evidence that the transition of 4D dynamical triangulation is first
order. Phys. Lett. B 389, 238 (1996). [arXiv:hep-lat/9603024]
203. S. Catterall, R. Renken, J.B. Kogut, Singular structure in 4D simplicial gravity. Phys. Lett. B
416, 274 (1998). [arXiv:hep-lat/9709007]
204. J.A. Wheeler, Geometrodynamics and the issue of the final state, in Relativity, Groups and
Topology, ed. by C. DeWitt, B.S. DeWitt (Gordon and Breach, New York, 1964)
205. T. Regge, General relativity without coordinates. Nuovo Cim. 19, 558 (1961)
206. R.M. Williams, P.A. Tuckey, Regge calculus: a bibliography and brief review. Class. Quantum
Grav. 9, 1409 (1992)
207. R.M. Williams, Discrete quantum gravity: the Regge calculus approach. Int. J. Mod. Phys. B
6, 2097 (1992)
208. R.M. Williams, Recent progress in Regge calculus. Nucl. Phys. Proc. Suppl. 57, 73 (1997).
209. F. David, What is the intrinsic geometry of two-dimensional quantum gravity? Nucl. Phys. B
368, 671 (1992)
210. T. Jonsson, J.F. Wheater, The spectral dimension of the branched polymer phase of two-
dimensional quantum gravity. Nucl. Phys. B 515, 549 (1998). [arXiv:hep-lat/9710024]
211. J.D. Correia, J.F. Wheater, The spectral dimension of non-generic branched polymer ensem-
bles. Phys. Lett. B 422, 76 (1998). [arXiv:hep-th/9712058]
212. C. Destri, L. Donetti, The spectral dimension of random trees. J. Phys. A 35, 9499 (2002).
[arXiv:cond-mat/0206233]
213. B. Durhuus, T. Jonsson, J.F. Wheater, Random walks on combs. J. Phys. A 39, 1009 (2006).
214. B. Durhuus, T. Jonsson, J.F. Wheater, The spectral dimension of generic trees. J. Stat. Phys.
128, 1237 (2007). [arXiv:math-ph/0607020]
215. J.J. Halliwell, J.B. Hartle, Wave functions constructed from an invariant sum over histories
satisfy constraints. Phys. Rev. D 43, 1170 (1991)
216. C. Teitelboim, Quantum mechanics of the gravitational field. Phys. Rev. D 25, 3159 (1982)
217. J.J. Halliwell, Derivation of the Wheeler–DeWitt equation from a path integral for minisuper-
space models. Phys. Rev. D 38, 2468 (1988)
218. J.B. Hartle, K.V. Kuchař, Path integrals in parametrized theories: the free relativistic particle.
Phys. Rev. D 34, 2323 (1986)
219. J.J. Halliwell, M.E. Ortiz, Sum-over-histories origin of the composition laws of rela-
tivistic quantum mechanics and quantum cosmology. Phys. Rev. D 48, 748 (1993).
220. R. Gurau, J.P. Ryan, Colored tensor models – a review. SIGMA 8, 020 (2012).
[arXiv:1109.4812]
221. V. Rivasseau, Quantum gravity and renormalization: the tensor track. AIP Conf. Proc. 1444,
18 (2011). [arXiv:1112.5104]
222. L. Freidel, S. Majid, Noncommutative harmonic analysis, sampling theory and the Duflo map
in 2+1 quantum gravity. Class. Quantum Grav. 25, 045006 (2008). [arXiv:hep-th/0601004]
223. E. Joung, J. Mourad, K. Noui, Three dimensional quantum geometry and deformed symmetry.
J. Math. Phys. 50, 052503 (2009). [arXiv:0806.4121]
224. C.J. Pethick, H. Smith, Bose–Einstein Condensation in Dilute Gases (Cambridge University
225. L.P. Pitaevskii, S. Stringari, Bose–Einstein Condensation (Clarendon Press, Oxford, 2003)
226. S. Giorgini, L.P. Pitaevskii, S. Stringari, Theory of ultracold atomic Fermi gases. Rev. Mod.
Phys. 80, 1215 (2008). [arXiv:0706.3360]
227. G. Calcagni, S. Gielen, D. Oriti, Group field cosmology: a cosmological field theory of
quantum geometry. Class. Quantum Grav. 29, 105005 (2012). [arXiv:1201.4151]
228. C.R. Contaldi, F. Dowker, L. Philpott, Polarization diffusion from spacetime uncertainty.
229. A. Lue, L. Wang, M. Kamionkowski, Cosmological signature of new parity-violating
interactions. Phys. Rev. Lett. 83, 1506 (1999). [arXiv:astro-ph/9812088]
230. J. Madore, An Introduction to Noncommutative Geometry and Its Physical Applications
(Cambridge University Press, Cambridge, 1999)
231. M.R. Douglas, N.A. Nekrasov, Noncommutative field theory. Rev. Mod. Phys. 73, 977 (2001).
232. R.J. Szabo, Quantum field theory on noncommutative spaces. Phys. Rep. 378, 207 (2003).
233. A. Connes, Noncommutative Geometry (Academic Press, San Diego, 2004)
234. P. Aschieri, M. Dimitrijevic, P. Kulish, F. Lizzi, J. Wess, Noncommutative Spacetimes
(Springer, Berlin, 2009)
235. A.P. Balachandran, A. Ibort, G. Marmo, M. Martone, Quantum fields on noncommutative
spacetimes: theory and phenomenology. SIGMA 6, 052 (2010). [arXiv:1003.4356]
236. H.-J. Matschull, M. Welling, Quantum mechanics of a point particle in .2 C 1/-dimensional
gravity. Class. Quantum Grav. 15, 2981 (1998). [arXiv:gr-qc/9708054]
237. M. Arzano, E. Alesci, Anomalous dimension in three-dimensional semiclassical gravity.
Phys. Lett. B 707, 272 (2012). [arXiv:1108.1507]
238. S. Deser, R. Jackiw, G. ’t Hooft, Three-dimensional Einstein gravity: dynamics of flat space.
Ann. Phys. (N.Y.) 152, 220 (1984)
239. L. Freidel, E. Livine, Ponzano–Regge model revisited III: Feynman diagrams and effective
field theory. Class. Quantum Grav. 23, 2021 (2006). [arXiv:hep-th/0502106]
240. L. Freidel, E.R. Livine, 3D quantum gravity and noncommutative quantum field theory. Phys.
Rev. Lett. 96, 221301 (2006). [arXiv:hep-th/0512113]
241. G. Amelino-Camelia, L. Smolin, A. Starodubtsev, Quantum symmetry, the cosmologi-
cal constant and Planck-scale phenomenology. Class. Quantum Grav. 21, 3095 (2004).
242. L. Freidel, J. Kowalski-Glikman, L. Smolin, 2+1 gravity and doubly special relativity. Phys.
Rev. D 69, 044001 (2004). [arXiv:hep-th/0307085]
243. S. Doplicher, K. Fredenhagen, J.E. Roberts, Space-time quantization induced by classical
gravity. Phys. Lett. B 331, 39 (1994)
244. S. Doplicher, K. Fredenhagen, J.E. Roberts, The quantum structure of space-time at the Planck
scale and quantum fields. Commun. Math. Phys. 172, 187 (1995). [arXiv:hep-th/0303037]
245. S. Majid, H. Ruegg, Bicrossproduct structure of -Poincaré group and non-commutative
geometry. Phys. Lett. B 334, 348 (1994). [arXiv:hep-th/9405107]
246. A. Agostini, G. Amelino-Camelia, F. D’Andrea, Hopf algebra description of noncommutative
space-time symmetries. Int. J. Mod. Phys. A 19, 5187 (2004). [arXiv:hep-th/0306013]
247. M. Arzano, G. Calcagni, D. Oriti, M. Scalisi, Fractional and noncommutative spacetimes.
Phys. Rev. D 84, 125002 (2011). [arXiv:1107.5308]
248. G. Calcagni, M. Ronco, Deformed symmetries in noncommutative and multifractional
spacetimes. arXiv:1608.01667
249. C.-S. Chu, P.-M. Ho, Non-commutative open string and D-brane. Nucl. Phys. B 550, 151
(1999). [arXiv:hep-th/9812219]
250. V. Schomerus, D-branes and deformation quantization. JHEP 9906, 030 (1999).
251. N. Seiberg, E. Witten, String theory and noncommutative geometry. JHEP 0909, 032 (1999).
252. A. Matusis, L. Susskind, N. Toumbas, The IR/UV connection in the non-commutative gauge
theories. JHEP 0012, 002 (2000). [arXiv:hep-th/0002075]
253. A.H. Chamseddine, G. Felder, J. Fröhlich, Gravity in noncommutative geometry. Commun.
Math. Phys. 155, 205 (1993). [arXiv:hep-th/9209044]
References 617
254. P. Aschieri, C. Blohmann, M. Dimitrijevic, F. Meyer, P. Schupp, J. Wess, A gravity theory on

noncommutative spaces. Class. Quantum Grav. 22, 3511 (2005). [arXiv:hep-th/0504183]
255. P. Aschieri, M. Dimitrijevic, F. Meyer, J. Wess, Noncommutative geometry and gravity. Class.
Quantum Grav. 23, 1883 (2006). [arXiv:hep-th/0510059]
256. E. Harikumar, V.O. Rivelles, Noncommutative gravity. Class. Quantum Grav. 23, 7551
(2006). [arXiv:hep-th/0607115]
257. A. Connes, M.R. Douglas, A.S. Schwarz, Noncommutative geometry and matrix theory:
compactification on tori. JHEP 9802, 003 (1998). [arXiv:hep-th/9711162]
258. J. Madore, S. Schraml, P. Schupp, J. Wess, Gauge theory on noncommutative spaces. Eur.
Phys. J. C 16, 161 (2000). [arXiv:hep-th/0001203]
259. P. Kosiński, J. Lukierski, P. Maślanka, Local field theory on -Minkowski space, star products
and noncommutative translations. Czech. J. Phys. 50, 1283 (2000). [arXiv:hep-th/0009120]
260. A. Agostini, F. Lizzi, A. Zampini, Generalized Weyl systems and -Minkowski space. Mod.
Phys. Lett. A 17, 2105 (2002). [arXiv:hep-th/0209174]
261. F. Lizzi, G. Mangano, G. Miele, M. Peloso, Cosmological perturbations and short distance
physics from noncommutative geometry. JHEP 0206, 049 (2002). [arXiv:hep-th/0203099]
262. H.J. Groenewold, On the principles of elementary quantum mechanics. Physica 12, 405
(1946)
263. J.E. Moyal, Quantum mechanics as a statistical theory. Proc. Camb. Philos. Soc. 45, 99 (1949)
264. A. Kempf, Mode generating mechanism in inflation with a cutoff. Phys. Rev. D 63, 083514
(2001). [arXiv:astro-ph/0009209]
265. A. Kempf, J.C. Niemeyer, Perturbation spectrum in inflation with a cutoff. Phys. Rev. D 64,
103501 (2001). [arXiv:astro-ph/0103225]
266. R. Easther, B.R. Greene, W.H. Kinney, G. Shiu, Inflation as a probe of short distance physics.
267. O. Bertolami, C.A.D. Zarro, Towards a noncommutative astrophysics. Phys. Rev. D 81,
025005 (2010). [arXiv:0908.4196]
268. H. García-Compeán, O. Obregón, C. Ramírez, Noncommutative quantum cosmology. Phys.
269. C. Bastos, O. Bertolami, N. Costa Dias, J.N. Prata, Phase-space noncommutative quantum
270. C. Bastos, O. Bertolami, N. Costa Dias, J.N. Prata, Black holes and phase space noncommu-
tativity. Phys. Rev. D 80, 124038 (2009). [arXiv:0907.1818]
271. C. Bastos, O. Bertolami, N. Costa Dias, J.N. Prata, The singularity problem and phase-space
noncanonical noncommutativity. Phys. Rev. D 82, 041502(R) (2010). [arXiv:0912.4027]
272. C. Bastos, O. Bertolami, N. Costa Dias, J.N. Prata, Noncanonical phase-space noncommuta-
tivity and the Kantowski–Sachs singularity for black holes. Phys. Rev. D 84, 024005 (2011).
[arXiv:1012.5523]
273. M. Rinaldi, A new approach to non-commutative inflation. Class. Quantum Grav. 28, 105022
(2011). [arXiv:0908.1949]
274. H. Perrier, R. Durrer, M. Rinaldi, Explosive particle production in non-commutative inflation.
JHEP 1301, 067 (2013). [arXiv:1210.5373]
275. S. Alexander, J. Magueijo, Noncommutative geometry as a realization of varying speed of
light cosmology. arXiv:hep-th/0104093
276. S. Alexander, R. Brandenberger, J. Magueijo, Noncommutative inflation. Phys. Rev. D 67,
081301 (2003). [arXiv:hep-th/0108190]
277. S. Koh, R.H. Brandenberger, Cosmological perturbations in non-commutative inflation. JCAP
0706, 021 (2007). [arXiv:hep-th/0702217]
278. U.D. Machado, R. Opher, Conceptual problem for noncommutative inflation and the new
approach for nonrelativistic inflationary equation of state. Phys. Rev. D 87, 123517 (2013).
[arXiv:1211.6478]
279. P.J.E. Peebles, Principles of Physical Cosmology (Princeton University Press, Princeton,
1993)
280. J. Magueijo, L. Pogosian, Could thermal fluctuations seed cosmic structure? Phys. Rev. D
67, 043518 (2003). [arXiv:astro-ph/0211337]
281. T. Biswas, R. Brandenberger, T. Koivisto, A. Mazumdar, Cosmological perturbations from
statistical thermal fluctuations. Phys. Rev. D 88, 023517 (2013). [arXiv:1302.6463]
282. J. Magueijo, P. Singh, Thermal fluctuations in loop cosmology. Phys. Rev. D 76, 023510
(2007). [arXiv:astro-ph/0703566]
283. F. Lizzi, G. Mangano, G. Miele, G. Sparano, Inflationary cosmology from noncommutative
geometry. Int. J. Mod. Phys. A 11, 2907 (1996). [arXiv:gr-qc/9503040]
284. O. Bertolami, L. Guisado, Noncommutative scalar field coupled to gravity. Phys. Rev. D 67,
025001 (2003). [arXiv:gr-qc/0207124]
285. E. Di Grezia, G. Esposito, A. Funel, G. Mangano, G. Miele, Spacetime noncommutativity
and antisymmetric tensor dynamics in the early Universe. Phys. Rev. D 68, 105012 (2003).
286. S.A. Alavi, F. Nasseri, Running of the spectral index in noncommutative inflation. Int. J.
Mod. Phys. A 20, 4941 (2005). [arXiv:astro-ph/0406477]
287. C.-S. Chu, B.R. Greene, G. Shiu, Remarks on inflation and noncommutative geometry. Mod.
Phys. Lett. A 16, 2231 (2001). [arXiv:hep-th/0011241]
288. E. Akofor, A.P. Balachandran, S.G. Jo, A. Joseph, B.A. Qureshi, Direction-dependent CMB
power spectrum and statistical anisotropy from noncommutative geometry. JHEP 0805, 092
(2008). [arXiv:0710.5897]
289. T.S. Koivisto, D.F. Mota, CMB statistics in noncommutative inflation. JHEP 1102, 061
(2011). [arXiv:1011.2126]
290. A. Nautiyal, Anisotropic non-gaussianity with noncommutative spacetime. Phys. Lett. B 728,
472 (2014). [arXiv:1303.4159]
291. E. Akofor, A.P. Balachandran, A. Joseph, L. Pekowsky, B.A. Qureshi, Constraints from CMB
on spacetime noncommutativity and causality violation. Phys. Rev. D 79, 063004 (2009).
[arXiv:0806.2458]
292. R. Brandenberger, P.-M. Ho, Noncommutative spacetime, stringy spacetime uncertainty
principle, and density fluctuations. Phys. Rev. D 66, 023517 (2002). [arXiv:hep-th/0203119]
293. S. Tsujikawa, R. Maartens, R. Brandenberger, Non-commutative inflation and the CMB.
294. Q.-G. Huang, M. Li, CMB power spectrum from noncommutative spacetime. JHEP 0306,
014 (2003). [arXiv:hep-th/0304203]
295. M. Fukuma, Y. Kono, A. Miwa, Effects of space-time noncommutativity on the angular power
spectrum of the CMB. Nucl. Phys. B 682, 377 (2004). [arXiv:hep-th/0307029]
296. Q.-G. Huang, M. Li, Noncommutative inflation and the CMB multipoles. JCAP 11, 001
(2003). [arXiv:astro-ph/0308458]
297. Q.-G. Huang, M. Li, Power spectra in spacetime noncommutative inflation. Nucl. Phys. B
713, 219 (2005). [arXiv:astro-ph/0311378]
298. H. Kim, G.S. Lee, Y.S. Myung, Noncommutative spacetime effect on the slow-roll period of
inflation. Mod. Phys. Lett. A 20, 271 (2005). [arXiv:hep-th/0402018]
299. H. Kim, G.S. Lee, H.W. Lee, Y.S. Myung, Second-order corrections to noncommutative
spacetime inflation. Phys. Rev. D 70, 043521 (2004). [arXiv:hep-th/0402198]
300. R.-G. Cai, A note on curvature fluctuation of noncommutative inflation. Phys. Lett. B 593, 1
(2004). [arXiv:hep-th/0403134]
301. G. Calcagni, Noncommutative models in patch cosmology. Phys. Rev. D 70, 103525 (2004).
302. G. Calcagni, Consistency relations and degeneracies in (non)commutative patch inflation.
Phys. Lett. B 606, 177 (2005). [arXiv:hep-ph/0406057]
304. Q.-G. Huang, M. Li, Running spectral index in noncommutative inflation and WMAP three
year results. Nucl. Phys. B 755, 286 (2006). [arXiv:astro-ph/0603782]
References 619
305. X. Zhang, F.-Q. Wu, Noncommutative chaotic inflation and WMAP three year results. Phys.
306. G. Calcagni, S. Kuroyanagi, J. Ohashi, S. Tsujikawa, Strong Planck constraints on braneworld
and non-commutative inflation. JCAP 1403, 052 (2014). [arXiv:1310.5186]
307. X. Bekaert, N. Boulanger, D. Francia, Mixed-symmetry multiplets and higher-spin curvatures.
J. Phys. A 48, 225401 (2015). [arXiv:1501.02462]
308. E. Tomboulis, 1=N expansion and renormalization in quantum gravity. Phys. Lett. B 70, 361
(1977)
309. E. Tomboulis, Renormalizability and asymptotic freedom in quantum gravity. Phys. Lett. B
97, 77 (1980)
310. M. Kaku, Strong-coupling approach to the quantization of conformal gravity. Phys. Rev. D
27, 2819 (1983)
311. E.T. Tomboulis, Unitarity in higher-derivative quantum gravity. Phys. Rev. Lett. 52, 1173
(1984)
312. M. Ostrogradski, Mémoire sur les équations différentielles relatives au problème des
isopérimètres. Mem. Act. St. Petersbourg VI 4, 385 (1850)
313. D.A. Eliezer, R.P. Woodard, The problem of nonlocality in string theory. Nucl. Phys. B 325,
389 (1989)
314. J.Z. Simon, Higher-derivative Lagrangians, nonlocality, problems, and solutions. Phys. Rev.
D 41, 3720 (1990)
315. R.P. Woodard, Ostrogradsky’s theorem on Hamiltonian instability. Scholarpedia 10, 32243
(2015). [arXiv:1506.02210]
316. G. Wataghin, Bemerkung über die Selbstenergie der Elektronen. Z. Phys. 88, 92 (1934)
317. F. Bopp, Lineare Theorie des Elektrons. II. Ann. Phys. (Berlin) 434, 573 (1943)
318. R.P. Feynman, A relativistic cut-off for classical electrodynamics. Phys. Rev. 74, 939 (1948)
319. A. Pais, G.E. Uhlenbeck, On field theories with non-localized action. Phys. Rev. 79, 145
(1950)
320. N. Shôno, N. Oda, Note on the non-local interaction. Prog. Theor. Phys. 8, 28 (1952)
321. W. Pauli, On the hamiltonian structure of non-local field theories. Nuovo Cim. 10, 648 (1953)
322. M. Chrétien, R.E. Peierls, Properties of form factors in non-local theories. Nuovo Cim. 10,
668 (1953)
323. C. Hayashi, Hamiltonian formalism in non-local field theories. Prog. Theor. Phys. 10, 533
(1953)
324. C. Hayashi, On field equations with non-local interaction. Prog. Theor. Phys. 11, 226 (1954).
325. M. Chrétien, R.E. Peierls, A study of gauge-invariant non-local interactions. Proc. R. Soc.
Lond. A 223, 468 (1954) [World Sci. Ser. 20th Cent. Phys. 19, 397 (1997)]
326. M. Meyman, The causality principle and the asymptotic behavior of the scattering amplitude.
Zh. Eksp. Teor. Fiz. 47, 1966 (1965) [Sov. Phys. JETP 20, 1320 (1965)]
327. G.V. Efimov, Non-local quantum theory of the scalar field. Commun. Math. Phys. 5, 42 (1967)
328. G.V. Efimov, On a class of relativistic invariant distributions. Commun. Math. Phys. 7, 138
(1968)
329. M.Z. Iofa, V.Ya. Fainberg, Wightman formulation for a nonlocalizable field theory. I. Zh.
Eksp. Teor. Fiz. 56, 1644 (1969) [Sov. Phys. JETP 29, 880 (1969)]
330. M.Z. Iofa, V.Ya. Fainberg, Wightman formulation for nonlocalizable field theories II. Theory
of asymptotic fields and particles. Teor. Mat. Fiz. 1, 187 (1969) [Theor. Math. Phys. 1, 143
(1969)]
331. V.A. Alebastrov, G.V. Efimov, A proof of the unitarity of S matrix in a nonlocal quantum field
theory. Commun. Math. Phys. 31, 1 (1973)
332. V.A. Alebastrov, G.V. Efimov, Causality in quantum field theory with nonlocal interaction.
Commun. Math. Phys. 38, 11 (1974)
333. G.V. Efimov, Nelokalnye vzaimodestvi kvantovannyh pole [Nonlocal Inter-
actions of Quantized Fields (in Russian)] (Nauka, Moscow, 1977)
334. V.Ya. Fainberg, M.A. Soloviev, How can local properties be described in field theories without
strict locality? Ann. Phys. (N.Y.) 113, 421 (1978)
335. V.Ya. Fainberg, M.A. Soloviev, Nonlocalizability and asymptotical commutativity. Teor. Mat.
Fiz. 93, 514 (1992) [Theor. Math. Phys. 93, 1438 (1992)]
336. R.L.P. do Amaral, E.C. Marino, Canonical quantization of theories containing fractional
powers of the d’Alembertian operator. J. Phys. A 25, 5183 (1992)
337. D.G. Barci, L.E. Oxman, M. Rocca, Canonical quantization of non-local field equations. Int.
338. N. Moeller, B. Zwiebach, Dynamics with infinitely many time derivatives and rolling
tachyons. JHEP 0210, 034 (2002). [arXiv:hep-th/0207107]
339. G. Calcagni, G. Nardelli, Tachyon solutions in boundary and cubic string field theory. Phys.
Rev. D 78, 126010 (2008). [arXiv:0708.0366]
340. G. Calcagni, M. Montobbio, G. Nardelli, Route to nonlocal cosmology. Phys. Rev. D 76,
126001 (2007). [arXiv:0705.3043]
341. G. Calcagni, M. Montobbio, G. Nardelli, Localization of nonlocal theories. Phys. Lett. B
662, 285 (2008). [arXiv:0712.2237]
342. G. Calcagni, G. Nardelli, Nonlocal instantons and solitons in string models. Phys. Lett. B
669, 102 (2008). [arXiv:0802.4395]
343. D.J. Mulryne, N.J. Nunes, Diffusing nonlocal inflation: solving the field equations as an initial
value problem. Phys. Rev. D 78, 063519 (2008). [arXiv:0805.0449]
344. G. Calcagni, G. Nardelli, Kinks of open superstring field theory. Nucl. Phys. B 823, 234
(2009). [arXiv:0904.3744]
345. G. Calcagni, G. Nardelli, Cosmological rolling solutions of nonlocal theories. Int. J. Mod.
Phys. D 19, 329 (2010). [arXiv:0904.4245]
346. G. Calcagni, G. Nardelli, String theory as a diffusing system. JHEP 1002, 093 (2010).
[arXiv:0910.2160]
347. N. Barnaby, N. Kamran, Dynamics with infinitely many derivatives: the initial value problem.
JHEP 0802, 008 (2008). [arXiv:0709.3968]
348. E.T. Tomboulis, Super-renormalizable gauge and gravitational theories.
arXiv:hep-th/9702146.
349. T. Biswas, A. Mazumdar, W. Siegel, Bouncing universes in string-inspired gravity. JCAP
0603, 009 (2006). [arXiv:hep-th/0508194]
350. G.V. Efimov, Amplitudes in nonlocal theories at high energies. Teor. Mat. Fiz. 128, 395 (2001)
[Theor. Math. Phys. 128, 1169 (2001)]
351. E.T. Tomboulis, Nonlocal and quasilocal field theories. Phys. Rev. D 92, 125037 (2015).
[arXiv:1507.00981]
352. N.V. Krasnikov, Nonlocal gauge theories. Teor. Mat. Fiz. 73, 235 (1987) [Theor. Math. Phys.
73, 1184 (1987)]
353. J.W. Moffat, Finite nonlocal Gauge field theory. Phys. Rev. D 41, 1177 (1990)
354. B.J. Hand, J.W. Moffat, Nonlocal regularization and the one-loop topological mass in three-
dimensional QED. Phys. Rev. D 43, 1896 (1991)
355. D. Evens, J.W. Moffat, G. Kleppe, R.P. Woodard, Nonlocal regularizations of gauge theories.
Phys. Rev. D 43, 499 (1991)
356. N.J. Cornish, New methods in quantum nonlocal field theory. Mod. Phys. Lett. A 07, 1895
(1992)
357. N.J. Cornish, Quantum nonlocal field theory: physics without infinities. Int. J. Mod. Phys. A
07, 6121 (1992)
358. N.J. Cornish, Quantum non-local gravity. Mod. Phys. Lett. A 07, 631 (1992)
359. J.W. Moffat, Ultraviolet complete quantum gravity. Eur. Phys. J. Plus 126, 43 (2011).
[arXiv:1008.2482]
360. J. Khoury, Fading gravity and self-inflation. Phys. Rev. D 76, 123513 (2007).
361. G. Calcagni, G. Nardelli, Nonlocal gravity and the diffusion equation. Phys. Rev. D 82,
123518 (2010). [arXiv:1004.5144]
362. T. Biswas, T. Koivisto, A. Mazumdar, Towards a resolution of the cosmological singularity in
non-local higher derivative theories of gravity. JCAP 1011, 008 (2010). [arXiv:1005.0590]
References 621
363. L. Modesto, Super-renormalizable quantum gravity. Phys. Rev. D 86, 044005 (2012).
[arXiv:1107.2403]
364. T. Biswas, E. Gerwick, T. Koivisto, A. Mazumdar, Towards singularity- and ghost-free
theories of gravity. Phys. Rev. Lett. 108, 031101 (2012). [arXiv:1110.5249]
365. S. Alexander, A. Marcianó, L. Modesto, The hidden quantum groups symmetry of super-
renormalizable gravity. Phys. Rev. D 85, 124030 (2012). [arXiv:1202.1824]
366. L. Modesto, Super-renormalizable multidimensional quantum gravity: theory and applica-
tions. Astron. Rev. 8, 4 (2013). [arXiv:1202.3151]
367. L. Modesto, Towards a finite quantum supergravity. arXiv:1206.2648
368. G. Calcagni, L. Modesto, Nonlocal quantum gravity and M-theory. Phys. Rev. D 91, 124059
(2015). [arXiv:1404.2137]
369. L. Modesto, L. Rachwał, Super-renormalizable and finite gravitational theories. Nucl. Phys.
B 889, 228 (2014). [arXiv:1407.8036]
370. M. Asorey, J.L. López, I.L. Shapiro, Some remarks on high derivative quantum gravity. Int.
371. K.S. Stelle, Renormalization of higher-derivative quantum gravity. Phys. Rev. D 16, 953
(1977)
372. A. Accioly, A. Azeredo, H. Mukai, Propagator, tree-level unitarity and effective nonrelativis-
tic potential for higher-derivative gravity theories in D dimensions. J. Math. Phys. 43, 473
(2002)
373. P. Van Nieuwenhuizen, On ghost-free tensor Lagrangians and linearized gravitation. Nucl.
Phys. B 60, 478 (1973)
374. S. Talaganis, T. Biswas, A. Mazumdar, Towards understanding the ultraviolet behavior of
quantum loops in infinite-derivative theories of gravity. Class. Quantum Grav. 32, 215017
(2015). [arXiv:1412.3467]
375. M.J. Duff, D.J. Toms, Kaluza–Klein–Kounterterms, in Unification of Fundamental Particle
Interactions II, ed. by J. Ellis, S. Ferrara (Springer, Amsterdam, 1983)
376. S. Deser, D. Seminara, Tree amplitudes and two loop counterterms in D D 11 supergravity.
377. A.S. Koshelev, Stable analytic bounce in non-local Einstein–Gauss–Bonnet cosmology. Class.
378. T. Biswas, A. Conroy, A.S. Koshelev, A. Mazumdar, Generalized ghost-free quadratic
curvature gravity. Class. Quantum Grav. 31, 015022 (2014); Erratum-ibid. 31, 159501 (2014).
[arXiv:1308.2319]
379. H.-J. Schmidt, Variational derivatives of arbitrarily high order and multi-inflation cosmologi-
cal models. Class. Quantum Grav. 7, 1023 (1990)
380. I.Ya. Aref’eva, Nonlocal string tachyon as a model for cosmological dark energy. AIP Conf.
Proc. 826, 301 (2006). [arXiv:astro-ph/0410443]
381. I.Ya. Aref’eva, L.V. Joukovskaya, Time lumps in nonlocal stringy models and cosmological
applications. JHEP 0510, 087 (2005). [arXiv:hep-th/0504200]
382. I.Ya. Aref’eva, A.S. Koshelev, S.Yu. Vernov, Stringy dark energy model with cold dark matter.
383. G. Calcagni, Cosmological tachyon from cubic string field theory. JHEP 0605, 012 (2006).
384. I.Ya. Aref’eva, A.S. Koshelev, Cosmic acceleration and crossing of w D 1 barrier from
cubic superstring field theory. JHEP 0702, 041 (2007). [arXiv:hep-th/0605085]
385. I.Ya. Aref’eva, I.V. Volovich, On the null energy condition and cosmology. Theor. Math.
Phys. 155, 503 (2008). [arXiv:hep-th/0612098]
386. N. Barnaby, T. Biswas, J.M. Cline, p-adic inflation. JHEP 0704, 056 (2007).
387. A.S. Koshelev, Non-local SFT tachyon and cosmology. JHEP 0704, 029 (2007).
388. I.Ya. Aref’eva, L.V. Joukovskaya, S.Yu. Vernov, Bouncing and accelerating solutions in
nonlocal stringy models. JHEP 0707, 087 (2007). [arXiv:hep-th/0701184]
389. I.Ya. Aref’eva, I.V. Volovich, Quantization of the Riemann zeta-function and cosmology. Int.
J. Geom. Methods Mod. Phys. 4, 881 (2007). [arXiv:hep-th/0701284]
390. J.E. Lidsey, Stretching the inflaton potential with kinetic energy. Phys. Rev. D 76, 043511
(2007). [arXiv:hep-th/0703007]
391. N. Barnaby, J.M. Cline, Large non-Gaussianity from non-local inflation. JCAP 0707, 017
(2007). [arXiv:0704.3426]
392. L.V. Joukovskaya, Dynamics in nonlocal cosmological models derived from string field
393. L. Joukovskaya, Rolling solution for tachyon condensation in open string field theory.
arXiv:0803.3484
394. I.Ya. Aref’eva, A.S. Koshelev, Cosmological signature of tachyon condensation. JHEP 0809,
068 (2008). [arXiv:0804.3570]
395. L. Joukovskaya, Dynamics with infinitely many time derivatives in Friedmann–Robertson–
Walker background and rolling tachyons. JHEP 0902, 045 (2009). [arXiv:0807.2065]
396. N. Barnaby, N. Kamran, Dynamics with infinitely many derivatives: variable coefficient
equations. JHEP 0812, 022 (2008). [arXiv:0809.4513]
397. N.J. Nunes, D.J. Mulryne, Non-linear non-local cosmology. AIP Conf. Proc. 1115, 329
(2009). [arXiv:0810.5471]
398. A.S. Koshelev, S.Yu. Vernov, Cosmological perturbations in SFT inspired non-local scalar
field models. Eur. Phys. J. C 72, 2198 (2012). [arXiv:0903.5176]
399. S.Yu. Vernov, Localization of non-local cosmological models with quadratic potentials in the
case of double roots. Class. Quantum Grav. 27, 035006 (2010). [arXiv:0907.0468]
400. S.Yu. Vernov, Localization of the SFT inspired nonlocal linear models and exact solutions.
Phys. Part. Nucl. Lett. 8, 310 (2011). [arXiv:1005.0372]
401. A.S. Koshelev, S.Yu. Vernov, Analysis of scalar perturbations in cosmological models with a
non-local scalar field. Class. Quantum Grav. 28, 085019 (2011). [arXiv:1009.0746]
402. A.S. Koshelev, S.Yu. Vernov, On bouncing solutions in non-local gravity. Phys. Part. Nucl.
43, 666 (2012). [arXiv:1202.1289]
403. T. Biswas, A.S. Koshelev, A. Mazumdar, S.Yu. Vernov, Stable bounce and inflation in non-
local higher derivative cosmology. JCAP 1208, 024 (2012). [arXiv:1206.6374]
404. F. Briscese, A. Marcianò, L. Modesto, E.N. Saridakis, Inflation in (super-)renormalizable
gravity. Phys. Rev. D 87, 083507 (2013). [arXiv:1212.3611]
405. G. Calcagni, L. Modesto, P. Nicolini, Super-accelerating bouncing cosmology in
asymptotically-free non-local gravity. Eur. Phys. J. C 74, 2999 (2014). [arXiv:1306.5332]
406. C. Bambi, D. Malafarina, L. Modesto, Terminating black holes in quantum gravity. Eur.
Phys. J. C 74, 2767 (2014). [arXiv:1306.1668]
407. N. Arkani-Hamed, S. Dimopoulos, G. Dvali, G. Gabadadze, Nonlocal modification of gravity
and the cosmological constant problem. arXiv:hep-th/0209227
408. M.E. Soussa, R.P. Woodard, A nonlocal metric formulation of MOND. Class. Quantum Grav.
20, 2737 (2003). [arXiv:astro-ph/0302030]
409. A.O. Barvinsky, Nonlocal action for long-distance modifications of gravity theory. Phys.
410. A.O. Barvinsky, On covariant long-distance modifications of Einstein theory and strong
coupling problem. Phys. Rev. D 71, 084007 (2005). [arXiv:hep-th/0501093]
411. H.W. Hamber, R.M. Williams, Nonlocal effective gravitational field equations and the running
of Newton’s constant G. Phys. Rev. D 72, 044026 (2005). [arXiv:hep-th/0507017]
412. S. Deser, R.P. Woodard, Nonlocal cosmology. Phys. Rev. Lett. 99, 111301 (2007).
[arXiv:0706.2151]
413. S. Nojiri, S.D. Odintsov, Modified non-local-F.R/ gravity as the key for the inflation and dark
energy. Phys. Lett. B 659, 821 (2008). [arXiv:0708.0924]
414. S. Jhingan, S. Nojiri, S.D. Odintsov, M. Sami, I. Thongkool, S. Zerbini, Phantom and non-
phantom dark energy: the cosmological relevance of non-locally corrected gravity. Phys.
Lett. B 663, 424 (2008). [arXiv:0803.2613]
References 623
415. T.S. Koivisto, Dynamics of nonlocal cosmology. Phys. Rev. D 77, 123513 (2008).
[arXiv:0803.3399]
416. T.S. Koivisto, Newtonian limit of nonlocal cosmology. Phys. Rev. D 78, 123505 (2008).
[arXiv:0807.3778]
417. S. Capozziello, E. Elizalde, S. Nojiri, S.D. Odintsov, Accelerating cosmologies from non-
local higher-derivative gravity. Phys. Lett. B 671, 193 (2009). [arXiv:0809.1535]
418. C. Deffayet, R.P. Woodard, Reconstructing the distortion function for nonlocal cosmology.
JCAP 0908, 023 (2009). [arXiv:0904.0961]
419. G. Cognola, E. Elizalde, S. Nojiri, S.D. Odintsov, S. Zerbini, One-loop effective action for
non-local modified Gauss–Bonnet gravity in de Sitter space. Eur. Phys. J. C 64, 483 (2009).
[arXiv:0905.0543]
420. S. Nojiri, S.D. Odintsov, M. Sasaki, Y.-l. Zhang, Screening of cosmological constant in non-
local gravity. Phys. Lett. B 696, 278 (2011). [arXiv:1010.5375]
421. S. Nojiri, S.D. Odintsov, Unified cosmic history in modified gravity: from F.R/ theory to
Lorentz non-invariant models. Phys. Rep. 505, 59 (2011). [arXiv:1011.0544]
422. A.O. Barvinsky, Dark energy and dark matter from nonlocal ghost-free gravity theory. Phys.
Lett. B 710, 12 (2012). [arXiv:1107.1463]
423. Y.-l. Zhang, M. Sasaki, Screening of cosmological constant in non-local cosmology. Int. J.
Mod. Phys. D 21, 1250006 (2012). [arXiv:1108.2112]
424. E. Elizalde, E.O. Pozdeeva, S.Yu. Vernov, Y.-l. Zhang, Cosmological solutions of a nonlocal
model with a perfect fluid. JCAP 1307, 034 (2013). [arXiv:1302.4330]
425. M. Jaccard, M. Maggiore, E. Mitsou, Nonlocal theory of massive gravity. Phys. Rev. D 88,
044033 (2013). [arXiv:1305.3034]
426. S. Deser, R.P. Woodard, Observational viability and stability of nonlocal cosmology. JCAP
1311, 036 (2013). [arXiv:1307.6639]
427. L. Modesto, S. Tsujikawa, Non-local massive gravity. Phys. Lett. B 727, 48 (2013).
[arXiv:1307.6968]
428. S. Foffa, M. Maggiore, E. Mitsou, Apparent ghosts and spurious degrees of freedom in non-
local theories. Phys. Lett. B 733, 76 (2014). [arXiv:1311.3421]
429. A. Conroy, T. Koivisto, A. Mazumdar, A. Teimouri, Generalised quadratic curvature, non-
local infrared modifications of gravity and Newtonian potentials. Class. Quantum Grav. 32,
015024 (2015). [arXiv:1406.4998]
430. G. ’t Hooft, Dimensional reduction in quantum gravity, in Salamfestschrift, ed. by A. Ali, J.
Ellis, S. Randjbar-Daemi (World Scientific, Singapore, 1993). [arXiv:gr-qc/9310026]
431. S. Carlip, Spontaneous dimensional reduction in short-distance quantum gravity? AIP Conf.
Proc. 1196, 72 (2009). [arXiv:0909.3329]
432. G. Calcagni, Fractal universe and quantum gravity. Phys. Rev. Lett. 104, 251301 (2010).
[arXiv:0912.3142]
433. S. Carlip, The small scale structure of spacetime, in [78]. [arXiv:1009.1136]
434. D. Benedetti, J. Henson, Spectral geometry as a probe of quantum spacetime. Phys. Rev. D
80, 124036 (2009). [arXiv:0911.0401]
435. T.P. Sotiriou, M. Visser, S. Weinfurtner, Spectral dimension as a probe of the ultraviolet
continuum regime of causal dynamical triangulations. Phys. Rev. Lett. 107, 131303 (2011).
[arXiv:1105.5646]
436. M.R. Atkin, G. Giasemidis, J.F. Wheater, Continuum random combs and scale dependent
spectral dimension. J. Phys. A 44, 265001 (2011). [arXiv:1101.4174]
437. G. Giasemidis, J.F. Wheater, S. Zohren, Dynamical dimensional reduction in toy models of
4D causal quantum gravity. Phys. Rev. D 86, 081503(R) (2012). [arXiv:1202.2710]
438. G. Giasemidis, J.F. Wheater, S. Zohren, Multigraph models for causal quantum gravity and
scale dependent spectral dimension. J. Phys. A 45, 355001 (2012). [arXiv:1202.6322]
439. F. Caravelli, L. Modesto, Fractal dimension in 3d spin-foams. arXiv:0905.2170
440. E. Magliaro, C. Perini, L. Modesto, Fractal space-time from spin-foams. arXiv:0911.0437
441. A. Connes, Noncommutative geometry and the standard model with neutrino mixing. JHEP
0611, 081 (2006). [arXiv:hep-th/0608226]
442. A.H. Chamseddine, A. Connes, M. Marcolli, Gravity and the standard model with neutrino
mixing. Adv. Theor. Math. Phys. 11, 991 (2007). [arXiv:hep-th/0610241]
443. D. Benedetti, Fractal properties of quantum spacetime. Phys. Rev. Lett. 102, 111303 (2009).
[arXiv:0811.1396]
444. M. Arzano, T. Trześniewski, Diffusion on -Minkowski space. Phys. Rev. D 89, 124024
(2014). [arXiv:1404.4762]
445. Anjana V., E. Harikumar, Spectral dimension of kappa-deformed spacetime. Phys. Rev. D 91,
065026 (2015). [arXiv:1501.00254]
446. Anjana V., E. Harikumar, Dimensional flow in the kappa-deformed spacetime. Phys. Rev. D
92, 045014 (2015). [arXiv:1504.07773]
447. G. Calcagni, D. Oriti, J. Thürigen, Spectral dimension of quantum geometries. Class.
448. G. Calcagni, D. Oriti, J. Thürigen, Dimensional flow in discrete quantum geometries. Phys.
Rev. D 91, 084047 (2015). [arXiv:1412.8390]
449. P. Hořava, Spectral dimension of the universe in quantum gravity at a Lifshitz point. Phys.
Rev. Lett. 102, 161301 (2009). [arXiv:0902.3657]
450. S. Carlip, D. Grumiller, Lower bound on the spectral dimension near a black hole. Phys. Rev.
D 84, 084029 (2011). [arXiv:1108.4686]
451. J.R. Mureika, Primordial black hole evaporation and spontaneous dimensional reduction.
Phys. Lett. B 716, 171 (2012). [arXiv:1204.3619]
452. M. Arzano, G. Calcagni, Black-hole entropy and minimal diffusion. Phys. Rev. D 88, 084017
(2013). [arXiv:1307.6122]
453. L. Modesto, P. Nicolini, Spectral dimension of a quantum universe. Phys. Rev. D 81, 104040
(2010). [arXiv:0912.0220]
454. G Calcagni, Geometry of fractional spaces. Adv. Theor. Math. Phys. 16, 549 (2012).
[arXiv:1106.5787]
455. G. Calcagni, Geometry and field theory in multi-fractional spacetime. JHEP 1201, 065 (2012).
[arXiv:1107.5041]
456. G. Calcagni, Multiscale spacetimes from first principles. arXiv:1609.02776
457. G. Calcagni, L. Modesto, Nonlocality in string theory. J. Phys. A 47, 355402 (2014).
[arXiv:1310.4957]
458. G. Calcagni, G. Nardelli, Quantum field theory with varying couplings. Int. J. Mod. Phys. A
29, 1450012 (2014). [arXiv:1306.0629]
459. G. Amelino-Camelia, M. Arzano, G. Gubitosi, J. Magueijo, Dimensional reduction in the sky.
Phys. Rev. D 87, 123532 (2013). [arXiv:1305.3153]
460. G. Amelino-Camelia, M. Arzano, G. Gubitosi, J. Magueijo, Rainbow gravity and scale-
invariant fluctuations. Phys. Rev. D 88, 041303 (2013). [arXiv:1307.0745]
461. G. Amelino-Camelia, M. Arzano, G. Gubitosi, J. Magueijo, Dimensional reduction in
momentum space and scale-invariant cosmological fluctuations. Phys. Rev. D 88, 103524
(2013). [arXiv:1309.3999]
462. G. Calcagni, Multi-scale gravity and cosmology. JCAP 1312, 041 (2013). [arXiv:1307.6382]
463. G. Calcagni, S. Kuroyanagi, S. Tsujikawa, Cosmic microwave background and inflation in
multi-fractional spacetimes. JCAP 1608, 039 (2016). [arXiv:1606.08449]
Chapter 12
String Theory
Quaedam processu priorem exuunt formam et in novam

transeunt. Ubi aliquid animus diu protulit et magnitudinem eius
sequendo lassatus est, infinitum coepit vocari; quod longe aliud
factum est quam fuit cum magnum videretur sed finitum. Eodem
modo aliquid difficulter secari cogitavimus: novissime crescente
hac difficultate insecabile inventum est.
— Seneca, Ad Lucilium Epistularum Moralium, XX, 118, 17
Certain things undergo a transformation from their prior form
into a new one. When the mind extends an object and gets tired
of following its magnitude, one begins to call it infinity; which is
by far a different thing than what it was when it was seen large
but finite. In the same way, we reflect upon something which is
difficult to dissect: as the difficulty of this sub-division increases,
we invent the indivisible.
Contents
12.1 Bosonic String. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 627
12.1.1 Classical Free Strings and Branes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 627
12.1.2 D-Branes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 631
12.1.3 Quantum Strings and Critical Dimension. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 632
12.1.4 Interactions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 637
12.1.5 Low-Energy Limit. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 638
12.1.6 String Field Theory. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 640
12.2 Superstring. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 642
12.2.1 Action. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 643
12.2.2 Quantization. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 645
12.2.3 Type-I Superstring. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 646
12.2.4 Type-II Superstrings. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 647
12.2.5 Interactions and Anomaly Cancellation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 648
12.2.6 Heterotic Superstrings. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 649
12.2.7 Massless Spectra and Low-Energy Limits. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 651
12.2.8 Branes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 653
12.2.9 Superstring Field Theory. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 655
12.3 Compactification. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 656
12.3.1 T-Duality. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 656
(continued)

DOI 10.1007/978-3-319-41127-9_12
626 12 String Theory
12.3.2 Spontaneous Compactification. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 658

12.3.3 Calabi–Yau Spaces and Orbifolds. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 658
12.3.4 Cycles and Fluxes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 662
12.3.5 Moduli. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 663
12.3.6 Stacking Branes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 668
12.3.7 Flux Compactification. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 669
12.3.8 String Theory and the Standard Model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 671
12.3.9 Anti-de Sitter Vacua. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 671
12.4 Dualities and M-Theory. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 678
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 683
Between the late 1960s and the early 1970s, an attempt was made to describe
the quantum interactions of hadrons via the so-called dual (or dual-resonance)
models [1–16]. These were essentially S-matrix models, whose dynamics was
encoded in effective scattering amplitudes characterized by certain symmetries and
spectra. It was soon recognized that such dynamics, symmetries and spectra could
be reproduced by a Hamiltonian-Lagrangian model of quantum mechanics where
the fundamental objects were not point-wise particles but oscillating strings [17–
29]. It did not take long for this framework to acquire an autonomous life as a
fundamental theory including not only quark interactions but also the electroweak
and the gravitational force [30–33].
In the widest acceptance of the term, string theory is not a model of Nature
but a theoretical framework. Just as quantum field theory is a set of tools much
wider than the Standard Model of electroweak and strong interactions, the immense
apparatus of strings, branes and extra dimensions is much more than what we would
need in order to describe the observed natural phenomena at high energies. At the
same time, however, so far we have been unable to extract enough information from
the theory to complete this task in a fully satisfactory way. String theory has more
than four dimensions, much more than gravitational and matter degrees of freedom,
more symmetries than SU.3/ ˝ SU.2/ ˝ U.1/ and diffeomorphisms, more than one
graviton and the 17 particles of the Standard Model (12 fermions, 4 gauge bosons
and one scalar). From all the bounty the theory can offer, it is difficult to reproduce
these degrees of freedom in the low-energy limit while maintaining an acceptably
low number of assumptions, unobserved side effects and undesired fine tunings.
Nevertheless, string theory continues to fascinate for its economy of thought, its
mathematical beauty and the tantalizing possibility of solving many fundamental
problems in a unified frame.
Contrary to the majority of the other theories so far examined in the book,
the broad goal of string theory is not to quantize gravity but, rather, to have all
interactions emerging from a more fundamental set-up that can be studied with
a language closer to that of particle physics than to that of general relativity.
Diffeomorphism invariance and the very concept of spacetime do not play a central
role because the gravitational action arises only in the low-energy limit. Spacetime
12.1 Bosonic String 627
and the graviton stem from the degrees of freedom of a two-dimensional conformal
field theory, while diffeomorphisms are a low-energy byproduct of a very large
symmetry group. In this loose sense, string theory is an extremely sophisticated
scenario of emergent gravity.
This chapter is meant to give the cosmologist a compact and intuitive bird’s
eye view of the main features of string theory. Sections 12.1 and 12.2 introduce
the classical bosonic and supersymmetric strings and their quantizations, D-branes,
fluxes and string field theory. Section 12.3 is dedicated to the compactification of
string theory on torii, Calabi–Yau spaces and orbifolds and discusses the important
problem of the stabilization of string moduli in flux compactifications. Section 12.4
deals with string dualities and M-theory.
We will omit several technical and conceptual details, which can be found in
dedicated textbooks [34–37] and reviews [38]. Such details are important to take
control of the theory. We therefore encourage the reader seriously interested in
strings to consult first of all the just-mentioned books and, as far as the historical
development of the theory is concerned, the references in the present text.
12.1 Bosonic String
12.1.1 Classical Free Strings and Branes
When a string propagates, its trajectory is not a one-dimensional world-line as for

a point particle but a two-dimensional world-sheet, which is a strip for open strings
and a tubular surface for closed strings (Fig. 12.1). The free propagation of the
bosonic string in D-dimensional Minkowski spacetime is governed by the Polyakov
action [39–41]
Z C1
1 p
SŒ ; X D 0
d d ab @a X @b X
4 ˛ 1
Z C1 Z l
1
D d d @ X @ X @ X @ X ; (12.1)
4 ˛ 0 1 0
where ˛ 0 is called Regge slope (the tension of the string is equal to .2 ˛ 0 /1 [24]),
ab is the world-sheet metric (a; b D 1; 2) with signature .; C/, D ; are
a
coordinates parametrizing, respectively, the (infinite) proper-time length and the
σ σ
0
τ τ
Fig. 12.1 Open- and closed-string world-sheet

(finite) spatial width of the world-sheet, X . ; / are D scalar fields and Greek
indices are contracted with the Minkowski metric. By convention, for the closed
string l D 2 while l D for the open string. Since ŒX D 1, the Regge slopep
has dimension Œ˛ 0 D 2; it is the only free parameter of the theory and ls :D ˛ 0
fixes the length scale of the string.
In the second line of (12.1), we gauge fixed the world-sheet metric to the
flat one ab by combining world-sheet diffeomorphisms, X 0 . 0 a / D X . a /
and @a 0 c @b 0 d cd
0
D ab , and two-dimensional Weyl (or conformal) invariance,
0 2 c
ab D ˝ . / ab [42]. In infinitesimal form,
ıX D a @a X ; ı ab D ra b C rb a C ! ab ; (12.2)
where a and ! are, respectively, a vector and a scalar. Different gauge choices
correspond to different ways to embed the world-sheet in spacetime, while con-
formally equivalent metrics represent the same embedding. In general, the metric
ab . ; / is dynamical but it has no kinetic term because in two dimensions the Ricci
.2/
tensor is proportional to the metric itself: Rab D C ab . From the trace, it stems that
.2/
C D R =2 and the world-sheet Ricci scalar is a constant. The energy-momentum
tensor

2 ıS 1 1 c
Tab :D p D @a X @b X ab @c X @ X (12.3)
ı ab 2 ˛ 0 2
.2/ .2/
vanishes identically: Tab D Rab .1=2/ ab R D 0. As a consequence of
conformal invariance,
Ta a D 0 ; (12.4)
while diffeomorphism invariance guarantees that the energy-momentum tensor is

covariantly conserved. These conditions will also be enforced at the quantum level.
From (12.1), the equation of motion for each X is
.@2 C @2 /X D 0 : (12.5)
We Fourier transform X in the world-sheet coordinate :

Z C1
dk ik
X . ; / D e g.k; / Xk ; (12.6)
1 2
where g.k; / is some kernel function determined by the boundary conditions

necessary to solve (12.5). The latter is .@2 C k2 /g.k; / D 0, the harmonic oscillator
with solution g.k; / D bk sin.k/ C ck cos.k/.
12.1.1.1 Open String
The free open string is described by the Neumann boundary conditions

ˇ ˇ
ˇ ˇ
@ X . ; /ˇ D 0 D @ X . ; /ˇ ; (12.7)
D0 D
which require, respectively, that bk D 0 and that k D n be integer. Setting ck D 1,

the free open string solution is

p X ˛n

XNeu . ; / D x C 2˛ 0 p C i 2˛ 0 ein cos.n/ ; (12.8)
n
0¤n2Z

where we split the zero mode X0 into
p a constant x and a linear term p and we

rewrote the other modes as Xn D: i 2˛ ˛n =n. The vectors x and p are interpreted
0
as the position and momentum of the center of mass of the string. Other boundary
conditions will be discussed in Sect. 12.1.2.
For open strings with Neumann boundary conditions, the Hamiltonian is
Z Z
1
HD d T D d @ X @ X C @ X @ X
0 4 ˛ 0 0
1 X

D ˛m ˛m D: L0 ; (12.9)
2 m2Z
where in the last line we reserved a special symbol for the quantity
C1
X
L0 D ˛ 0 p2 C
˛m ˛m D: ˛ 0 p2 C N : (12.10)
mD1
R
The Hamiltonian H and the momentum P D 0 d T generate world-sheet
reparametrizations of, respectively, the coordinate and . Due to the trace con-
dition (12.4), there are no other independent components of the energy-momentum
tensor. The conserved charges associated with reparametrizations of the string
world-sheet are given by the Fourier (or, by recasting the system as a conformal
field theory on the complex plane [43–48], the Laurent) modes of such components.
We are at liberty to choose the linear combinations TCC :D .T C T /=2 and
T :D .T T /=2, so that instead of Fourier transforming P we can consider
Z X
1
Ln :D d .ein T C ein TCC / D
˛nm ˛m : (12.11)
0 2 m2Z
Here we used the mode decomposition (12.8) and the definition

p of the Kronecker
1
R 2 in
delta ın;0 :D .2/ 0
. By definition, ˛0 D 2˛ p . The charges Ln
0 d e
obey the Witt algebra
fLm ; Ln g D .m n/LmCn : (12.12)
The classical equations of motion (12.5) then correspond to H D 0 D P, i.e., Ln D 0

for all n 2 Z. The condition P D 0 reflects the fact that we can choose a gauge (the
so-called static gauge X 0 D t D ) where lines of constant and constant are
orthogonal, @ X @ X D 0. The latter condition allows one to write the equations
of motion (12.5) as [37]
X ˙
0 D .@ X ˙ @ X /.@ X ˙ @ X / D 4˛ 0 Ln ein ; ˙ :D ˙:
n2Z
(12.13)
This is another way to state that Ln D 0 classically.
12.1.1.2 Closed String
The closed string does not have end-points at which to specify boundary conditions.
On the other hand, it has two sets of wave modes, one moving clockwise (“right”, R)
and the other moving counter-clockwise (“left”, L). The general solution of (12.5)
reads

X . ; / D XR . / C XL . C / ; (12.14a)
r
˛ 0 X in ˛R;L n

x ˛0
X R ;L D C p C i e : (12.14b)
2 2 2 n
0¤n2Z
In the following and as in the standard literature, we will call ˛ and ˛Q the
oscillators ˛R and ˛L , respectively. Using the coordinates ˙ , one finds that
TCC D @C X @C X =.2 ˛ 0 / and T D @ X @ X =.2 ˛ 0 /, which are the only
non-vanishing components of the world-sheet energy-momentum tensor. All the
open-string expressions hold for the closed string,
p with some differences. One

is the definition of the zero modes, ˛0 D ˛ 0 =2 p D ˛Q 0 . The integration
range of the Hamiltonian and of the total momentum is now Œ0; 2, so that
R 2 R 2
H D 0 d .TCC C T / and P D 0 d .TCC T /, which also reflects
in (12.11):
Z 2 X
1
Ln :D d ein T D
˛nm ˛m ; (12.15)
0 2 m2Z
together with its left-moving counterpart LQ n defined with TCC . The closed-string
Hamiltonian and total momentum are then given by H D L0 C LQ 0 and P D L0 LQ 0 ,
where
C1
X
˛0 2 ˛0 2
L0 :D p C ˛m ˛m D p CN (12.16)
4 mD1
4
and LQ 0 D ˛ 0 p2 =4 C N.
Q Although we use the same symbols L0 and Ln as in the
open-string case, there will be no danger of confusion.
12.1.2 D-Branes
If one imposes the end-points of the open string to be fixed, then (12.7) is replaced
by the Dirichlet boundary conditions
ˇ ˇ
ˇ ˇ
X . ; /ˇ D x1 ; X . ; /ˇ D x2 : (12.17)
D0 D
At the end-points, the variation ıX is zero. Then, the general solution is
p X ˛n

XDir D x1 C .x2 x1 / C 2˛ 0 ein sin.n/ : (12.18)
n
0¤n2Z
Each of the string end-points spans a spacetime object called D-brane [49] (see
Fig. 12.2; “D” stands for Dirichlet). Branes can have different spatial dimensionality
p, in which case they are called Dp-branes.
For string systems to be dynamical, the time direction must satisfy a Neumann
boundary condition, otherwise X 0 D t D const. Therefore, ¤ 0 in (12.17) and
Fig. 12.2 Open strings with Dirichlet boundary conditions. The end-points are fixed on two D-
branes or on the same brane, while the string can oscillate in the bulk. Closed strings are absorbed
by branes which then acquire open-string excitations
(12.18). Branes do not preserve the global Lorentz invariance and the Lorentz group
is broken into SO.1; D 1/ ! SO.1; p/ ˝ SO.D 1 p/.
An effective action for branes can be written down as an extension of (12.1)
which includes a U.1/ gauge field Aa (a D 0; 1; : : : ; p). The Dirac–Born–Infeld
(DBI) action for a Dp-brane with induced metric ab D @a X @b X is
Z q
1
SpDBI D Tp dpC1 det.ab C 2l2s Fab / ; Tp D pC1
;
.2/p gs ls
(12.19)
where Tp is the brane tension and Fab D @a Ab @b Aa is the field strength of the gauge
field. The dependence of Tp on the inverse string coupling gs , to be introduced later,
indicates that branes are intrinsically non-perturbative. Dp-branes move through
spacetime and interact with strings; therefore, they are fully dynamical objects [35].
12.1.3 Quantum Strings and Critical Dimension
There are different ways to quantize the string and they are all mutually consistent:
old covariant quantization, light-cone quantization, path-integral covariant quanti-
zation and Becchi–Rouet–Stora–Tyutin (BRST) quantization.
In the so-called old covariant quantization of the string (e.g., [10, 11, 22, 24]),
general covariance is preserved. After defining the classical momentum ˘ :D
ıS=.ı@ X / D @ X =.2 ˛ 0 / and promoting X and ˘ to operators, one imposes
the commutation relation
ŒXO . ; /; ˘O . ; 0 / D i ı. 0 / : (12.20)
Thus, the center of mass and the momentum are conjugate, ŒOx ; pO D i , and

so are the oscillators. For the open string, Œ˛n ; ˛m D mın;m and, for X to be
"
self-adjoint, .˛n / D ˛n . Operators with negative and positive n are therefore
interpreted as, respectively, creating and annihilating string modes. The vacuum
jN D 0I p D 0i of the Fock space represents a configuration with no excitations and
no momentum. The eigenvalues of pO are indicated as p .
To guarantee that all annihilation operators be on the right, one imposes a normal
ordering on the Virasoro operators:
C1 1
!
1X 1 X X
LO n :D
W˛nm ˛m W D
˛nm ˛m C ˛m ˛nm

: (12.21)
2 m2Z 2 mD0 mD1
For n ¤ 0 the normal ordering is immaterial, since operators ˛ with different index
commute. On the other hand, for n D 0 the ordering is important and distinguishes
LO 0 D WL0 W, which is the quantization of (12.10), from the operator
C1
DX
LO 00 D ˛ 0 pO 2 C NO A ; AD m; (12.22)
2 mD1
which is the quantization of (12.9). The difference is in a formally divergent constant

term, LO 0 D LO 00 C A. In general, there are infinitely many choices for the quantum
Hamiltonian, which is parametrized by an arbitrary constant:
O :D LO 0 a :
H (12.23)
The operators LO n obey the Virasoro algebra (duplicated for the closed string) [4]
D 3
ŒLO m ; LO n D .m n/LO mCn C .m m/ım;n ; (12.24)
12
which is the central extension of the classical Witt algebra (12.12). The central
charge (i.e., the delta term in the right-hand side of (12.24)) is determined by using
the Jacobi identity for nested commutators and calculating the vacuum expectation
value of ŒLO 2 ; LO 2 . A particularly important sub-algebra of (12.24) is generated by
LO 0 and LO ˙1 and has vanishing central charge:
ŒLO 1 ; LO 1 D 2LO 0 ; ŒLO 0 ; LO ˙1 D LO ˙1 : (12.25)
This is the special linear (or conformal-anomaly) algebra sl.2; R/.

To get the spectrum of string excitations, one imposes conditions on physical (or
admissible [37]) states j˚i corresponding to the vanishing of the energy-momentum
tensor, TOCC j˚i D 0 D TO j˚i. In terms of the Virasoro operators, these
conditions are compatible with the commutation relations (12.24) provided we do
not impose them for all integer n:
LO n j˚i D 0 8n > 0 ; .LO 0 a/ j˚i D 0 : (12.26)
The further condition LO n<0 j˚i D 0 would be in contradiction with the Virasoro
algebra and it is not even necessary: the expectation values h˚1 j LO n<0 j˚2 i D
"
h˚1 j LO n>0 j˚2 i on two physical states is zero by virtue of the first constraint in
(12.26). States annihilated by LO n>0 are called primary. There are also spurious states
of the form j˚ 0 i D LO n>0 j˚i which are orthogonal to all the others. Two physical
states are isomorphic if their difference is a null state, i.e., a spurious admissible
state.
Physical states are eigenvectors of both operators pO and NO in (12.22), with
eigenvalue p (the momentum of the mode) and N (the number of excitations in
the state). The mass spectrum of physical states is given by the dispersion relation
p2 C Mopen
2
D 0, which yields
2 Na
Mopen D ; (12.27)
˛0
while for the closed string another copy of (12.26) holds for the LQ n and
2 4.N a/ 4.NQ a/
Mclosed D 0
D : (12.28)
˛ ˛0
This implies that the number of right- and left-moving modes is the same for
physical states, N D NQ (level-matching condition).
Consider now the open-string spectrum.
• The ground state
ND0 W j0I pi D j0iosc ˝ jpi (12.29)
contains no oscillations, it is an eigenfunction of the string momentum,

pO j0I pi D p j0I pi, and has a mass Mopen2
D a=˛ 0 . It has no Lorentzian
indices and corresponds to a scalar particle. If a > 0, as it will soon turn out,
then this is a negative-mass mode, i.e., a tachyon.
• The next state in the tower is

ND1 W j1I pi D . p/ ˛1 j0I pi ; (12.30)
where is a polarization vector. This state is admissible if LO 1 j1I pi D

p p
. 2˛ 0 p ˛1 C : : : / j1I pi D 0, requiring p D 0. Since LO 1 j0I pi D 2˛ 0 p

˛1 j0I pi, j1I pi is a spurious state when / p . Such state is null (i.e.,
spurious and admissible) when LO 1 LO 1 j0I pi D 0, which requires p2 D 0. In this
case, which is the only one compatible with the light-cone quantization discussed
below, there is an isomorphism class of polarization vectors, Š Cconstp ,
the mass of the state is zero and the state (12.30) corresponds to a massless
Lorentz vector particle identified with the photon. The arbitrary constant in
(12.23) is thus fixed to
a D 1: (12.31)
The two conditions p D 0 and p2 D 0 remove pure gauge modes and one
is left with D 2 degrees of freedom. These correspond to the “little group” of
spatial rotations SO.D 2/ [8], the sub-group leaving this state invariant at rest.
2
• The N D 2 state ˛1 ˛1 j0I pi is a rank-2 tensor with mass Mopen D 1=˛ 0 ,
invariant under SO.D 1/. The number of degrees of freedom is the dimension
of this group, .D 1/.D 2/=2.
The quantum algebra of the system is not canonical, since commutators with
; D 0 are negative definite. These are associated with an infinite tower of
0
negative-norm states ˛n j0; pi, n > 0. Fortunately, such ghosts are non-physical,
as a gauge fixing shows. Even if we have fixed ab as the flat metric, there is
still a residual gauge invariance: a conformal transformation of the coordinates
X yields a conformal factor that can be compensated by a Weyl transformation.
This residual freedom can be spent by selecting the so-called light-cone gauge [9],
which breaks manifest covariance. The advantage in doing so is that negative-norm
states disappear from the game. In fact, thep D 0 and D D 1 coordinates are
combined together as X ˙ :D .X 0 ˙ X D1 /= 2, which is nothing but a frame whose
axes coincide with the local light cone. The transverse coordinates X i are indexed by
i D 1; : : : ; D 2. The light-cone gauge consists in setting to zero all the oscillators
of X C , which becomes X C D xC C pC . Therefore, only D 2 oscillators survive
(per sector, in the case of the closed string) [8]. All that has been said above can be
easily recast in light-cone gauge, resulting in the so-called light-cone quantization
of the string [24]. The transverse operators
1X
L?
n :D ıij W˛nm
i
˛j W ; i; j D 1; : : : ; D 2 ; (12.32)
2 m2Z
and in particular
C1
X
L?
0 D ˛ pO i pO C NO ? ;
0 i ?
N :D ıij ˛m
i
˛mj ; (12.33)
mD1
obey the Virasoro algebra (12.24) with the replacement D ! D 2. The same
replacement
P happens in the normal-ordering constant in (12.22), A D .D
2/ C1 mD1 m=2, as well as in the condition (12.26) for admissible states: L?
n j i D0
for all n > 0, .L?
0 a ?
/ j i D 0. A calculation of the commutator of the generators
of the Lorentz algebra fixes the value of both a? and D [12, 24, 25]:
a? D 1 ; D D 26 : (12.34)
To explain this result about the critical dimension of spacetime, for the sake of
simplicity we sketch a heuristic argument. First of all, we identify the arbitrary
constant a? in the Hamiltonian constraint with the zero-point energy A of the
oscillators. The latter is the normal-ordering constant present
PC1 in the alternative
definition L00 ? D L?0 A, so that a ?
D A D .D 2/ mD1 m=2. Next, intro-
P PC1 m
ducing a regularization parameter , one has C1 mD1 m e m
D @ mD1 e D
@ Œe =.1e / D @ Œ1=.1e / D e =.1e /2 D 1= 2 1=12CO. 2/. The first
term can be reabsorbed by adding a counter-term in the action, an operation which
preserves Weyl invariance. Sending to zero, the final result is a? D .D 2/=24.
Then, one notices that the N ? D 1 state ˛1
i
j0I pi contains D2 degrees of freedom.
Vectors in D dimensions have such counting only if they are massless, which implies
2
Mopen D .1 a? /=˛ 0 D 0, hence a? D 1, hence D D 26.
The critical dimension of string theory is determined by the simultaneous
requirement that the theory maintain its classical symmetries when quantized and
that ghosts decouple from the physical spectrum. The latter is known as the no-ghost
theorem: the bosonic-string spectrum is ghost free only if D D 26 [10, 11]. This can
be proven also in the BRST quantization [50–53] of the bosonic string [48, 54–
56].1 Conformal invariance is of paramount importance for ensuring that the theory
is unitary and, in fact, the techniques of conformal field theory (CFT) make these
results particularly transparent.
Another way to quantize the string is via a covariant path integral (or, more
precisely, partition function) of the form
Z
ZD ŒDhab ŒDX eSŒh;X : (12.35)
Here, the world-sheet metric hab in the functional measure and in S has Euclidean
signature and replaces ab . This modification is convenient to make the path integral
convergent and for the study of interactions [27, 28], but it does not change the
physics. From now on, we will employ a Riemannian world-sheet.
Due to the redundancy of world-sheet diffeomorphisms and Weyl transforma-
tions (12.2), there are unwanted gauge degrees of freedom in (12.35). These can be
canceled out by promoting a and ! to auxiliary fermionic fields with integer spin,
i.e., Faddeev–Popov ghosts. R The Jacobian resulting fromR integrating out the ghosts
is something of the form ŒD a ŒD! / expŒ.D 26/C d d .@!/2 , where C is
some constant. If we want Weyl invariance to be preserved at the quantum level, this
quantity must be independent of the Weyl factor and, hence, of ghosts. Therefore, it
must be D D 26 [41]. For D ¤ 26, there is a conformal anomaly and the commutator
of the energy-momentum tensor T with itself does not vanish.
For the closed string, the arguments on the spacetime dimensionality are mainly
unchanged, only doubly copied for the two sectors of the theory. In the old covariant
quantization, physical states are determined by the constraints LO n>0 j˚i D 0 D
LOQ n>0 j˚i, .LO 0 1/ j˚i D 0 D .LOQ 0 1/ j˚i. The Hamiltonian is H
O D LO 0 C LOQ 0 2.
f osc ˝ jpi is a tachyon with
Q pi :D j0iosc ˝ j0i
• The N D NQ D 0 ground state j0; 0;
2
mass Mclosed D 4=˛ 0 .
• The N D NQ D 1 state j1; 1; f osc ˝ jpi (gauge modes
Q pi D ˛ i j0iosc ˝ ˛Q j j0i
1 1
have been removed already) is a rank-2 tensor given by the product of two
massless vectors. It is invariant under the group SO.D 2/ ˝ SO.D 2/,
1
Strictly speaking, the no-ghost theorem forbids ghosts for the Veneziano dual model [1, 4] when
D 6 26. However, for D < 26 the physical spectrum is not the one of bosonic string theory since
it contains also the longitudinal modes removed in the light-cone quantization [57, 58].
which can be decomposed in three irreducible representations [32]: a scalar

called dilaton, a symmetric traceless tensor hij (the graviton [30, 31]) and an
anti-symmetric tensor Bij (the Kalb–Ramond field [26]). This is the massless
sector of bosonic string theory. The total number of degrees of freedom is
1 C Œ.D 1/.D 2/=2 1 C .D 2/.D 3/=2 D .D 2/2 . The tensor
hij is identified with the graviton because gauge modes are modded out by a
transformation hij ! hij C pi j C pj i , which is nothing but the spatial part of an
infinitesimal diffeomorphism.
The above spectra are characteristic of oriented string theory, where the parame-
ter fixes a direction along the string. There also exist unoriented strings which are
invariant under a twist transformation
! l: (12.36)
In this case, both the photon (12.30) of the open string and the anti-symmetric tensor
B of the closed string disappear in the unoriented version of the theory. Even after
removing these fields and the gauge group they carry, there is another non-trivial
gauge group associated with the open string. Symmetry considerations require to
introduce, at each end-point of the string, fermionic fields in the irreducible spinor
representation of SO.2N=2 /, where N is even. This representation has dimension
2N=2 . Thus, each end-point is labelled by an index with 2N=2 values, called Chan–
Paton factor [59, 60], which determines the type of brane attached to the string.
One-loop tadpole divergences cancel for N D D and the gauge group associated
with the theory is therefore SO.8192/ [61, 62].
12.1.4 Interactions
In the old covariant quantization, the rigorous (and, actually, first) derivation of
the critical dimension of spacetime entails interactions. When strings meet, their
interactions are not point-wise but extended to a finite region of space; this feature
is at the core of the ultraviolet finiteness that the theory is believed to have. Feynman
diagrams consist of strips (for open strings) and tubular Riemann surfaces (for
closed strings), endowed with handles (the number of handles is called genus). The
Feynman expansion is governed by the dimensionless string coupling gs / lPl =ls .
To string vertices, there correspond operators with which one can construct
transition amplitudes between in- and out-states. It turns out that, while closed string
theory is self-consistent, open string theory is not complete, since some diagrams
correspond to closed-string features (an example is given in Fig. 12.3) and an open
string can join its end-points to form a closed one. In other words, closed strings can
be produced in open-string scattering. Therefore, closed strings must be included in
open string theory. Interactions are consistent only when D D 26 [63].
Fig. 12.3 The annulus, a one-loop diagram that can be seen either as an open string propagating
around (left) or as a closed string expanding or contracting (right)
Like most of the features of string theory, the study of the Riemannian surfaces
describing string interactions is conveniently approached [48] in the covariant
formalism of conformal field theory [43–48]. In CFTs, massless interacting fields
live in two dimensions (parametrized as a complex plane) and their dynamics
is conformally invariant. The conformal structure of Riemann surfaces is fully
encoded in a set of parameters called moduli [64–67], so that the set of Riemannian
surfaces is also known as moduli space. Compared to the old covariant or light-cone
formalisms we are adopting here, CFT is arguably a more economic and elegant way
to introduce string theory. However, it is technically more involved with respect to
the average presentation of this book and, for this reason, we will omit it from the
chapter.
12.1.5 Low-Energy Limit
If we allow closed strings to move in a non-flat background, we have to replace the

Minkowski metric in (12.1) with an X-dependent rank-2 tensor. This tensor
decomposes into a symmetric part gN .X/, encoding the graviton mode, and an
anti-symmetric tensor B .X/ (Kalb–Ramond field) accounting for the other closed-
string excitations:
Z p
1
S D d d h hab .Ng C iB /@a X @b X C ˛ 0 R.2/ ˚ ; (12.37)
4 ˛ 0
where the imaginary unit comes from the analytic continuation to the Euclidean
world-sheet [35] and we included the world-sheet Ricci scalar term, now coupled
to a non-constant field ˚.X/. This field ˚ gives the dilaton when combined with
the trace of gN , but often it is simply called dilaton. The kinetic term for X in the
action is no longer quadratic; while dynamical systems with such property are called
k-essence in cosmology, in field and string theory they are dubbed non-linear sigma
models [68–70].
At the quantum level, the trace of the energy-momentum tensor is
2˛ 0 Ta a D .ˇ
gN ab
s C iˇ /@a X @b X ˛ˇ ˚ R.2/ ;
B ab
(12.38)
where we have split the world-sheet metric into a symmetric and an anti-symmetric
part, hab D sab C ab . At the one-loop level, the beta functions are [71]
˛0 2
gN
ˇ D ˛ 0 RN C 2˛ 0 rN rN ˚ H H C O.˛ 0 / ; (12.39a)
4
˛0 N 2
ˇ
B
D r H C ˛ 0 H rN ˚ C O.˛ 0 / ; (12.39b)
2
˛0 N ˛0 D 26 2
ˇ ˚ D ˚ C ˛ 0 rN ˚ rN ˚ H H C C O.˛ 0 / ;
2 24 6
(12.39c)
where the 3-form H D @ B C@ B C@ B (H3 in short) is the Kalb–Ramond
field strength. This is the string analogue of the electromagnetic field strength for
point particles. Notice that ŒH3 D 1 and Œ˚ D 0. In order to have conformal
invariance, all beta functions must vanish simultaneously, thus giving the equations
of motion for gravity, for the anti-symmetric tensor and for the dilaton at first order
gN
in ˛ 0 : ˇ D 0, ˇ
B
D 0, ˇ ˚ D 0. These equations are obtained also from the
effective spacetime action [71–73]
Z
1 p 2.26 D/
SD dD x Ng e2˚ C N 1 H32 C 4.r˚/
R N 2 C O.˛ 0 / ;
2D2 3˛ 0 2
(12.40)
where D2 is the D-dimensional Newton’s constant and H32 D H H =3Š. In D D
26 dimensions, a solution to the dynamical equations is the Minkowski metric gN D
, B D 0 and a constant dilaton. The action (12.40) describes a scalar field non-
minimally coupled to gravity. The metric gN plays the role of the Jordan metric of
Sect. 7.4, hence the use of a bar. For this reason, the Jordan frame is often called
string frame. After a conformal transformation g D expŒ4˚=.D 2/Ng , one
can recast (12.40) in the Einstein frame:
Z
1 p 2.26 D/ 4˚ =.D2/ 1
SD 2 d x g
D
e C R e8˚ =.D2/ H32
2D 3˛ 0 2

4
.r˚/2 C O.˛ 0 / ; (12.41)
D2
where the indices in H32 are now contracted with the Einstein-frame metric g .
On a background with non-zero curvature and fields, the brane action (12.19) is
modified as (fermionic part ignored)
Z q
SpDBI ;Ng D TQp dpC1 e˚ det.Ngab C Bab C 2l2s Fab / ; (12.42a)
1
TQp D gs Tp D pC1
; (12.42b)
.2/p ls
where gN ab D gN @a X @b X and a similar expression for Bab are the projections of,
respectively, the metric and the Kalb–Ramond field on the brane. By definition, the
vacuum expectation value of the dilaton measures the strength of the string coupling:
e˚ D g s : (12.43)
12.1.6 String Field Theory
The Polyakov action (12.1) is the string extension of the classical and quantum
mechanical action for a free point particle. The main dynamical object is, in these
cases, the target coordinate X . ; / parametrized by the world-sheet and the
particle position q . / parametrized by the world-line. Multi-particle and multi-
string interactions are described by n-point correlation functions. For the particle,
interactions among infinitely many fundamental objects are included by moving
from classical or quantum mechanics to classical or quantum field theory. A similar
generalization is possible also for strings: string field theory (SFT).
A string field is a mathematical object given by the superposition (called level
expansion) of string oscillators applied to the Fock vacuum (or vacua, in the case of
the closed string). The coefficients of the level expansion are particle-field modes of
progressively higher mass. The interaction and precise field content depend on the
type of string and on the composition rules of the theory. Open SFT is the simplest
example of this construction [74, 75]. The bosonic open SFT action is [76]
Z
1 1 1
SD Q C ; (12.44)
g2o 2˛ 0 3
where go is the open string coupling,2 Q is the BRST operator (which annihilates
the field in the free theory, Q D 0) and * is a non-commutative product describing
how strings interact. A visual picture of the cubic interaction is obtained by gluing
the halves of two different open strings, while the other two halves merge into a
third string.
2
In several circumstances like here, gs will be further differentiated into the open-string and closed-
string coupling go and gc , respectively.
The equation of motion from (12.44) is Q C ˛ 0 D 0. The action is

invariant under the infinitesimal gauge transformation ı D Q C ,
where is a zero-ghost-number state. The gauge group of string field theory is very
large and not completely known. It includes at least world-sheet reparametrizations,
spacetime diffeomorphisms and (in the supersymmetric extensions) supersymmetry
transformations [77, 78].
Choosing a CFT representation of (12.44), fixing the gauge and keeping only
some of the particle fields in (a technical assumption called level truncation),
one can obtain from (12.44) an effective spacetime action for the fields which
exhibits a non-local interaction. We only show the final result of this procedure
when truncating up to the tachyonic mode [79, 80]:
Z " 0 2
#
1 1 1 e3=.˛ Mo / Q 3
SD 2 D
d x C 0 ; (12.45)
go 2 ˛ 3
where
3=2
Q :D e=Mo2 ; 1 0 3
D ˛ ln 0:2616 ˛ 0 : (12.46)
Mo2 4
Notice the appearance of the correct open-string tachyonic mass 1=˛ 0 in the free
part. The value of Mo is dictated by conformal invariance, which is partly broken by
level truncation. The equation of motion from (12.45) is

1 2 0 2
C 0 e2=Mo Q e3=.˛ Mo / Q 2 D 0 : (12.47)
˛
The other fields of the level expansion, as well as a toy model called p-adic or non-
Archimedean string [81], display the same type of non-locality.
The degrees of freedom corresponding to the graviton are included in the
spectrum of closed SFT [82–96]. The action is
Z C1
X .gc ˛ 0 /N3
1
SD ˚ ? Qb 0 ˚ C g c ˚ ? Œ˚ N1 ; (12.48)
˛0 ND3
2 N3 NŠ
where ˚ is the closed-string field, ? is a non-commutative product, Q is the BRST

operator, b N
0 D .b0 b0 /=2 is a combination of anti-ghost zero modes, gc is the
closed-string coupling and the symbol Œ˚ N1 D Œ˚1 ? ? ˚N1 represents the
string field obtained combining N 1 terms ˚1 ; : : : ; ˚N1 using the N-string vertex
function. Again, for physical processes dominated by light states one can truncate
the string field in terms of oscillators and particle fields, ˚ D ˚phys C ˚aux , where
˚aux is made of various auxiliary fields,

˚phys D c
0 C A ˛1 ˛N 1 C : : : j0i ; (12.49)
c
0 D c0 c N 0 is a combination of the left and right ghost zero mode, is the
bosonic tachyon field and the symmetric and anti-symmetric parts of the tensor A
are, respectively, the graviton gN and the two-form B . After gauge fixing and
a rearrangement of the auxiliary fields, the resulting effective Lagrangian for the
physical modes contains a kinetic and an interaction part, L D Lfree C Lint . The
kinetic and mass terms read
1 1 2
Lfree D @ A @ A @ @ 0 2
; (12.50)
2 2 ˛
while there are about 50 cubic interaction terms:
gc Q 3 gc ˛ 0 Q 2 Q gc Q Q 2 gc ˛ 0 Q Q 2
Lint D C .@ / A C A C 4 .@A/
3Š 3 23 2 2 2
gc ˛ 0 2 Q 2 Q 2 gc ˛ 0 Q 2 2 Q
C .@ A/ 3 A .@ A/ C : : : ; (12.51)
27 2
where D 24 =33 and fields with a tilde are “dressed” by the same exponential non-
local operator (12.46) except that the constant Mo2 ! Mc2 D 2Mo2 has an extra factor
of 2, which can be removed by a coordinate rescaling.
The same non-local structure in effective spacetime actions is expected also when
open and closed strings are considered simultaneously, as proposals for the p-adic
tachyonic action indicate [97–101]. One can conclude that, after a field redefinition
[74, 79, 80, 85], the effective field actions on target spacetime stemming from string
field theory have the following schematic structure for bosonic fields fQi :
XZ
1Q 2 =.ci m2i / Q Q
Seff D d x fi . mi /e
D
fi U. fi ; @/ ; (12.52)
i
2
where the ci are constants and U is a cubic potential of the dressed fields fQi and their
derivatives.
12.2 Superstring
A pressing problem of the bosonic case is that the ground state j0I pi is unstable
since it corresponds to a scalar tachyon with mass / 1=˛ 0 . Therefore, the theory
has been quantized on a false vacuum. Superstring theory can address this issue.
Tachyons will appear only for a class of unstable D-branes, which will decay into
lower-dimensional stable branes.
12.2 Superstring 643
12.2.1 Action
In the Ramond–Neveu–Schwarz formulation of the theory [5–7, 102, 103], the

gauge-fixed action (12.1) is augmented by a fermionic sector [39, 40, 104, 105]:
Z
1
SŒX; D d d @a X @a X i N a @a
; (12.53)
4 ˛ 0
where are D Majorana fermions made of two components and a are 2 2

matrices obeying the algebra
fa ; b g D 2ab : (12.54)
They are related to the Pauli matrices (5.208) as 0 D 2 and 1 D i 1 . Since ia is
real, so are the two-component spinors .
The system described by (12.53) in invariant under: (a) the local world-sheet dif-
feomorphism and Weyl transformations (12.2), augmented by ı D a @a ; (b)
the global Poincaré invariance of the bosonic string; (c) the global supersymmetry
transformations ıX D N , ı D ia @a X and ı N D iN a @a X , where
is a constant Majorana spinor. An action more general than (12.53), invariant
under both world-sheet reparametrizations and local spacetime supersymmetry, is
also possible [106, 107]. It can be quantized in the light-cone gauge and gives the
same physics.
Using the equations of motion
@a @a X D 0 ; ia @a
D 0; (12.55)
one can easily show that the supercurrent J a D b a @b X =2 is conserved

on shell (Problem 12.1). Here and in what follows, represents contraction of
spacetime indices via . Supersymmetry transformations are then expressed by
Rl
the commutator ŒQN ; X D N , where Q WD 0 d J 0 is the supercharge. The
energy-momentum tensor is also conserved, @a T ab D 0, and its trace vanishes,
Ta a D 0.
In terms of the left- and right-moving coordinates defined in (12.13), the

dynamics (12.55) is recast as @C @ X D 0 and @ C D 0 D @C , where we
t
split the 2-spinors D . ; C / into two Grassmann fields ˙ D ˙ . ˙ /.

The constrains on the energy-momentum tensor components and the supercurrent

are then T˙˙ D Œ@˙ X @˙ X C .i=2/ ˙ @˙ ˙ =.2 ˛ 0 / D 0 and J˙ D
0
˙ @˙ X=.2 ˛ / D 0.
These equations of motion are obtained by varying the action (12.53) and
imposing suitable conditions to make boundary terms vanish. The conditions on
the scalars X are the same of Sect. 12.1 and define open or closed strings. The
boundary condition C ı C ı D 0 on the spinorial fields defines distinct

fermionic sectors of the theory, so that C D ˙ at the end-points (coincident, in

the closed case) of the string. At D 0, by convention we set C. ; 0/ D . ; 0/,
while for all

Neveu–Schwarz (NS) sector: C. ; l/ D . ; l/ ; (12.56a)

Ramond (R) sector: C. ; l/ D C . ; l/ : (12.56b)
For the open string there are only two sectors (NS or R), while for the closed string
there are four (NS-NS, NS-R, R-NS and R-R) since we can impose the boundary
conditions on the left- and right-moving parts of (12.14) independently.
The most general solutions of are expanded in terms of harmonic oscillators.
For the open string, they are
p
˛ 0 X in ˙
NS sector: ˙ D bn e ; (12.57a)
2 1
n 2 2Z
p
˛ 0 X ir ˙
R sector: ˙ D d e : (12.57b)
2 r2Z r
The Virasoro generators are defined as in (12.11), augmented by a fermionic part:

2 3
1 X X n
NS sector: L(NS) D 4
˛nm ˛m C kC bk bnCk 5 ;
n
2 1
2
m2Z k 2 2Z
(12.58a)
" #
1 X X n

R sector: n D
L(R) ˛nm ˛m C rC d d ; (12.58b)
2 m2Z r2Z
2 r nCr
where n 2 N. Fermionic generators are obtained by Fourier expanding the

supercurrent
p R in the periodic and anti-periodic
P sectors. In the Neveu–Schwarz sector,
Gk D 2 0 d .eik J C eik JC / D m ˛m bkCm for k 2 Z C 1=2. In the
p R P
Ramond sector, Fr D 2 0 d .eir J C eir JC / D m ˛m drCm for r 2 Z.
For the closed string,
p X
X C
NS sector:
D 2˛ 0 bk e2ik ; C D bQ k e2ik ;
kC 12 2Z kC 12 2Z
(12.59a)
p X
X
2ir C
R sector:
D 2˛ 0 dr e2ir ; C D dQ r e ; (12.59b)
r2Z r2Z
and the Virasoro generators follow from the definition (12.15) (the right-hand side
having also fermionic oscillators) and its right-moving counterpart.
12.2.2 Quantization
The old covariant quantization proceeds by imposing (12.20) and equal-time anti-

commutation relations for : f A . ; /; B . ; 0 /g D ı. 0 /ıAB , where

A; B D ˙. This results in the algebras fdr ; ds g D ır;s , fbk ; bl g D ık;l
and their counterparts in the right sector for the closed string. The Fock vacuum is
annihilated by

˛n>0 j0iosc D 0 ; bk>0 j0iosc; NS D 0 ; dr>0 j0iosc; R D 0 : (12.60)
In the case of the NS oscillators b, there is no zero mode as k is semi-integer. There

is a unique ground state j0I piNS corresponding to a Lorentz scalar; the NS sector
of the theory is therefore called also bosonic sector. In the case of d, the r D 0
oscillator is excluded from the definition of the vacuum due to the non-vanishing

anti-commutator fd0 ; d0 g D . The irreducible representations of this Clifford
algebra are SO.1; 9/ spinors and, therefore, the vacuum j0I piR of the Ramond sector
is a fermionic state.
The Virasoro super-algebras of the NS and R sector generated, respectively, by
the normal-ordered operators LO (NS)
n , GOk and LO (R)
n , Fr can be found in [34]. As in
the bosonic theory, these operators are applied to Fock-space states to define a set
of physicality conditions. The counterpart of (12.26) in the NS and R sector are,
respectively,

NS sector: LO (NS)
n j˚i D 0 D Gk j˚i 8n; k > 0 ; LO (NS)
0 1
2
j˚i D 0 ;
(12.61a)
R sector: LO (R)
n j˚i D 0 D Fr j˚i 8n; r > 0 ; (12.61b)
where, in the case of the open string,

!
X X
LO (NS)
0
0 2
D ˛ pO C
˛m ˛m C k bk bk D: ˛ 0 pO 2 C NO NS ;
m>0 k>0
(12.62a)
!
X X
LO (R) 0 2
0 D ˛ p
O C
˛m ˛m C
r dr dr D: ˛ 0 pO 2 C NO R :
m>0 r>0
(12.62b)
On physical states, the number operators take eigenvalues NNS D 0; 1=2; 1; 3=2; : : :
and NR D 0; 1; 2; : : : . For the closed string, the coefficient in front of pO 2 changes as
˛ 0 ! ˛ 0 =4 and there is another copy of operators LOQ (NS,R) , GOQk and FOQ r for right-moving
n
modes. Therefore, the open and closed superstring spectra are
2 2NNS 1 2 NR
Mopen; NS D ; Mopen; R D ; (12.63)
2˛ 0 ˛0
2 2.2NNS 1/ 2 4NR
Mclosed; D ; Mclosed; D 0 ; (12.64)
NS
˛0 R
˛
plus independent copies of the second line in the right-moving sector.
The critical dimension of spacetimes where the superstring can live in is
D D 10 : (12.65)
This result can be obtained from the extension to superstrings of the proofs given
in the bosonic case: among these, there are the no-ghost theorem for the string
spectrum [10, 13] also in BRST quantization [47, 48], the requirement for the
Lorentz generators of the theory to obey the Lorentz algebra [12, 24] and the
cancellation of the conformal anomaly in the covariant path-integral quantization
[105].
12.2.3 Type-I Superstring
The theory of open strings contains both fermions and bosons but also a tachyon.
Furthermore, there are states that do not combine to supersymmetric multiplets.
Both the tachyon and this class of states can be eliminated by imposing that all
physical states are eigenvectors with eigenvalue 1 of the Gliozzi–Scherk–Olive
(GSO) operators [108]
P P
1 .1/ k>0 bk bk 1 .1/ r>0 dr dr
PNS :D ; PR :D : (12.66)
2 2
The lowest levels in the open-string spectrum (12.63) are:
• NS sector:
2 0
– NNS D 0, Mopen; NS D 1=.2˛ /: the ground state j0I piNS of the NS sector is
a tachyon, which is projected out since PNS j0I piNS D 0.
2
– NNS D 1=2, Mopen; NS D 0: a massless vector b1=2 j0I piNS , corresponding to
a non-Abelian gauge field A . It survives the GSO projection, since .PNS
a

1/b1=2 j0I piNS D 0.
2 0
– NNS D 1, Mopen; NS D 1=.2˛ /: a massive vector ˛1 j0I piNS and a massive

rank-2 tensor b1=2 b1=2 j0I piNS .
• R sector:
2
– NR D 0, Mopen; R D 0: a massless spinor j0I piR , the ground state of the
Ramond sector, associated with a fermionic field we will denote as . This
state survives the GSO projection, since .PR 1/ j0I piR D 0. In D D 10,
the spinor representation of SO.D/ has dimension 210=2 D 32, corresponding
to a complex 32-component spinor. The reality condition reduces it to a 32-
component Majorana spinor, while the chirality condition (Weyl spinor) cuts
it to 16 components. These components are not independent since they are
constrained by the Dirac equation. Therefore, j0I piR encodes 8 degrees of
freedom. Also, defining the chirality matrix 11 D 0 1 9 in terms of the
SO.1; 9/ gamma matrices, the ground state is defined to have positive chirality,
11 j0I piR D C j0I piR .
2 0
– NR D 1, Mopen; R D 1=˛ : D massive spinors ˛1 j0I piR . D extra spinors of the

form d1 j0I piR are projected out: PR d1 j0I piR D 0. The number of physical
spinors is reduced to D 2 in light-cone quantization [102, 109, 110].
One can show that the states that survive the projection form multiplets. For
instance, we have seen that j0I piR has 8 real independent components (symbolically
indicated with 8s , where “s” stands for spinor), while it is easy to see that

b1=2 j0I piNS represents 8 degrees of freedom corresponding to the transverse
directions of a vector (8v , where “v” stands for vector). This hints to the fact that
the massless spectrum is a supersymmetric multiplet 8v ˝ 8s [108, 111], actually
describing the super-Yang–Mills theory with gauge group SO.32/ [108, 112]. A
proof of supersymmetry at all mass levels exists [109].
When combined with the type-IIB closed-string modes described below, a deeper
analysis of the spectrum leads to the formulation of type-I string theory (see
also Sect. 12.3.3). This is a theory of unoriented open and closed strings with
N D 1 supersymmetry. Its low-energy limit is SO.32/ super-Yang–Mills theory
[113] coupled to a supergravity sector [110, 114].
12.2.4 Type-II Superstrings
The GSO projection works also for the closed string and must be applied on the left-
and right-moving sectors separately [108]. This results in type-II theories for ori-
ented closed strings with 32 supercharges [108, 110]. The massless spectrum is:
• NS-NS sector:
– NNS D 0 D NQ NS , Mclosed;
2 0 Q
NSNS D 2=˛ : the tachyon j0; 0I piNS :D j0iNS ˝
f
j0iNS ˝ jpi is projected out. From now on, we omit the momentum label in
kets.
1 1Q
– NNS D 12 D NQ NS , Mclosed;2
NSNS D 0: the rank-2 tensor j 2 ; 2 iNS D
f NS is decomposed into a scalar ˚, a symmetric tensor
b1=2 j0iNS ˝ bQ 1=2 j0i

gN and an anti-symmetric tensor B . In terms of transverse components in

10 dimensions, 8v ˝ 8v D 1 ˚ 35v ˚ 28a , where “a” stands for anti-symmetric.
This supergravity sector is called universal because it is present in all closed
superstring theories.
• R-R sector: NR D 0 D NQ R , Mclosed;
2 Q
RR D 0: the ground state is j0; 0iR :D
f
j0iR ˝ j0iR . If one defines the right sector to have positive chirality (11 j0iR D
C j0iR ), one can choose whether the left sector has same or opposite chirality,
f R D ˙j0i
11 j0i f R . If one takes opposite chirality, the theory is called type-IIA,
otherwise type-IIB.
– Type-IIA: In terms of representations of SO.8/, the non-chiral ground state
Q R is decomposed as 8s ˝ 8c D 8v ˚ 56v (“c” stands for conjugate
j0; 0i
spinor), a 1-form C and an anti-symmetric 3-form C .3 These fields are
called Ramond–Ramond potentials.
– Type-IIB: The decomposition of the chiral ground state is 8s ˝ 8s D 1 ˚ 28a ˚
35s and the R-R potentials are a pseudo-scalar C0 (axion), an anti-symmetric
.C/
2-form C (also dubbed B0 in the literature) and the self-dual 4-form C .
• NS-R and R-NS sectors:
1 Q
– NNS D 12 , NQ R D 0, Mclosed;
2
NSR D 0: the fermionic state j 2 ; 0iNSR D
f R is decomposed as 8v ˝ 8s D 8c ˚ 56c . These degrees
b1=2 j0iNS ˝ j0i
of freedom correspond to a dilatino and a gravitino, super-partners of,
respectively, the dilaton and the graviton.
– NR D 0, NQ NS D 12 , Mclosed;
2
RNS D 0: The same analysis holds for the R-NS
1Q Q f R . In type-IIA theory, the decomposition is
state j0; 2 iRNS D j0iNS ˝ b1=2 j0i
8c ˝ 8v D 8s ˚ 56s and one has a dilatino and a gravitino of opposite chirality
with respect to those in the NS-R sector. In type-IIB theory, the spectrum is
chiral since 8s ˝ 8v D 8c ˚ 56c as in the NS-R sector.
12.2.5 Interactions and Anomaly Cancellation
String interactions [102, 114–116] (also formulated with superfields [117, 118]) play
an essential role in determining whether superstring theory is perturbatively finite.
This was checked at one loop [102, 103, 119, 120] and up to two loops and four
external states in string scattering amplitudes [121–127]. At arbitrary loop order,
there are non-renormalization theorems stating that loop corrections to the vacuum
amplitude and to the massless n-point functions for n D 0; 1; 2; 3 vanish to all orders
in D D 10 flat backgrounds [48, 128, 129]. Although no formal proof exists, many
independent arguments encourage the belief that superstring theory is indeed finite.
8
3
The number of degrees of freedom of a p-form for the SO.8/ group is p
D 8Š=ŒpŠ.8 p/Š.
The recent super-moduli space formalism [130] and a new formalism using picture-
changing operators [131, 132] permit to compute multi-loop off-shell amplitudes
and, therefore, to have renormalization properties under a much better control.
Ultimately, the problem of renormalizability is tightly related to the formulation of a
supersymmetric field theory of strings, to which Sect. 12.2.9 is dedicated. Sen [133]
offers a valuable pedagogical overview on the subject of divergences in superstring
perturbation theory.
Two other applications of interactions are the establishing of the low-energy
limits of superstrings, summarized in Sect. 12.2.7, and the calculation of anomalies.
A theory free from quantum anomalies is not only mathematically viable but also
a useful framework wherein phenomenological predictions are made possible. If
superstrings aim to describe sensible low-energy particle and gravitational physics,
then unitarity, Lorentz invariance and general covariance must be preserved by
quantum interactions. This may not be the case if anomalous Feynman diagrams
involving chiral fields give a non-zero net contribution to scattering amplitudes. It
is therefore necessary (and non-trivial) to check that the low-energy limit of type-I
and type-II string theories be free from these anomalies, which are called chiral or
gauge anomalies. The case of type-IIA is the simplest: the theory conserves parity
and there are no anomalies. Less obvious is the case of type-IIB theory. Indeed,
cancellations between fields of different spin guarantee that D D 10, N D 2 chiral
SUGRA is the only case free from one-loop gravitational anomalies coming from
the coupling of Weyl fermions and of the anti-symmetric tensor with gravity [134].
Regarding type-I theory, we recall that the vector multiplet 8v ˝ 8s of super-
Yang–Mills theories (a gauge vector field plus its super-partner) lives in the adjoint
representation of some group G. When super-Yang–Mills is coupled to D D 10
chiral SUGRA, the so-called Green–Schwarz mechanism is enforced: the only two
groups which allow the cancellation of gauge, gravitational and mixed anomalies
are G D SO.32/ and G D E8 ˝ E8 [135–138]. The low-energy limit of the type-I
open-string sector has precisely G D SO.32/.
12.2.6 Heterotic Superstrings
Remarkably, both anomaly-free groups E8 ˝ E8 (henceforth denoted as E8 E8 ) and

SO.32/ have realizations in terms of closed string theories known as heterotic [139–
141]. In type-II theory, there are 16 supercharges from the left sector and 16 from
the right one, for a total of 32 supercharges and N D 1 C 1 D 2 supersymmetries.
In the heterotic string, one picks a right sector with both bosons and fermions and
a purely bosonic left sector with 16 directions compactified on a torus (more on
compact directions later), for a total of 32 .26 10/ D 16 supercharges (GSO
projection) and N D 1 supersymmetry. Due to this mismatch, which can always
be prescribed since the left- and right-moving sectors are decoupled, this theory
is a hybrid between bosonic and superstring theory, hence the name “heterotic.”
The use of the critical dimensionality in each separate sector guarantees anomaly
cancellation.
Depending on whether the right sector is NS or R, the spectrum of SO.32/ and
E8 E8 heterotic string theory is determined by the mass formulæ
˛0 2 3
Mclosed NS D NNS C NQ C ˛ 0 p2L ; (12.67)
2 2
0
˛ 2
M D NR C NQ 1 C ˛ 0 p2L ; (12.68)
2 closed R
together with a condition guaranteeing that there is no distinguished point in the
closed string:
NS right sector: NQ 1 C ˛ 0 p2L D NNS 12 ; (12.69)

R right sector: NQ 1 C ˛ 0 p2L D NR : (12.70)
P
Here p2L D 25 2
ID10 pI is the momentum of the compact directions in the left sector,
which takes discrete values, ˛ 0 p2L D 0; 1; : : : .
Q 0iNS;R :D jNiNS;R ˝ jNI
Let jN; NI Q p2L i . The NNS D 0 tachyon states
L
Q 0iNS D j0iNS ˝ jNI
j0; NI Q 0iL (with zero winding number) do not respect the
condition (12.69) for any NQ 2 N and are removed from the picture.
2
The massless (Mclosed D 0) spectrum levels are the following.
• NS sector, NNS D 12 :
Q 0i
– NQ D 1, p2L D 0: the 10 10 rank-2 tensor j 12 ; 1I NS
D b1=2 j0iNS ˝

˛Q 1 Q 0iL is split into the components of the universal sector of the closed
j0I
superstring.
– NQ D 1, p2L D 0 and NQ D 0, p2L D 1=˛ 0 : the (NQ D 1) 16 neutral vectors
Q 0iL and the (NQ D 0) 480 charged vectors b j0iNS ˝
b1=2 j0iNS ˝ ˛Q 1
I
j0I 1=2
Q 1=˛ 0 iL are 496 gauge bosons in the adjoint representation of the group
j0I
E8 E8 or SO.32/.
• R sector, NR D 0:
Q 0iR D j0iR ˝ ˛Q j1I
– NQ D 1, p2L D 0: the fermionic state j0; 1I Q 0iL decomposes
1
into the gravitino and the dilatino, super-partners of the universal sector.
– NQ D 1, p2L D 0 and NQ D 0, p2L D 1=˛ 0 : the fermionic states j0iR ˝ ˛Q 1
I Q 0iL
j0I
Q 1=˛ iL are gauginos, super-partners of the gauge bosons in the
and j0iR ˝ j0I 0
same irreducible adjoint representation.

12.2.7 Massless Spectra and Low-Energy Limits
Let us succintly denote with a number the indices of p-forms: Cp D C1 :::p . Then,
to summarize the massless content of the theory:
• Type-IIA superstring theory is non-chiral (spacetime parity is conserved), it
has N D 2 ten-dimensional supersymmetries, its R-R fields are forms of odd
order (a 1-form C1 and a 3-form C3 ) and its gauge group is U.1/ as for the
bosonic closed string (one Abelian vector field). Therefore, the type-IIA low-
energy limit [110] is D D 10, N D 2 non-chiral SUGRA [142, 143], with
universal bosonic sector ˚, gN 2 and B2 . It is important to note that this model of
supergravity arises via dimensional reduction [142–144] of D D 11, N D 1
SUGRA [144], the only theory of supergravity in eleven dimensions. We will
sketch this relation in Sect. 12.4.
• Type-IIB superstring theory is chiral (spacetime parity is violated), it has N D
2 ten-dimensional supersymmetries of the same chirality, its R-R fields are forms
.C/
of even order (a 0-form C0 , a 2-form C2 and the self-dual 4-form C4 ) and
there is no gauge field. Its low-energy limit [110] is D D 10, N D 2 chiral
SUGRA [145–148], with universal bosonic sector ˚, gN 2 and B2 . Contrary to the
non-chiral model, this one does not descend from the dimensional reduction of
D D 11 supergravity.
• Type-I superstring theory has a gauge vector A1 in the bosonic open spectrum.
The closed-string sector is unoriented type-IIB theory and is therefore chiral.
After the identification ! , only ˚ and gN 2 (from the NS-NS sector) and
C2 (from the R-R sector) survive in the bosonic closed-string spectrum, plus their
super-partners. The low-energy limit [110, 113, 114] of, respectively, the open-
and closed-string sectors are D D 10 super-Yang–Mills [108, 112] with gauge
group SO.32/ [113] and D D 10, N D 2 chiral SUGRA.
• Heterotic superstring theories have N D 1 ten-dimensional supersymmetry
and their low-energy limit [139, 140] is D D 10, N D 1 SUGRA [149] (with
universal bosonic sector ˚, gN 2 and B2 ) coupled with super-Yang–Mills theory
with gauge group E8 E8 or SO.32/.
The supersymmetric generalization of the string action (12.37) on curved back-
grounds, of the beta functions (12.39) and of the low-energy spacetime action
(12.40) for these string theories was found in [71, 72, 150–152]. The massless
spectrum of the five superstring theories is summarized in Table 12.1.
The SUGRA actions in D D 10 with this p-form population can be found in
any textbook, e.g. [35]. Here we report only the type-II cases because they have
applications in Sect. 12.4 and Chap. 13. The NS-NS universal sector is
Z
1 p 1 2
SNS-NS ŒNg; ˚; B2 D 2 10
d x Ng e 2˚ N 2
R C 4.r˚/ H3 ; (12.71)
210 2
652
Table 12.1 Massless spectrum (p-forms) of the five superstring theories. Gauge groups are indicated in boldface
I IIA IIB Heterotic

SUGRA sector ˚ (dilaton), ˚ (dilaton), gN 2 (graviton), B2 (Kalb–Ramond)
(NS-NS type-II or NS heterotic) gN 2 (graviton)
NS gauge sector A1 496 vectors
[SO.32/] [E8 E8 or SO.32/]
.C/
R-R sector C2 C1 [U.1/], C3 C0 (axion), C2 , C4
(R-R potentials)
R sector (gaugino) dilatino, gravitino,
496 gauginos
NS-R and R-NS sectors dilatino, gravitino dilatino, gravitino
(non-chiral) (chiral)
12 String Theory
while the R-R sectors are

Z p
1
IIA
SR-R ŒB2 ; C1 ; C3 D 2
d10 x Ng F22 C FQ 42 ; (12.72)
410
Z
.C/ 1 p 1
IIB
SR-R ŒB2 ; C0 ; C2 ; C4 D 2
d10 x Ng F12 C FQ 32 C FQ 52 : (12.73)
410 2
Here Fp2 :D F1 :::p F 1 :::p =pŠ, FQ 3 :D F3 C0 ^ H3 , FQ 4 :D F4 C C1 ^ H3 and the

field strength FQ 5 :D F5 .1=2/C2 ^ H3 C .1=2/B2 ^ F3 is constrained to be self-dual
(FQ 5 D FQ 5 ) in theR equations of motion. Including alsoRthe Chern–Simons actions
2 1 2 1
IIA
SCS D .410 / B2 ^ F4 ^ F4 and SCS IIB
D .410 / C4 ^ H3 ^ F3 , the bosonic
part of the low-energy type-II string actions reads
SIIA D SNS-NS ŒNg; ˚; B2 C SR-R

IIA
ŒB2 ; C1 ; C3 C SCS
IIA
; (12.74)
.C/
SIIB D SNS-NS ŒNg; ˚; B2 C SR-R
IIB
ŒB2 ; C0 ; C2 ; C4 C SCS
IIB
: (12.75)
Notice that these expressions are in the string (Jordan) frame, hence the use of bars
as in (12.40).
12.2.8 Branes
Strings are not the only content of the theory. In general, type-IIA and type-IIB
theory display, respectively, forms of odd and even order, corresponding to branes
of even and odd order. In the above table, we appreciate that type-IIA theory has two
odd-order forms, C1 and C3 , while type-IIB theory has three even-order forms, C0 ,
.C/
C2 and C4 . This pattern in the p-form content is extended to all admissible orders.
For a . p C 1/-form, let FpC2 D dCpC1 be its strength (we use the compact notation
for differential forms introduced in Sect. 9.1.1). Higher R-R fields are Hodge dual to
lower R-R fields, since FpC2 D FQ D2p D dCD3p M
. To distinguish fundamental
E
forms from their duals, we call the first “electric” CpC1 forms and the second
M
“magnetic” CnC1 forms.
A . p C 1/-form naturally relates Rto a . p C 1/-dimensional world-volume
VpC1 , in the sense that the integral VpC1 CpC1 is covariant and well defined.
Therefore, electric and magnetic forms are associated with extended objects [153–
155]. In string theory, these are, respectively, Dp-branes and Dn-branes with n D
D 4 p. The strength of an electric CpC1 E
form defines the Dp-brane charge
R R
Qp :D SD2p FpC2 and its magnetic dual gD4p :D SpC2 FpC2 , where these
1 1
integrals are defined on a hypersphere at infinity (the ideal boundary of the brane).
The DBI brane action (12.19) is augmented by the coupling between R the brane and
its . p C 1/-form, represented by the Chern–Simons action iTQp VpC1 CpC1 . On a
curved background with the forms ˚, B and F , one has
Z
. p/ . p/
C2 ˛ 0 F2
Sp D SpDBI ;Ng C iTQp CpC1 ^ eB2 ; (12.76)
VpC1
. p/ . p/
where SpDBI ;Ng is given by (12.42) and B2 and F2 are the projections Bab and Fab
of B and F on the brane.
For each D D 10 type-II theory, the relations
Dp-brane $ FpC2 D dCpC1

E
! FQ 8p D dC7p
M
$ D.6 p/-brane
(12.77)
eventually translate into the presence of all possible stable Dp-branes with world-
volume VpC1 . For instance, for p D 2 one has .dC3E / D F4 D FQ 6 D dC5M , so that
C5 is dual to C3 and both D2- and D4-branes are automatically included in type-
IIA theory. The dynamics of branes is governed by the action (12.76) appropriately
decorated with fermionic fields.
To summarize, string theories are populated by the following branes [156]:
• Type-I string: D1-, D5- and D9-branes. The D9-brane is the whole target
spacetime and is associated with freely propagating open strings.
• Type-IIA string: D0-, D2-, D4-, D6- and D8-branes (p even). The D8-brane is
associated with a 9-form with constant field strength.
• Type-IIB string: D.1/-, D1-, D3-, D5-, D7- and D9-branes (p odd). The
D.1/-brane is an instanton localized in space and time. D1-branes are .1 C 1/-
dimensional objects called D-strings to differentiate them from fundamental
(or F-) strings. D3-branes are called dyons, objects coupling electrically and
magnetically at the same time.
• Heterotic string: no D-branes.
The 2-form associated with F-strings is the NS-NS Kalb–Ramond field B2 .
The magnetic dual of an F-string is a five-dimensional soliton called Neveu–
Schwarz
p 5-brane, NS5-brane
p in short. Its action is similar to (12.42) but with
det.Ng C B2 / ! det.Ng C gs C2 /.
The dynamical degrees of freedom of all these branes are encoded in fundamental
open strings attached to them. Branes of odd (respectively, even) order in type-IIA
(type-IIB) string theory are characterized by a tachyonic scalar mode in the open-
string spectrum which marks an instability. For instance, D5-branes are stable in
type-IIB theory but unstable in type-IIA theory. Unstable branes naturally decay
into branes of lower dimensionality according to some specific rules [157, 158].4
This phenomenon is called tachyon condensation [161–180] and is described by the
rolling of the tachyon from a local maximum of its potential (corresponding to the
unstable brane configuration) down to the local minimum representing the lower-
dimensional stable brane, where the tachyon decays into particles. A possible role
of the tachyon in cosmology will be discussed in Sect. 13.7.2.
12.2.9 Superstring Field Theory
While there is only one proposal for the bosonic open string field, there are many
for open superstring field theory. The first by Witten [181] was later modified in
[182–184], depending on whether a certain picture-changing operator in the action
is chiral and local [183, 184] or non-chiral and bilocal [182, 185, 186]. A non-
polynomial version of open super-SFT is due to Berkovits [187, 188]. The Ramond
sector has been implemented only very recently in a new theory with connections
both with Witten’s and Berkovits’ proposals [189–193].
The non-local effective action for the tachyon, representing an unstable brane,
has been constructed only for the non-chiral version [182, 185, 186]. In that case, at
lowest order the interaction is slightly more complicated than that of (12.47):
0 2
1 2 e4=.˛ Mo / Q 2=Mo2 Q 2
C e2=Mo Q e D 0: (12.78)
2˛ 0 9
The general form of the action (12.52) is the same as for the bosonic case, except that
now the potential U depends on the fields fQi further dressed by non-local operators.
For Berkovits’ SFT, the low-energy spacetime limit has never been derived
while retaining non-locality, so that it is not obvious whether the exponential
operator would arise also in that case. However, several dualities between different
open SFTs are known, including a mapping between supersymmetric and bosonic
classical solutions [194, 195], cubic and Berkovits’ supersymmetric SFTs (where
classical solutions are mapped onto one another) [196, 197] and between different
polynomial supersymmetric SFTs [198, 199]. In this sense, we can regard the non-
locality (12.78) as a generic feature of open superstring field theory at the effective
level.
Supersymmetric versions of both type-II and heterotic closed SFT have been
studied comparatively less [200–203] but very recently there have been exciting
developments. The computation of multi-loop off-shell amplitudes in covariant
superstring theory is at a closer hand than before [131, 132] and so is a complete
4
These rules, called Sen descent relations, were born as conjectures but they have been proven both
numerically and analytically in string field theory [159, 160].
formulation of type-II and heterotic closed super-SFT [204–208]. One expects to

find the same non-local structure in the low-energy effective spacetime actions.
The type of non-locality in (12.46) and (12.52) is particularly benign. Not only
is the Cauchy problem of the initial conditions well defined in general [209, 210],
but it also allows one to study tachyon condensation in all its phases [211–213].
Exact solutions of the full theory are available only for marginal deformations,
i.e., configurations where the brane is starting to decay and the tachyon has just
begun to roll down the potential [214]. At the level of target spacetime, one can go
beyond marginal deformations and there exist approximate but very accurate non-
perturbative solutions which describe the whole brane decay [212]. These properties
are a direct reflection of the gauge symmetries of SFT [213].5
12.3 Compactification
The multiple derivation of the value of the spacetime critical dimension

(Sects. 12.1.3 and 12.2.2) is of utmost importance since it is a prediction of a well-
defined quantum theory of strings. The bosonic string can be consistently quantized
in a Lorentz-invariant spacetime only when such spacetime has D D 1 C 25
directions, while the quantum superstring is consistent in D D 1 C 9. Such a
prediction, unavailable in most of the quantum-gravity frameworks of the preceding
chapters, is accessible thanks to the rigidity of CFTs and the virtual absence of
fundamental free parameters. However, one has to explain the discrepancy with
respect to the observed D D 1 C 3.
12.3.1 T-Duality
The most immediate possibility is to compactify 22 (in the bosonic case) or 6 (in the
supersymmetric one) spatial directions. Consider for instance the compactification
of one such direction on a circle S1 with radius r, so that the direction D 1 is
periodic with period 2r:
X D1 Š X D1 C 2rn ; n 2 Z: (12.79)
5
All known exact solutions in open SFT are superpositions of special states in the Fock space,
called surface states, which obey a “diffusion” equation involving only Virasoro and ghost
operators. This diffusion equation performs a change of gauge which reparametrizes a trivial non-
normalizable solution of the equation of motion into a non-trivial normalized solution. At the
level of spacetime fields, the same structure survives in (12.52) and, in fact, a spacetime diffusion
equation holds for non-perturbative approximate solutions [213]. States similar to the open-string
surface states exist also in the closed case [215], which is responsible of the fact that the closed-
string non-locality in (12.51) is the same as in (12.52).
12.3 Compactification 657
In this sub-section, we focus on the bosonic sector. The momentum of the center of
mass of the string takes discrete values on a compact manifold, so that pD1 D n=r.
Closed strings also have an extra feature: they can wrap around the circle. This gives
rise to another discrete quantity w 2 Z, the winding number, which counts how
many times a closed string winds around the compact direction: X D1 . ; C 2/ D
X D1 . ; / C 2rw. Overall,
n
X D1 . ; / D xD1 C ˛ 0 C wr C (oscillations) : (12.80)
r
Again, one can split the closed string into left-moving and right-moving modes as
in (12.14). In particular, the quantum momentum operator is
˛0 2 ˛0 2
PO D LO 0 LOQ 0 ; LO 0 D p C NO ; LOQ 0 D p C NOQ ; (12.81)
4 L 4 R
where
n wr
pL;R D ˙p : (12.82)
r ˛0
On a physical state and for a non-zero winding number, PO j i D 0 no longer implies

the level-matching condition: N NQ D nw ¤ 0. These relations reveal a surprising
symmetry of closed string theory which goes under the name of T-duality [216]:
the system is invariant under the simultaneous interchange of the compactification
radius with its inverse and of the winding number with momentum:
˛0
T-duality: r $ r0 D ; w $ n: (12.83)
r
R D 2˚ .D/
RWhen compactifying (12.40) or (12.71) on a circle, d xe R /
dD1 x r e2˚ R.D1/ C : : : . T-duality must preserve the physical couplings,
2˚ 0
implying that
p re D r0 e2˚ and that the dilaton must transform as ˚ 0 D
˚ ln.r= ˛ 0 /.
The limit of decompactification r ! C1 is physically
p equivalent to sending the
radius to zero. Therefore, the self-dual radius r D ˛ 0 D ls sets the characteristic
scale of perturbative string theory, as can also be seen via other arguments [217].
Below this scale, there is structure accessible with non-perturbative techniques [35].
The same arguments can be replicated for the other directions and T-duality is
extended to the compact manifold TD4 D S1 S1 , the .D 4/-dimensional
torus.
Open strings do not have a conserved winding number because they can always
be unwrapped from the compact space. Since w is T-dual to momentum, the center-
of-mass momentum is not conserved either, which corresponds to a breaking of
translation invariance. This breaking is due to the confinement of string end-points
to branes. The limit r ! 0 is dual to the decompactification limit, so that open
strings compactified on a torus of infinitely small radius are dual to strings in D

dimensions attached to Dp-branes. This picture can be made rigorous in several
ways.
12.3.2 Spontaneous Compactification
Although compactification can reduce the number of large spacetime dimensions

down to four, one should also find a mechanism which somehow forced the
extra dimensions to become compact. In other words, a viable compactification
scheme must occur in the theory in a “natural” manner. This requirement can be
characterized, first of all, by classical solutions to the equations of motion such
that fields acquire position-dependent vacuum expectation values along certain
directions where spacetime is strongly curved. In this case, the line element can be
decomposed into a four-dimensional line element (times an overall positive function
of the coordinates) plus a contribution from the extra dimensions:
d2 s2D D f 2 .x/ ds24 C e2u.x/ ds2D4 ; (12.84)
for some functions f and u. If such solutions exist and are stable, the compactifica-
tion is said to be spontaneous [218, 219].
However, this is not sufficient to obtain a realistic model of Nature. At the
classical level, the compactified theory must have the correct global and local
symmetries: it must recover both general-covariant Einstein gravity and the SU.3/˝
SU.2/ ˝ U.1/ Lorentz-invariant Standard Model, eventually embedded in a super-
symmetric extension, with the correct number of generations of fermions and
bosons. Undesired features such as a large effective cosmological constant should
also be avoided. At the quantum level, the dimensionally reduced theory should
be free from tachyons (meaning that quantization has been performed on a stable
vacuum) as well as gauge and gravitational anomalies, which would result in a
spoiling of symmetries and the appearance of negative-norm states. We will come
back to spontaneous compactification and its relation with the observed value D D 4
in Sect. 12.4.
12.3.3 Calabi–Yau Spaces and Orbifolds
The stringent requirements listed in Sect. 12.3.2 are not met in the bosonic string
and they provide further motivation for the superstring. There, not only is the
dimensionality of target spacetime reduced from 26 to 10, but the remaining 6 extra
directions can be compactified with a wealth of tools that allow for a comprehensive
classification of those spaces which comply with our wish-list. The structure of
the compact spatial manifold depends on a set of non-fundamental parameters for
which there is a freedom of choice. This freedom will make string theory lose some
predictive power but one will be able to get realistic phenomenological models of
Nature.
The simplest compactification of the low-energy limit of superstring theories to
four dimensions is on the flat hypertorus [102, 110, 220],
M10 Š M4 T6 : (12.85)
When dimensionally reduced to D D 4, SO.32/ super-Yang–Mills theory (i.e., type-

I superstring when ˛ 0 ! 0) generates N D 4 super-Yang–Mills, while both non-
chiral and chiral D D 10, N D 2 SUGRA (type-IIA and IIB superstrings when
˛ 0 ! 0) both collapse to the N D 8 SUGRA limit [142, 143, 221].6 These relations
are valid also beyond tree level [102]. If one does not require the dynamical effective
field equations to be solved exactly, these four-dimensional limits can contain any
number of fermion families.
Other types of compactifications, more involved than that on a 6-torus, are
necessary to obtain realistic effective quantum field theories complying with
the requirements of Sect. 12.3.2, valid at energies above those probed so far in
accelerators (.10 TeV) but much lower than the string mass scale ms D l1 s . In
particular, it is important to find compactification schemes where the classical string
equations are exactly soluble and the four-dimensional theory is anomaly-free. An
early instance where anomaly cancellation is preserved in the dimensional reduction
[136] consisted in compactifying four of the ten dimensions as M10 Š M6 K3,
where the compact space K3 is Kummer’s quartic surface [222, 223]. To achieve
this, it was not necessary to solve the classical string equations exactly. Two
dimensions of the manifold M6 can be further wrapped but simple compact spaces
such as T2 K3 do not give rise to viable four-dimensional phenomenology.
A powerful way to restrict the allowed compact spaces is to sharpen the
phenomenological demands imposed on the four-dimensional manifold M4 . In
particular, we may want to obtain the minimal extension of the Standard Model on
a maximally symmetric spacetime (Minkowski, de Sitter or anti-de Sitter). These
conditions are equivalent to have unbroken N D 1 supersymmetry on M4 and
Racbd / gab gcd gad gbc , where a; b; : : : are the coordinates on M4 . Then, starting
from heterotic SUGRA in ten dimensions, the compactification scheme is uniquely
selected as [224, 225]
M10 Š M4 C3 (12.86)
(up to overall conformal factors [226]), where M4 D M4 is Minkowski spacetime

(thus, the cosmological constant is zero) and C3 is a compact six-dimensional
6
A different compactification reduces D D 11, N D 1 SUGRA to D D 10, N D 1 SUGRA, in
turn reduced to D D 4, N D 4 SUGRA [149].
Calabi–Yau space. Calabi–Yau spaces Cn are Kähler manifolds (Sect. 5.12) with n
complex dimensions (2n real dimensions) endowed with an su.n/ connection (SU.n/
holonomy).7 In the present case, n D 3. The Riemannian and complex structures
are mutually compatible and the metric can be defined via a Kähler potential, as
described in Sect. 5.12. A consequence of having an su.3/ connection is that the
Ricci tensor Rmn of C3 vanishes identically. Applying the dimensional reduction
(12.86) to type-II theories, one ends up with D D 4, N D 2 SUGRA; the four-
manifold M4 in (12.86) can be more general than Minkowski spacetime and have
a non-vanishing cosmological constant. Calabi–Yau spaces can be compact or non-
compact.
K3 is a Calabi–Yau 2-fold and, in fact, it is the only compact simply-connected
2-fold. If the definition of Calabi–Yau spaces is extended also to manifolds with
proper sub-groups of SU.n/, then T6 and T2 K3 are compact 3-folds (which we
have already discarded on phenomenological grounds). A less trivial example is
the following. Let CPnC1 be the complex projective hyperplane, the space of n C 2
complex variables zi ¤ 0 identified under dilations, zi Š zi for any 2 C n f0g.
This is a Kähler manifold and so is any sub-space defined by analytic equations
on the zi . In particular, K3 Š CP3 . The space of all the zeros of a homogeneous
polynomial PnC2 .zi / of order n C 2 in CPnC1 is a compact Calabi–Yau n-fold.
This and other Calabi–Yau spaces constructed from complex projective hyper-
planes are discussed in the literature, together with the following example not
stemming from CPnC1 [224]. Let us parametrize the torus T2 with a complex
coordinate z under the identifications z Š z C 1 Š z C ei=3 . This set is invariant
under the Z3 symmetry generated by the transformation p ˛.z/ Dpe2i=3 z. On T2 , this
transformation has three fixed points, z D 0; e = 3; 2e = 3. On the 6-torus
i=6 i=6
T6 Š T2 T2 T2 , the transformation ˛.z1 /˛.z2 /˛.z3 / generates a Z3 symmetry

and has a total of 33 D 27 fixed points. Let F be the set of these fixed points and
TQ D T6 n F . Then, the quotient T=Z
Q 3 is a non-compact complex manifold which
can be rendered compact by gluing, at the position of each singularity (former fixed
point), copies of a non-compact manifold with certain properties and the correct
asymptotic behaviour. The result, called Z, is a simply-connected Calabi–Yau 3-
fold.
12.3.3.1 Orbifolds and Orientifolds
The explicit construction of Calabi–Yau spaces may pass through an intermediate

step known as orbifold, a space M=G (not necessarily a manifold) obtained by the
quotient of a smooth manifold M with a discrete group G [231, 232]. The name
orbifold comes from the fact that, by definition, for any group element g 2 G and
point y 2 M there is a congruence class y Š gy D ˛.y/, such that each point in
7
Kähler n-folds have a U.n/ holonomy. The group U.n/ Š SU.n/ ˝ U.1/ can be reduced to SU.n/
under special provisions, as conjectured by Calabi [227, 228] and proved by Yau [229, 230].
M=G is identified with its orbit fgy j g 2 Gg under the action of the group. Here
and in the following, yl are real coordinates on an orbifold or Calabi–Yau space. In
physical applications, l; m; n D 4; 5; : : : ; 9 and the four-fold M4 is spanned by the
coordinates xa .
A trivial example of a smooth orbifold is the hypertorus TN Š RN =Z N , where
N
Z is the discrete group of translations. Often, a less broad definition of orbifold
is employed where the action of G is not free. In this case, the transformation
˛.y/ has fixed points which constitute conical singularities in M=G. For instance,
consider the discrete group Z2 defined by reflections, ˛.y/ D y. On a unit circle
S1 parametrized by y 2 Œ0; 2, the only two fixed points are y D 0 and y D , so
that S1 =Z2 is an orbifold with two singularities.8
Amusingly, the flat tetrahedron so much used in quantum gravity is an orbifold
[233]: it is T2 =Z2 , where the origin of the reflection is the centre of the fundamental
Q 3 is another example of orbifold. Asymmetric
lattice cell defining the torus. T=Z
orbifolds where the left- and right-moving sectors live on different spaces can also
be conceived [234].
A generalization of orbifold of importance for type-I theory is the orientifold,
an orbifold where the twist symmetry (12.36) is implemented [235]. Thus, strings
on orientifolds are unoriented. Type-I superstring theory is obtained by a Z2 twist
on the left and right modes of the closed string [235–238]. The twist mixes the
two sectors of the theory, which have therefore to be symmetric. In particular, left
and right sectors must have the same chirality and the closed string from which the
open one descends must be of type IIB. Colloquially, the procedure to get the type-
I superstring may be described as placing type-IIB theory on an orientifold. The
bosonic open string descends in a similar way from the bosonic closed string.
The regularization (or “blowing up”) of the singularities of orbifolds can lead to
Calabi–Yau spaces, as in the case of Z. Similarly, K3 can be obtained by regularizing
the 16 singularities of the orbifold .T4 n F /=Z2 , where the group (isomorphic to)
Z2 is generated by the transformation ˛.yl / D yl on the periodic coordinates yl Š
yl C 1 on the four-torus [231]. Strings can wrap around singularities, which gives
rise to twisting modes apart from the usual winding modes in non-trivial topologies.
The regularization of singularities is a procedure more general than its applica-
tion to orbifolds and can produce compact Calabi–Yau spaces from non-compact
ones. An important example is the Klebanov–Strassler throat [239], a conical sin-
gularity [240] on a non-compact Calabi–Yau space. Deforming the throat to a cone
with a rounded tip, one obtains a compact Calabi–Yau n-fold with smooth geometry
[241]. The deformed Klebanov–Strassler throat is an important ingredient in flux
compactification and cosmological scenarios, as we will describe in Sects. 12.3.7
and 13.5.
An advantage of dealing with orbifolds is that, on one hand, there are not many
six-dimensional orbifolds compatible with the string and spacetime symmetries.
8
The discrete group Z N is defined by rotations of a vector by an angle of 2=N around the origin,
so that it has N fixed points on a circle.
On the other hand, the study of compactification schemes becomes fairly tractable
and singularities are relatively innocuous insofar as the string spectrum, dynamics
[232, 233] and interactions [233, 242] are concerned. On Z, on its generalizations
and on other Calabi–Yau spaces, the string classical equations are soluble with
arbitrarily good approximation (i.e., exactly soluble on the corresponding orbifolds
[231–233, 243]), which is one of the requisites for a spontaneous compactification.
When compactifying the E8 E8 heterotic string on some of these spaces, one can
obtain effective four-dimensional models with an almost realistically low number of
fermion generations (2, 4 or 5, but not 3) [224, 225, 232].
In the light of the web of dualities described in Sect. 12.4 between string theories
on one hand and an 11-dimensional theory of supergravity and membranes on the
other hand, compactification schemes from 11 to 4 dimensions are also of interest.
In that case, the compact space can be constructed from a Calabi–Yau four-fold C4
in the limit where one of the extra dimensions is unfolded [244]. For certain choices
of the seven-dimensional compact manifold, one can obtain N D 1 or N D 2
supersymmetry in D D 4 [245].
12.3.4 Cycles and Fluxes
Calabi–Yau spaces usually have compact n-dimensional sub-spaces n called

n-cycles. The integrals
Z
TQp
FpC2 D kF 2 Z (12.87)
2 pC2
of field strengths of the NS-NS sector and of the R-R potentials, performed on non-
trivial cycles pC2 of the compact manifold, are called . p C 2/-form fluxes and play
a crucial role in compactification schemes.9 An analogue of the Dirac quantization
condition on the electric and magnetic charge of Maxwell’s theory holds for
branes, so that fluxes are quantized by an integer kF (the “quantum number” of the
brane flux). This explains the appearance of the Dp-brane tension TQp D gs Tp D
Œ.2/p ls 1 in (12.87) (the left-hand side of (12.87) is dimensionless, since
pC1
ŒFpC2 D 1).
In the universal supergravity sector, one has a 1-form flux associated with the
dilaton and the 3-form flux of the Kalb–Ramond field strength H3 D dB2 ; this is the
quantum number associated with fundamental strings. The R-R fluxes are sourced
by the p-branes of the theory. For instance, in type-IIB string theory, the R-R fluxes
are the surface integrals of the 1-, 3- and 5-forms associated with, respectively, the
0-, 2- and 4-form R-R potentials. D3-branes are a source for the R-R electric and
9
An informal use of the term “flux” often applies to the field strengths themselves.
magnetic 5-form flux, D5-branes (more precisely, their dual D1-branes) source the
R-R magnetic 3-form flux, and so on.
12.3.5 Moduli
The shape and size of Calabi–Yau spaces are governed by a set of parameters called,
as in the case of Riemannian surfaces, moduli [249]. These parameters are actually
fields, since they depend on the position on C3 . In the most general stringy use of the
word, moduli are spacetime-dependent fields in 10 dimensions with a well-defined
Lorentz structure. Upon compactification, they usually reduce to a number of scalars
on the four-dimensional manifold M4 . In multi-brane configurations, moduli are
also the separations of branes. These will be extensively used by some important
cosmological models of Chap. 13.
We left the discussion in Sect. 5.12 at the point where one had a supergravity no-
scale model of the form (5.230) but apparently no justification for its origin from first
principles. Supersymmetry is spontaneously broken at the tree level at a scale which
cannot be determined due to an invariance of the potential under field rescalings,
hence the name “no-scale.” String theory completes the picture in a remarkable
way: compactifying the D D 10 supergravity action on a Calabi–Yau space C3 ,
the resulting low-energy four-dimensional model is precisely (5.230). This was first
shown for E8 E8 SUGRA [246–248]. The classical scale invariance of the SUGRA
action is an inheritance of conformal invariance of the superstring vertex diagrams.
The fields which appear in the super- and Kähler potential are the string moduli.
In this sub-section, we overview the main types of moduli without entering into
precise technical details, which the reader can find in [246, 247, 249, 250] (a more
pedagogical introduction is in [251]). To classify the spacetime moduli on M4 , we
consider the metric decomposition
ds210 D gN dx dx D e6u.x/ ds24 C e2u.x/ gmn .y/dym dyn ; (12.88a)
ds24 D gab .x/dx dx ;
a b
(12.88b)
where we have assumed for simplicity that all compactification radii are equal and
u is a scalar related to the radius by r D ls exp u. The metric gab is the one in the
Einstein frame (hence the Weyl prefactor e6u ). Energy units are absorbed in the
definitions so that all moduli are dimensionless.
12.3.5.1 Axions
String axions are real pseudo-scalar fields which arise from the p-forms (the NS-NS
and R-R potentials) of the theory [252]. In type-IIB string compactifications, where
most of the cosmological model we will see in Chap. 13 live, we have:
• The R-R 0-form C0 .
• The axion Q B of the universal SUGRA sector, coming from the D D 4 Hodge
.4/
dual of the four-dimensional NS-NS 2-form B2 D Bab dxa ^ dxb : d Q B D e8u
.4/
dB2 .
• The axion QC coming form the D D 4 Hodge dual of the four-dimensional R-R
.4/ .4/
2-form C2 D Cab dxa ^ dxb : d Q C D e8u dC2 .
.C/ .C/
• The dimensionally reduced R-R 4-form C4 produces a 2-form aab / Cabmn and
8u
one defines an axion as the Hodge dual e d D da2 . The factor exp.8u/
pops up after transforming from the string to the Einstein frame (see below).
• The axions arising from the fluxes of the D D 10 forms B2 and C2 over 2-cycles
.C/
of M6 and of C4 over 4-cycles. For N2 and N4 such cycles,
Z Z Z
1 1 1 .C/
B;i2 D B2 ; C;i2 D C2 ; i4 D C4 ;
2 ˛ 0 i
2 2 2 ˛ 0 i
2 2 2˛ 0 2 i
4 4
(12.89)
where i2 D 1; : : : ; N2 and i4 D 1; : : : ; N4 .
Some of these axions (e.g., Q B and Q C ) are absent in type-IIB orientifold compact-
ifications. Type-IIA and the heterotic strings have other axionic spectra. In general,
we will collect all axions of a model under the same symbol i , i D 1; : : : ; Naxions
where the total number of axions Naxions can be rigorously determined for a given
compact manifold and flux population. Typically, Naxions is very large.
The actions (12.74) and (12.75) depend only on the field strength of the p-forms,
so that axions enjoy the classical shift symmetry
i ! i C const ; (12.90)
where the constant is arbitrary. Actually, the shift (12.90) is the definition of classical
axion. However, at the quantum level non-perturbative instantonic effects, such as
Euclidean branes and the wrapping of string world-sheets on non-trivial cycles,
break this continuous symmetry to a discrete one, i ! i C 2n [253, 254]. The
period 2 can be intuitively understood by noting that the world-sheet wrapping on
a 2-cycle is governed by a Euclidean instanton action S given by the exponentiation
of the world-sheet action of the 2-form B2 , the imaginary term in (12.37). In
practice, the exponent is a 2-form flux on the world-sheet ˙2 where the integrand
is a combination of B2 and another (the Kähler) R 2-form. With the normalization
convention as in (12.89), S / expŒi.2 ˛ 0 /1 ˙2 B2 D ei B . Therefore, B has
a periodic potential with period 2 (restoring length units, 2ls ). We have seen an
example of such potential already in Chap. 5: it is the cosine profile (5.90) of natural
inflation. We will come back to string axions and their potentials in Sect. 13.4.
Axions appear not only as independent moduli but also as components of the
complex-structure moduli, the axio-dilaton and the Kähler moduli.
12.3.5.2 Complex-Structure Moduli
For a Calabi–Yau space, shape moduli or complex-structure moduli are coordinates

zi .x/ locally parametrizing the complex structure. If C3 has handles, cycles or other
non-trivial topology features, the number of complex-structure moduli can be very
large.
The Kähler potential associated with the shape moduli is [241, 249]
Z
Kcs .zi / D ln i ˝ ^ ˝" ; (12.91)
C3
where ˝lmn is the holomorphic .3; 0/-form of the Calabi–Yau space. In special
geometry, Kcs depends only on the imaginary part of shape moduli and the real
parts Rezi are axions (see, e.g., [255]).
12.3.5.3 Axio-Dilaton
String axions can be combined with other real scalars into complex moduli, so that
the low-energy effective action becomes a simple expression. Here we will see a
first example, the type-IIB axio-dilaton (or simply dilaton)
.x/ D e˚.x/ C iC0 .x/ : (12.92)
Note that, thanks to the choice of warping factors in (12.88), the ten-dimensional
dilaton ˚ is the same as the four-dimensional dilaton ˚4D (Problem 12.2). Making
the conformal transformation g D exp.˚=2/Ng and defining
G3 :D F3 i H3 ; (12.93)
the type-IIB low-energy action (12.75) in the Einstein frame becomes

Z
1 p @ @ jG3 j2
SIIB D 2
d10 x g R C : : : ; (12.94)
210 2.Re /2 2Re
where we wrote only the part of interest for the cosmological models of Chap. 13.
Other moduli which we will not meet during our journey are the 2-form scalars
Gi2 D C;i2 C i B;i2 .
The axio-dilaton contributes to the D D 10 Kähler potential with the term [241,
246–249]

K. / D ln. C /: (12.95)
Often in the literature, the definition ! i is employed as well as the symbol S.

We will reserve S (as the D D 4 chiral-supergravity scalar which appears in (5.230))
to the axio-dilaton of the heterotic string. In this case, there is no axion C0 and the
imaginary part of S is supplemented by the axion Q B of the universal SUGRA sector.
The heterotic axio-dilaton is
S.x/ D e˚.x/ C i QB .x/ : (12.96)
The Kähler potential is as in (12.95), K.S/ D ln.S C S /.
12.3.5.4 Kähler Moduli
The Kähler moduli determine the scales and the total volume of C3 . A Kähler
modulus % is made of a size modulus and an axion .10 The square of the radius
in (12.88), u itself or functions of u such as are indistinctly called radion. In
the case of the isotropic metric (12.88), there is only
R onepradion u and one size
modulus D exp.4u/, related to the volume V6 D C3 d6 y g.6/ of the Calabi–Yau
manifold by
V D .Re%/3=2 D 3=2 ; V :D l6

s V6 : (12.97)
Then, the Kähler modulus is
%.x/ D e4u.x/ C i .x/ D .x/ C i .x/ ; (12.98)
where was defined among the axions in the list above as the dualized 4-form. In
the heterotic case (chiral supergravity), the field % is called T. The Kähler potential
10
About nomenclature in the literature. The term “size modulus” is often used as a synonym for
the composite modulus C i , while “Kähler modulus” is often (and more correctly) used as a
synonym for the size modulus , since aR general definition of Kähler modulus is the flux of the
Kähler form over a 2-cycle, ti :D ˛ 0 1 2 J (the i below are functions of ti ). In this book, we
distinguish these two denominations as per our declaration above (Kähler modulus D C i ),
which appears in the literature as often as the others.
is augmented by the term [241, 246–248]
Q
K.%/ D 3 ln.% C % / D 2 ln V ; (12.99)
and the total Kähler potential is
Q
K D Kcs .zi / C K. / C K.%/ : (12.100)
With these ingredients, it is not difficult to see that the four-dimensional action is
Z q
4D 1 @ @ @ %@ %
SIIB D d4 x g.4/ R.4/ 2 6 C : : : ; (12.101)
242 .Re /2 .Re%/2
plus the complex-structure term. Consistently, this action can also be obtained by
compactifying the D D 10 expression (12.94) on (12.88).
Calabi–Yau spaces can be anisotropic and have six different radions. Moreover,
in type-IIB theory they have 2-cycles and 4-cycles and the total volume V is made
of all these contributions. Ignoring 2-cycle terms dependent on the 2-form scalars
Gi2 , Kähler moduli take the general form
%i4 .x/ D i4 .x/ C i i4 .x/ ; i4 D 1; : : : ; N4 ; (12.102)
where i4 is the volume of the i4 -th 4-cycle and i4 is defined in (12.89). The Kähler
Q
potential remains the same as in (12.99), K.V/ D 2 ln V, where V D V.%i4 /.
12.3.5.5 Stabilization of the Moduli
One of the major goals of modern string theory and string cosmology is to stabilize
the moduli. Moduli that do not acquire a fixed value can lead to trouble. The
compact manifold can decompactify and a higher-dimensional spacetime unfold.
The evolution of the universe can be disrupted at very early stages as well as at
late times. A running dilaton, for instance, would not respect the tight observational
constraints on the variation of Newton’s coupling G.
It turns out that there are two non-perturbative mechanisms that can perform
the task of freezing these fields: flux compactification and gaugino condensation.
The complex-structure moduli and the axio-dilaton are stabilized when the topology
of the Calabi–Yau space allows several flux fields to acquire non-zero expectation
values (Sect. 12.3.7). Kähler moduli are not stabilized by fluxes and the problem
of radion stabilization requires a different approach: the condensation of the
supersymmetric partners of the gauge fields (Sect. 12.3.9).
12.3.6 Stacking Branes
A technique to obtain a four-dimensional field theory with little or no super-

symmetry is to stack D-branes and other localized extended objects on compact
spaces and invoke the extremely powerful AdS/CFT correspondence. The AdS/CFT
correspondence [256–258] and its extensions (see [259–261] for reviews) state that
a given superstring theory on a suitable background induces a given supersymmetric
conformal field theory on the boundary of spacetime. To pass from one theory to the
other, one relates the string background to some brane configuration.
Let us first make a short digression about gauge groups and branes. In the case
of a single D-brane, one immediately notices that there is a U.1/ gauge theory
associated with it. Let gab D ab C .2l2s /2 @a m @b m be the induced metric on the
brane in flat spacetime, where m are D 1 p scalars representing the transverse
coordinates of the embedding. Expanding (12.19) in the fields, the quadratic terms
combine into the Lagrangian L.2/ D g2 YM .Fab F =4 C @a
ab
@ m =2/, where the
m a
2 p3
effective Yang–Mills coupling is gYM D 2.2/ ls gs . For D3-branes,
p2
g2YM D 4gs : (12.103)
In superstring theory, additional spinor fields in the DBI action provide the fermionic
degrees of freedom of the gauge theory. For N branes, the fields A and m
become matrices which transform in the adjoint representation of U.N/. The U.1/
excitations on an individual brane decouple from inter-brane modes (the non-
diagonal elements of the matrix fields) and the gauge group is therefore SU.N/. In
the large-N limit, the product g2YM N is kept fixed, which leads to the identification
gs 1=N. Thus, on a stack of branes one has a gauge theory. Embedding this stack
in a D D 10 spacetime is equivalent to consider a string theory in a spacetime
(usually the product of anti-de Sitter and a five-manifold) with boundaries. The
AdS/CFT insight consists in recognizing that the degrees of freedom on the
boundary are described by the gauge theory.
The first established AdS/CFT correspondence is between type-IIB strings on
AdS5 S5 (five-dimensional anti-de Sitter (AdS) spacetime times the 5-sphere) and
the large-N limit of SU.N/ super-Yang–Mills gauge theory in D D 4 with N D 4
supersymmetries (hence 8 supercharges) [256–258]. It is easy to see that AdS5
S5 spacetime is conveniently described by N parallel D3-branes on a smooth ten-
dimensional manifold. An SU.N/ super-CFT lives on the brane stack and, in the
low-energy limit, it decouples from bulk degrees of freedom. Since the Minkowski
spacetime M4 on which the CFT lives is identified (up to some points at infinity) with
the boundary of AdS5 , the AdS/CFT correspondence is holographic: in the large-N
limit, the physics in the bulk is described by that of its boundary. The dimension of
operators in the gauge theory are then related to the modes of the string spectrum,
which is of type IIB in this case.
More typically, branes are placed at the singularity of the transverse space [262–
266]. In one explicit example, AdS5 times a certain five-dimensional manifold M5
is equivalent to a stack of parallel D3-branes at the conical singularity of a non-

compact Calabi–Yau 3-fold, which eventually helps to show that type-IIB string
theory on AdS5 M5 is dual to an N D 1 super-CFT in four dimensions [265].
These correspondences are carried out by using classical SUGRA actions, since
string loop corrections go as gs 1=N and are sub-dominant in the large-N limit.
12.3.7 Flux Compactification
Since D-branes carry fluxes, the stacking of branes on a compact manifold can be
regarded as a case of flux compactification [226, 241, 267–270]. If the expectation
value of gauge forms is non-zero in the compact space, background fluxes are turned
on and supersymmetry is partially broken.11 This helps in reducing the number
of supercharges in the final configuration. Moreover, fluxes generate effective
potentials for the axio-dilaton and the shape moduli of the compact space. The
Gukov–Vafa–Witten superpotential associated with G3 is [244, 272, 273]
Z
W D WG :D ˝ ^ G3 ; (12.104)
C3
where ˝ is the holomorphic form in (12.91) and G3 is the complex 3-flux (12.93)
stemming from the combination of the NS-NS and R-R 3-fluxes together with the
D D 10 axio-dilaton in type-IIB string theory. In the flux compactification of [241],
WG combines with the Kähler potential (12.100) (given by (12.91), (12.95) and
(12.99)) into the analogue of the potentials (5.225) and (5.231):
44 V D eK G ij Di WG .Dj WG / 3jWG j2

X
D eK G ij Di WG .Dj WG / : (12.105)
i;j¤%
In the second equality, the sum is over all moduli except %, since WG is independent

of % and the Kähler potential (12.99) is such that the F-term G %% D% WG .D% WG / D
2 Q D .j%j =3/j 3WG =%j D 3jWG j exactly cancels the term
.j%j =3/jWG @% Kj 2 2 2 2
3jWG j2 . We follow the convention (5.224), so that ŒK D 0 D ŒW and all moduli
fields are dimensionless.
Supersymmetry is spontaneously broken by WG : for a given flux background,
WG D W0 D const ¤ 0 : (12.106)
11
In D D 11, vacuum solutions that respect the full supersymmetry on M4 times a compact space
do not admit non-trivial fluxes [271].
Once the flux is fixed, the complex structure zi and the axio-dilaton adjust to
minimize the F-terms (12.105) associated with (12.104) [241, 268]. The moduli
space is therefore partially stabilized.
The presence of fluxes is important but not sufficient to guarantee a finite scale
hierarchy which could explain why the Planck mass mPl is much larger than the
electroweak energy scale. In addition, one must also ensure that spacetime be
warped. A warped spacetime [274], either non-compact [239, 275] or compact
[226, 269, 276–280], is a special case of (12.88) such that the normalization of the
four-dimensional metric gab .xa ; ym / D expŒ2!.ym /Qgab .xa / varies in the transverse
directions ym :
ds2D D e2!.y/.D4/u.x/ ds24 C e2!.y/C2u.x/ gmn .y/dym dyn ; (12.107a)

ds24 D gQ ab .x/ dx dx :
a b
(12.107b)
We have already mentioned an example of warped space: the deformed Klebanov–

Strassler throat [241], which we will discuss in Sect. 13.5.2. Typically, Calabi–Yau
3-folds possess several such warped regions and it is easy to place fluxes on a warped
background without fine tuning.
The electroweak scale can be generated by the suppression of the D-dimensional
Planck scale by the position-dependent warping factor !. This mechanism can be
seen in action, for instance, by plugging the Ansatz (12.107) with gQ ab D ab in the
equations of motion of classical D D 10 SUGRA with the R-R and NS-NS fluxes
(F3 D dC2 and H3 D dB2 ) of type-IIB supergravity, and then finding solutions
!.y/ and gmn .y/ on a compact manifold [241]. For a smooth energy-momentum
tensor, all solutions on compact warped spacetimes are such that all fluxes vanish
and ! is constant [276, 277]. To obtain non-trivial solutions, it is necessary to
include localized matter sources such as D-branes and objects with negative tension
called Op-planes (the “O” stands for orientifold); their . p C 1/-dimensional world-
volume actions are hence added to the low-energy SUGRA action. The solutions
of [241] have vanishing four-dimensional cosmological constant and are invariant
under the rescaling gmn ! 2 gmn . Non-supersymmetric solutions (more generic
than supersymmetric ones, which have N D 1 in this case) with D 0 and a
dynamical radial modulus u (in practice, the Kähler modulus %) are nothing but no-
scale models, as it can be checked by noting that the superpotential generated by the
3-fluxes of the solutions obeys the condition (5.223).
No-scale models are, in general, scenarios where size (Kähler) moduli are
unfixed. In all solutions, a large hierarchy of scales can be generated by rea-
sonably small flux quantum numbers, since the warp factors in (12.107) depend
exponentially on these. Both the flux content and the presence of localized sources
is determined by the string and brane spectra of superstring theory, which thus
provides a concrete realization of the models of Sect. 5.12. Notice, however, that
˛ 0 -corrections to the SUGRA action, quantum corrections and non-perturbative
effects can spoil the no-scale properties of solutions and hence stabilize the radion
(Sect. 12.3.9).
There exist a wealth of flux compactifications involving different compact spaces

and superstring theories (often related by the dualities described in Sect. 12.4)
[226, 241, 267, 268, 281–296]. The goal, in general, is to promptly stabilize all the
moduli and to obtain as much a realistic low-energy four-dimensional field theory
as possible.
12.3.8 String Theory and the Standard Model
In the quest for the Standard Model, much of the attention has been drawn to
models in type-I and heterotic string theory [297–303] and, especially, to type-II
orientifold compactifications with branes intersecting at angles. Orientifolds are a
somewhat special class of Calabi–Yau spaces but they permit to obtain controllable
realizations of both the non-supersymmetric Standard Model with three generations
[304–318] and its minimal N D 1 supersymmetric extension [304, 319–335]. At
the intersection of D-branes [336], chiral spinorial degrees of freedom emerge which
mimic the fermionic families of the SU.3/ ˝ SU.2/ ˝ U.1/ Standard Model.
Usually, the result is not completely satisfactory due to the presence of a “hidden
sector” of extra particles that cannot be removed from the spectrum. When the
hidden sector has masses above the present observational limits, it becomes either
innocuous or a valuable asset, inasmuch as it can give characteristic predictions
that could be tested in accelerators. However, if these particles have light masses,
the model is regarded only as semi-realistic. In their open-string sector, non-
supersymmetric models closely resemble the Standard Model but they have the
additional problem that their vacuum is not protected from instabilities by exact or
softly-broken supersymmetry. The consequence is that some moduli in the closed-
string sector (including the dilaton) are not stabilized [308]. Moreover, it is difficult
to reconcile particle-physics models with cosmology, mostly because the parameter
ranges required by moduli stabilization and by a small positive do not overlap.
All these reasons stimulate the further study of the embedding of the Standard
Model in string theory in compactification schemes with background fluxes and
brane stacks, intersecting branes or branes at singularities [289, 295, 337–352]. A
very important element of the discussion, related to the way moduli are stabilized, is
how to determine the vacuum of the theory in such schemes. In the next sub-section,
we will tackle this problem in general terms, without reference to specific proposals
for a Standard Model.
12.3.9 Anti-de Sitter Vacua
In type-IIA superstring theory, all moduli can be stabilized by fluxes [291–296].

The resulting D D 4 stable vacua are either non-supersymmetric or N D 1
supersymmetric, with a negative cosmological constant. In type-IIB and heterotic
theories, classical fluxes alone cannot give a mass to Kähler moduli and one has
to include quantum correction to the super- and Kähler potentials. String loop
corrections to the Kähler potential [353] can give a mass to all the moduli in type-IIB
theory to yield an AdS vacuum [354], but at the price of some fine tuning. Barring
this perturbative effect, it is necessary to introduce a non-perturbative mechanism
which we describe presently. As in the type-IIA case, the resulting stable vacua are
anti-de Sitter.
12.3.9.1 Gaugino Condensation and String Instantons
We have seen examples of condensates in Sects. 7.6.4, 9.4.2 and 11.5.2. Bilinears
(fermionic, in the first two cases) acquired a non-trivial expectation value h N i ¤
0, leading to a more favourable vacuum configuration and the breaking of certain
symmetries of the system. The second of these properties can play a role in
the stabilization of Kähler moduli and it is based on the observation that local
supersymmetry of SUGRA models can be broken by the condensation of gauginos,
the super-partners of the gauge bosons [355–359]. When this happens, a mass m3=2
for the gravitino and a superpotential for the Kähler moduli are induced [355, 358].
Condensation is also a rather generic feature of string theory. For instance, among
the type-IIB brane configurations appearing in the AdS/CFT correspondence of
Sect. 12.3.6, consider the N D 1 SU.N/ super-Yang–Mills theory living on a stack
of N D7-branes. The only modulus not fixed by the flux background is the radion
D Re% and it determines the gauge coupling via
8 2
D 2 : (12.108)
g2YM
If gluino condensation takes place, then the superpotential term Wcond D

A exp.2%=N/ is generated, where A is a positive constant determined by the
UV cut-off scale of the gauge theory. The non-perturbative nature of this correction
is clear from (12.103), since exp.Re%/ D exp.1=gs/. A similar superpotential W
arises also from the contribution of string instantons [254, 360–363], in particular
type-IIB Euclidean D3-branes [361, 362], and gaugino condensation in four
dimensions actually descends from an M5-brane instanton in M-theory [364].
In general, the normalization A D A.zi ; / depends on the other moduli and it
becomes a constant when these are stabilized.
In E8 E8 heterotic string theory, the breaking of one of the E8 groups down to a
product of smaller non-Abelian groups is sufficient to trigger gaugino condensation,
which, in turn, breaks supersymmetry via an effective superpotential that features
the axio-dilaton S rather than the modulus T [365–373],
Wcond .S/ D AeˇS ; (12.109)

where ˇ is a constant. In general, one sees the emergence of multi-exponential

superpotentials of the form [368, 369]
X X P
Wcond .S/ D cm eˇm S ; Wcond .%i / D cn e l ˛l;n %l ; (12.110)
m n
which are called racetrack.
12.3.9.2 KKLT Stabilization Scenarios
The stabilization of Kähler moduli by these non-perturbative mechanisms is well

illustrated by the Kachru–Kallosh–Linde–Trivedi model (KLT in short) [374]. For
simplicity, one considers only one Kähler modulus % in type-IIB string theory
and combines the tree-level contribution (12.106) with Wcond into the generic
superpotential
W.%/ D W0 C Wcond .%/ D W0 C A e˛% ; (12.111)
where W0 < 0, A > 0 and ˛ > 0 are real-valued constants. Remarkably,

both (12.111) and a two-field racetrack potential arise, independently from flux
compactification, as a phenomenological requirement to address the moduli problem
(Sect. 13.2) in the supergravity scenario with a heavy gravitino [375].
One assumes that the dilaton and complex-structure moduli have already been
stabilized by fluxes, so that the potential only depends on %:
n o
Q
44 V D eK.%/ G %% D% W.%/ŒD% W.%/ 3jW.%/j2 ; (12.112)
where KQ is given by (12.99) and we have absorbed an overall numerical normal-

ization factor in the definition of W. From @% W D ˛A exp.˛%/ and @% KQ D
3=.2Re%/, the definition (5.221) is

3W0 3
D% W D @% W C W@% KQ D A e˛% ˛ C : (12.113)
2Re% 2Re%
After decomposing % as
%D Ci ; (12.114)

where D Re% and D Im%, and noting that G %% D .% C % /2 =3 D 4 2 =3, the
potential (12.112) becomes
˛A 2˛ h ˛ i
44 V D e 1 C A C W0 e ˛
cos.˛ / : (12.115)
2 2 3
For the sole purpose of keeping the presentation pedagogical, take now the axionic
part of the Kähler modulus to be zero, % D , and repeat the discussion around
(7.24). The condition D% W D 0 for a supersymmetric vacuum is met at some critical
value min > 0 such that

2˛min
W0 D 1 C A e˛min : (12.116)
3
Correspondingly, the minimum of (12.115) with D 0 is negative:
Q .˛A/2 2˛min
44 Vmin D 3jW.min /j2 eK.min / D e < 0: (12.117)
6min
Q
Notice that the gravitino mass m3=2 :D eK=2 jWjj Dmin MPl is determined by the scale
of the minimum
Vmin D 3m23=2 MPl2 ; (12.118)
where MPl D 41 is the reduced Planck mass in four dimensions.

For the SUGRA approximation to be valid and to maintain ˛ 0 -corrections to
the Kähler potential suppressed, the volume of compact space must be sufficiently
large, 1 (small Yang–Mills coupling gYM , perturbative field-theory regime).
Also, non-perturbative corrections to W are under control provided ˛ > 1. Both
conditions are met in (12.116) if jW0 j 1, which can be obtained by tuning the
fluxes to preserve supersymmetry. The potential (12.115) for typical values of the
parameters is depicted in Fig. 12.4. For jW0 j 1013 , the gravitino mass is O.TeV/
and the string scale is close to the grand-unification scale.
To summarize, instantonic effects or gaugino condensation in the KLT model
preserve supersymmetry but lower the minimum of the potential to negative values,
thus violating the property Vmin D 0 of no-scale solutions. This results in a four-
manifold M4 with a negative cosmological constant: the supersymmetric vacuum
is anti-de Sitter spacetime.
Realistic type-IIB models with many light moduli are more complicated than
the original KLT construction [376–400] (see [377, 378, 381, 389] for especially
hands-on constructions). In the KLT model, we have integrated out the axio-dilaton
and the complex-structure moduli zi before including non-perturbative quantum
effects. This procedure, which resulted in the effective superpotential (12.111), is
not justified in general. For instance, when one tries to stabilize the dilaton and
the Kähler modulus % simultaneously, the resulting AdS minimum is unstable;
Fig. 12.4 The K LT potential (12.115) multiplied by 1015 and in Planck-mass units MPl D 41 D
1, with W0 D 104 , A D 1 and ˛ D 0:1, in the .; / plane (top panel) and for D 0 (bottom
panel). The local minimum is at min 113:6
then, one has to turn on the complex-structure moduli as well [378, 388, 392].
Other moduli arise as follows. Chiral fields mimicking the Standard Model can be
generated in three steps, by placing a singularity near the tip of the throat, modding
out the discrete symmetry associated with such a singularity and placing an anti-D3-
brane (indicated as D3-brane) on one of the fixed points [387]. For consistency, the
resulting Calabi–Yau space must be populated by D7-branes [401, 402] and other
D3- and D3-branes. The presence of gauge fields on the branes at the fixed points
introduces new moduli to be stabilized.
In many of these generalizations, the contribution Wcond from gaugino conden-
sation is of the racetrack form (12.110) and the potential V may acquire several
minima.
Gaugino condensation and instantonic branes have also been considered in
heterotic string theory [376, 386, 392, 403, 404]. Also in these cases the stable vacua
are anti-de Sitter or, under some tuning, Minkowski. All models where moduli are
fixed to an AdS or Minkowski minimum will be called KLT stabilization scenarios.
12.3.9.3 Large-Volume Stabilization Scenarios
An important offspring of the KLT construction are type-IIB large-volume sta-

bilization scenarios [382, 383, 390, 393, 395, 405–411]. The combined presence
of ˛ 0 3 -corrections [412] and many Kähler moduli (as well as gs loop correc-
tions to the Kähler potential, in certain cases [408, 410]) generically leads to a
non-supersymmetric AdS vacuum corresponding to a Calabi–Yau space with an
exponentially large volume V6 . The leading ˛ 0 -corrections modify the total Kähler
potential (12.100) as [412]

0 3
K D Kcs .zi / C K. / C KQ .˛ / .%/ D Kcs .zi / C K. / 2 ln V C gs 2 ; (12.119)
where 0 < D O.1/ is a constant dependent on the characteristics of C3 . These

corrections are perturbative in the string coupling and hence in a large-volume
expansion, since gs D 1=Re% / V 2=3 for one size modulus. The next-to-leading
term in the logarithm in (12.119) is O.V 1=3 / [395]. Viable models have more than
one size modulus. The case of the orientifold P4Œ1;1;1;6;9 (a Calabi–Yau manifold
constructed in [389, 413]) features just two fields % and % , where the labels
stand for large and small, respectively. The “small” modulus is assumed to give a
negative contribution to the total dimensionless volume
3 3 3 3
V / .Re% / 2 .Re% / 2 D 2
2
: (12.120)
If
> 1 ; (12.121)
then V6 is positive and large compared to the string scale. More small moduli
would represent holes in a “Swiss-cheese” compact space. The superpotential is of
racetrack type. Assuming, as in the KLT scenario, that the dilaton and the complex-
structure moduli have been fixed previously, one has
W D W0 C A e˛ % C A e˛ % ' W0 C A e˛ % : (12.122)
Using (12.112), (12.119), (12.121) and (12.122), one finds the effective potential
p
jW0 j ˛ jW0 j2
44 V D ˇ1 e2˛ C ˇ2 e cos.˛ C / C ˇ3 3=2
;
V V2 gs V 3
(12.123)
where ˇ1;2;3 are model-dependent numerical coefficients that also depend on

˛ A , is defined as exp.i / :D A W0 =jA W0 j and D Im% is
the axion of the “small” mode. Minimizing V with respect to the axion at the global
minimum min
D . /=˛ , we get
p
2˛ jW0 j ˛ jW0 j2
44 V D ˇ1 e ˇ2 e C ˇ3 3=2
: (12.124)
V V2 gs V 3
The range of the constants is such that the potential has a local negative
minimum, as in the KLT case. A qualitative estimate runs as follows. The minimum
of V in the direction is found by solving the transcendental equation @V=@ D
0. If ˛ is moderately large but .ˇ2 =ˇ1 /2 D O.1/, one has ˛ min
/ ln V
p
and e˛ / jW0 j=V approximately. Plugging this back into (12.124) and
minimizing in V (i.e., ), one gets
!
2=3 ˛ min
min / ; V min / jW0 j exp ; (12.125)
gs gs
while the potential mininum is
jW0 j2 3=2
Vmin D Vnp C V˛0 D 1 .ln V/ 2 ; V D V min ; (12.126)
44 V 3
where 1;2 are constants. Thus, with a relatively small string coupling gs we can
obtain a moderately large modulus , an exponentially large volume of the
compact space and, hence, an exponentially small negative cosmological constant
Vmin .
The string and gravitino mass are related to the four-dimensional reduced Planck
mass by
gs MPl g2 W0 MPl
ms D p p ; m3=2 D ps ; (12.127)
4 V 4 V
and a large hierarchy of scales is generated. Therefore, string scenarios with large
extra dimensions [414–416] can be embedded concretely in this large-volume
stabilization scheme. In specific type-IIB examples, V6 D 104 –1015 l6s and, for gs D
0:1 and W0 D 102 , the string scale and the gravitino mass are ms 1015 –109 GeV
and m3=2 1014 –103 GeV. Very large volumes V6 D 1015 l6s explain the Planck-
to-electroweak hierarchy with a supersymmetry breaking scale MSUSY m3=2 D
O.TeV/.
A difference with respect to KLT scenarios is that there is no strong theoretical

constraint on W0 , which can take non-large values W0 < 10–100. The fine tuning
jW0 j 1 in the KLT construction is mainly due to the limited range of applicability
of the approximations, where one must balance the size between tree-level and
non-perturbative contributions relative to ignored ˛ 0 -corrections. Supersymmetry
breaking is also managed differently. In KLT stabilization scenarios, the AdS min-
imum is supersymmetric; supersymmetry will be broken only later, by the uplifting
mechanisms described in Sect. 13.1.1. On the other hand, the AdS minimum of
large-volume scenarios is already non-supersymmetric, due to the F-terms of Kähler
moduli.
12.4 Dualities and M-Theory
The five D D 10 superstring theories are equivalent to one another by dualities,

relations that hold in certain compactification limits or manipulations of the
parameters. These equivalences are physical, meaning that the mass spectrum, the
brane population, the low-energy effective action, the partition function and other
features of two dual theories coincide.
A hint of a yet deeper structure is the existence of dualities also between D D 10
and D D 11 supergravity. The bosonic sector of D D 11 SUGRA is sketched as
Z
1 11 p 1 2
S11 D 2 d x g R 2 F4 C L.A3 / ; (12.128)
211 2
where is a bare cosmological constant, F4 is the field strength of the three-form A3

and L.A3 / is the Chern–Simons Lagrangian. Notice the absence of a dilaton field.
Both D D 10 and D D 11 SUGRA are special for different reasons. Nahm showed
that D D 9 C 1 is the highest spacetime dimensionality where super-Yang–Mills
theories can be constructed, while supergravity cannot exist in more than 10 C 1
dimensions [145]. Moreover, an elegant theorem by Freund and Rubin shows that
D-dimensional Einstein gravity with a rank-.s 1/ anti-symmetric tensor field As1
compactifies spontaneously (in the sense of Sect. 12.3.1) as Ms MDs , where one
of the manifolds is non-compact and contains the time direction and the other is
a Riemannian compact manifold [417]. In the absence of supersymmetry, nothing
more can characterize the number of compact dimensions, since one can introduce
tensors of arbitrary rank 1 6 s 6 D. However, in D D 11 supergravity there is
only one s-form, with s D 4. Therefore, there is a preferential compactification
M4 M7 and the compact space is either four- or seven-dimensional. Notably,
in the second case one has just a four-dimensional spacetime. Eleven dimensions
are special also in a cosmological context not related to supersymmetry, for two
independent reasons. First, D D 11 marks a transition in the way to approach
12.4 Dualities and M-Theory 679
the big-bang singularity in general relativity without matter, oscillatory chaotic á

la BKL for D 6 10 and monotonic otherwise [418–420] (Sect. 6.3.5). Second, in
general relativity with p-forms the approach to the singularity is chaotic for D > 3
and the structure of chaos depends on D: it cannot be a Kac–Moody billiard for
D > 11 and, in the presence of just one 3-form field, it is a Coxeter polyhedron only
in D D 11 [421].
Below, we will see that different compactifications of D D 11 SUGRA coincide
with the compactifications (in one dimension less) of D D 10 SUGRA. Given that
N D 2 and N D 1 supergravity in D D 10 are the low-energy limit of superstring
theories, one may suspect a connection between the latter and an 11-dimensional
structure, conventionally called M-theory. The “M” [422] stands for “mother” (it
is supposed to be the fundamental theory unifying all superstrings), “magic” (for
its marvels), “mysterious” (because it is known only through certain features, for
instance its low-energy limit and its brane content), “membrane” (as its building
blocks are identified with branes [423–425]) and “matrix” (as there is evidence that
its degrees of freedom are the same of a matrix model [426]).
In particular, the low-energy limit of M-theory is postulated to be D D 11
E M
SUGRA. Recall that, to a CpC1 form, there corresponds a dual form CD3p and
that these are interpreted, in string theory, as Dp- and D.D 4 p/-branes. Then,
one can conjecture that, just like D D 10 SUGRA is the low-energy limit of
superstring theory with a certain brane population, the structure whose low-energy
limit is D D 11 SUGRA is a quantum theory of super-membranes [423, 424].
Since in (12.128) there is a 3-form and its magnetic dual is a 6-form, M-theory
is characterized by M2- and M5-branes.12 The classical and quantum dynamics of
these branes can be studied via the construction of a class of . p C 1/-dimensional
actions [423–425] and via the dualities between D D 10 superstrings and the eleven-
dimensional theory.
To help the reading, we mark a duality between D D 10 superstrings with a bullet
and a duality between a D D 10 superstring theory and the D D 11 theory with a
triangle I.
The first type of relation between string theories is T-duality, introduced in
Sect. 12.3.1. (a) The E8 E8 and SO.32/ heterotic strings are related by a twist
in the boundary conditions [232]. Upon toroidal compactification (in particular, on
M 9 S1 ), they collapse into each other [427–429]. Since taking large and small radii
corresponds to perform gauge transformations, this equivalence can be interpreted
as the fact that the two heterotic superstrings are different ground states of the same
theory. (b) Also type-IIA and type-IIB theories are mutually equivalent when
dimensionally reduced on a hypertorus (in particular, on S1 ) [49, 430]. As in the
previous case, type-II theories are interpreted as different ground states of a larger
structure.
12
We reserve the “D” for the Dirichlet branes of string theories, while the p-super-membranes of
the D D 11 theory are called Mp-branes.
Another relation between superstrings is S-duality, which maps one theory with
coupling gs (the parameter governing the loop expansion) to another (or the same)
theory with coupling g1
s [431–435]:
1
S-duality: gs ! : (12.129)
gs
The letter “S” indicates that this duality is at the level of string states rather than of
target spacetime. T-duality is a target duality. While T-duality is perturbative in gs
and non-perturbative in ˛ 0 , S-duality is non-perturbative in gs and is valid order by
order in ˛ 0 (so that it can be checked at the level of low-energy effective actions).13
Finally, the generalization of S-duality to type-II theories is called U-duality
[436]. “U” stands for unified, since it involves both S and T dualities. However, to
stress the non-perturbative relation (12.129) between the strong- and weak-coupling
sector of mathematically different theories, U-dualities are often called S-dualities.
(c) Type-I string theory and SO.32/ heterotic string theory are S-dual [437],
while (d) type-IIB string theory is S-self-dual [436]. I (e) A U-duality also holds
between type-IIA theory and 11-dimensional SUGRA on M 10 S1 [436, 438, 439],
and I ( f) between E8 E8 heterotic theory and 11-dimensional SUGRA on M 10
S1 =Z2 [422, 440–442].14
To give a flavour of (e) and of how the low-energy limit of type-IIA string
theory connects with 11-dimensional gravity, we compactify the action (12.128)
on a circle S1 with radius r. The eleven-dimensional metric gMN decomposes into
a scalar ˚.x /, a vector A .x / and the ten-dimensional metric gN .x /, with
; ; D 0; : : : ; 9:
2 4
ds2 D gMN dxM dxN D e 3 ˚ gN dx dx C e 3 ˚ .dx10 C A dx /2 ; (12.130)
where M; N D 0; : : : ; 10. Replacing this decomposition into (12.128) and ignoring

, we obtain exactly (12.74). All the low-energy ingredients of type-IIA string
theory are in place: the universal SUGRA sector (graviton gN 2 , dilaton ˚ and Kalb–
Ramond field B2 ) and the R-R fields C1 and C3 , corresponding to D0-, D6-, D2- and
D4-branes. Intuitively, one direction of the M2- and M5-brane of the D D 11 theory
wraps around the circle to give, respectively, the D1- and D4-brane in D D 10.
If the M2-brane has toroidal spatial sections, the resulting D1-branes are tubular
surfaces interpreted as type-IIA strings, as the dimensional reduction of the three-
dimensional brane action to the string action strongly indicates [425]. More involved
13
Originally, T and S referred to the T and S fields in the chiral multiplet composing the
superpotential and Kähler potential (5.230) in D D 4 SUGRA (regarded as the dimensional
reduction of D D 10 heterotic string on a six-torus). The decompactification limit corresponds
to ReT ! 1, while the real part of the complex dilaton field S is 1=g2c [431].
14
An early study of solitons in the low-energy limit of the heterotic string suggested that, at strong
coupling, the latter admits a dual description as a weakly interacting theory of D5-branes [443].
12.4 Dualities and M-Theory 681
arguments at the quantum level confirm the conclusion that the strong-coupling
behaviour of type-IIA string theory is weakly-coupled D D 11 supergravity.
In ten dimensions, the gravitational coupling constant for all closed string
theories is
2
210 D .2/7 l8s g2c ; (12.131)
while the gravitational coupling in eleven and ten dimensions are related by
2 2
211 D .2r/210 D .2/8 l9s g3c : (12.132)
2
Defining the 11-dimensional Planck mass and length as 211 D .2/8 M11
9
D
8 9
.2/ l11 , we get
1 3 l311
l11 D gc3 ls ; gc D .M11 r/ 2 ; l2s D : (12.133)
r
The D D 11 theory is the non-perturbative decompactification limit gc ! 1 (r !
1) of the D D 10 theory.
Different compactifications give rise to a wealth of U- and S-dualities between
the heterotic theories (which coincide when dimensionally reduced, according to
(a)) and type-II strings [436, 438, 444–449]. I (g) On M 7 T3 , heterotic string
theory is dual to D D 11 SUGRA on M 7 K3 (here, a closed string is obtained by
wrapping an M5-brane with topology K3 S1 around K3); (h) on M 6 T4 , it is
dual to type-IIA string theory on M 6 K3; (i) on M 5 T5 , it is dual to type-IIB
string theory on M 6 K3; (j) on M 4 T6 and consistently with (a), it is dual to
itself. The E8 E8 heterotic theory on M 6 K3 is both (k) dual to itself and I (l)
dual to the D D 11 theory compactified on M 6 K3 S1 =Z2 , consistently with ( f).
Some derived dualities follow from those above. By virtue of (e), (g) and (h), I
(m) both heterotic and type-IIA string theory in six dimensions are dual to a model
of super-membranes in D D 11 on M 6 K3 S1 [450]. Also, by virtue of (b) and
(e), I (n) type-II string theory on M 9 S1 is dual to the D D 11 theory on M 9 T2
[451–453].
Finally, Calabi–Yau compactifications give rise to more threads of the web of
dualities constituting M-theory. Compatibly with (e), I (o) the low-energy limit of
type-IIA theory compactified on M 4 C3 is equivalent to D D 11 SUGRA on M 4
C3 S1 , where C3 is the same Calabi–Yau space [454, 455]. Moreover, I ( p) D D 11
SUGRA on M 5 C3 [454, 455] is dual to heterotic theory on M 5 K3S1 [456–458].
The fundamental heterotic string is identified with the M5-brane wrapped around a
four-cycle of the Calabi–Yau space. I (q) The D D 11 theory on M 5 C3 is also
equivalent to type-II SUGRA on M 4 C3 when one of the compact directions is
appropriately decompactified [457]. Last but not least, we recall from Sect. 12.3.3
that (r) type-I theory is type-IIB theory on an orientifold [235–238].
Fig. 12.5 M theory: the web of dualities among string theories (solid lines) and between string
theories and the D D 11 super-membrane model (dot-dashed lines). Along each connection, the
topology of the compactification scheme at each end-point is indicated. The dualities (a) and (b)
are also valid on M 9d T d for d > 2
These 18 dualities, and many more we have not discussed here, marked one of
the deepest discoveries in string theory and a breakthrough in our understanding of
its non-perturbative structure. The partial list (a)-(i), (k) and ( p) is summarized in
Fig. 12.5.
The flux compactification of M-theory [269, 270, 459–470] follows a scheme
similar to that in string theory. Non-perturbative effects stabilize the moduli on
Minkowski or anti-de Sitter vacua.
12.1 Conserved supercurrent. Show that @a J a D 0 on shell, where J a D

b a .@b X / =2 and the a obey the Clifford algebra (12.54).
References 683
Solution From (12.54), one notes that a b a D b a a 2ab a D 2b C

2b D 0, so that
2@a J a D b a .@a @b X / C b a .@b X /.@a / D 0:

(12.55)

12.2 Dilaton in 10 and 4 dimensions. Modify (12.88) with a generic warp

factor e2u gmn ! e2ˇu gmn in the six-dimensional metric, with ˇ a constant.
Show that for ˇ ¤ 1 the ten-dimensional dilaton ˚ is not the same as the
four-dimensional dilaton ˚4D .
Solution The D D 10 Ricci scalar in the low-energy action (12.75) decomposes

N g D RN.4/ ŒNgab C RN.6/ ŒNgmn C : : : D e6u R.4/ Œgab C e2ˇu R.6/ Œgmn C : : :
as RŒN
(from (7.150), with ˝ D e3u in the four-dimensional part and ˝ D eˇu p in the
six-dimensional
p part),p while the ten-dimensional
p p volume factor is split as Ng D
12u .4/ 6ˇu .6/ 6.ˇ2/u .4/ .6/
.e g /.e g /De g g , so that
p q q
Nge2˚ RN D g.6/ g.4/ e2˚ e6.ˇ1/u R.4/ C e4.3ˇ/u R.6/ C : : : :
This expressions defines the four-dimensional dilaton ˚4D :D ˚ 3.ˇ 1/u.

Therefore, a choice of warping factors different than (12.88) leads, in general, to
˚ ¤ ˚4D .
References
1. G. Veneziano, Construction of a crossing-simmetric, Regge-behaved amplitude for linearly

rising trajectories. Nuovo Cim. A 57, 190 (1968)
2. S. Fubini, G. Veneziano, Level structure of dual-resonance models. Nuovo Cim. A 64, 811
(1969)
3. M.A. Virasoro, Alternative constructions of crossing-symmetric amplitudes with Regge
behavior. Phys. Rev. 177, 2309 (1969).
4. M.A. Virasoro, Subsidiary conditions and ghosts in dual-resonance models. Phys. Rev. D 1,
2933 (1970)
5. P. Ramond, Dual theory for free fermions. Phys. Rev. D 3, 2415 (1971)
6. A. Neveu, J.H. Schwarz, Factorizable dual model of pions. Nucl. Phys. B 31, 86 (1971)
7. A. Neveu, J.H. Schwarz, Quark model of dual pions. Phys. Rev. D 4, 1109 (1971)
8. E. Del Giudice, P. Di Vecchia, S. Fubini, General properties of the dual resonance model.
Ann. Phys. (N.Y.) 70, 378 (1972)
9. E. Del Giudice, P. Di Vecchia, S. Fubini, R. Musto, Light-cone physics and duality. Nuovo
Cim. A 12, 813 (1972)
10. P. Goddard, C.B. Thorn, Compatibility of the dual Pomeron with unitarity and the absence of
ghosts in the dual resonance model. Phys. Lett. B 40, 235 (1972)
11. R.C. Brower, Spectrum-generating algebra and no-ghost theorem for the dual model. Phys.
Rev. D 6, 1655 (1972)
12. P. Goddard, C. Rebbi, C.B. Thorn, Lorentz covariance and the physical states in dual-
resonance models. Nuovo Cim. A 12, 425 (1972)
13. R.C. Brower, K.A. Friedman, Spectrum-generating algebra and no-ghost theorem for the
Neveu–Schwarz model. Phys. Rev. D 7, 535 (1973)
14. J.H. Schwarz, Dual resonance theory. Phys. Rep. 8, 269 (1973)
15. G. Veneziano, An introduction to dual models of strong interactions and their physical
motivations. Phys. Rep. 9, 199 (1974)
16. J. Scherk, J.H. Schwarz, Dual field theory of quarks and gluons. Phys. Lett. B 57, 463 (1975)
17. Y. Nambu, Quark model and the factorization of the Veneziano amplitude, in Symmetries and
Quark Models, ed. by R. Chand (Gordon and Breach, New York, 1970) [Reprinted in Broken
Symmetry. Selected Papers of Y. Nambu, ed. by T. Eguchi, K. Nishijima (World Scientific,
Singapore, 1995)]
18. L. Susskind, Structure of hadrons implied by duality. Phys. Rev. D 1, 1182 (1970)
19. T. Takabayasi, Relativistic quantum mechanics of a mechanical continuum underlying the
dual amplitude. Prog. Theor. Phys. 44, 1429 (1970)
20. O. Hara, On origin and physical meaning of Ward-like identity in dual-resonance model. Prog.
Theor. Phys. 46, 1549 (1971)
21. T. Gotō, Relativistic quantum mechanics of one-dimensional mechanical continuum and
subsidiary condition of dual resonance model. Prog. Theor. Phys. 46, 1560 (1971)
22. L.N. Chang, F. Mansouri, Dynamics underlying duality and gauge invariance in the dual-
resonance models. Phys. Rev. D 5, 2535 (1972)
23. F. Mansouri, Y. Nambu, Gauge conditions in dual resonance models. Phys. Lett. B 39, 375
(1972)
24. P. Goddard, J. Goldstone, C. Rebbi, C.B. Thorn, Quantum dynamics of a massless relativistic
string. Nucl. Phys. B 56, 109 (1973)
25. S. Mandelstam, Interacting-string picture of dual-resonance models. Nucl. Phys. B 64, 205
(1973)
26. M. Kalb, P. Ramond, Classical direct interstring action. Phys. Rev. D 9, 2273 (1974)
27. M. Kaku, K. Kikkawa, Field theory of relativistic strings. I. Trees. Phys. Rev. D 10, 1110
(1974)
28. M. Kaku, K. Kikkawa, Field theory of relativistic strings. II. Loops and Pomerons. Phys.
Rev. D 10, 1823 (1974)
29. C. Rebbi, Dual models and relativistic quantum strings. Phys. Rep. 12, 1 (1974)
30. T. Yoneya, Quantum gravity and the zero-slope limit of the generalized Virasoro model. Lett.
Nuovo Cim. 8, 951 (1973)
31. T. Yoneya, Connection of dual models to electrodynamics and gravidynamics. Prog. Theor.
Phys. 51, 1907 (1974)
32. J. Scherk, J.H. Schwarz, Dual models for non-hadrons. Nucl. Phys. B 81, 118 (1974)
33. J. Scherk, J.H. Schwarz, Dual models and the geometry of space-time. Phys. Lett. B 52, 347
(1974)
34. M.B. Green, J.H. Schwarz, E. Witten, Superstring Theory (Cambridge University Press,
Cambridge, 1987)
Press, Cambrdige, 2007)
37. B. Zwiebach, A First Course in String Theory (Cambridge University Press, Cambridge,
2009)
38. D. Tong, String theory. arXiv:0908.0333
39. L. Brink, P. Di Vecchia, P.S. Howe, A locally supersymmetric and reparametrization invariant
action for the spinning string. Phys. Lett. B 65, 471 (1976)
40. S. Deser, B. Zumino, A complete action for the spinning string. Phys. Lett. B 65, 369 (1976)
41. A.M. Polyakov, Quantum geometry of bosonic strings. Phys. Lett. B 103, 207 (1981)
References 685
42. A.M. Polyakov, Conformal symmetry of critical fluctuations. Pisma Zh. Eksp. Teor. Fiz. 12,
538 (1970) [JETP Lett. 12, 381 (1970)]
43. A.M. Polyakov, Non-Hamiltonian approach to conformal quantum field theory. Zh. Eksp.
Teor. Fiz. 66, 23 (1974) [Sov. Phys. JETP 39, 10 (1974)]
44. A.A. Belavin, A.M. Polyakov, A.B. Zamolodchikov, Infinite conformal symmetry in two-
dimensional quantum field theory. Nucl. Phys. B 241, 333 (1984)
45. D. Friedan, Z. Qiu, S. Shenker, Conformal invariance, unitarity, and critical exponents in two
dimensions. Phys. Rev. Lett. 52, 1575 (1984)
46. D. Friedan, Z. Qiu, S. Shenker, Superconformal invariance in two dimensions and the
tricritical Ising model. Phys. Lett. B 151, 37 (1985)
47. D. Friedan, S. Shenker, E. Martinec, Covariant quantization of superstrings. Phys. Lett. B
160, 55 (1985)
48. D. Friedan, E. Martinec, S. Shenker, Conformal invariance, supersymmetry and string theory.
Nucl. Phys. B 271, 93 (1986)
49. J. Dai, R.G. Leigh, J. Polchinski, New connections between string theories. Mod. Phys. Lett.
A 04, 2073 (1989)
50. C. Becchi, A. Rouet, R. Stora, The abelian Higgs Kibble model, unitarity of the S-operator.
Phys. Lett. B 52, 344 (1974)
51. C. Becchi, A. Rouet, R. Stora, Renormalization of the abelian Higgs–Kibble model. Commun.
Math. Phys. 42, 127 (1975)
52. C. Becchi, A. Rouet, R. Stora, Renormalization of gauge theories. Ann. Phys. (N.Y.) 98, 287
(1976)
53. I.V. Tyutin, Gauge invariance in field theory and statistical physics in operator formalism,
unpublished (1975). arXiv:0812.0580
54. K. Fujikawa, Path integral of relativistic strings. Phys. Rev. D 25, 2584 (1982)
55. M. Kato, K. Ogawa, Covariant quantization of string based on BRS invariance. Nucl. Phys.
B 212, 443 (1983)
56. S. Hwang, Covariant quantization of the string in dimensions D 6 26 using a Becchi–Rouet–
Stora formulation. Phys. Rev. D 28, 2614 (1983)
57. J.L. Gervais, A. Neveu, The dual string spectrum in Polyakov’s quantization (I). Nucl. Phys.
B 199, 59 (1982)
58. J.L. Gervais, A. Neveu, Dual string spectrum in Polyakov’s quantization (II). Mode separa-
tion. Nucl. Phys. B 209, 125 (1982)
59. J.E. Paton, H.M. Chan, Generalized Veneziano model with isospin. Nucl. Phys. B 10, 516
(1969)
60. N. Marcus, A. Sagnotti, Group theory from “quarks” at the ends of strings. Phys. Lett. B 188,
58 (1987)
61. M.R. Douglas, B. Grinstein, Dilaton tadpole for the open bosonic string. Phys. Lett. B 183,
52 (1987); Erratum-ibid. B 187, 442 (1987)
62. S. Weinberg, Cancellation of one-loop divergences in SO.8192/ string theory. Phys. Lett. B
187, 278 (1987)
63. C. Lovelace, Pomeron form factors and dual Regge cuts. Phys. Lett. B 34, 500 (1971)
64. L. Bers, Uniformization, moduli, and Kleinian groups. Bull. Lond. Math. Soc. 4, 257 (1972)
65. S. Wolpert, On the homology of the moduli space of stable curves. Ann. Math. 118, 491
(1983)
66. D. Friedan, S. Shenker, The integrable analytic geometry of quantum string. Phys. Lett. B
175, 287 (1986)
67. D. Friedan, S. Shenker, The analytic geometry of two-dimensional conformal field theory.
Nucl. Phys. B 281, 509 (1987)
68. D. Friedan, Nonlinear models in 2 C dimensions. Phys. Rev. Lett. 45, 1057 (1980)
69. C. Lovelace, Strings in curved space. Phys. Lett. B 135, 75 (1984)
70. D.H. Friedan, Nonlinear models in 2 C " dimensions. Ann. Phys. (N.Y.) 163, 318 (1985)
71. C.G. Callan, D. Friedan, E.J. Martinec, M.J. Perry, Strings in background fields. Nucl. Phys.
B 262, 593 (1985)
72. E.S. Fradkin, A.A. Tseytlin, Effective field theory from quantized strings. Phys. Lett. B 158,
316 (1985)
73. E.S. Fradkin, A.A. Tseytlin, Quantum string theory effective action. Nucl. Phys. B 261, 1
(1985)
74. K. Ohmori, A review on tachyon condensation in open string field theories.
75. E. Fuchs, M. Kroyter, Analytical solutions of open string field theory. Phys. Rep. 502, 89
(2011). [arXiv:0807.4722]
76. E. Witten, Non-commutative geometry and string field theory. Nucl. Phys. B 268, 253 (1986)
77. Z.-a. Qiu, A. Strominger, Gauge symmetries in (super)string field theory. Phys. Rev. D 36,
1794 (1987)
78. Y. Okawa, L. Rastelli, B. Zwiebach, Analytic solutions for tachyon condensation with general
projectors. arXiv:hep-th/0611110
79. V.A. Kostelecký, S. Samuel, The static tachyon potential in the open bosonic string theory.
Phys. Lett. B 207, 169 (1988)
80. V.A. Kostelecký, S. Samuel, On a nonperturbative vacuum for the open bosonic string. Nucl.
Phys. B 336, 263 (1990)
81. L. Brekke, P.G.O. Freund, M. Olson, E. Witten, Nonarchimedean string dynamics. Nucl.
Phys. B 302, 365 (1988)
82. M. Saadi, B. Zwiebach, Closed string field theory from polyhedra. Ann. Phys. (N.Y.) 192,
213 (1989)
83. T. Kugo, H. Kunitomo, K. Suehiro, Nonpolynomial closed string field theory. Phys. Lett. B
226, 48 (1989)
84. T. Kugo, K. Suehiro, Nonpolynomial closed string field theory: action and its gauge
invariance. Nucl. Phys. B 337, 434 (1990)
85. V.A. Kostelecký, S. Samuel, Collective physics in the closed bosonic string. Phys. Rev. D 42,
1289 (1990)
86. B. Zwiebach, Closed string field theory: quantum action and the Batalin–Vilkovisky master
equation. Nucl. Phys. B 390, 33 (1993). arXiv:hep-th/9206084
87. A. Sen, B. Zwiebach, A proof of local background independence of classical closed string
field theory. Nucl. Phys. B 414, 649 (1994). [arXiv:hep-th/9307088]
88. A. Sen, B. Zwiebach, Quantum background independence of closed string field theory. Nucl.
89. A. Sen, B. Zwiebach, A note on gauge transformations in Batalin–Vilkovisky theory. Phys.
90. Y. Okawa, B. Zwiebach, Twisted Tachyon condensation in closed string field theory. JHEP
0403, 056 (2004). [arXiv:hep-th/0403051]
91. H. Yang, B. Zwiebach, Dilaton deformations in closed string field theory. JHEP 0505, 032
(2005). [arXiv:hep-th/0502161]
92. H. Yang, B. Zwiebach, A closed string tachyon vacuum? JHEP 0509, 054 (2005).
93. Y. Michishita, Field redefinitions, T-duality and solutions in closed string field theories. JHEP
0609, 001 (2006). [arXiv:hep-th/0602251]
94. N. Moeller, Closed bosonic string field theory at quintic order: five-tachyon contact term and
dilaton theorem. JHEP 0703, 043 (2007). [arXiv:hep-th/0609209]
95. N. Moeller, Closed bosonic string field theory at quintic order. II: marginal deformations and
effective potential. JHEP 0709, 118 (2007). [arXiv:0705.2102]
96. N. Moeller, A tachyon lump in closed string field theory. JHEP 0809, 056 (2008).
[arXiv:0804.0697]
97. L. Brekke, P.G.O. Freund, p-adic numbers in physics. Phys. Rep. 233, 1 (1993)
98. N. Moeller, M. Schnabl, Tachyon condensation in open-closed p-adic string theory. JHEP
0401, 011 (2004). [arXiv:hep-th/0304213]
99. V. Vladimirov, Nonlinear equations for p-adic open, closed, and open-closed strings. Theor.
Math. Phys. 149, 1604 (2006). [arXiv:0705.4600]
References 687
100. K. Ohmori, Toward open-closed string theoretical description of rolling tachyon. Phys. Rev.
D 69, 026008 (2004). [arXiv:hep-th/0306096]
101. T. Biswas, J.A.R. Cembranos, J.I. Kapusta, Thermal duality and Hagedorn transition from
p-adic strings. Phys. Rev. Lett. 104, 021601 (2010). [arXiv:0910.2274]
102. J.H. Schwarz, Superstring theory. Phys. Rep. 89, 223 (1982)
103. M.B. Green, Supersymmetrical dual string theories and their field theory limits —a review.
Surveys High Energy Phys. 3, 127 (1983)
104. L. Brink, J.H. Schwarz, Local complex supersymmetry in two dimensions. Nucl. Phys. B
121, 285 (1977)
105. A.M. Polyakov, Quantum geometry of fermionic strings. Phys. Lett. B 103, 211 (1981)
106. M.B. Green, J.H. Schwarz, Covariant description of superstrings. Phys. Lett. B 136, 367
(1984)
107. M.B. Green, J.H. Schwarz, Properties of the covariant formulation of superstring theories.
Nucl. Phys. B 243, 285 (1984)
108. F. Gliozzi, J. Scherk, D.I. Olive, Supersymmetry, supergravity theories and the dual spinor
model. Nucl. Phys. B 122, 253 (1977)
109. M.B. Green, J.H. Schwarz, Supersymmetrical dual string theory. Nucl. Phys. B 181, 502
(1981)
110. M.B. Green, J.H. Schwarz, Supersymmetrical string theories. Phys. Lett. B 109, 444 (1982)
111. F. Gliozzi, J. Scherk, D.I. Olive, Supergravity and the spinor dual model. Phys. Lett. B 65,
282 (1976)
112. L. Brink, J.H. Schwarz, J. Scherk, Supersymmetric Yang–Mills theories. Nucl. Phys. B 121,
77 (1977)
113. N. Marcus, A. Sagnotti, Tree-level constraints on gauge groups for type I superstrings. Phys.
Lett. B 119, 97 (1982)
114. M.B. Green, J.H. Schwarz, Supersymmetric dual string theory: (II). Vertices and trees. Nucl.
Phys. B 198, 252 (1982)
115. M.B. Green, J.H. Schwarz, Supersymmetric dual string theory: (III). Loops and renormaliza-
tion. Nucl. Phys. B 198, 441 (1982)
116. M.B. Green, J.H. Schwarz, Superstring interactions. Nucl. Phys. B 218, 43 (1983)
117. M.B. Green, J.H. Schwarz, L. Brink, Superfield theory of type (II) superstrings. Nucl. Phys.
B 219, 437 (1983)
118. M.B. Green, J.H. Schwarz, Superstring field theory. Nucl. Phys. B 243, 475 (1984)
119. M.B. Green, J.H. Schwarz, Infinity cancellations in SO.32/ superstring theory. Phys. Lett. B
151, 21 (1985)
120. E. D’Hoker, D.H. Phong, Momentum analyticity and finiteness of the 1-loop superstring
amplitude. Phys. Rev. Lett. 70, 3692 (1993). [arXiv:hep-th/9302003]
121. E. D’Hoker, D.H. Phong, Two-loop superstrings. I. Main formulas. Phys. Lett. B 529, 241
(2002). [arXiv:hep-th/0110247]
122. E. D’Hoker, D.H. Phong, Two-loop superstrings. II. The chiral measure on moduli space.
123. E. D’Hoker, D.H. Phong, Two-loop superstrings. III. Slice independence and absence of
ambiguities. Nucl. Phys. B 636, 61 (2002). [arXiv:hep-th/0111016]
124. E. D’Hoker, D.H. Phong, Two-loop superstrings. IV. The cosmological constant and modular
forms. Nucl. Phys. B 639, 129 (2002). [arXiv:hep-th/0111040]
125. E. D’Hoker, D.H. Phong, Two-loop superstrings. V. Gauge slice independence of the N-point
function. Nucl. Phys. B 715, 91 (2005). [arXiv:hep-th/0501196]
126. E. D’Hoker, D.H. Phong, Two-loop superstrings VI: Non-renormalization theorems and the
4-point function. Nucl. Phys. B 715, 3 (2005). [arXiv:hep-th/0501197]
127. E. D’Hoker, D.H. Phong, Two-loop vacuum energy for Calabi–Yau orbifold models. Nucl.
Phys. B 877, 343 (2013). [arXiv:1307.1749]
128. E.J. Martinec, Nonrenormalization theorems and fermionic string finiteness. Phys. Lett. B
171, 189 (1986)
129. E.P. Verlinde, H.L. Verlinde, Multiloop calculations in covariant superstring theory. Phys.
Lett. B 192, 95 (1987)
130. E. Witten, Superstring perturbation theory revisited. arXiv:1209.5461
131. A. Sen, Off-shell amplitudes in superstring theory. Fortsch. Phys. 63, 149 (2015).
[arXiv:1408.0571]
132. A. Sen, E. Witten, Filling the gaps with PCO’s. JHEP 1509, 004 (2015). [arXiv:1504.00609]
133. A. Sen, Ultraviolet and infrared divergences in superstring theory. arXiv:1512.00026
134. L. Alvarez-Gaumé, E. Witten, Gravitational anomalies. Nucl. Phys. B 234, 269 (1984)
135. M.B. Green, J.H. Schwarz, Anomaly cancellation in supersymmetric D D 10 gauge theory
and superstring theory. Phys. Lett. B 149, 117 (1984)
136. M.B. Green, J.H. Schwarz, P.C. West, Anomaly-free chiral theories in six dimensions. Nucl.
Phys. B 254, 327 (1985)
137. M.B. Green, J.H. Schwarz, The hexagon gauge anomaly in type 1 superstring theory. Nucl.
Phys. B 255, 93 (1985)
138. A. Sagnotti, A note on the Green–Schwarz mechanism in open string theories. Phys. Lett. B
294, 196 (1992). [arXiv:hep-th/9210127]
139. D.J. Gross, J.A. Harvey, E. Martinec, R. Rohm, Heterotic string. Phys. Rev. Lett. 54, 502
(1985)
140. D.J. Gross, J.A. Harvey, E. Martinec, R. Rohm, Heterotic string theory: (I). The free heterotic
string. Nucl. Phys. B 256, 253 (1985)
141. D.J. Gross, J.A. Harvey, E. Martinec, R. Rohm, Heterotic string theory: (II). The interacting
heterotic string. Nucl. Phys. B 267, 75 (1986)
142. E. Cremmer, B. Julia, The N D 8 supergravity theory. I. The Lagrangian. Phys. Lett. B 80,
48 (1978)
143. E. Cremmer, B. Julia, The SO.8/ supergravity. Nucl. Phys. B 159, 141 (1979)
144. E. Cremmer, B. Julia, J. Scherk, Supergravity in theory in 11 dimensions. Phys. Lett. B 76,
409 (1978)
145. W. Nahm, Supersymmetries and their representations. Nucl. Phys. B 135, 149 (1978)
146. M.B. Green, J.H. Schwarz, Extended supergravity in ten dimensions. Phys. Lett. B 122, 143
(1983)
147. J.H. Schwarz, P.C. West, Symmetries and transformations of chiral N D 2, D D 10
supergravity. Phys. Lett. B 126, 301 (1983)
148. P.S. Howe, P.C. West, The complete N D 2, d D 10 supergravity. Nucl. Phys. B 238, 181
(1984)
149. A.H. Chamseddine, N D 4 supergravity coupled to N D 4 matter and hidden symmetries.
Nucl. Phys. B 185, 403 (1981)
150. E.S. Fradkin, A.A. Tseytlin, Effective action approach to superstring theory. Phys. Lett. B
160, 69 (1985)
151. A. Sen, Equations of motion for the heterotic string theory from the conformal invariance of
the sigma model. Phys. Rev. Lett. 55, 1846 (1985)
152. A. Sen, Heterotic string in an arbitrary background field. Phys. Rev. D 32, 2102 (1985)
153. R.I. Nepomechie, Magnetic monopoles from antisymmetric tensor gauge fields. Phys. Rev.
D 31, 1921 (1985)
154. C. Teitelboim, Gauge invariance for extended objects. Phys. Lett. B 167, 63 (1986)
155. C. Teitelboim, Monopoles of higher rank. Phys. Lett. B 167, 69 (1986)
156. J. Polchinski, Dirichlet branes and Ramond–Ramond charges. Phys. Rev. Lett. 75, 4724
(1995). [arXiv:hep-th/9510017]
157. A. Sen, Descent relations among bosonic D-branes. Int. J. Mod. Phys. A 14, 4061 (1999).
158. A. Sen, Non-BPS states and branes in string theory. Class. Quantum Grav. 17, 1251 (2000).
159. M. Schnabl, Analytic solution for tachyon condensation in open string field theory. Adv.
Theor. Math. Phys. 10, 433 (2006). [arXiv:hep-th/0511286]
References 689
160. T. Erler, C. Maccaferri, String field theory solution for any open string background. JHEP
1410, 029 (2014). [arXiv:1406.3021]
161. A. Sen, SO(32) spinors of type I and other solitons on brane-antibrane pair. JHEP 9809, 023
(1998). [arXiv:hep-th/9808141]
162. A. Sen, BPS D-branes on non-supersymmetric cycles. JHEP 9812, 021 (1998).
163. A. Sen, Supersymmetric world-volume action for non-BPS D-branes. JHEP 9910, 008 (1999).
164. A. Sen, Universality of the tachyon potential. JHEP 9912, 027 (1999).
165. M.R. Garousi, Tachyon couplings on non-BPS D-branes and Dirac–Born–Infeld action. Nucl.
166. E.A. Bergshoeff, M. de Roo, T.C. de Wit, E. Eyras, S. Panda, T-duality and actions for non-
BPS D-branes. JHEP 0005, 009 (2000). [arXiv:hep-th/0003221]
167. J. Klusoň, Proposal for non-Bogomol’nyi–Prasad–Sommerfield D-brane action. Phys. Rev.
D 62, 126003 (2000). [arXiv:hep-th/0004106]
168. G.W. Gibbons, K. Hori, P. Yi, String fluid from unstable D-branes. Nucl. Phys. B 596, 136
(2001). [arXiv:hep-th/0009061]
169. D. Kutasov, M. Mariño, G.W. Moore, Some exact results on tachyon condensation in string
field theory. JHEP 0010, 045 (2000). [arXiv:hep-th/0009148]
170. D. Kutasov, M. Mariño, G.W. Moore, Remarks on tachyon condensation in superstring field
theory. arXiv:hep-th/0010108
171. P. Kraus, F. Larsen, Boundary string field theory of the DDN system. Phys. Rev. D 63, 106004
(2001). [arXiv:hep-th/0012198]
172. T. Takayanagi, S. Terashima, T. Uesugi, Brane-antibrane action from boundary string field
theory. JHEP 0103, 019 (2001). [arXiv:hep-th/0012210]
173. A. Sen, Rolling tachyon. JHEP 0204, 048 (2002). [arXiv:hep-th/0203211]
174. A. Sen, Tachyon matter. JHEP 0207, 065 (2002). [arXiv:hep-th/0203265]
175. A. Sen, Field theory of tachyon matter. Mod. Phys. Lett. A 17, 1797 (2002).
176. F. Leblond, A.W. Peet, SD-brane gravity fields and rolling tachyons. JHEP 0304, 048 (2003).
177. N.D. Lambert, H. Liu, J.M. Maldacena, Closed strings from decaying D-branes. JHEP 0703,
014 (2007). [arXiv:hep-th/0303139]
178. M.R. Garousi, Off-shell extension of S-matrix elements and tachyonic effective actions. JHEP
0304, 027 (2003). [arXiv:hep-th/0303239]
179. M.R. Garousi, Slowly varying tachyon and tachyon potential. JHEP 0305, 058 (2003).
180. A. Sen, Tachyon dynamics in open string theory. Int. J. Mod. Phys. A 20, 5513 (2005).
181. E. Witten, Interacting field theory of open superstrings. Nucl. Phys. B 276, 291 (1986)
182. C.R. Preitschopf, C.B. Thorn, S.A. Yost, Superstring field theory. Nucl. Phys. B 337, 363
(1990)
183. I.Ya. Aref’eva, P.B. Medvedev, A.P. Zubarev, Background formalism for superstring field
theory. Phys. Lett. B 240, 356 (1990)
184. I.Ya. Aref’eva, P.B. Medvedev, A.P. Zubarev, New representation for string field solves the
consistency problem for open superstring field theory. Nucl. Phys. B 341, 464 (1990)
185. I.Ya. Aref’eva, A.S. Koshelev, D.M. Belov, P.B. Medvedev, Tachyon condensation in cubic
superstring field theory. Nucl. Phys. B 638, 3 (2002). [arXiv:hep-th/0011117]
186. I.Ya. Aref’eva, L.V. Joukovskaya, A.S. Koshelev, Time evolution in superstring field theory
on nonBPS brane. 1. Rolling tachyon and energy momentum conservation. JHEP 0309, 012
(2003). [arXiv:hep-th/0301137]
187. N. Berkovits, Super-Poincaré invariant superstring field theory. Nucl. Phys. B 459, 439
(1996). [arXiv:hep-th/9503099]
188. N. Berkovits, A. Sen, B. Zwiebach, Tachyon condensation in superstring field theory. Nucl.
189. T. Erler, S. Konopka, I. Sachs, Resolving Witten’s superstring field theory. JHEP 1404, 150
(2014). [arXiv:1312.2948]
190. H. Kunitomo, Y. Okawa, Complete action of open superstring field theory. Prog. Theor. Exp.
Phys. 2016, 023B01 (2016). [arXiv:1508.00366]
191. T. Erler, Y. Okawa, T. Takezaki, Complete action for open superstring field theory with cyclic
A1 structure. JHEP 1608, 012 (2016). [arXiv:1602.02582]
192. H. Matsunaga, Comments on complete actions for open superstring field theory.
arXiv:1510.06023
193. S. Konopka, I. Sachs, Open superstring field theory on the restricted Hilbert space. JHEP
1604, 164 (2016). [arXiv:1602.02583]
194. T. Erler, Marginal solutions for the superstring. JHEP 0707, 050 (2007). [arXiv:0704.0930]
195. E. Fuchs, M. Kroyter, Marginal deformation for the photon in superstring field theory. JHEP
0711, 005 (2007). [arXiv:0706.0717]
196. E. Fuchs, M. Kroyter, On the classical equivalence of superstring field theories. JHEP 0810,
054 (2008). [arXiv:0805.4386]
197. I.Ya. Aref’eva, R.V. Gorbachev and P.B. Medvedev, Pure gauge configurations and solutions
to fermionic superstring field theories equations of motion. J. Phys. A 42, 304001 (2009).
[arXiv:0903.1273]
198. M. Kroyter, Superstring field theory equivalence: Ramond sector. JHEP 0910, 044 (2009).
[arXiv:0905.1168]
199. M. Kroyter, Comments on superstring field theory and its vacuum solution. JHEP 0908, 048
(2009). [arXiv:0905.3501]
200. Y. Okawa, B. Zwiebach, Heterotic string field theory. JHEP 0407, 042 (2004).
201. N. Berkovits, Y. Okawa, B. Zwiebach, WZW-like action for heterotic string field theory. JHEP
0411, 038 (2004). [arXiv:hep-th/0409018]
202. B. Jurčo, K. Münster, Type II superstring field theory: geometric approach and operadic
description. JHEP 1304, 126 (2013). [arXiv:1303.2323]
203. H. Matsunaga, Construction of a gauge-invariant action for type II superstring field theory.
arXiv:1305.3893
204. T. Erler, S. Konopka, I. Sachs, NS-NS sector of closed superstring field theory. JHEP 1408,
158 (2014). [arXiv:1403.0940]
205. T. Erler, S. Konopka, I. Sachs, Ramond equations of motion in superstring field theory. JHEP
1511, 199 (2015). [arXiv:1506.05774]
206. A. Sen, BV master action for heterotic and type II string field theories. JHEP 1602, 087
(2016). [arXiv:1508.05387]
207. A. Sen, Covariant action for type IIB supergravity. JHEP 1607, 017 (2016).
[arXiv:1511.08220]
208. A. Sen, Wilsonian effective action of superstring theory. arXiv:1609.00459
209. N. Barnaby, N. Kamran, Dynamics with infinitely many derivatives: the initial value problem.
JHEP 0802, 008 (2008). [arXiv:0709.3968]
210. G. Calcagni, M. Montobbio, G. Nardelli, Localization of nonlocal theories. Phys. Lett. B
662, 285 (2008). [arXiv:0712.2237]
211. G. Calcagni, G. Nardelli, Tachyon solutions in boundary and cubic string field theory. Phys.
Rev. D 78, 126010 (2008). [arXiv:0708.0366]
212. G. Calcagni, G. Nardelli, Kinks of open superstring field theory. Nucl. Phys. B 823, 234
(2009). [arXiv:0904.3744]
213. G. Calcagni, G. Nardelli, String theory as a diffusing system. JHEP 1002, 093 (2010).
[arXiv:0910.2160]
214. N. Moeller, B. Zwiebach, Dynamics with infinitely many time derivatives and rolling
tachyons. JHEP 0210, 034 (2002). [arXiv:hep-th/0207107]
References 691
215. I. Kishimoto, Y. Matsuo, E. Watanabe, A universal nonlinear relation among boundary states
in closed string field theory. Prog. Theor. Phys. 111, 433 (2004). [arXiv:hep-th/0312122]
216. A. Giveon, M. Porrati, E. Rabinovici, Target space duality in string theory. Phys. Rep. 244,
77 (1994). [arXiv:hep-th/9401139]
217. G. Calcagni, L. Modesto, Nonlocality in string theory. J. Phys. A 47, 355402 (2014).
[arXiv:1310.4957]
218. E. Cremmer, J. Scherk, Spontaneous compactification of space in an Einstein–Yang–Mills–
Higgs model. Nucl. Phys. B 108, 409 (1976)
219. E. Cremmer, J. Scherk, Spontaneous compactification of extra space dimensions. Nucl. Phys.
B 118, 61 (1977)
220. M.B. Green, J.H. Schwarz, L. Brink, N D 4 Yang–Mills and N D 8 supergravity as limits of
string theories. Nucl. Phys. B 198, 474 (1982)
221. B. de Wit, D.Z. Freedman, On SO.8/ extended supergravity. Nucl. Phys. B 130, 105 (1977)
222. G.W. Gibbons, S.W. Hawking, Classification of gravitational instanton symmetries. Commun.
Math. Phys. 66, 291 (1979)
223. D.N. Page, A physical picture of the K3 gravitational instanton. Phys. Lett. B 80, 55 (1978)
224. P. Candelas, G.T. Horowitz, A. Strominger, E. Witten, Vacuum configurations for super-
strings, Nucl. Phys. B 258, 46 (1985)
225. A. Strominger, E. Witten, New manifolds for superstring compactification. Commun. Math.
Phys. 101, 341 (1985)
226. A. Strominger, Superstrings with torsion. Nucl. Phys. B 274, 253 (1986)
227. E. Calabi, The space of Kähler metrics. Proc. Int. Congr. Math. 2, 206 (1954)
228. E. Calabi, On Kähler manifolds with vanishing canonical class, in Algebraic Geometric and
Topology: A Symposium in Honor of S. Lefschetz, ed. by R.H. Fox et al. (Princeton University
Press, Princeton, 1957)
229. S.-T. Yau, Calabi’s conjecture and some new results in algebraic geometry. Proc. Natl. Acad.
Sci. 74, 1798 (1977)
230. S.-T. Yau, On the Ricci curvature of a compact Kähler manifold and the complex Monge–
Ampère equation, I. Commun. Pure Appl. Math. 31, 339 (1978)
231. L. Dixon, J.A. Harvey, C. Vafa, E. Witten, Strings on orbifolds. Nucl. Phys. B 261, 678
(1985)
232. L. Dixon, J.A. Harvey, C. Vafa, E. Witten, Strings on orbifolds (II). Nucl. Phys. B 274, 285
(1986)
233. L. Dixon, D. Friedan, E. Martinec, S. Shenker, The conformal field theory of orbifolds. Nucl.
Phys. B 282, 13 (1987)
234. K.S. Narain, M.H. Sarmadi, C. Vafa, Asymmetric orbifolds. Nucl. Phys. B 288, 551 (1987)
235. A. Sagnotti, Open strings and their symmetry groups, in Nonperturbative Quantum Field
Theory, ed. by G. ’t Hooft, A. Jaffe, G. Mack, P.K. Mitter, R. Stora (Plenum, New York,
1988). [arXiv:hep-th/0208020]
236. M. Bianchi, A. Sagnotti, On the systematics of open-string theories. Phys. Lett. B 247, 517
(1990)
237. M. Bianchi, A. Sagnotti, Twist symmetry and open string Wilson lines. Nucl. Phys. B 361,
519 (1991)
238. M. Bianchi, G. Pradisi, A. Sagnotti, Toroidal compactification and symmetry breaking in open
string theories. Nucl. Phys. B 376, 365 (1992)
239. I.R. Klebanov, M.J. Strassler, Supergravity and a confining gauge theory: duality cascades
and SB-resolution of naked singularities. JHEP 0008, 052 (2000). [arXiv:hep-th/0007191]
240. P. Candelas, X.C. de la Ossa, Comments on conifolds. Nucl. Phys. B 342, 246 (1990)
241. S.B. Giddings, S. Kachru, J. Polchinski, Hierarchies from fluxes in string compactifications.
242. S. Hamidi, C. Vafa, Interactions on orbifolds. Nucl. Phys. B 279, 465 (1987)
243. E. Witten, New issues in manifolds of SU.3/ holonomy. Nucl. Phys. B 268, 79 (1986)
244. S. Gukov, C. Vafa, E. Witten, CFT’s from Calabi–Yau four-folds. Nucl. Phys. B 584, 69
(2000); Erratum-ibid. B 608, 477 (2001). [arXiv:hep-th/9906070]
245. P. Kaste, R. Minasian, A. Tomasiello, Supersymmetric M theory compactifications with fluxes

on seven-manifolds and G structures. JHEP 0307, 004 (2003). [arXiv:hep-th/0303127]
246. E. Witten, Dimensional reduction of superstring models. Phys. Lett. B 155, 151 (1985)
247. S. Ferrara, C. Kounnas, M. Porrati, General dimensional reduction of ten-dimensional
supergravity and superstring. Phys. Lett. B 181, 263 (1986)
248. M. Cvetič, J. Louis, B.A. Ovrut, A string calculation of the Kähler potentials for moduli of
ZN orbifolds. Phys. Lett. B 206, 227 (1988)
249. P. Candelas, X.C. de la Ossa, Moduli space of Calabi–Yau manifolds. Nucl. Phys. B 355, 455
(1991)
250. T.W. Grimm, J. Louis, The effective action of N D 1 Calabi–Yau orientifolds. Nucl. Phys. B
699, 387 (2004). [arXiv:hep-th/0403067]
252. P. Svrček, E. Witten, Axions in string theory. JHEP 0606, 051 (2006). [arXiv:hep-th/0605206]
253. X.-G. Wen, E. Witten, World-sheet instantons and the Peccei–Quinn symmetry. Phys. Lett.
B 166, 397 (1986)
254. M. Dine, N. Seiberg, X.-G. Wen, E. Witten, Nonperturbative effects on the string world sheet.
Nucl. Phys. B 278, 769 (1986)
255. R. Blumenhagen, D. Herschmann, E. Plauschinn, The challenge of realizing F-term axion
monodromy inflation in string theory. JHEP 1501, 007 (2015). [arXiv:1409.7075]
256. J.M. Maldacena, The large-N limit of superconformal field theories and supergrav-
ity. Adv. Theor. Math. Phys. 2, 231 (1998) [Int. J. Theor. Phys. 38, 1113 (1999)].
257. S.S. Gubser, I.R. Klebanov, A.M. Polyakov, Gauge theory correlators from non-critical string
theory. Phys. Lett. B 428, 105 (1998). [arXiv:hep-th/9802109]
258. E. Witten, Anti de Sitter space and holography. Adv. Theor. Math. Phys. 2, 253 (1998).
259. O. Aharony, S.S. Gubser, J.M. Maldacena, H. Ooguri, Y. Oz, Large N field theories, string
theory and gravity. Phys. Rep. 323, 183 (2000). [arXiv:hep-th/9905111]
260. A.V. Ramallo, Introduction to the AdS/CFT correspondence. Springer Proc. Phys. 161, 411
(2015). [arXiv:1310.4319]
261. V.E. Hubeny, The AdS/CFT correspondence. Class. Quantum Grav. 32, 124010 (2015).
[arXiv:1501.00007]
262. S. Kachru, E. Silverstein, 4D conformal field theories and strings on orbifolds. Phys. Rev.
Lett. 80, 4855 (1998). [arXiv:hep-th/9802183]
263. A.E. Lawrence, N. Nekrasov, C. Vafa, On conformal field theories in four dimensions. Nucl.
264. A. Kehagias, New type IIB vacua and their F-theory interpretation. Phys. Lett. B 435, 337
(1998). [arXiv:hep-th/9805131]
265. I.R. Klebanov, E. Witten, Superconformal field theory on threebranes at a Calabi–Yau
singularity. Nucl. Phys. B 536, 199 (1998). [arXiv:hep-th/9807080]
266. B.S. Acharya, J.M. Figueroa-O’Farrill, C.M. Hull, B.J. Spence, Branes at conical singularities
and holography. Adv. Theor. Math. Phys. 2, 1249 (1998). [arXiv:hep-th/9808014]
267. J. Polchinski, A. Strominger, New vacua for type II string theory. Phys. Lett. B 388, 736
(1996). [arXiv:hep-th/9510227]
268. O. DeWolfe, S.B. Giddings, Scales and hierarchies in warped compactifications and brane
worlds. Phys. Rev. D 67, 066008 (2003). [arXiv:hep-th/0208123]
269. K. Becker, M. Becker, M theory on eight manifolds. Nucl. Phys. B 477, 155 (1996).
270. K. Dasgupta, G. Rajesh, S. Sethi, M-theory, orientifolds and G-flux. JHEP 9908, 023 (1999).
271. P. Candelas, D.J. Raine, Compactification and supersymmetry in d D 11 supergravity. Nucl.
Phys. B 248, 415 (1984)
References 693
272. T.R. Taylor, C. Vafa, RR flux on Calabi–Yau and partial supersymmetry breaking. Phys. Lett.
B 474, 130 (2000). [arXiv:hep-th/9912152]
273. M. Haack, J. Louis, M theory compactified on Calabi–Yau fourfolds with background flux.
Phys. Lett. B 507, 296 (2001). [arXiv:hep-th/0103068]
274. B. de Wit, H. Nicolai, A new SO(7) invariant solution of d D 11 supergravity. Phys. Lett. B
148, 60 (1984)
275. L. Randall, R. Sundrum, An alternative to compactification. Phys. Rev. Lett. 83, 4690 (1999).
276. B. De Wit, D.J. Smit, Residual supersymmetry of compactified d D 10 supergravity. Nucl.
Phys. B 283, 165 (1987)
277. J.M. Maldacena, C. Nuñez, Supergravity description of field theories on curved manifolds
and a no go theorem. Int. J. Mod. Phys. A 16, 822 (2001). [arXiv:hep-th/0007018]
278. L. Randall, R. Sundrum, Large mass hierarchy from a small extra dimension. Phys. Rev. Lett.
83, 3370 (1999). [arXiv:hep-ph/9905221]
279. H.L. Verlinde, Holography and compactification. Nucl. Phys. B 580, 264 (2000).
280. C.S. Chan, P.L. Paul, H.L. Verlinde, A note on warped string compactification. Nucl. Phys.
B 581, 156 (2000) [arXiv:hep-th/0003236]
281. M. Graña, J. Polchinski, Gauge-gravity duals with a holomorphic dilaton. Phys. Rev. D 65,
126005 (2002). [arXiv:hep-th/0106014]
282. S. Kachru, M.B. Schulz, S. Trivedi, Moduli stabilization from fluxes in a simple IIB
orientifold. JHEP 0310, 007 (2003). [arXiv:hep-th/0201028]
283. A.R. Frey, J. Polchinski, N D 3 warped compactifications. Phys. Rev. D 65, 126009 (2002).
284. S. Gurrieri, J. Louis, A. Micu, D. Waldram, Mirror symmetry in generalized Calabi–Yau
compactifications. Nucl. Phys. B 654, 61 (2003). [arXiv:hep-th/0211102]
285. S. Kachru, M.B. Schulz, P.K. Tripathy, S.P. Trivedi, New supersymmetric string compactifi-
cations. JHEP 0303, 061 (2003). [arXiv:hep-th/0211182]
286. P.K. Tripathy, S.P. Trivedi, Compactification with flux on K3 and tori. JHEP 0303, 028 (2003).
287. K. Becker, M. Becker, K. Dasgupta, P.S. Green, Compactifications of heterotic theory on
non-Kähler complex manifolds, I. JHEP 0304, 007 (2003). [arXiv:hep-th/0301161]
288. R. Blumenhagen, D. Lüst, T.R. Taylor, Moduli stabilization in chiral type IIB orientifold
models with fluxes. Nucl. Phys. B 663, 319 (2003). [arXiv:hep-th/0303016]
289. J.F.G. Cascales, A.M. Uranga, Chiral 4d string vacua with D-branes and NSNS and RR fluxes.
JHEP 0305, 011 (2003). [arXiv:hep-th/0303024]
290. K. Becker, M. Becker, P.S. Green, K. Dasgupta, E. Sharpe, Compactifications of het-
erotic strings on nonKahler complex manifolds II. Nucl. Phys. B 678, 19 (2004).
291. J.-P. Derendinger, C. Kounnas, P.M. Petropoulos, F. Zwirner, Superpotentials in IIA compact-
ifications with general fluxes. Nucl. Phys. B 715, 211 (2005). [arXiv:hep-th/0411276]
292. S. Kachru, A.-K. Kashani-Poor, Moduli potentials in type-IIA compactifications with RR and
NS flux. JHEP 0503, 066 (2005). [arXiv:hep-th/0411279]
293. G. Villadoro, F. Zwirner, N D 1 effective potential from dual type-IIA D6/O6 orientifolds
with general fluxes. JHEP 0506, 047 (2005). [arXiv:hep-th/0503169]
294. O. DeWolfe, A. Giryavets, S. Kachru, W. Taylor, Type IIA moduli stabilization. JHEP 0507,
066 (2005). [arXiv:hep-th/0505160]
295. P.G. Cámara, A. Font, L.E. Ibáñez, Fluxes, moduli fixing and MSSM-like vacua in a simple
IIA orientifold. JHEP 0509, 013 (2005). [arXiv:hep-th/0506066]
296. B.S. Acharya, F. Benini, R. Valandro, Fixing moduli in exact type IIA flux vacua. JHEP 0702,
018 (2007). [arXiv:hep-th/0607223]
297. R. Donagi, B.A. Ovrut, T. Pantev, D. Waldram, Standard models from heterotic M theory.
Adv. Theor. Math. Phys. 5, 93 (2001). [arXiv:hep-th/9912208]
298. R. Blumenhagen, L. Görlich, B. Körs, D. Lüst, Noncommutative compactifications

of type I strings on tori with magnetic background flux. JHEP 0010, 006 (2000).
299. R. Blumenhagen, B. Körs, D. Lüst, Type I strings with F- and B-flux. JHEP 0102, 030 (2001).
300. V. Braun, Y.-H. He, B.A. Ovrut, T. Pantev, A heterotic standard model. Phys. Lett. B 618,
252 (2005). [arXiv:hep-th/0501070]
301. V. Braun, Y.-H. He, B.A. Ovrut, T. Pantev, A standard model from the E8 E8 heterotic
superstring. JHEP 0506, 039 (2005). [arXiv:hep-th/0502155]
302. V. Braun, Y.-H. He, B.A. Ovrut, T. Pantev, Vector bundle extensions, sheaf coho-
mology, and the heterotic standard model. Adv. Theor. Math. Phys. 10, 525 (2006).
303. H.P. Nilles, S. Ramos-Sánchez, M. Ratz, P.K.S. Vaudrevange, From strings to the MSSM.
Eur. Phys. J. C 59, 249 (2009). [arXiv:0806.3905]
304. G. Aldazabal, L.E. Ibáñez, F. Quevedo, A.M. Uranga, D-branes at singularities: a bottom-up
approach to the string embedding of the standard model. JHEP 0008, 002 (2000).
305. G. Aldazabal, S. Franco, L.E. Ibáñez, R. Rabadán, A.M. Uranga, D D 4 chi-
ral string compactifications from intersecting branes. J. Math. Phys. 42, 3103 (2001).
306. G. Aldazabal, S. Franco, L.E. Ibáñez, R. Rabadán, A.M. Uranga, Intersecting brane worlds.
JHEP 0102, 047 (2001). [arXiv:hep-ph/0011132]
307. L.E. Ibáñez, F. Marchesano, R. Rabadán, Getting just the standard model at intersecting
branes. JHEP 0111, 002 (2001). [arXiv:hep-th/0105155]
308. R. Blumenhagen, B. Körs, D. Lüst, T. Ott, The Standard Model from stable intersecting brane
world orbifolds. Nucl. Phys. B 616, 3 (2001). [arXiv:hep-th/0107138]
309. D. Bailin, G.V. Kraniotis, A. Love, Standard-like models from intersecting D4-branes. Phys.
310. D. Cremades, L.E. Ibáñez, F. Marchesano, Intersecting brane models of particle physics and
the Higgs mechanism. JHEP 0207, 022 (2002). [arXiv:hep-th/0203160]
311. D. Cremades, L.E. Ibáñez, F. Marchesano, Standard model at intersecting D5-branes:
lowering the string scale. Nucl. Phys. B 643, 93 (2002). [arXiv:hep-th/0205074]
312. C. Kokorelis, New Standard Model vacua from intersecting branes. JHEP 0209, 029 (2002).
313. C. Kokorelis, Exact Standard Model compactifications from intersecting branes. JHEP 0208,
036 (2002). [arXiv:hep-th/0206108]
314. C. Kokorelis, Exact Standard Model structures from intersecting D5-branes. Nucl. Phys. B
677, 115 (2004). [arXiv:hep-th/0207234]
315. D. Bailin, G.V. Kraniotis, A. Love, New standard-like models from intersecting D4-branes.
316. D. Bailin, G.V. Kraniotis, A. Love, Standard-like models from intersecting D5-branes. Phys.
317. D. Bailin, G.V. Kraniotis, A. Love, Intersecting D5-brane models with massive vector-like
leptons. JHEP 0302, 052 (2003). [arXiv:hep-th/0212112]
318. C. Kokorelis, Standard model compactifications of IIA Z3 Z3 orientifolds from intersecting
D6-branes. Nucl. Phys. B 732, 341 (2006). [arXiv:hep-th/0412035]
319. D. Berenstein, V. Jejjala, R.G. Leigh, Standard model on a D-brane. Phys. Rev. Lett. 88,
071602 (2002). [arXiv:hep-ph/0105042]
320. M. Cvetič, G. Shiu, A.M. Uranga, Three-family supersymmetric standardlike models from
intersecting brane worlds. Phys. Rev. Lett. 87, 201801 (2001). [arXiv:hep-th/0107143]
321. M. Cvetič, G. Shiu, A.M. Uranga, Chiral four-dimensional N D 1 supersymmet-
ric type IIA orientifolds from intersecting D6-branes. Nucl. Phys. B 615, 3 (2001).
References 695
322. M. Cvetič, P. Langacker, G. Shiu, Phenomenology of a three-family standardlike string model.

323. M. Cvetič, P. Langacker, G. Shiu, A three-family standard-like orientifold model: Yukawa
couplings and hierarchy. Nucl. Phys. B 642, 139 (2002). [arXiv:hep-th/0206115]
324. R. Blumenhagen, L. Görlich, T. Ott, Supersymmetric intersecting branes on the type 2A
T 6 =Z 4 orientifold. JHEP 0301, 021 (2003). [arXiv:hep-th/0211059]
325. M. Cvetič, I. Papadimitriou, G. Shiu, Supersymmetric three family SU.5/ grand unified
models from type IIA orientifolds with intersecting D6-branes. Nucl. Phys. B 659, 193
(2003); Erratum-ibid. B 696, 298 (2004). [arXiv:hep-th/0212177]
326. D. Cremades, L.E. Ibáñez, F. Marchesano, Yukawa couplings in intersecting D-brane models.
JHEP 0307, 038 (2003). [arXiv:hep-th/0302105]
327. G. Honecker, Chiral supersymmetric models on an orientifold of Z 4 Z 2 with intersecting
D6-branes. Nucl. Phys. B 666, 175 (2003). [arXiv:hep-th/0303015]
328. M. Cvetič, I. Papadimitriou, Conformal field theory couplings for intersecting D-branes on
orientifolds. Phys. Rev. D 68, 046001 (2003); Erratum-ibid. D 70, 029903(E) (2004).
329. M. Cvetič, I. Papadimitriou, More supersymmetric standardlike models from intersecting D6-
branes on type IIA orientifolds. Phys. Rev. D 67, 126006 (2003). [arXiv:hep-th/0303197]
330. R. Blumenhagen, D. Lüst, S. Stieberger, Gauge unification in supersymmetric intersecting
brane worlds. JHEP 0307, 036 (2003). [arXiv:hep-th/0305146]
331. M. Cvetič, T. Li, T. Liu, Supersymmetric Pati–Salam models from intersecting D6-branes: a
road to the Standard Model. Nucl. Phys. B 698, 163 (2004). [arXiv:hep-th/0403061]
332. T.P.T. Dijkstra, L.R. Huiszoon, A.N. Schellekens, Chiral supersymmetric standard
model spectra from orientifolds of Gepner models. Phys. Lett. B 609, 408 (2005).
333. G. Honecker, T. Ott, Getting just the supersymmetric standard model at intersecting branes
on the Z 6 orientifold. Phys. Rev. D 70, 126010 (2004); Erratum-ibid. D 71, 069902 (2005).
334. D. Lüst, P. Mayr, R. Richter, S. Stieberger, Scattering of gauge, matter, and moduli fields from
intersecting branes. Nucl. Phys. B 696, 205 (2004). [arXiv:hep-th/0404134]
335. M. Cvetič, P. Langacker, T. Li, T. Liu, D6-brane splitting on type IIA orientifolds. Nucl.
336. M. Berkooz, M.R. Douglas, R.G. Leigh, Branes intersecting at angles. Nucl. Phys. B 480,
265 (1996). [arXiv:hep-th/9606139]
338. P.G. Cámara, L.E. Ibáñez, A.M. Uranga, Flux-induced SUSY-breaking soft terms on D7-D3
brane systems. Nucl. Phys. B 708, 268 (2005). [arXiv:hep-th/0408036]
339. F. Marchesano, G. Shiu, Minimal supersymmetric standard model string vacua from flux
compactifications. Phys. Rev. D 71, 011701(R) (2005). [arXiv:hep-th/0408059]
340. L.E. Ibáñez, Fluxed minimal supersymmetric standard model. Phys. Rev. D 71, 055005
(2005). [arXiv:hep-ph/0408064]
341. M. Cvetič, T. Liu, Supersymmetric standard models, flux compactification and moduli
stabilization. Phys. Lett. B 610, 122 (2005). [arXiv:hep-th/0409032]
342. F. Marchesano, G. Shiu, Building MSSM flux vacua. JHEP 0411, 041 (2004).
343. M. Cvetič, T. Li, T. Liu, Standard-like models as type IIB flux vacua. Phys. Rev. D 71, 106008
(2005). [arXiv:hep-th/0501041]
344. R. Blumenhagen, M. Cvetič, P. Langacker, G. Shiu, Toward realistic intersecting D-brane
models. Ann. Rev. Nucl. Part. Sci. 55, 71 (2005). [arXiv:hep-th/0502005]
345. H. Verlinde, M. Wijnholt, Building the standard model on a D3-brane. JHEP 0701, 106
(2007). [arXiv:hep-th/0508089]
346. F. Marchesano, Progress in D-brane model building. Fortsch. Phys. 55, 491 (2007).
347. J.P. Conlon, A. Maharana, F. Quevedo, Towards realistic string vacua. JHEP 0905, 109 (2009).
[arXiv:0810.5660]
348. M.J. Dolan, S. Krippendorf, F. Quevedo, Towards a systematic construction of realistic D-
brane models on a del Pezzo singularity. JHEP 1110, 024 (2011). [arXiv:1106.6039]
349. M. Cicoli, C. Mayrhofer, R. Valandro, Moduli stabilisation for chiral global models. JHEP
1202, 062 (2012). [arXiv:1110.3333]
350. M. Cicoli, S. Krippendorf, C. Mayrhofer, F. Quevedo, R. Valandro, D-branes at del
Pezzo singularities: global embedding and moduli stabilisation. JHEP 1209, 019 (2012).
[arXiv:1206.5237]
351. M. Cicoli, D. Klevers, S. Krippendorf, C. Mayrhofer, F. Quevedo, R. Valandro, Explicit
de Sitter flux vacua for global string models with chiral matter, JHEP 1405, 001 (2014).
[arXiv:1312.0014]
352. R. Blumenhagen, A. Font, M. Fuchs, D. Herschmann, E. Plauschinn, Y. Sekiguchi, F. Wolf, A
flux-scaling scenario for high-scale moduli stabilization in string theory. Nucl. Phys. B 897,
500 (2015). [arXiv:1503.07634]
353. M. Berg, M. Haack, B. Körs, String loop corrections to Kähler potentials in orientifolds. JHEP
0511, 030 (2005). [arXiv:hep-th/0508043]
354. M. Berg, M. Haack, B. Körs, Stabilization of the compactification volume by quantum
corrections. Phys. Rev. Lett. 96, 021601 (2006). [arXiv:hep-th/0508171]
355. S. Ferrara, L. Girardello, H.P. Nilles, Breakdown of local supersymmetry through gauge
fermion condensates. Phys. Lett. B 125, 457 (1983)
356. I. Affleck, M. Dine, N. Seiberg, Supersymmetry breaking by instantons. Phys. Rev. Lett. 51,
1026 (1983)
357. I. Affleck, M. Dine, N. Seiberg, Dynamical supersymmetry breaking in supersymmetric QCD.
Nucl. Phys. B 241, 493 (1984)
358. I. Affleck, M. Dine, N. Seiberg, Dynamical supersymmetry breaking in four dimensions and
its phenomenological implications. Nucl. Phys. B 256, 557 (1985)
359. M.A. Shifman, A.I. Vainshtein, On gluino condensation in supersymmetric gauge theories
with SU(N) and O(N) groups. Sov. Phys. JETP 66, 1100 (1987) [Nucl. Phys. B 296, 445
(1988)]
360. M. Dine, N. Seiberg, X.-G. Wen, E. Witten, Nonperturbative effects on the string world sheet
(II). Nucl. Phys. B 289, 319 (1987)
361. K. Becker, M. Becker, A. Strominger, Fivebranes, membranes and non-perturbative string
theory. Nucl. Phys. B 456, 130 (1995). [arXiv:hep-th/9507158]
362. E. Witten, Non-perturbative superpotentials in string theory. Nucl. Phys. B 474, 343 (1996).
363. J.A. Harvey, G.W. Moore, Superpotentials and membrane instantons. arXiv:hep-th/9907026
364. S.H. Katz, C. Vafa, Geometric engineering of N D 1 quantum field theories. Nucl. Phys. B
497, 196 (1997). [arXiv:hep-th/9611090]
365. J.-P. Derendinger, L.E. Ibáñez, H.P. Nilles, On the low energy d D 4, N D 1 supergravity
theory extracted from the d D 10, N D 1 superstring. Phys. Lett. B 155, 65 (1985)
366. M. Dine, R. Rohm, N. Seiberg, E. Witten, Gluino condensation in superstring models. Phys.
Lett. B 156, 55 (1985)
367. J.-P. Derendinger, L.E. Ibáñez, H.P. Nilles, On the low-energy limit of superstring theories.
Nucl. Phys. B 267, 365 (1986)
368. N.V. Krasnikov, On supersymmetry breaking in superstring theories. Phys. Lett. B 193, 37
(1987)
369. J.A. Casas, Z. Lalak, C. Muñoz, G.G. Ross, Hierarchical supersymmetry breaking and
dynamical determination of compactification parameters by non-perturbative effects. Nucl.
Phys. B 347, 243 (1990)
370. A. Font, L.E. Ibáñez, D. Lüst, F. Quevedo, Supersymmetry breaking from duality invariant
gaugino condensation. Phys. Lett. B 245, 401 (1990)
References 697
371. S. Ferrara, N. Magnoli, T.R. Taylor, G. Veneziano, Duality and supersymmetry breaking in
string theory. Phys. Lett. B 245, 409 (1990)
372. T.R. Taylor, Dilaton, gaugino condensation and supersymmetry breaking. Phys. Lett. B 252,
59 (1990)
373. B. de Carlos, J.A. Casas, C. Muñoz, Supersymmetry breaking and determination of the
unification gauge coupling constant in string theories. Nucl. Phys. B 399, 623 (1993).
374. S. Kachru, R. Kallosh, A.D. Linde, S.P. Trivedi, de Sitter vacua in string theory. Phys. Rev.
D 68, 046005 (2003). [arXiv:hep-th/0301240]
375. M. Endo, M. Yamaguchi, K. Yoshioka, Bottom-up approach to moduli dynamics in heavy
gravitino scenario: superpotential, soft terms, and sparticle mass spectrum. Phys. Rev. D 72,
015004 (2005). [arXiv:hep-ph/0504036]
376. G. Curio, A. Krause, D. Lüst, Moduli stabilization in the heterotic/IIB discretuum. Fortsch.
Phys. 54, 225 (2006). [arXiv:hep-th/0502168]
377. F. Denef, M.R. Douglas, B. Florea, A. Grassi, S. Kachru, Fixing all moduli in a simple F-
theory compactification. Adv. Theor. Math. Phys. 9, 861 (2005). [arXiv:hep-th/0503124]
378. D. Lüst, S. Reffert, W. Schulgin, S. Stieberger, Moduli stabilization in type IIB orientifolds
(I). Nucl. Phys. B 766, 68 (2007). [arXiv:hep-th/0506090]
379. E. Dudas, S.K. Vempati, Large D-terms, hierarchical soft spectra and moduli stabilisation.
380. G. Aldazabal, P.G. Cámara, A. Font, L.E. Ibáñez, More dual fluxes and moduli fixing. JHEP
0605, 070 (2006). [arXiv:hep-th/0602089]
381. D. Lüst, S. Reffert, E. Scheidegger, W. Schulgin, S. Stieberger, Moduli stabilization in type
IIB orientifolds (II). Nucl. Phys. B 766, 178 (2007). [arXiv:hep-th/0609013]
382. J.P. Conlon, S.S. Abdussalam, F. Quevedo, K. Suruliz, Soft SUSY breaking terms for chiral
matter in IIB string compactifications. JHEP 0701, 032 (2007). [arXiv:hep-th/0610129]
383. R. Blumenhagen, S. Moster, E. Plauschinn, Moduli stabilisation versus chirality for MSSM
like type IIB orientifolds. JHEP 0801, 058 (2008). [arXiv:arXiv:0711.3389]
384. A.R. Frey, M. Lippert, B. Williams, Fall of stringy de Sitter spacetime. Phys. Rev. D 68,
046008 (2003). [arXiv:hep-th/0305018]
385. C. Escoda, M. Gómez-Reino, F. Quevedo, Saltatory de Sitter string vacua. JHEP 0311, 065
(2003). [arXiv:hep-th/0307160]
386. C.P. Burgess, R. Kallosh, F. Quevedo, de Sitter string vacua from supersymmetric D-terms.
JHEP 0310, 056 (2003). [arXiv:hep-th/0309187]
387. J.F.G. Cascales, M.P. García del Moral, F. Quevedo, A.M. Uranga, Realistic D-brane models
on warped throats: fluxes, hierarchies and moduli stabilization. JHEP 0402, 031 (2004).
388. R. Brustein, S.P. de Alwis, Moduli potentials in string compactifications with fluxes: mapping
the discretuum. Phys. Rev. D 69, 126006 (2004). [arXiv:hep-th/0402088]
389. F. Denef, M.R. Douglas, B. Florea, Building a better racetrack. JHEP 0406, 034 (2004).
390. V. Balasubramanian, P. Berglund, Stringy corrections to Kähler potentials, SUSY breaking,
and the cosmological constant problem. JHEP 0411, 085 (2004). [arXiv:hep-th/0408054]
391. R. Kallosh, A.D. Linde, Landscape, the scale of SUSY breaking, and inflation. JHEP 0412,
004 (2004). [arXiv:hep-th/0411011]
392. K. Choi, A. Falkowski, H.P. Nilles, M. Olechowski, S. Pokorski, Stability of flux
compactifications and the pattern of supersymmetry breaking. JHEP 0411, 076 (2004).
393. V. Balasubramanian, P. Berglund, J.P. Conlon, F. Quevedo, Systematics of moduli stabilisa-
tion in Calabi–Yau flux compactifications. JHEP 0503, 007 (2005). [arXiv:hep-th/0502058]
394. K. Choi, A. Falkowski, H.P. Nilles, M. Olechowski, Soft supersymmetry breaking in KKLT
flux compactification. Nucl. Phys. B 718, 113 (2005). [arXiv:hep-th/0503216]
395. J.P. Conlon, F. Quevedo, K. Suruliz, Large-volume flux compactifications: moduli spectrum
and D3/D7 soft supersymmetry breaking. JHEP 0508, 007 (2005). [arXiv:hep-th/0505076]
396. S.B. Giddings, A. Maharana, Dynamics of warped compactifications and the shape of the
warped landscape. Phys. Rev. D 73, 126003 (2006). [arXiv:hep-th/0507158]
397. G. Villadoro, F. Zwirner, de Sitter vacua via consistent D terms. Phys. Rev. Lett. 95, 231602
(2005). [arXiv:hep-th/0508167]
398. J.J. Blanco-Pillado, R. Kallosh, A.D. Linde, Supersymmetry and stability of flux vacua. JHEP
0605, 053 (2006). [arXiv:hep-th/0511042]
399. A. Achúcarro, B. de Carlos, J.A. Casas, L. Doplicher, de Sitter vacua from uplift-
ing D-terms in effective supergravities from realistic strings. JHEP 0606, 014 (2006).
400. O. Lebedev, H.P. Nilles, M. Ratz, de Sitter vacua from matter superpotentials. Phys. Lett. B
636, 126 (2006). [arXiv:hep-th/0603047]
401. P. Ouyang, Holomorphic D7 branes and flavored N D 1 gauge theories. Nucl. Phys. B 699,
207 (2004). [arXiv:hep-th/0311084]
402. S. Kuperstein, Meson spectroscopy from holomorphic probes on the warped deformed
conifold. JHEP 0503, 014 (2005). [arXiv:hep-th/0411097]
403. G.L. Cardoso, G. Curio, G. Dall’Agata, D. Lüst, Heterotic string theory on non-
Kähler manifolds with H-flux and gaugino condensate. Fortsch. Phys. 52, 483 (2004).
404. B. de Carlos, S. Gurrieri, A. Lukas, A. Micu, Moduli stabilisation in heterotic string
compactifications. JHEP 0603, 005 (2006). [arXiv:hep-th/0507173]
405. J.P. Conlon, F. Quevedo, Gaugino and scalar masses in the landscape. JHEP 0606, 029 (2006).
406. J.P. Conlon, Moduli stabilisation and applications in IIB string theory. Fortsch. Phys. 55, 287
(2007). [arXiv:hep-th/0611039]
407. A. Westphal, de Sitter string vacua from Kähler uplifting. JHEP 0703, 102 (2007).
408. M. Berg, M. Haack, E. Pajer, Jumping through loops: on soft terms from large volume
compactifications. JHEP 0709, 031 (2007). [arXiv:0704.0737]
409. J.P. Conlon, C.H. Kom, K. Suruliz, B.C. Allanach, F. Quevedo, Sparticle spectra and LHC sig-
natures for large volume string compactifications. JHEP 0708, 061 (2007). [arXiv:0704.3403]
410. M. Cicoli, J.P. Conlon, F. Quevedo, General analysis of LARGE volume scenarios with string
loop moduli stabilisation. JHEP 0810, 105 (2008). [arXiv:0805.1029]
411. D. Ciupke, J. Louis, A. Westphal, Higher-derivative supergravity and moduli stabilization.
JHEP 1510, 094 (2015). [arXiv:1505.03092]
412. K. Becker, M. Becker, M. Haack, J. Louis, Supersymmetry breaking and ˛ 0 corrections to
flux induced potentials. JHEP 0206, 060 (2002). [arXiv:hep-th/0204254]
413. P. Candelas, A. Font, S.H. Katz, D.R. Morrison, Mirror symmetry for two-parameter
models—II. Nucl. Phys. B 429, 626 (1994). [arXiv:hep-th/9403187]
414. N. Arkani-Hamed, S. Dimopoulos, G.R. Dvali, The hierarchy problem and new dimensions
at a millimeter. Phys. Lett. B 429, 263 (1998). [arXiv:hep-ph/9803315]
415. N. Arkani-Hamed, S. Dimopoulos, G.R. Dvali, Phenomenology, astrophysics and cosmology
of theories with submillimeter dimensions and TeV scale quantum gravity. Phys. Rev. D 59,
086004 (1999). [arXiv:hep-ph/9807344]
416. I. Antoniadis, N. Arkani-Hamed, S. Dimopoulos, G.R. Dvali, New dimensions at a millimeter
to a Fermi and superstrings at a TeV. Phys. Lett. B 436, 257 (1998). [arXiv:hep-ph/9804398]
417. P.G.O. Freund, M.A. Rubin, Dynamics of dimensional reduction. Phys. Lett. B 97, 233
(1980)
J. Theor. Phys. 28, 1067 (1989)
References 699
421. T. Damour, S. de Buyl, M. Henneaux, C. Schomblond, Einstein billiards and overextensions

of finite-dimensional simple Lie algebras. JHEP 0208, 030 (2002). [arXiv:hep-th/0206125]
422. P. Hořava, E. Witten, Heterotic and type I string dynamics from eleven dimensions. Nucl.
423. E. Bergshoeff, E. Sezgin, P.K. Townsend, Supermembranes and eleven-dimensional super-
gravity. Phys. Lett. B 189, 75 (1987)
424. E. Bergshoeff, E. Sezgin, P.K. Townsend, Properties of the eleven-dimensional supermem-
brane theory. Ann. Phys. (N.Y.) 185, 330 (1988)
425. M.J. Duff, P.S. Howe, T. Inami, K.S. Stelle, Superstrings in D D 10 from supermembranes
in D D 11. Phys. Lett. B 191, 70 (1987)
426. T. Banks, W. Fischler, S.H. Shenker, L. Susskind, M theory as a matrix model: a conjecture.
427. K.S. Narain, New heterotic string theories in uncompactified dimensions < 10. Phys. Lett. B
169, 41 (1986)
428. K.S. Narain, M.H. Sarmadi, E. Witten, A note on toroidal compactification of heterotic string
theory. Nucl. Phys. B 279, 369 (1987)
429. P.H. Ginsparg, On toroidal compactification of heterotic superstrings. Phys. Rev. D 35, 648
(1987)
430. M. Dine, P. Huet, N. Seiberg, Large and small radius in string theory. Nucl. Phys. B 322, 301
(1989)
431. A. Font, L.E. Ibáñez, D. Lüst, F. Quevedo, Strong-weak coupling duality and non-perturbative
effects in string theory. Phys. Lett. B 249, 35 (1990)
432. J.H. Schwarz, A. Sen, Duality symmetric actions. Nucl. Phys. B 411, 35 (1994).
433. J.H. Schwarz, A. Sen, Duality symmetries of 4D heterotic strings. Phys. Lett. B 312, 105
(1993). [arXiv:hep-th/9305185]
434. A. Sen, Strong-weak coupling duality in four-dimensional string theory. Int. J. Mod. Phys. A
9, 3707 (1994). [arXiv:hep-th/9402002]
435. M.J. Duff, Strong/weak coupling duality from the dual string. Nucl. Phys. B 442, 47 (1995).
436. C.M. Hull, P.K. Townsend, Unity of superstring dualities. Nucl. Phys. B 438, 109 (1995).
437. J. Polchinski, E. Witten, Evidence for heterotic – type I string duality. Nucl. Phys. B 460,
525 (1996). [arXiv:hep-th/9510169]
438. E. Witten, String theory dynamics in various dimensions. Nucl. Phys. B 443, 85 (1995).
439. P.K. Townsend, The eleven-dimensional supermembrane revisited. Phys. Lett. B 350, 184
(1995). [arXiv:hep-th/9501068]
440. P. Hořava, E. Witten, Eleven-dimensional supergravity on a manifold with boundary. Nucl.
441. E. Witten, Strong coupling expansion of Calabi–Yau compactification. Nucl. Phys. B 471,
135 (1996). [arXiv:hep-th/9602070]
442. T. Banks, M. Dine, Couplings and scales in strongly coupled heterotic string theory. Nucl.
443. A. Strominger, Heterotic solitons. Nucl. Phys. B 343, 167 (1990); Erratum-ibid. B 353, 565
(1991)
444. N. Seiberg, Observations on the moduli space of superconformal field theories. Nucl. Phys.
B 303, 286 (1988)
445. J.A. Harvey, A. Strominger, The heterotic string is a soliton. Nucl. Phys. B 449, 535 (1995);
Erratum-ibid. B 458, 456 (1996). [arXiv:hep-th/9504047]
446. C. Vafa, E. Witten, A one-loop test of string duality. Nucl. Phys. B 447, 261 (1995).
447. M.J. Duff, J.T. Liu, R. Minasian, Eleven-dimensional origin of string-string duality: a one-
loop test. Nucl. Phys. B 452, 261 (1995). [arXiv:hep-th/9506126]
448. E. Witten, Some comments on string dynamics, in Future Perspectives in String Theory:
Strings’95, ed. by I. Bars, P. Bouwknegt, J. Minahan, D. Nemeschansky, K. Pilch, H. Saleur,
N.P. Warner (World Scientific, Singapore, 1996). [arXiv:hep-th/9507121]
449. M.J. Duff, R. Minasian, E. Witten, Evidence for heterotic/heterotic duality. Nucl. Phys. B
465, 413 (1996). [arXiv:hep-th/9601036]
450. P.K. Townsend, String-membrane duality in seven dimensions. Phys. Lett. B 354, 247 (1995).
451. J.H. Schwarz, An SL.2; Z/ multiplet of type IIB superstrings. Phys. Lett. B 360, 13 (1995);
Erratum-ibid. B 364, 252 (1995). [arXiv:hep-th/9508143]
452. P.S. Aspinwall, Some relationships between dualities in string theory. Nucl. Phys. Proc. Suppl.
46, 30 (1996). [arXiv:hep-th/9508154]
453. J.H. Schwarz, The power of M theory. Phys. Lett. B 367, 97 (1996). [arXiv:hep-th/9510086]
454. M. Bodner, A.C. Cadavid, S. Ferrara, .2; 2/ vacuum configurations for type IIA superstrings:
N D 2 supergravity Lagrangians and algebraic geometry. Class. Quantum Grav. 8, 789 (1991)
455. A.C. Cadavid, A. Ceresole, R. D’Auria, S. Ferrara, Eleven-dimensional supergravity com-
pactified on Calabi–Yau threefolds. Phys. Lett. B 357, 76 (1995). [arXiv:hep-th/9506144]
456. G. Papadopoulos, P.K. Townsend, Compactification of D D 11 supergravity on spaces of
exceptional holonomy. Phys. Lett. B 357, 300 (1995). [arXiv:hep-th/9506150]
457. I. Antoniadis, S. Ferrara, T.R. Taylor, N D 2 heterotic superstring and its dual theory in five
dimensions. Nucl. Phys. B 460, 489 (1996). [arXiv:hep-th/9511108]
458. S. Ferrara, R.R. Khuri, R. Minasian, M-theory on a Calabi–Yau manifold. Phys. Lett. B 375,
81 (1996). [arXiv:hep-th/9602102]
459. G. Curio, A. Krause, G-fluxes and nonperturbative stabilization of heterotic M-theory. Nucl.
460. B.S. Acharya, A moduli fixing mechanism in M theory. arXiv:hep-th/0212294
461. E.I. Buchbinder, B.A. Ovrut, Vacuum stability in heterotic M theory. Phys. Rev. D 69, 086010
(2004). [arXiv:hep-th/0310112]
462. M. Becker, G. Curio, A. Krause, de Sitter vacua from heterotic M-theory. Nucl. Phys. B 693,
223 (2004). [arXiv:hep-th/0403027]
463. E.I. Buchbinder, Raising anti-de Sitter vacua to de Sitter vacua in heterotic M theory. Phys.
Rev. D 70, 066008 (2004). [arXiv:hep-th/0406101]
464. B.S. Acharya, F. Denef, R. Valandro, Statistics of M theory vacua. JHEP 0506, 056 (2005).
465. P.S. Aspinwall, R. Kallosh, Fixing all moduli for M-theory on K3K3. JHEP 0510, 001
(2005). [arXiv:hep-th/0506014]
466. L. Anguelova, K. Zoubos, Flux superpotential in heterotic M-theory. Phys. Rev. D 74, 026005
(2006). [arXiv:hep-th/0602039]
467. V. Braun, B.A. Ovrut, Stabilizing moduli with a positive cosmological constant in heterotic
M-theory. JHEP 0607, 035 (2006). [arXiv:hep-th/0603088]
468. G. Curio, A. Krause, S-Track stabilization of heterotic de Sitter vacua. Phys. Rev. D 75,
126003 (2007). [arXiv:hep-th/0606243]
469. A. Krause, Supersymmetry breaking with zero vacuum energy in M-theory flux compactifi-
cations. Phys. Rev. Lett. 98, 241601 (2007). [arXiv:hep-th/0701009]
470. F. Paccetti Correia, M.G. Schmidt, Moduli stabilization in heterotic M-theory. Nucl. Phys. B
797, 243 (2008). [arXiv:0708.3805]
Chapter 13
String Cosmology
Así que, casi me es forzoso seguir por su camino, y por él tengo

de ir a pesar de todo el mundo, y será en balde cansaros en
persuadirme a que no quiera yo lo que los cielos quieren, la
fortuna ordena y la razón pide, y, sobre todo, mi voluntad desea.
— Miguel de Cervantes, Segunda Parte del Ingenioso Cavallero
Don Quixote de la Mancha, II – 6
Therefore, it is almost forced upon me to follow its path, and

because of it I must go despite everybody, and in vain will you
tire yourselves out to persuade me not to want what heavens
want, fate orders and reason demands and, most of all, what my
will desires.
Contents
13.1 String Landscape. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 703
13.1.1 de Sitter Vacua. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 705
13.1.2 Cosmological Constant. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 709
13.1.3 Open Problems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 714
13.2 Inflation in the Landscape. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 715
13.2.1 Single-Field Inflation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 716
13.2.2 Large-Field Models and the Weak Gravity
Conjecture. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 718
13.2.3 Multi-field Inflation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 719
13.2.4 Moduli Problem and -Problem. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 720
13.3 Size Moduli Inflation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 720
13.3.1 Large-Volume Inflation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 721
13.3.2 Volume-Modulus Inflation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 725
13.3.3 Fluxless Inflation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 727
13.4 Axion Inflation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 728
13.4.1 Racetrack Axion Inflation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 729
13.4.2 Axion Valley. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 730
13.4.3 N-flation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 732
13.4.4 Aligned and Hierarchical Axion Inflation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 734
13.4.5 Monodromy Inflation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 735
(continued)

DOI 10.1007/978-3-319-41127-9_13
702 13 String Cosmology
13.4.6 Problems with Axion Inflation and Ways Out. . . . . . . . . . . . . . . . . . . . . . . . . . 740

13.5 Slow-Roll D-Brane Inflation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 742
13.5.1 Early Brane-Inflation Models. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 742
13.5.2 Warped D-Brane Inflation and K LM T Model. . . . . . . . . . . . . . . . . . . . . . . . . 744
13.5.3 Cosmological K LM T Dynamics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 748
13.5.4 Refinements and Related Models. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 750
13.5.5 Why the Tensor Spectrum Is Small. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 753
13.6 DBI Inflation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 754
13.6.1 Setting. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 754
13.6.2 UV Model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 755
13.6.3 IR Model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 758
13.7 Other Models. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 761
13.7.1 Braneworld. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 761
13.7.2 Cosmological Tachyon. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 765
13.7.3 Modified Gravity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 767
13.7.4 Non-local Models. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 769
13.7.5 Pre-Big-Bang and Dilaton Cosmology. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 770
13.7.6 String-Gas Cosmology. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 772
13.7.7 Cyclic Ekpyrotic Universe. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 776
13.8 Inflation and Alternatives: Compact Summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 783
13.9 Big-Bang Problem. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 786
13.9.1 Big Bang in String Theory. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 786
13.9.2 Classical Billiards: Living with the Singularity. . . . . . . . . . . . . . . . . . . . . . . . 788
13.9.3 Quantum Billiards: Avoiding the Singularity?. . . . . . . . . . . . . . . . . . . . . . . . . 792
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 795
After introducing some basic aspects of strings and branes in Chap. 12, we move
to cosmological models arising in or motivated by the theory, concentrating on
those based upon KLT and large-volume uplifting scenarios [1–28] embedded in
the string landscape [29–59] (reviews are [60–64]). Uplifting scenarios realize a
de Sitter spacetime in mechanisms of moduli stabilization in M-theory and string
theory, with two main consequences:
• The old cosmological constant problem is reinterpreted in a manner quite unique
to string theory and a possible resolution is proposed [65, 66] (Sect. 13.1). A
dynamical approach to the problem where string axions play the role of dark
energy is also possible [67–69] (Sect. 13.4.5).
• Inflation can be realized by several models: (i) size moduli inflation (Sect. 13.3),
divided into large-volume models [70–80] (Sect. 13.3.1), volume-modulus infla-
tion [81–83] (Sect. 13.3.2) and fluxless inflation [84, 85] (Sect. 13.3.3); (ii) axion
inflation (Sect. 13.4), divided into racetrack axion inflation [86, 87] (Sect. 13.4.1),
the axion valley [88, 89] (Sect. 13.4.2), N-flation [89–105] (Sect. 13.4.3), aligned
and hierarchical axion inflation [106–119] (Sect. 13.4.4; see also [104, 105, 120])
and monodromy inflation [121–143] (Sect. 13.4.5; see also [116]); (iii) warped
D-brane inflation, divided into slow-roll D-brane inflation [144–167] (Sect. 13.5)
and DBI inflation [168–191] (Sect. 13.6); some references discuss both classes
[192–194]. The scenarios (i)–(iii) are reviewed in [88, 195–202].
13.1 String Landscape 703
The above list is not exhaustive and neglects many other proposals, some of which
are reported in Sect. 13.7:
• Models with standard acceleration mechanisms (inflation or dark energy):
– Braneworld (Sect. 13.7.1). Inflation is realized in a brane and the cosmological
constant may relax to small values naturally. The big-bang problem is not
addressed.
– Cosmological tachyon (Sect. 13.7.2). The inflaton is identified with the string
tachyon. Neither the cosmological-constant nor the big-bang problem are
solved.
– Higher-order gravity models (Sect. 13.7.3), stringy or string-inspired realiza-
tions of some of the scenarios of Sect. 7.5.
– Non-local models (Sect. 13.7.4). These are essentially a sub-set of the models
discussed in Sects. 11.8.2 and 11.8.3. The big-bang problem is addressed
already at the classical level.
– Pre-big-bang cosmology (Sect. 13.7.5), an early attempt to resolve the big-
bang singularity with the string low-energy effective action of gravity. Infla-
tion is driven by the dilaton.
• Models with alternatives to inflation:
– String-gas cosmology (Sect. 13.7.6). The big-bang problem is solved by T-
duality while the early-universe spectra acquire a characteristic prediction on
the sign and magnitude of the tensor spectral index.
– Cyclic ekpyrotic universe (Sect. 13.7.7). Inflation is replaced by a mechanism
of cyclic contractions and expansions. The cosmological-constant and big-
bang problems may be addressed.
All these scenarios are summarized in Sect. 13.8. The big-bang problem in string
theory is discussed in Sect. 13.9 under the perspective of chaotic billiards.
13.1 String Landscape
The topology, geometry, shape and size of the compact space on which the theory is
dimensionally reduced are parametrized by moduli. The symmetry groups and the
number of moduli are enlarged when space is compactified and non-perturbative
gauge symmetries (for instance, from branes) are included. The main effect of all
these hidden (i.e., not directly observable) sectors is to increase the multiplicity
of vacua in the four-dimensional effective, low-energy, particle-field-theory limit
of string or M-theory. By “vacuum,” we mean a local minimum in the effective
potential of such effective limit. The increase in the number of vacua happens
independently of whether the hidden sectors induce a change in the number of
particles (matter or gauge carriers) in the low-energy limit. If there are nhid hidden
sectors each giving rise to nvac distinct vacua not in causal contact, the number of
vacua goes as
Nvac .nvac /nhid : (13.1)
The set of vacua with cardinality Nvac is called string landscape [29].
The landscape of string and M-theory can be mapped by cataloguing Calabi–Yau
spaces, fluxes and moduli fields systematically [30, 31, 34, 37, 40, 41, 43]. Under
several assumptions and approximations, one can even come to estimate the number
of vacua that can give rise to the Standard Model of particles [42, 49, 52–54].
Another requirement one can impose on the moduli space is that it reproduces a
cosmological background with a small but non-vanishing cosmological constant.
In practice, it is much easier to ask whether, how and how often string theory
realizes a de Sitter background in 3 C 1 non-compact dimensions. Even if a de Sitter
universe does not describe what we observe, it is a first step towards understanding
the cosmological constant problem in string theory.
In Sects. 12.3.7 and 12.3.9, we have illustrated how flux compactification and
gaugino condensation fix all the moduli to an AdS vacuum. The number of such
vacua can be estimated in each model. For instance, in type-IIA string theory the
number of vacua with a cosmological constant larger than a certain value N is
Nvac .jj > jj/N jj N 2=9 : minima with a small negative cosmological constant
are favoured [203]. In M-theory, vacua are not uniformly distributed in and
statistical results depend on the type of compactification [46].
Vacua with > 0 do exist, as we will see in Sect. 13.1.1, and Nvac is possibly
very large. Within a given set of Nvac minima with a roughly uniform distribution
of , one expects to find some vacua with a cosmological constant as small as
=42 m4Pl =Nvac . The observed vacuum would then be realized if Nvac & 10124 .
This cardinality can be easily reached by flux compactification (Sect. 12.3.7), where
there can be as many as nhid 10 – 500 fluxes wrapping around Calabi–Yau spaces,
each allowing for nvac 10 phases:
Nvac 1010 – 10500 ; (13.2)
with yet larger numbers for specific compactification schemes. Most of these
vacua do not realize the observed universe but, still, the number of minima that
can describe physics with the observed value of the constants of Nature can be
impressively large. In this context, the concept of naturalness of couplings in
quantum field theory gives way to another perspective: a field-theory limit A is more
natural in string theory than another limit B if the number of phenomenologically
acceptable vacua leading to A is larger than the number of vacua leading to B [30].
In the absence of a guiding principle of super-selection of all these vacua, a
deterministic resolution of the old cosmological constant problem is out of the
question. Nevertheless, a probabilistic resolution would be at hand if one had a
sufficiently large number of vacua, so that vacua with obs 10123 m2Pl would
occur with high frequency. The smallness of would then amount to a statistical-
selection effect rather than to some unwanted fine tuning [29, 30].
13.1.1 de Sitter Vacua
We have seen in Sects. 12.3.3 and 12.3.7 that the cosmological constant on the
visible four-manifold M4 vanishes identically in Calabi–Yau compactifications
(equation (12.86)). Complete moduli stabilization is achieved by gaugino conden-
sation or instantonic branes, which lower the minimum to an AdS configuration
(Sect. 12.3.9). The lifting of the minimum to a vanishing or positive value is
achieved by combining these ingredients with a third one, described later, which
typically relies on placing branes at some key points of the Calabi–Yau manifold.
The resulting KLT uplifting scenarios [1–4, 6, 7, 9, 11, 12, 14, 17–20] and their
cousins the large-volume uplifting scenarios [10, 13, 15, 21, 22] can be summarized
as the realization of a four-dimensional de Sitter background in string theory, with
applications to inflation and to late-time acceleration. They are the next step beyond,
respectively, KLT and large-volume stabilization scenarios.
The original KLT construction [1] is based on the warped compactification
of type-IIB superstring theory with 3-form fluxes (Sect. 12.3.7). In this model,
the dilaton and complex-structure moduli are stabilized by fluxes (Sect. 12.3.5)
[204], while the only Kähler modulus, the axio-radion %, is stabilized by gaugino
condensation (Sect. 12.3.9). At this point, all non-supersymmetric and N D 1
supersymmetric moduli are stabilized at the classical level to an anti-de Sitter
vacuum configuration.
To break supersymmetry and lift the minimum to positive values, it is sufficient to
add a small number of anti-D3-branes to the deformed Klebanov–Strassler throat in
the Calabi–Yau space C3 [1] (Fig. 13.1). While D-branes preserve supersymmetries
of the same chirality as the supersymmetries of the compactification, by definition
D-branes preserve supersymmetries with opposite chirality, since they carry charges
opposite to those of the background fluxes. Therefore, anti-branes break the
supersymmetries of the system. D3-branes, which are transverse to C3 and are
completely embedded in the four-manifold, must be added to cancel tadpoles of
the NS-NS B2 field [205] when too many fluxes are turned on.
Just like in the non-compact version of this construction [206], the volumes of
the D3-branes are stabilized by the background fluxes and no additional moduli are
introduced. Let us split the Kähler modulus % as in (12.114). The positive tension of
the anti-branes over-compensates the AdS minimum [1] by a contribution [144, 206]
ˇ1
V1 D ; (13.3)
2
where the coefficient ˇ1 / T3 > 0 depends on the number of D3-branes, on

their tension T3 and on the warp factor in the throat [206, 207]. Supersymmetry
is explicitly broken by V1 .
A similar inverse-power-law potential arises from wrapping a D7-brane (or more)
around a 4-cycle 4 of the Calabi–Yau manifold and turning on the fluxes of the
NS Fluxes
Wrapped D7 Brane
Throat
RR Fluxes
Anti D3 Branes
Fig. 13.1 A deformed Klebanov–Strassler throat embedded in a Calabi–Yau space with fluxes
around non-trivial cycles. In the figure, a stack of D3-brane are trapped near the tip of the throat,
while a D7-brane is wrapped around a 4-cycle (Credit: [6])
gauge fields living inside the brane [4]. These induce a D-term potential
Z
g2 p ˇ2
V2 D YM D2 D T7 d4 y g8 Fmn F mn D 3 ; (13.4)
2 4
where T7 is the brane tension, g8 is the determinant of the brane metric, Fmn is
the field strength of the gauge field and the magnitude of ˇ2 > 0 depends on the
intensity of the flux. Also matter fields charged under U.1/ may contribute to (13.4)
but we shall ignore them. Contrary to (13.3), the D-term (13.4) is supersymmetric
and one can employ all the tools of D D 4 supergravity. These tools show that the
D-term (13.4) can improve the stability of the AdS minimum but cannot uplift it
[14], unless ˛ 0 -corrections are also included.
Since D3-branes and wrapped D7-branes with fluxes coexist in the same
scenario, in general one will have both contributions, which we parametrize with
a generic inverse power-law
ˇ 1
Vn1 D / 2n=3 ; n > 2; (13.5)
n V
Fig. 13.2 The K LT potential (13.6) multiplied by 1015 and in Planck-mass units MPl D 41 D 1,
with n D 3, W0 D 104 , A D 1, ˛ D 0:1 (as in Fig. 12.4) and ˇ D 3 109 , in the .; /
plane (top panel) and for D 0 (bottom panel). The global minimum is at min 114:9, not far
away from the AdS minimum min 113:6 of (12.115)
where we used (12.97) and (12.99).The total potential is the sum of (12.115) and
Vn1 :1
˛A 2˛ h ˛ i ˇ
44 VKKLT D 2
e 1C A C W0 e˛ cos.˛ / C n ; (13.6)
2 3
where W0 < 0. The last contribution is positive and the AdS minimum can be
lifted to a positive value for a certain critical range of ˇ (Fig. 13.2). Below this
range, the local minimum stays negative, while above it it disappears and only
1
Yet another way to generate this class of potentials is to find local minima in the complex-structure
and dilaton directions in moduli space. Around these points, V develops a de Sitter minimum in
the direction [208].
the runaway minimum at ! C1 remains. Inside the critical range, > 0

and the non-compact four-manifold where the observable universe lives in is de
Sitter spacetime. Note that the position of the minimum does not change much
from the value (12.117), provided ˇ is sufficiently small; therefore, the large-volume
approximation 1 maintains its validity.
An important feature of de Sitter vacua, which will bear consequences for the
cosmological constant problem, is that there is an asymptotic minimum at V D 0
in the limit ! C1, separated from the de Sitter minimum by a potential barrier.
Due to quantum fluctuations of moduli fields, one expects de Sitter vacua to decay to
more favourable ground states (not necessarily Minkowski spacetime) via quantum
tunneling. The tunneling process is described by a Coleman–de Luccia instanton
[209, 210]. However, the lifetime of de Sitter vacua is much greater than the age
of the universe and a full cosmic evolution can take place in each vacuum [1].2 A
vacuum state which is unstable but on a very large time scale is called metastable.
This result of string theory is in agreement with general thermodynamic arguments
that suggest that de Sitter spacetime is metastable in any theory of quantum gravity
where the entropy inside the dS horizon is finite [211].
As we have seen in Sect. 12.3.9, the KLT model of moduli stabilization can
be heavily affected by the details of the concrete Calabi–Yau compactification
and of the dynamical stabilization mechanism. In general, models with only one
light modulus % do not work [7, 12, 212] and one has to stabilize up to O.10/
Kähler moduli and O.102 / complex-structure moduli. In particular, the complex-
structure fields should be stabilized simultaneously with the other light moduli, since
otherwise both the AdS minimum and the corresponding uplifted de Sitter minimum
are unstable [12]. The range of admissible values for W0 can only be determined by
a detailed study of a multi-moduli stabilization scheme; in many situations, such
range does not include values for which a de Sitter minimum is produced.
The uplifting mechanism can also be adapted to heterotic string theory [4, 7, 12]
(where gaugino condensation generates the potential (12.109) or its racetrack
extension (12.110) for the axio-dilaton S), type-IIA string theory [213] (where the
AdS minimum, obtained only via fluxes, is raised to positive values by perturbative
corrections to the Kähler potential) and M-theory [5, 8, 214, 215].
Moreover, alternative or concomitant sources breaking supersymmetry can
appear, such as corrections to the F-term (5.222) [7, 13, 15, 216], the already men-
tioned D-terms such as (13.4) [4, 17, 19, 216, 217], or perturbative ˛ 0 3 -corrections
to the Kähler potential [13]. Depending on the model, supersymmetry breaking can
take place before or after lifting the minimum.3 The first case corresponds to large-
volume stabilization and uplifting scenarios. We have already mentioned that W0 /
2
The decay rate has the same expression of the tunneling probability (10.48), which is of order of
exp.10123 /.
3
Thanks to the interplay of non-perturbative and ˛ 0 3 -corrections to the Kähler potential, it is also
possible to hit a de Sitter minimum without passing through an AdS vacuum (thus avoiding the
introduction of anti-branes) [23].
V6 =l6s can be much larger than in the original KLT scenario. This datum is important
when considered in the string landscape, where (assuming the complex-structure
moduli have already been integrated out) eK.%l ; / W02 is uniformly distributed in the
choices of flux [34, 43]. The number of type-IIB vacua with a supersymmetry
breaking scale MSUSY smaller than some given scale M N is Nvac .MSUSY < M/
N M N 12
[43] and low-MSUSY models are, so to speak, disfavoured. On the other hand, in
large-volume scenarios MSUSY m3=2 can be high, as we saw below (12.127).
Therefore, large values of W0 are more frequent in string theory and large-volume
scenarios are, in this sense, favoured in the landscape. However, the number of
vacua with small W0 can still be large enough to support also models with a low
supersymmetry scale. Backed by the typicality of KLT and large-volume uplifting
scenarios, the string landscape is pocked with enough de Sitter vacua to sustain an
interesting hypothesis on the origin of the old cosmological constant problem.
13.1.2 Cosmological Constant
In supergravity, the strategy towards a solution of the cosmological constant

problem is to overshoot the value (7.11) to zero or to some negative value and then
to lift the potential by some non-perturbative mechanism. String theory starts with
the same procedure but does not stop there.
The value of D 2 VKKLT obtained in the previous sub-section depends on the
free parameters of the model. The KLT scenario has two types of free parameters:
the number of D3-branes at the deformed Klebanov–Strassler throat and the number
of flux quanta at the 3-cycles. Fluxes push D3-branes towards the end of the throat,
where the warp factor (and, consequently, ˇ in (13.6)) is exponentially small. By
tuning the number of fluxes not too severely, one can even obtain a very fine-tuned
value for V, thus leaving a very small positive cosmological constant. This tuning
cannot be arbitrarily high, since the number of branes and fluxes can only take
discrete values. Then, one could obtain the observed value of for certain choices
of the parameters in KLT and large-volume string models.
Such a mildening of the fine tuning in the cosmological constant problem is not
dissimilar to analogous situations in some of the models of Chap. 7. The many ways
one can compactify the theory give rise to free parameters. Perhaps, for this reason
the above conclusion would not be regarded as particularly elegant, even if it is the
result of a top-down construction from a theory of everything. However, so far we
have neglected the fact that the de Sitter minimum is metastable. As it happens, the
decay of the vacuum has such a dramatic consequence on the physical picture as to
change the approach to the problem completely.
Assume that the Universe has reached a vacuum state with some D 1 ¤ 0.
As we have seen, moduli fields do not sit indefinitely at this vacuum and eventually
tunnel to another minimum with 0 < D 2 < 1 . Here they stay until they
tunnel again to a neighboring valley with a yet smaller minimum 3 < 2 ; and so
on. This scenario bears resemblance with eternal inflation (Sect. 5.6.5), to the point
where one can describe both with the same picture; vice versa, eternal inflation is
naturally realized by the string landscape, at least qualitatively. A local observer
would see the cosmological constant in their causal patch decrease in a series
of events [29]. Adopting a global view, causal patches (or metastable “bubbles”)
with progressively smaller form within a larger patch with larger cosmological
constant. The bubbles expand at a slower rate than the space between them, so that
they evolve independently from one another and never enter into causal contact.
Although the probability that one patch goes through a phase with 10123 m2Pl
is very small, among the infinitely many transitions occurring in the Universe there
will most likely be some bubbles in an acceptable vacuum.
A concrete model of quantum fluctuations driving the decay of the vacuum
can be understood as follows. We have seen in Chap. 7 and Sect. 10.2.4 that the
cosmological constant is effectively determined by the contribution of different
sources, from symmetry-breaking effective potentials to dynamical fields, including
anti-symmetric tensor fields [218–220]. As we have remarked in the course of
Chap. 12, p-forms are associated with branes, so that one can expect that quantum
fluctuations of a p-form field gives rise to the non-perturbative creation of virtual
pairs of branes–anti-branes. Once these pairs are created, they do not annihilate
because the space separating them expands fast enough. The net consequence is
that the energy density of the field decreases, thus reducing the value of the effective
cosmological constant [221, 222]. The process can be described in terms of compact
spatial 2-branes arising from a 3-form field (i.e., a 4-flux), which form the walls of
the de Sitter bubbles continuously nucleating via quantum tunneling.
This mechanism can be adapted to and improved in M-theory, where the presence
of many 4-fluxes in compactified configurations gives rise to a discrete spectrum
with thinly spaced energy levels [65, 66]. Consider the compactification of the action
(12.128) on a 7-manifold with volume V7 :
Z
1 p 1
S11 D d4 x g R 20 F42 C : : : ; (13.7)
242 2
where we keep the symbols 0 and F4 to denote the bare cosmological constant
and the 4-flux after dimensional reduction. The four-dimensional Newton’s constant
is 42 D 112
V7 . A solution for the gauge-field equations of motion r F4 D
0 is F D c , where c is a mass coefficient. This produces an effective
sol
cosmological constant eff D 0 .F4sol /2 =4 D 0 C c2 =4, which can be positive

if 0 6 0. At the quantum level, .F4sol /2 ! h.F4sol /2 i becomes the expectation value
of the gauge field strength. In string theory, this expectation value is quantized due
to the simultaneous presence of electric and magnetic sources and for a consistent
quantization of branes [65, 223]. Then, c D 242 en, where e is the charge of the M2-
brane (with dimension Œe D 3) and n 2 Z. Creation of brane pairs changes the value
of c, which jumps to lower and lower values due to the repeated tunneling towards
more favourable vacua. Notice that these branes are not the fundamental ones of the
theory; they correspond to instantonic solutions of the Euclidean action [224].
A first requisite one demands to obtain a solution of the old cosmological

constant problem is that the spacing between the allowed values of eff be of order
of 10123 m2Pl or smaller. Such a requisite translates into asking that the minimal
spacing in the cosmological-constant spectrum be no larger than the observed
value,
. obs : (13.8)
This cannot be achieved by a single 4-flux, but many fluxes arise if the compact
manifold contains non-trivial 3-cycles. If there are N 1 3-cycles, one produces N
fluxes associated with N different types of branes with charges ci / 42 ei : these are
either M2-branes completely embedded in the non-compact manifold M4 or M5-
branes with three directions wrapped around a 3-cycle. The effective cosmological
constant reads
X
N
eff D 0 C n2i q2i ; (13.9)
iD1
where qi :D ci =.2ni / is a constant with dimension Œqi D 1.

We can represent the flux spectrum as an N-dimensional Q regular lattice with
spacing qi ; the volume of an elementary cell is Vq i qi . On this grid, the
observational range of the cosmological constant p is represented by a thin shell
centered at ni D 0 of average radius r D j0 j and thickness r D obs =r
(Fig. 13.3). The volume of the shell is V D ˝N1 rN1 r, where ˝N1 D
2 N=2 = .N=2/ is the area of a unit N-sphere. The inequality (13.8) is satisfied if at
least one point of the lattice falls in the shell, which occurs if dVq . V, where d
is the typical degeneracy of states per charge level. This relation yields the minimal
spacing [65]
QN
d qi
D iD1
: (13.10)
˝N1 j0 j 2 1
N
For 0 D O.m2Pl / and small degeneracy d D O.1/, only about N D 102 fluxes are
sufficient to produce the desired spacing if the charges qi 101 –100 mPl are not
much smaller than the Planck mass.
In the simple case of a 7-torus with radii rl (l D 1; : : : ; 7), there are 35
3-cycles with volume V3;i D .2/3 rmi rm0i rm00i (one for each of the unordered
triplets .m; m0 ; m00 /). If one writes the gravitational coupling in 11 dimensions as
2 9
11 D 4M11 , where M11 is the fundamental mass scale of M-theory, then the
four-dimensional reduced Planck mass is MPl2 D 42 D 4M11 9
V7 , the tension of
an M5-brane wrapped around the i-th cycle and the tension of the M2-brane are,
6 3
respectively, Ti D 2M11 V3;i for i < N and TN D 2M11 , and the brane charges are
3
e D 2M11 . By taking into proper account the volume factors of the 3-cycles and
n2 q2 0
–2
–4
–4 –2 0 2 4
n1 q1
Fig. 13.3 The allowed values of the energy density of N D 2 4-forms as described in the text. A
negative bare cosmological constant 0 is compensated up to the p observed value obs if at least
one point lies in the shell (shaded region) of average radius r D j0 j
of the compact 7-manifold, one finds that qi D 42 Ti . Assuming V3;i l3Pl , we have
6 3
2 M11 2 M11
qi¤N ' .4/ mPl ; qN ' .4/ mPl (13.11)
mPl mPl
and, to obtain (13.8), the 11-dimensional mass must be close to the Planck mass.
For large degeneracies d or small N, the value of the charges is reduced, but still
to a level acceptable in string phenomenology. Clearly, one flux is not enough, as the
brane charge would be fine-tuned in that case, q . obs . This gap (or small-step)
problem [221, 222] is not present in string and M-theory [65], where the compact
space usually has many 3-cycles. Furthermore, the nucleation rate of bubbles with
small cosmological constant is tiny compared to the age of the universe. Therefore,
the landscape mechanism of quantum dynamical suppression of the cosmological
constant is capable of producing an observationally viable phase with obs
which is stable on cosmological time scales.
Even if there were enough realizations of the obs 10123 m4Pl vacuum, to
address the cosmological constant problem one should ask whether and why such
vacuum should be a preferred outcome for an observer. The anthropic principle
[225–227] has been proposed as a selection rule coming to the fore at the time of
answering these questions [29, 65]. In its weakest form, it is based on the following
observation. If the cosmological constant were too large, the accelerated expansion
at large redshift would hinder gravitational collapse and the formation of galaxies.
On the other hand, if were too negative, the universe would have recollapsed
before galaxies and stars could form. These two conditions give, respectively, an
upper and a lower bound for the observationally admissible values of [226]:
10123 m4Pl < < 10121 m4Pl : (13.12)
The weak anthropic principle states that the observed is so small because galaxy
and star formation, both prerequisites for the emergence of life (and, eventually, of
an observer), would not have taken place otherwise. This is not a mere tautology: the
interval (13.12), which includes the point D obs = 2 , was computed before any
evidence of dark energy was known and may be regarded as an actual prediction
on the existence and range of values of a cosmological constant. This argument
can be made more compelling, and the range (13.12) further restricted, by more
sophisticated analyses [228, 229] or by stronger versions of the anthropic principle.
Coming back to the string landscape, the Universe is made of patches (cosmolog-
ically large, causally disconnected regions) with different cosmological constants,
but only in those regions where the cosmological constant is sufficiently small does
galaxy and star formation take place. Therefore, we observe a small cosmological
constant simply because we inhabit one of the statistically allowed bubbles where
organic life is made possible.
Similar reasonings can also be applied to the parameters of the Standard
Model, in particular the Higgs mass. In this case, the requirement that atomic
nuclei be stable (atomic principle) yields a prediction for the range of the Higgs
mass [230]. Applying the weak anthropic principle and the atomic principle to
string theory, one can scan the cosmological constant and the Higgs mass across
the landscape and address the cosmological constant problem and the hierarchy
problem simultaneously [45]. This gives rise to stimulating consequences on the
supersymmetry breaking scale MSUSY (typically very high) and the observability of
new O.TeV/ particle phenomenology at the LHC, on which we will comment in
Chap. 14.
An alternative which does not rely on the anthropic principle is statistical selec-
tion, which aims at extracting predictions from the analysis of joint probabilities
in the landscape [30, 40, 63]. An intuitive example is the following. Imagine there
were two sets of vacua A and B, one with Nvac A
10160 elements and a low-scale
supersymmetry (say, MSUSY D 10 TeV) and the other with Nvac B
10120 elements
9 17
and a high-scale supersymmetry (say, MSUSY D 10 –10 GeV). Suppose that both
sets reproduce all known couplings except for the observed cosmological constant
obs . By comparing NvacA B
and Nvac , one cannot conclude convincingly that “string
theory predicts low-scale supersymmetry,” since both numbers are very large.
However, the requirement that these vacua also reproduce obs would make the
argument more stringent. If both sets realized a uniform distribution of cosmological
constants, then we would expect about Nvac A;
1040 low-supersymmetry viable
3
vacua but only Nvac 1–10 viable vacua with high-energy supersymmetry. Both
B;
alternatives would describe the observed universe but only the one with low-scale
supersymmetry would be realized with significant frequency in the string landscape.
In actual procedures to survey the landscape, in reality one devises a set of
vacua with given supersymmetry breaking scale and determines the probability
distribution of these models as a function of . Conversely, the distribution of
models with obs and the observed couplings of the Standard Model can be analyzed
as a function of the supersymmetry scale.
Actual determinations of the vacua distribution in scenarios with flux com-
pactification are much more refined that the crude estimates based on (13.1)
[40]. Notwithstanding the problems related to an incomplete control over the
moduli space and to a lack of experimental evidence for supersymmetry (which
considerably increases the number of possibilities), these studies fuel the hope that
statistical selection in string theory may be a viable explanation of the smallness
of .
13.1.3 Open Problems
The impact of the string landscape on epistemology and physics has been compared,
in its importance, to the one caused by the discovery of a large number of solar
systems apart from ours [45]. Observers believing in the uniqueness of our Solar
System would try to understand why the Earth-Sun distance takes exactly the value
we measure. As soon as the observational range is expanded to galactic scales and a
landscape of 1011 stars is found, it becomes clear that just historic accidents, rather
than any fundamental principle, have determined our planetary distances. The deep
mystery of the value of the Astronomical Unit dissolves into a reassuring statistical
explanation, gladly helped by astrobiology considerations on the conditions for
carbon-based life in habitable orbital belts.
In the opinion of some, however, the existence of as many vacua as (13.2) places
a high stake on the possibility that string theory may not be falsifiable. Of course,
unique predictivity is not required for a theory to be falsifiable experimentally; the
Standard Model and general relativity are strikingly predictive despite the presence
of a number of free parameters. In the case of string theory, born as a fundamental
framework with no free parameters, demands and expectations are perhaps higher,
especially now that we still lack a distinctive signature of stringy effects in particle
physics or cosmology. We do not wish to enter a debate about the feasibility of
string theory, since it would involve philosophical issues and technical caveats on
the assumptions leading to (13.2); these are discussed, for instance, in [30, 32].
Suffice it to say that, as we have seen, a large statistics such as (13.2) can actually
help in reinterpreting the cosmological constant problem in an exceptional way.
Two basic requirements for a genuine solution of the problem are a robust
counting of the vacua and a coherent description of the evolution of an observer
in the string landscape. Much speculation is involved in the present understanding
of these features. On one hand, the absence of a rigorous global description of
bubble nucleation (due to the presence of de Sitter horizons, which define causal
13.2 Inflation in the Landscape 715
patches) makes the models discussed above somewhat too heuristic to extract solid
information on the probabilities involved in the nucleation process. This is nothing
but the measure problem in eternal inflation [231] (Sect. 5.6.5) or in multiverse
scenarios [232, 233]. Estimates of such probability distribution have been made
nonetheless and, in the absence of better theoretical grounds, they represent the
state of the art. When they do not involve anthropic arguments, these estimates work
towards a solution of the problem by statistical selection; however, they are not
conclusive yet. On the other hand, there is no compelling and universally accepted
principle or mechanism preferring one vacuum over another. Originally invoked to
supply a guidance in the vast string landscape, the anthropic principle has been the
subject, in virtually all of its forms, of diversified and complex criticism (a non-
exhaustive sample can be tasted in [32, 44, 234]), its actual effectiveness has been
questioned [235] and alternative criteria for vacuum selection have been proposed
[236].
Rather than an intrinsic flaw in string theory, these issues demonstrate the need to
extend our knowledge farther than the present point. In this respect, and until such
progress is made, any comparison between the string framework and the theories of
quantum gravity of Chap. 11 will be forcefully limited by the degree of development
of the single proponents.
13.2 Inflation in the Landscape
Within the landscape, there are several models of string cosmology that include
an early phase of acceleration. Although the step from a perfect de Sitter vacuum
to inflation seems small, it entails a number of subtleties concerning moduli
stabilization. In fact, the latter interferes with a viable period of inflation and Kähler
and superpotentials must be hand-picked with care. For any given vacuum in the
landscape, we will describe two classes of inflationary scenarios: moduli inflation
(Sects. 13.3 and 13.4) and D-brane inflation (Sects. 13.5 and 13.6). Different regions
of the landscape may offer a favourable habitat for any of these scenarios. In some
cases, the phenomenology of string inflation was developed some years before its
actual embedding in the theory.
Taken as a whole, these models are the string representative of quintessential
inflation, where one or more scalar fields are responsible for both the early-universe
and late-time acceleration (Sect. 7.3.6). As we will see, a modulus or several moduli
drive the inflationary era and then relax, at late times, at a metastable de Sitter
minimum of their potential, without relying on severely fine-tuned initial conditions.
Before going into the details of how inflation takes place in flux compactifica-
tions, let us discuss the general phenomenology of inflation in the big picture of the
string landscape, where the dynamical field or fields pass through many metastable
vacua. Concrete realizations of inflation are the subject of Sects. 13.3, 13.4, 13.5
and 13.6. In the rest of the chapter, the inflaton will be denoted as when identified
with any of the moduli except for the position of branes, in which case we will
employ the symbol '.
13.2.1 Single-Field Inflation
The string landscape can relate the string or M-theory couplings to the inflationary
energy scale, thus connecting fundamental micro-physics with observations of the
early universe. To argue this, we introduce a novel ingredient with respect to the
picture described in Sect. 13.1: we assume that inflation takes place between a
.1/ .0/ .1/
given vacuum VdS and the next one VdS D < VdS with the value of
the cosmological constant observed today. The sequence of tunnelings through the
preceding metastable vacua is not important in what follows and one can consider
the simplified context of a false-vacuum decay followed by a period of inflation. A
potential barrier separates the inflationary trough in the landscape from the higher
.1/ .0/
vacuum VdS . The final inflaton vacuum is VdS .
The single-field case is one of the simplest representatives of landscape inflation.
One assumes (or shows, in specific models) that all moduli but the inflaton scalar
have been stabilized at an early stage. Multi-field constructions support such a
dynamical situation to some extent. Many viable single-field models have three main
properties: a very small negative curvature, just enough e-foldings and suppression
of the power spectrum at low multipoles. The scalar spectrum is suppressed at
large scales due to a steep feature at the beginning of inflation, the potential barrier
.0/ .1/
separating VdS from VdS .
Let the bare cosmological constant be constituted by the inflaton potential, 0 '
42 V. /. During inflation, the cosmological dynamics is governed by the effective
Friedmann equation
eff
H2 ' ; (13.13)
3
where eff is given by (13.9). Consider a large-field model with V; > 0. As
we know from (5.131) in Sect. 5.6.2, in quasi-de Sitter spacetime the quantum
fluctuation of the scalar field during a Hubble time t H 1 is jı j ' H=.2/.
On the other hand, during the same interval the field displacement due to classical
slow rolling is j j ' V; t=.3H/ ' V; =.3H 2 /. Quantum fluctuations dominate
the evolution of the inflaton if j j < jı j, which gives an upper bound for V; :
3=2

V; < peff : (13.14)
2 3
In this case, bubble (i.e., brane) nucleation takes place and the effective cosmolog-
ical constant undergoes a sequence of suppressions. When the inflaton approaches
the minimum of its potential, inequality (13.14) is violated and reheating occurs. At
this point, the nucleation stops and the cosmological constant (the actual constant
in eff ) takes its final value final . Let i D N be the bubble representing our causal
patch. Inside it, the value of nN in (13.9) is lowered by 1, so that the value of the
cosmological constant is lowered by
Œn2N .n2N 1/2 q2N D .2nN 1/q2N D: : (13.15)
This is the cosmological constant at the penultimate nucleation. If final obs ,

we have seen that q2 102 – 101 m2Pl , so that m2Pl and it dominates over any
other contribution in eff at the end of inflation. By this approximation and using
(13.11), inequality (13.14) reads
9
.2nN 1/3=2 3 .4/5 M11
V; . e / 0:3 1=3
: (13.16)
mPl mPl
The quadratic potential V D m2 2 =2 is under strong observational pressure and

toroidal compactification is not the best choice one can make in string theory, but the
combination of these ingredients can give the reader an idea of the typical energy
scale of this scenario. According to (5.87c), a minimum numberpof 60 e-foldings
is guaranteed if m 0:5mPl . Then, V; . e / D m2 e 0:5m3Pl = 4 and M11 >
0:2mPl is around or somewhat above the grand-unification scale. In models with large
compact dimensions [237–239], this scale can be lowered as much as O.TeV/ [65].
Scenarios more detailed than (13.13) take into account the consequences of
a tunneling origin of the universe in the landscape [240–244]. The Coleman–
de Luccia instanton has the following peculiarity: an observer inside the newly
nucleated bubble would locally see an infinite open universe [209, 210]. Anthropic
bounds coupled with statistical analyses show that an inflationary period with just
the minimum number of e-foldings Ne 60 may be theoretically favoured [240].4
In fact, on one hand a long stage of inflation may be improbable in this scenario
and curvature would not be washed away. On the other hand, an ˝0 1 would
impede regular structure formation. The weak anthropic principle then places an
upper bound on the curvature ˝K close to the current experimental bound (2.128).
The combination of these arguments gives just enough inflation.
These are special features of landscape-based single-field inflation that lie at the
border of observability and can place strong constraints on the parameter space
of the models. For instance, despite the tight bound (2.128), there may be a non-
negligible but strongly measure-dependent theoretical chance to detect a negative
small curvature [240, 242]. Moreover, the CMB temperature and polarization
4
Throughout this chapter, we use the definition (2.48) for the number of e-folds and omit the
subscript a. This quantity coincides with the improved definition (5.12) during inflation.
spectra C`TT , C`BB and CÈE can be suppressed or enhanced at large scales depending
on whether the phase just after nucleation is, respectively, of fast or slow roll
[241, 243, 244]. The second case is already ruled out by observations, while the
first [240, 243] is compatible with the loss of power that has actually been found
since WMAP.5 The theoretical chance to observe such suppression is larger than the
one of seeing curvature [243]. Although cosmic-variance effects are most prominent
at these scales and the observed suppression is not in contradiction with the standard
CDM model, new polarization measurements can improve our knowledge of the
low-` region.
13.2.2 Large-Field Models and the Weak Gravity Conjecture
Any string model of inflation where the effective inflaton takes values comparable
with or larger than the Planck energy falls into the class of large-field models intro-
duced in Sect. 5.5.1. However, large-field inflation in the landscape can potentially
suffer from a severe constraint coming from the weak gravity conjecture [246–250].
This conjecture, stemming from the assumption that the number of stable particles
not protected by a symmetry is finite, elevates the observation that gravity is the
weakest force in Nature to the status of principle. Then, in any theory of quantum
gravity bending to such principle, elementary charged objects must feel a gauge
force stronger than their mutual gravitational attraction; in other words, the mass-
to-charge ratio jm=qj is bounded from above. For a dimensionless q,
ˇ ˇ
ˇmˇ
ˇ ˇ 6 MPl : (13.17)
ˇqˇ
(The right-hand side should not be taken literally; it can be mPl or O.1/ variations of
the same.) There are a mild and a strong version of the principle. In the mild version,
the above inequality is enforced only on the charged object which minimizes the
mass-to-charge ratio, so that jm=qjmin 6 MPl . In the strong version, which implies
the mild one, (13.17) holds for the lightest particle, so that jmmin =qj 6 MPl . The
bound (13.17) is motivated by several facts in string theory and, to date, there are
no string counter-examples to the conjecture. Evidence for the mild version is more
abundant than for the strong one.
The weak gravity conjecture has consequences that go against the intuition of
traditional low-energy effective field theory. For instance, in four dimensions and
in the presence of a U.1/ gauge field with coupling q0 , the conjecture implies
that there exists a UV cut-off scale UV q0 MPl , smaller than the Planck energy,
beyond which the effective field theory breaks down. If q0 is one of the Standard-
5
A fast-rolling inflaton is a possible source for large-scale power suppression independently of its
realization in the string landscape [245].
Model gauge couplings near the grand-unification scale, then UV is close to the
heterotic string scale 1017 GeV. Moreover, the sub-millimeter observation of any
tiny gauge coupling q0 1 would imply per force the existence of a low-energy
cut-off scale UV MPl well below the Planck or even the GUT scale. A cut-off
UV < MPl has an impact on the class of large-field models of inflation, where the
effective theory is expected to receive corrections at the Planck scale, not below it.
Axion monodromy models coming from consistent string-theory compactifications
(Sect. 13.4.5) provide a UV completion of large-field inflation but they are heavily
penalized for exactly the same reason (Sect. 13.4.6).
Recalling the great effort spent in Chaps. 9, 10 and 11, it is amusing to note that
the weak gravity conjecture, if true, excludes all theories of quantum gravity where
matter fields are introduced by hand after formulating the gravity sector. In those
cases (that include loop quantum gravity, spin foams, asymptotic safety and CDT in
their current formulation, but perhaps not group field theory), one has the freedom
to consider gauge fields with arbitrarily small coupling.
13.2.3 Multi-field Inflation
If moduli are not stabilized before inflation, it is possible to have several dynamical
scalar fields and a multi-dimensional potential. Multi-field inflation (Sect. 5.5.3)
can occur in an ample variety of situations within string theory and models can
range from landscape-related [161–165, 251] and generic moduli-related [252]
(including N-flation [90] and chain inflation [253–257]) to generic string-inspired
ones [258–260].
From a statistical point of view, small-field potentials with slow-roll near a saddle
point are more likely to be realized in the landscape [161–163, 251, 259, 260].
Moduli inflation (Sect. 13.3) and the infrared DBI model (Sect. 13.6) are examples
of string inflation at inflection points. As the inflaton evolves, however, it becomes
unlikely to get trapped in a metastable vacuum with positive energy for a sufficient
time [163]. To avoid this problem, which would partly invalidate the single-field
scenario outlined above, anthropic arguments can select the characteristics of the
saddle point so that the latter is in the vicinity of a hole in the landscape with a
positive-valued bottom. Inflation terminates when the scalar field rolls down the
hole and one ends up with a small > 0. The same anthropic arguments also help
to realize enough e-foldings and a sufficiently long inflationary period.
These models do not generate a high level of non-Gaussianity during slow rolling
[164, 260] but the hole must be shallow in order to avoid production of large non-
Gaussianities at the end of inflation [165]. To avoid fine tuning, one must conclude
that the inflaton potential has mild slopes and that ragged landscapes with deep
minima and steep slopes do not lead, in general, to viable scenarios without a certain
amount of fine tuning.
13.2.4 Moduli Problem and -Problem
The models we will describe below exemplify the subtle points involved in moduli
stabilization and multi-field inflation in string theory. Excepting a few cases, the
underlying Calabi–Yau space is not known and there is no way to determine
the total number of complex-structure and Kähler moduli: one must make a
phenomenological choice on their number and type. Then, to make the problem
tractable one stabilizes all but a few fields, either by mere assumption or by KLT -
like mechanisms. Here are the first two caveats to bear in mind in string cosmology.
One, which we already had occasion to appreciate, is that stabilizing a modulus
for real usually leads to very different results from assuming that it has been
stabilized by some unknown mechanism. The second is that stabilizing n moduli
simultaneously can lead to very different outcomes from doing so at stages (i.e.,
stabilizing first k moduli and then the other n k). Both points will be illustrated by
the various shapes the moduli C inflaton potential V will take in each model.
Even retaining as many dynamical fields as possible with an act of good will,
one must make sure to avoid the moduli problem, which is in fact a set of problems
[261–265]. Light moduli can disrupt the inflaton dynamics and bar the way to
flat directions; or, if inflation happens, they may freeze too late and spoil the
standard and well-constrained nucleosynthesis scenario; or, when they decay they
can produce too much entropy and, again, jeopardize nucleosynthesis. A task of
any string model is to produce a potential which avoids such problems. In general,
models of string inflation are successful in this respect but in a non-unique way:
each individual potential comes from a different cascade of stabilizations.
Finally, whether and how moduli are stabilized before or during inflation may
determine the onset of the notorious -problem plaguing generic supergravity
models of inflation (Sect. 5.12.3).
13.3 Size Moduli Inflation
Historically, the presence of light degrees of freedom in supergravity and string

compactifications suggested quite early to identify the inflaton with one of the
moduli, typically the dilaton or the Kähler modulus, while all the other fields were
assumed to be stabilized by some unspecified mechanism [266–270]. Despite its
attractive simplicity, the idea did not work due to the difficulty to obtain slow
rolling in the typical non-flat potentials of the moduli. The main culprit of such
steepness was the exponential form (12.109) and (12.110) of the non-perturbative
superpotential. Thanks to the recent progress in understanding moduli stabilization
in KLT and related constructions, it has been possible to revisit moduli inflation as
a candidate model of the early universe.
In this section, we mainly explore scenarios, each based on its own set of
assumptions, where the inflaton is a size modulus or a combination of size moduli
13.3 Size Moduli Inflation 721
in type-IIB and heterotic string theory. In general, these belong to the class of small-
field models introduced in Sect. 5.5.2, where the inflaton rolls down a flat local
maximum.
13.3.1 Large-Volume Inflation

13.3.1.1 Blow-Up Inflation
When many Kähler moduli %i are stabilized explicitly, one can identify the last
one to freeze with the inflaton. In type-IIB flux compactifications, this can be
achieved [70–72, 74] in the large-volume stabilization and uplifting scenarios
[10, 13, 15] described in Sects. 12.3.9.3 and 13.1.1. One takes a simplified racetrack
superpotential (12.110),
X
WD Ai e˛i %i ; (13.18)
i
and generalizes the potential (12.124) accordingly. While in the models of

Sect. 13.4.1 the inflaton will be the axionic part of a Kähler modulus
%infl D C i , here it is the last of the “small” or “blow-up” moduli i
(describing the size of the 4-cycles, the “holes” of C6 ) to roll down to the minimum.
All the Kähler moduli except the inflaton are stabilized dynamically. These models
are of small-field type (Sect. 5.5.2) and nowadays they are called of blow-up
inflation.
After freezing the dilaton, the potential for many Kähler moduli is the general-
ization of (12.123):
2 q 3
V.V; i
/ X
I
6 i
i
jW0 j
i
i 7
e2˛ C e˛ cos.˛
i i i i
4
D 4ˇ1 ˇ2i i i
C /5
M Pl iD1
V V2
jW0 j2
Cˇ3 3=2
; (13.19)
gs V 3
where exp.i i
/ :D Ai W0 =jAi W0 j and Ai is the amplitude in (13.18)
associated with the modes. Here we follow [70] for simplicity and stabilize
also the axionic part . (In [71], is switched on and a scenario intermediate
between the axionic and the large-volume model arises.) Minimizing with respect
to all the axions at i
D . i
/=˛
i
for all i and all
i
for i ¤ I,
one obtains a constant (and partly non-perturbative) contribution (12.126) for
Adding also the uplifting term (13.5) with n D 9=2, we get a constant
each i. P
V0 D i¤I Vmin C
i
V / .V min /3 , where V min is the minimized Calabi–Yau
volume. Calling D I
the surviving modulus (not normalized canonically) and
ignoring the heavily suppressed exp.2˛I / term in (13.19), the total potential
reads schematically
V. / D V0 V1 e˛I ; (13.20)
where ˛I > 0 and 0 < V1 / .V min /2 . The shape of (13.20) is shown in Fig. (13.4).
Viable inflation does not require to fine tune the parameters of the potential [70].
The slow-roll parameters at horizon-crossing are < 1012 and ' 2=Nk . The
observed normalization of the scalar spectrum restricts the size of the Calabi–Yau
volume,
105 6 V min 6 107 : (13.21)
For 50 < Nk < 70, the scalar spectral index ns ' 1 C 2 is
0:96 6 ns 6 0:97 : (13.22)
The index running is negative and small, ˛s D O.104 /. Since the inflationary scale
is rather low, V 1=4 1013 GeV, the tensor-to-scalar ratio (5.156) is unobservable,
r . 1010 . The tensor index nt ' 2 is nearly zero, too. These numbers persist
in a complete model with axions [71]. In general, two Kähler moduli may be not
enough to stabilize the volume.
Fig. 13.4 The potential (13.20) of large-volume moduli inflation in MPl D 1 units and for casual
choices of the parameters V0;1 and ˛I
13.3.1.2 Fibre Inflation
Blow-up inflation can suffer from the -problem because loop corrections can
dominate over non-perturbative effects, as we will see presently. A scenario capable
of producing a larger tensor spectrum and to avoid this problem is fibre inflation
[73, 75, 76, 79, 80]. Fibre moduli are Kähler moduli whose potential is generated at
one loop in string quantum corrections. There are two such moduli 1;2 in the model
representative of this class [73], living on a Calabi–Yau space with a K3 fibration
in a type-IIB large-volume compactification. An ordinary Kähler modulus is
p 3=2
also included. The internal volume is V D ˛. 1 2 /, where ˛ > 0 and
> 0 are model-dependent constants. Stabilization is sought in the large-volume
p 3=2
limit V 1, so that 1 2 . In the absence of quantum corrections, the
potential V.V; / depends only on two of the three moduli. The flat direction can
be parametrized by and 1 at fixed V. Therefore, while in the blow-up case the
inflaton is the size of a small cycle, in fibre inflation it is the size of a large one. Let
us call V D V.V min ; min
/ the value of the potential at the minimum.
One-loop corrections modify V.V; / with a term of the form

W02 A B C1
Vgs D 2 p C 2 ; (13.23)
V 12 V 1 V
where A; C / g2s > 0 and B can be of either sign. This contribution deforms the
flat direction and stabilizes 1 . In the single-field approximation, one freezes V and
and studies the dynamics of 1 in the effective potential V C Vgs . If 0 <
A; C p B, the potential V. / D V C Vgs . / for the canonically normalized field
4 D 3=4 ln 1 is
p p p
V. / ' Vgs 3 4 e4 = 3
C e44 = 3
C ˇ e24 = 3
; (13.24)
where ˇ D O.g2s / 1 and the constant
1
Vgs / ; V D V min ; (13.25)
V 10=3
depends on V min and min
. All the other coefficients of the potential are independent
of the compactification parameters. This potential has a global minimum reached
from a flat plateau, where inflation occurs (Fig. 13.5).
In the slow-roll phase, one can neglect the second and third exponentials in
(13.24) and compute the primordial spectra. One finds V ' 32V =2 and r '
6.ns 1/2 . For 50 6 Nk 6 60 e-foldings, the theoretical points in the .ns ; r/ plane
are well within the 1-level likelihood contour,
0:965 6 ns 6 0:970 ; 0:005 6 r 6 0:007 : (13.26)

Fig. 13.5 The potential (13.24) in k41 D MPl D 1 units with ˇ D 106
The tensor-to-scalar ratio is not as small as in the other string models of inflation
and, in fact, the inflationary scale is relatively high, V 1=4 1015 –1016 GeV.
Relaxing the single-field approximation and including the dynamics of the moduli
V and , one obtains similar results.
Post-inflationary conversion of isocurvature perturbations into adiabatic modes
can generate observable non-Gaussianities of the local form (Sect. 4.6.3.1), if the
minimum internal volume is not too large. The non-linear parameter is
.32ˇ3 W02 =3/1=3

local
fNL ' 105 1=6
; (13.27)
gs V min
where the constants ˇ3 and are those of equation (13.19). To realize an efficient
curvaton scenario, 103 6 V min 6 108 . Taking V min D 103 , gs D 102 , W0 D 101
and D 101 (ˇ3 depends on , on the parameters of the superpotential and on the
local
expectation value of one of the fields), one obtains fNL D O.10/. Choosing a larger
6 2
volume V min
D 10 and gs D 10 , W0 D 10 and D 1, non-Gaussianities in the
squeezed limit are within the experimental bound, fNL local
D O.1/.
Blow-up and fibre inflation yield attractive predictions but, like most string
cosmological models, they meet with some issues at the time of reading the fine
print. The inflaton tends to reheat into the hidden sector of the theory unless the
latter is severely constrained. This problem restricts the viable parameter space of
both classes of models [76] without ruling them out.
A third large-volume model (poly-instanton inflation) produces an index
ns 0:96 and a tensor-to-scalar ratio r 105 [77]; it may also serve as a
quintessence scenario [271]. In a fourth model, the ˛ 0 3 -corrections considered so
far (Sect. 12.3.9.3) can be combined with other ˛ 0 3 -contributions coming from
higher-order curvature terms in N D 1 SUGRA superspace, to give Starobinsky-
like potentials for the fibre modulus [78]. String loop gs -corrections do not drive
inflation but generate post-inflationary minima for the potential.
13.3.2 Volume-Modulus Inflation
In this sub-section, we present a scenario complementary to large-volume inflation:

while in the latter case the inflaton was a blow-up or a fibre modulus, in the present
one the inflaton is the volume modulus V D V6 =l6s of the Calabi–Yau internal space.
We already had this modulus play the role of the inflaton in several of the SUGRA
models of Sect. 5.12.4, where we called it T. The similarities are many: K is a
no-scale Kähler potential or a related deformation and the inflaton potential V is
the sum of exponential terms. However, both the details of the deformations and
the coefficients in V differ. While the models of SUGRA inflation proceed from
hand-picked super- and Kähler potentials, volume-modulus inflation finds its own
microscopic origin in string scenarios of flux compactification, which give rise to
very specific potentials and phenomenology.6
The concrete embedding of this mechanism in the SUGRA limit of string theory
can proceed in different ways [81–83]. We will see in Sect. 13.4.1 that, in type-
IIB flux compactification with a racetrack superpotential, the flattest directions in
the total potential are typically along D Im%. However, for certain values of the
parameters [11, 18, 51], inflation can be driven by D Re% D V 2=3 rather than
the axionic part of the Kähler modulus %, at an inflection point of the potential
[81]. Predictions range in the interval 0:93 . ns < 1, while the tensor-to-scalar
ratio is negligible, r < 106 . Viable inflation is characterized by some strong fine
tuning of the initial conditions which, however, might not be a problem since the
model supports eternal inflation. Regions with exponentially growing volume are
continuously produced and quantum fluctuations make the inflaton start always at
the saddle point of the potential. For long or eternal inflation, however, the scalar
spectral index is either ns 0:93 (if V is not Z2 -symmetric) or ns 0:95 (if V is
Z2 -symmetric), somewhat redder than the preferred PLANCK 2015 value 0.97 (see
(4.70) and (4.72)).
Another possibility is to perform, as in the large-volume scenarios, the V 1
expansion of the SUGRA limit of string theory [82, 83]. In this case, the volume
becomes large after, not before, inflation. Consider first a single-field
p scenario
where all moduli but V have been stabilized and let 4 D 2=3 ln V be the
canonically normalized p volume field. The total potential is given by six terms: the
03
V˛0 / V 3 D exp. 27=24 / perturbativep O.˛ / contribution in (12.126), the
Vnp / V 3 .ln V/3=2 / 3=2 exp. 27=24 / non-perturbative contribution p in
10=3
(12.126) (but now with V not minimized), the Vgs / V D expŒ.10= 6/4
6
A somewhat hybrid between theory and phenomenology is the no-scale SUGRA model of
[272–275], which deploys the Kähler potential K D 3 ln.KCK /ln.SCS /Cj˚j2 .T CT /n
and the superpotential W / ˚.T const/. Here, S is the dilaton and ˚ is a chiral field with modular
weight n D 1; 2; 3; 4; 5. The two no-scale logarithmic contributions [276–278] and the last term
[279–281] stem from the orbifold compactification of the heterotic string to four dimensions. The
inflaton can be either the imaginary or the real part of T. However, to the best of our knowledge
the superpotential W does not arise naturally in string theory.
perturbative string-loop
p correction (13.25) (again, with variable V), the V2 /
V 2 D exp. 64 / D-term correction (13.4) ((13.5) with n D 3) and two p more
terms we have ignored so far, a correction VF 4 / V 11=3 D expŒ.11= 6/4
from the ˛ 0 3 R4 higher-derivative term in the DpD 10 type-IIB SUGRA action [28]
and the contribution Vhid / V 8=3 D expŒ.8= 6/4 of a possible hidden sector
of charged matter fields [27, 282]. Overall [83],
V. / D V˛0 C Vnp C Vgs C V2 C VF 4 C Vhid

h p 27 10 p
p
D V0 .1 cnp 3=2 /e 2 4 C cgs e 6 4 C c Ve
64
(13.28)
11 i
p p8
CcF 4 e 6 4 C chid e 6 4 ;
where V0 / jW0 j2 ; cnp ; cgs ; c V ; cF 4 and chid are tunable constant coefficients. In the
absence of blow-up modes , the 3=2 term in the first contribution disappears,
while in the presence of one frozen blow-up mode , one can neglect the last term
in (13.28) (or the D-term V2 , with qualitatively similar results). The typical shape
of (13.28) can be appreciated in Fig. 13.6.
Requiring V to have a de Sitter minimum and an inflection point where inflation
lasts about Ne D 60 e-foldings, one can work out the cosmological primordial
spectra. For single-field inflation, the first slow-roll parameter is always much
smaller than , so that ns ' 1 C 2 and tensor modes are negligible. The normal-
ization of the scalar spectrum fixes V0 in (13.28) and, hence, W0 . The magnitude
Fig. 13.6 Log-linear plot of the volume-modulus potential (13.28) [82, 83] in k41 D MPl D
1 units, with V0 D 1010 (for convenience), cnp D 0 (this is the case without frozen blow-up
moduli), cgs 3:485, c V 4:7 104 , cF4 4:282 and chid 0:107. The parameters have
been determined by fixing the position of the inflection point and of the de Sitter minimum at,
respectively, infl D 3:5 and min D 4:2. Inflation takes place at the inflection point at an energy
density Vinfl 3:7 1010 , while the dark-energy era is around min . These values are not realistic
but allow us to see the main qualitative features of the potential. In general, the barrier between
min and the runaway Minkowski minimum at infinity is much smaller compared with Vinfl
of W0 depends on whether the visible matter sector is in communication with the

hidden sources of supersymmetry breaking. If it is (non-sequestered models), the
soft terms are of order of the gravitino mass, while if it is not (sequestered models)
these terms are much smaller than m3=2 and MSUSY D O.TeV/. In particular, there
are two cases [83]:
• High-scale supersymmetry. This is a non-sequestered model where the volume is
of order 100 both during inflation (i.e., at the inflection point) and at late times,
Vinfl 102 V min . Since V is not very large, the validity of the effective-field-
theory approximation is not guaranteed and this case is not under good control.
Still, one can explore its main features in the absence of frozen blow-up moduli.
The flux contribution to the superpotential is W0 D O.105 /, the inflationary
1=4
scale is Vinfl 1014 GeV, the tensor-to-scalar ratio is r 1010 and the spectral
index is ns 0:960.
• Low-scale supersymmetry. In the non-sequestered case, the value of the de Sitter
minimum is V min 1012 – 1015 , while V min 106 – 107 in the sequestered case.
All the other details are the same, with or without the frozen blow-up modulus:
the coefficients cgs and chid must be negative, the flux term is W0 102 – 102 ,
during inflation Vinfl 103 – 105 , the tensor-to-scalar ratio is r 109 and the
scalar index is ns 0:967.
To check the actual level of fine tuning on the parameter range required to obtain
the right amount of e-foldings, one can turn on the blow-up mode and consider
two-field inflation. It turns out that the fine tuning is about 103 .
13.3.3 Fluxless Inflation
Another model of moduli inflation [84, 85] is in the context of the heterotic string
at weak coupling without fluxes (W0 D 0 in the superpotential). The inflaton ˚ is a
generic complex modulus field and its dynamics is described by N D 1 SUGRA in
four dimensions. The Kähler potential is
K D 3 ln.T C T / ln.S C S / C j˚j2 ; (13.29a)
while the superpotential is
W D A1 eS=N1 A2 eS=N2 .˛ C ˇ˚/ı : (13.29b)
The Kähler modulus T is stabilized by the kinetic terms of matter fields, which
induce a mass breaking the flatness of the T direction. The inflationary mechanism
and gaugino condensates can trap also T. We do not write these non-perturbative
contributions here and assume that T has been frozen already, or can be frozen
in a more complicated scenario. The axio-dilaton S has a double exponential
superpotential coming from the gaugino condensation of two hidden SUGRA

sectors with gauge groups SU.N1 / and SU.N2 /. The constants A1;2 D O.1/ and
0 < ı . O.1/ depend on N1;2 and on the string cut-off scale. The term ˛ C ˇ˚,
with ˛ D O.1/ D ˇ, is the mass of a non-singlet gauge field, which depends on
the vacuum expectation value of moduli coupled with such field. In this way, W
acquires a dependence from the inflaton.
The field ˚ starts near a saddle point (V;˚ ˚ D 0) where the slow-roll conditions
hold, then it rolls down a local minimum. This configuration guarantees that the
model is of eternal inflation. This detail is important inasmuch as it solves a problem
typical of models where inflation occurs before stabilizing the dilaton [84, 269]:
the barrier between the KLT vacuum and the Minkowski vacuum at ReS ! 1
(Fig. 13.2) is several orders of magnitude smaller than the Planck-scale energy of
generic initial conditions set at some time ti after the big bang. An inflaton starting
from ˚.ti / D O.mPl / corresponds to a dilaton ReS starting at about the same energy,
which overshoots the KLT minimum and rolls over the barrier. This observation is
known as the runaway dilaton problem or, more generally, overshooting problem.7
Eternal inflation sets the initial conditions near a saddle point of the total effective
potential V.S; ˚/. Consequently, S does not acquire enough kinetic energy to climb
the barrier and it gets trapped in a minimum away from the asymptotic vacuum at
jSj ! 1. Thus is the runaway dilaton problem avoided. Without fine tuning, the
scalar spectrum is compatible with observations. For 30 < Ne < 60,
0:97 < ns . 0:98 : (13.30)
The running of the scalar index is negligible, ˛s < 105 . The tensor spectrum is too
small to be observed.
13.4 Axion Inflation
In this section, we present some ways to employ SUGRA and string axions in
cosmology. After checking on models of racetrack potentials in Sect. 13.4.1, we will
see cases of axion inflation capable of reproducing, under suitable approximations,
the single-field cosine potential (5.90) of natural inflation. To generate viable
perturbation spectra, the axion decay constant f must be greater than the Planck
mass (Sect. 5.9.1) but f < MPl in superstring theory [284]. How do strings face this
interesting challenge?
7
We already had occasion to comment on the necessity of stabilizing the dilaton, or any other
non-minimally coupled scalar, at early times to respect stringent “fifth-force” constraints on the
variation of Newton’s coupling [283] (Sects. 7.4.5 and 12.3.5.5).
13.4 Axion Inflation 729
13.4.1 Racetrack Axion Inflation
The first model of axionic inflation we examine [86, 87] is based on a minimal
modification of the KLT flux-compactification scenario of Sect. 12.3.9.2 [3]. The
first difference is that the superpotential is not (12.111) but of racetrack type
(12.110) in the Kähler modulus,
W D W0 C Ae˛% C Beˇ% : (13.31)
The constant term W0 represents the contribution of previously-stabilized moduli,

including the dilaton. Second, the axion D Im% is not assumed to be stabilized
and, in fact, it is identified with the inflaton. The reason for this choice is that the
Kähler potential (12.99) does not depend on , so that the total potential V.; / may
have flat regions in the direction. Eventually, this helps to address the -problem.
The details of the stabilization mechanism, absent in old models of moduli inflation,
mitigate the steepness of the racetrack superpotential to acceptable levels.
The potential V.; / can be readily written down; we leave this task to the
reader, plugging (13.31) in (12.112). It has several positive or negative minima
(Fig. 13.7). The kinetic term for the fields can be obtained from the kinetic matrix
in (5.220a).
As in the fluxless model of Sect. 13.3.3, eternal inflation avoids the overshooting
problem arising when fields with large kinetic energy reach the asymptotic limit
Re% ! C1, where space rapidly decompactifies. The number of regions with
initial conditions .ti / mPl , are replenished near the saddle point V; D 0 by
the mechanism of eternal reproduction and slow roll is then guaranteed. Inflation
takes place near the saddle point, away from the KLT minimum. Since the effective
dynamics of the system is well described by a field theory on a curved background,
all the inflationary observables are calculated as in Chap. 5. For Ne > 40 e-foldings,
the scalar spectral index is
ns . 0:97 : (13.32)
The inflationary scale V 1=4 1014 GeV is too low to produce an observable tensor
spectrum. Cosmic strings are not produced, either; see, in contrast, the case (13.63)
of warped D-brane inflation.
Other criteria that successful inflation should satisfy are a low amplitude for
density perturbations, ı= 105 , and the observed value of the vacuum,
10123 m4Pl . Both criteria are met by a fine tuning of the parameters of the
potential of one part over 103 , provided the ratio ˛=ˇ is irrational [86].
An explicit model on the orientifold P4Œ1;1;1;6;9 has a similar fine tuning, of
order of percent [87]. On this space, there are only two Kähler moduli, whose
superpotential is
W D W0 C Ae˛%1 C Beˇ%2 : (13.33)

Fig. 13.7 Top: the total potential V.; / for the axionic inflaton model of [86], in MPl D 1 units
and for the parameter choice of [86]. The potential is periodic and nearly flat in the D Im%
(inflaton) direction, while it is steep along the D Re% direction. Bottom: a zooming in of the
potential at a saddle point where inflation occurs; the inflaton rolls down one of the two minima
The dilaton and the complex-structure moduli have been frozen by fluxes into the
constant contribution W0 . A region of the potential V.1 ; 1 ; 2 ; 2 / is shown in
Fig. 13.8.
13.4.2 Axion Valley
As a further example of how the inflaton potential changes when a different cascade
of stabilizations is enforced, we mention the axion valley model of [88, 89], an
interesting and very simple way to obtain natural inflation (Sect. 5.5.2) in N D 1
supergravity. Whether this model can be derived from the low-energy limit of string
theory is still unclear [89].
Fig. 13.8 The total potential V.1;min ; 1 ; 2;min ; 2 / for the axionic-inflaton model of [87], in
MPl D 1 units and for the parameter choice of [87]. The size moduli 1;2 D Re%1;2 have been
fixed to their minimum. The role of the inflaton is played by the flattest direction, in this case 1
Here, the dilaton and all complex-structure and Kähler moduli are assumed to
be stabilized by the KLT mechanism. As in the fluxless model (13.29), we have a
complex modulus ˚, this time with a Kähler potential symmetric under shifts and a
generic non-perturbative superpotential:
K D 14 .˚ C ˚ /2 ; W D W0 C A e˛˚ ; (13.34)
where W0 is real. The superpotential can be generalized to a racetrack type in order

to keep multi-instanton corrections under control [89] but for illustrative purposes
(13.34) will suffice. Splitting the scalar as ˚ D C i and noting that D˚ W D
@˚ W C W@˚ K D W0 C A. ˛/ e˛.Ci / , G˚ ˚ D @2 K=.@˚@˚ / D 1=2 D

.G ˚ ˚ /1 and jWj2 D W02 C A2 e2˛ C 2W0 Ae˛ cos.˛ /, the F-term (12.105)
reads
V.; / D eK .2jD˚ Wj2 3jWj2 / D V1 ./ C V2 ./ cos.˛ / ; (13.35a)
where
2
V1 ./ D W02 .22 3/ e C A2 Œ2. ˛/2 3 e.2˛/ ; (13.35b)
.˛/
V2 ./ D 2W0 AŒ3 2. ˛/ e : (13.35c)
Fig. 13.9 The axion valley (13.35) [88] in MPl D 1 units, with W0 D 103 , A D 1 and
˛ D 0:05. The potential is periodic along the direction
The KLT uplifting mechanism adds a positive constant contribution V0 to (13.35),

so that the minima Vmin D V.min ; min / C V0 of the total potential are positive and
close to zero. Since it is possible to obtain a potential steep in and nearly flat in ,
the latter plays the role of the inflaton (Fig. 13.9).
Both the axion-valley potential in Fig. 13.9 and the axionic model in Fig. 13.7
are steep in the direction and nearly flat and periodic in the axion (the inflaton)
direction, while the KLT model in Fig. 13.2a is relatively steep both in the radion
(the inflaton) and the axion direction. Despite the similarity, the axion-valley
potential (13.35) is very different from the one in Fig. 13.7 inasmuch as it has a
sharp minimum in the direction, while there is no valley in the axionic-moduli
models.
The cosine potential (5.90) of natural inflation is obtained by fixing the param-
eters V0 , W0 , A and ˛ so that V1 .min / C V0 D V2 .min / D 4 =2 and ˛ D 1=f .
A normalization of order of the GUT scale and values above the PLANCK 2015
constraint f & 6:9 MPl [285] can be easily obtained in the context of supergravity.
13.4.3 N-flation
Another model populated by axions is N-flation [89–105]. We have already seen,

both in the axion-valley example and in the other scenarios of moduli inflation,
how a cosine dependence on axion moduli arises in supergravity. One can take N
axions n each with a periodic potential similar to the potential (5.90) of the pseudo-
Nambu–Goldstone boson in natural inflaton [90]:
X
N
Vn . n / D 4n cos
n
V. 1 ; : : : ; N/ D Vn . n / ; : (13.36)
nD1
fn
Instanton corrections proportional to cos.2 n =fn / and cos. n =fn / cos. n0 =fn0 / are
dropped from (13.36) (we will see such corrections more in detail in Sect. 13.4.6).
For n =fn , we have Vn . n / ' m2n n2 =2, where mn D 2n =fn . In the typical string
landscape, no one of the individual scalars n rolls slowly for P enough e-foldings
but, taken collectively, the radial degree of freedom 2 :D 2
n n behaves as an
inflaton. If all masses are equal, mn D m, then V. 1 ; : : : ; N / ' m2 2 =2 D V. /
and the predictions of N-flation are very close to standard chaotic inflation with a
quadratic potential [90]. The number of e-foldings is proportional to the number
of axions. Take an initial condition n;ip D ˛MPl and fn D f for all n, where
p
the constant ˛ D ˛ 0 3=2 = V6 D = V 6 f =MPl . O.1/ is determined by
the string compactification and it is bounded from above by the decay constant
f . The initial condition for the inflaton is i2 D N˛ 2 MPl2 . From (5.85e), one has
Ne ' i2 =.4MPl2 / D N˛ 2 =4.
Contrary to its phenomenological cousin (5.87), N-flation is not fine tuned
because radiative corrections are under control. Cross-couplings between the axions,
coming from quantum string corrections, are suppressed. An interesting feature is
that the maximum number of e-folds is limited by N and the number of complex-
structure moduli in the Calabi–Yau compactification. In general, both this upper
bound and N are large enough to fuel a long inflationary era.
For different masses mn , the analysis of the dynamics is more complicated and
shows that N-flation usually produces a redder scalar spectrum [91, 93]. Nearly
scale-invariant cosmological scalar perturbations are compatible with observations
only if N & O.103 / (easily realized in string theory) and the axion mass spectrum
is densely packed [92]. Parametrizing this spectrum as m2n D m2 expŒ.n 1/=,
one has & 300; this restriction is confirmed [97] in a more sophisticated
parametrization of mn [91].
Inflationary observables do not deviate from the single-field case and are,
therefore, compatible with the data. The first and second slow-roll parameters are
of the same size, D O.102 /, while the spectral index and tensor-to-scalar
ratio are
0:93 < ns 6 0:95 ; r . 103 : (13.37)
r is smaller than the current bounds (4.71), (4.73) and (4.74) but larger than in
other moduli-inflation scenarios and in D-brane inflation. Non-Gaussianity is too
small to be observed if inflation takes place at small field values (quadratic-potential
approximation) [94–96], while it can be much larger near the hilltop [100, 101].
The main assumption underlying N-flation is that radions are stabilized before
inflation. However, we have seen that in most examples of moduli inflation radions
and axions are stabilized more or less simultaneously, since they have masses of the
same order. Near a local minimum, the steepness of the multi-field potential in the
radion and axion directions is similar. This will be true also for the KLMT scenario
(Sect. 13.5) and it implies a fine tuning.
One exception is the axion valley but this model has never been fully derived
from string theory [89]. Reheating in N-flation may also be problematic [98]. The
N-flaton couples with all matter sectors, including the hidden ones originated
by supersymmetry breaking. Unless the couplings are fine tuned, can decay
into hidden-sector particles and give rise to cosmological relics incompatible with
constraints on dark matter; we have mentioned a similar reheating problem for blow-
up and fibre inflation. Other issues are discussed in [104, 196].
Another possibility is to consider the large-volume potential (13.19). In this case,
the axion masses can be hierarchically smaller than the other moduli and, with some
tuning, one can stabilize the volume and the blow-up modes in advance, leaving an
effective potential which differs from (13.36) by a constant additive term and phases:
X
I
/ D V0 C C / :
i i i
V. ci cos.˛ (13.38)
iD1
The parameters in (13.38) are tunable to give viable acceleration with essentially
the same properties as N-flation [102]. The volume is stabilized at V D O.102 /
with a large number of axions N 105 . Then, during inflation H 1015 GeV.
Unfortunately, too many axions correspond to a large density of cycles per unit
volume, which may give rise to a loss of control of the string compactification.
Moving away from the large-volume expansion entails other problems [99] which
are, in fact, only the tip of a grim iceberg. An analysis on simulated landscapes
show that N-flation is under strong theoretical pressure in type-II and heterotic string
compactifications, unless extreme fine tuning is invoked [104].
13.4.4 Aligned and Hierarchical Axion Inflation
From what discussed above, one can evince that N-flation can be safely regarded
as a supergravity model of inflation but its embedding in string theory is subject to
many theoretical constraints [89, 99, 102–105, 120]. However, this is not the end of
the story for string axions. The N-flationary potential (13.36) (sum of single-field
cosines) is a special case of the more general multi-axion profile [106–117]
" !#
X
J X
N
pjn
4j 1 cos
n
V. 1 ; : : : ; N/ D : (13.39)
jD1 nD1
fn
Specific relations among the decay constants pjn =fn < MPl (all smaller than the
Planck mass) give rise to flat directions in the axion space. This can be achieved in
two ways, which we illustrate for two axions and J D N D 2: either by an alignment
mechanism such that jp11 =p12 p21 =p22 j 1 [106] (for perfect alignment, the
inflaton is p11 1 C p12 2 ) or by establishing a hierarchy of the form p12 D 0,

p11 p22 [107, 111]. Integrating out the most massive modes, one obtains the
effective single-field inflationary potential (5.90) with an effective super-Planckian
decay constant feff & MPl and the usual properties of natural inflation (almost scale
invariance, small but non-negligible tensor-to-scalar ratio r, small non-Gaussianity,
and so on).
The potential (13.39) can be embedded in four-dimensional supergravity
and in type-IIB string theory compactified on a Calabi–Yau space with fluxes
[107, 111–115], as well as in type-IIA theory [119]. The axions n in (13.39) are
given by linear combinations of the Kähler-moduli axions i , while the coefficients
j depend on the expectation values of the size moduli i . Just like for N-flation, the
way in which size moduli are stabilized (with the KLT mechanism or with the large-
volume scheme) and the region where inflation occurs determine the level of tuning
of the parameters (of order of 104 – 102 ) required to suppress unwanted features
such as the large non-Gaussianities typical of multi-field and axionic models [286].
However, the actual moduli space in the landscape giving rise to viable inflation is
likely to be quite restricted, at least in type-II theory [104, 105, 116, 119, 120]. This
result does not rule out the alignment and hierarchical mechanisms but it limits their
scope to certain types of compactifications.
13.4.5 Monodromy Inflation
13.4.5.1 D-Term Monodromy
A careful inspection of flux compactifications suggests a way to modify the axion

potential to obtain large-field inflation. We saw from (12.89) that the axions B;i2 ,
C;i2 and i4 arise in type-IIB Calabi–Yau compactifications from the integration of,
.C/
respectively, the forms B2 , C2 and C4 over the 2- and 4-cycles of the Calabi–Yau
space (recall that the indices i2 and i4 , omitted from now on, run over the cycles). If
we also wrap branes around these cycles, one discovers that, when moving around
a cycle, the brane energy increases. This phenomenon is called a monodromy.
We omit the details of the model, which involve many extended sources and
stabilization ingredients. We shall just say that the effect of the wrapped brane is to
break the periodicity (i.e., the shift symmetry) of the axionic potential.
p For instance,
a D5-brane wrapped around a 2-cycle 2 of length L.x/ D ˛ 0 .x/ generates
q crit
a potential for B proportional to V. B / / .gs ˛ 0 2 /1 B2 C crit

2
B for B
crit . This happens because the DBI action (12.42) is evaluated over the wrapping
configuration and the flux of the Kalb–Ramond field Bab through 2 is nothing but
B . The linear potential [122] dominates over the periodic contributions. Similarly,
a wrapped NS5-brane gives a dominant linear potential to C , for C larger than
the critical value crit (assuming the length of the cycle to be the same). If L is of
p
the same order of the compactification scale R D ˛ 0 l.x/, the relations between
the axions and the canonically normalized inflaton are MPl B = crit and
MPl gs C = crit , so that
V. / : (13.40)
The potential can be further flattened by the back-reaction of the inflaton on the
geometry. If the cycle length L is much smaller than R (for instance, when the 2-
3=2 3=2
forms are localized in a throat), then MPl B l1=2 = crit and MPl gs C l1=2 = crit .
By back-reaction of the flux, the size R depends on the axion and increases
5=4
dynamically. In particular, one can argue that l2 / B , so that B and the
inflaton potential becomes [126]
4
V. / B. / 5 : (13.41)
This is only one example of flattening mechanism.

So far in this chapter, we have discussed models in type-IIB string theory and
one heterotic case. Type-IIA moduli inflation [287], based on stabilization scenarios
in the corresponding theory [203, 288, 289], is difficult to achieve and it will not be
discussed here. However, instead of compactifying on a warped Calabi–Yau space
one can decide to consider other internal manifolds with special properties that give
rise to a rather different phenomenology. On a fairly wide class of orbifolds, branes
can wrap around cycles as usual, but the length of the cycle increases while the
brane moves and so is the brane energy. Again we have monodromies, although not
necessarily of axionic type.
For instance, de Sitter vacua in type-IIA flux compactification can be obtained
in one such special orbifold M6 D N3 NQ 3 , the product of two identical Nil 3-
manifolds (also known as twisted tori) [290]. Each twisted torus is parametrized
by, respectively, a triplet of coordinates .y1 ; y2 ; y3 / and .Qy1 ; yQ 2 ; yQ 3 / [121]. The cross
section of N3 along a certain direction y1 is isomorphic to a 2-torus T2 . The
direction y1 is rendered compact by periodically identifying tori at different points
by a discrete group. The direction y2 is a 1-cycle of M6 whose length depends on y1 .
Having this background, we can wrap a D4-brane on the above 1-cycle and
note that, while the brane moves along y1 , the cycle becomes longer and the
brane tension increases [121]. After compactifying the DBI 4-brane action on M6
and expanding it to lowest order in the derivatives, one identifies the canonically
normalized inflaton .x/ D Œ y1 .x/ with a function of y1 . Once all the moduli but
are stabilized, the resulting total Lagrangian is L ' R=.242 / .r /2 =2 V. /,
where the effective potential V. / has a global minimum at D 0 and grows as
2 near the origin and as 2=3 at large . The transition scale crit depends on
the parameters of the string compactification and it is sub-Planckian. However, the
region crit near the origin cannot sustain chaotic inflation because the field
can attain trans-Planckian values only by violating the consistency requirement that
V 1=4 does not exceed the energy used to stabilize the moduli.8
In the other asymptotic region crit , one realizes the large-field model of
inflation (5.81) with n D 2=3, while wrapping the D4-brane around a different
direction results in an effective potential with n D 2=5 [121]. Yet different brane
wrappings induce a linear potential for the C3 axion [122], just like in type-IIB
theory.
An n D 2=3 or n D 4=3 potential arises also by flattening a quadratic potential
in the case where inflation (driven by the Kalb–Ramond axion B ) happens during,
not after, moduli stabilization [133]. The vacuum expectation value of the moduli
changes adiabatically while deforming the inflationary potential. The cases n D 2
and n D 3 are also obtained, from the flattening of a quartic potential [133] or when
the candidate inflaton is a complex-structure modulus [131, 133–135].
Therefore, in general, monodromy inflation [121–143] is a class of large-field
models giving rise to a monomial potential (5.81):
4 2 2 4 4
V. / D V0 n
C cos C ' V0 n
; nD ; ; ; 1; :
2 f 5 3 5 3
(13.42)
All the constants V0 , n, and f are determined by the parameters of the string
compactification (including the vacuum expectation values of the stabilized moduli)
and are such that the oscillatory term does not spoil the slow-roll approximation.
That is because the instantonic effects giving rise to the periodic potential are
exponentially suppressed with respect to the leading monomial term. In practice,
aligned and hierarchical axion inflation (Sect. 13.4.4) can be regarded as the multi-
field generalization of monodromy inflation [107, 108], since the periodicity of the
inflaton potential is broken by the mixing with heavier axions.
In the region crit , all these models are self-consistent inasmuch as moduli,
the inflationary trajectory and the effective background are not destabilized by the
evolution of the inflaton field, by the monodromy correction V. / to the total moduli
potential or by the back-reaction of the wrapped brane.9 The inflationary potential
is also robust against ˛ 0 -corrections in the brane curvature and against string loop
contributions, which are highly damped.
8
In other words, V must be smaller than the potential barrier separating the system from the
runaway vacuum of weak coupling and of the decompactification limit.
9
By back-reaction, one means the effect of the relative size of the brane core with respect to
the curvature radius of the internal manifold. For instance, in the n D 2=3 model on twisted
tori discussed here, the size rbrane of the D4-brane core can be determined by evaluating the
gravitational potential / rbrane =jEr j, where Er 2 f.y1 ; Qy1 ; Qy2 /g is a vector in the sub-space spanned
by three of the six internal directions (the other three are dimensionally reduced). jErj is the
distance from the intersection point with the D4-brane. The constant rbrane is proportional to the
7-dimensional Newton constant; if it is smaller than the other sizes of M6 , then the brane is a good
probe of the geometry [121].
Brane (or D-term) monodromy faces a number of issues. In the n D 2=3 type-
IIA model, large corrections to the slow-roll parameters can come from an arbitrary
orientation of the D4-brane from other extended sources. A symmetric orientation
solves the problem but it seems difficult, if not impossible, to construct a global
embedding with this feature. The model does not admit anti-de Sitter or Minkowski
vacua and the spectrum of allowed de Sitter vacua has a lower bound higher than
the observed value obs [140].
In the n D 1 type-IIB model, stabilization of the Kähler moduli steepens the
potential of B (roughly speaking, non-perturbative superpotentials such as (13.18)
create an -problem for B ) but not that of C . Thus, C can play the role of the
inflaton more likely than B . However, the NS5-brane must be paired (to ensure
tadpole cancellation) with a distant NS5-brane and their interaction is logarithmic
with the distance. This interaction term is not negligible with respect with V. / and
it can destabilize the inflationary dynamics.
13.4.5.2 F-Term Monodromy
An inflationary potential generated by an F-term can bypass these and other issues
[129–132, 134, 135, 137, 138]. In this case, supersymmetry is broken spontaneously
and there is an effective SUGRA description, not available in previous models where
supersymmetry is broken explicitly at the string scale. Using D7-branes instead of
5-branes in type-IIB theory yields a model with an n D 2 quadratic inflationary
potential, where the inflaton is an axion among the complex-structure moduli [131,
132]. The n D 2 case also emerges from the F-term scenarios of [129, 130]. One
identifies the axio-inflaton with specific integrals of a p-form over a p-cycle, called
massive Wilson lines. Then, the effective potential V. / behaves as a power law n
asymptotically and interpolates between n D 2=3 and n D 2.
Other mechanisms involve background fluxes or torsion and produce polynomial
potentials with n > 2. In a type-IIB toy model, one is able to move away from
the polynomial trend of F-term axion monodromy and to obtain, in the large-field
regime, the Starobinsky potential (5.234), where M depends on the parameters of the
flux compactification [137]. In type-IIA theory, it is possible to obtain an inflaton
potential interpolating between n D 1 and n D 2 [139, 141].
13.4.5.3 Inflationary Predictions
The parameter space of monodromy models is the set of vacuum expectation values
of the moduli, in turn dependent on brane charges and quantized fluxes. Within
the theoretically allowed region in this parameter space, there is a window for
generating observationally acceptable observables. For instance, in the type-IIA
n D 2=3 case the fine tuning on the parameters is O.102 / andRcorresponds to an
O.100/ anisotropy in the radii of M6 and to a reasonable kF / M6 F6 D O.100/
number of flux units of the R-R 6-form F6 . Type-IIB brane models have the same
amount of tuning and so do F-term monodromy scenarios; the issue of tuning can
be explored also by counting viable vacua in the landscape [135].
Within these windows of viability, one has almost scale-invariant spectra and
a non-negligible tensor-to-scalar ratio. As we saw in Sect. 5.5.1.2, the slow-roll
parameters (5.85a) and (5.85b) are small as long as mPl , a condition respected
by monodromy scenarios. Equation (5.197) with Nk n D O.1/ are
nC2 4n
ns ' 1 ; r' : (13.43)
2Nk Nk
For the values of n given in (13.42) and for Nk D 60,
0:97 < ns 6 0:98 ; 0:03 < r < 0:09 ; (13.44)
compatible with PLANCK 2015 data [285]. The linear case n D 1 corresponds to
.ns ; r/ .0:975; 0:07/. The sub-dominant periodic part of the potential (13.42) can
create an oscillatory modulation of the perturbative correlation functions (spectrum,
bispectrum, trispectrum, and so on) [123–125, 136, 286]. These oscillations are
strongly constrained by observations and are virtually undetectable in the CMB
temperature spectrum, but they are also responsible for non-Gaussianities of
“resonant” type, encoded in a non-linear parameter fNLres
10. The present sensitivity
res
on fNL does not allow to check this prediction yet. With an eye to the future, we
also mention that a coupling between the axion and gauge matter breaks the shift
symmetry spontaneously and can generate a TB polarization pattern in the CMB,
as explained in Sect. 5.9.2. In turn, observations of non-Gaussian signals constrain
the magnitude of such coupling and of the axion decay constant f gs MPl =V 1=3
in (13.42) [285] (the expression for f is more complicated in the presence of many
moduli).
13.4.5.4 Quintessence
Finally, axion monodromy has been also proposed as a source of quintessence

[68, 69]. Placing 5-branes in a Klebanov–Strassler throat not containing the Stan-
dard Model breaks the shift symmetry, while the warp factor at the bottom of the
throat provides the 10120 damping of the potential necessary for dark energy.
There is no tracking behaviour for the linear potential and, hence, the coincidence
problem is not solved. To get viable dark energy (but indistinguishable from the
CDM model [69]), one must fine tune the initial conditions, a plague of the
quintessence models of Sect. 7.3 that might find an accommodation in some corners
of the string landscape. More general axion-driven realizations of quintessence in
string theory have a similar issue [67].
13.4.6 Problems with Axion Inflation and Ways Out
Let us now come back to the question, raised at the beginning of this section, of how
string theory can manage to get large axion decay constants. The bottom line is that
it cannot.
For each of the axion-inflation models listed here, we have mentioned the
presence of a variety of issues and their proposed cures. Unfortunately, these
problems might be not point-wise occurrences due to the technical limitations of
individual models, but a reflection of an obstacle endemic to the string landscape:
large axion decay constants are in contrast with the weak gravity conjecture
(Sect. 13.2.2).
We know from Sect. 12.2.8 that, given a R-R . p C 1/-form CpC1 in ten
dimensions, we have a p-brane with charge qp (this is the dimensionless version
of Qp ). For p D 0 in type-IIB theory, the D.1/-brane is an instanton with mass
m > 0 and action Sinst m=MPl . This instanton couples with the 0-form axion
D =f by a positive U.1/ coupling q0 MPl =f which breaks the shift symmetry
of the four-dimensional potential:
X
l
V. / ' 4l elSinst cos : (13.45)
l
f
The weak gravity conjecture states that m=q0 6 MPl , so that Sinst m=MPl 6 q0
MPl =f . Therefore, MPl =f cannot be too small (a condition for successful inflation) lest
the higher-order l-terms in (13.45) become more important and the effective theory
fail, at energies much lower than desired by theoretical consistency [104, 246]. The
extension of the conjecture to many copies of U.1/ [248, 249] leads to a similar
hindrance for multi-axion models such as N-flation, aligned and hierarchical axion
inflation [104, 105, 116, 119, 120, 291]. In these cases, there is an instanton for
each axion and the axion decay constants are individually bounded by the Planck
mass, fn < MPl for all n. If the weak gravity conjecture holds, the constraint
feff < MPl obtained in these models for a collective, effective decay constant feff must
eventually receive yet-ignored corrections that bring down the collective bound to
the individual one. Performing such a check would contribute to verify the validity
of the conjecture.
A possible loophole in the above arguments would open up if only the mild
version of the conjecture held. Then, the constraints would apply only to the particle
minimizing jm=qj, while another axion would be free to drive inflation. Consider the
case of two axions 1;2 such that 2 couples with an instanton with mass m2 > m1
and charge q2 D kq1 > q1 , where k is a positive integer. If the lighter particle does
not obey the weak-gravity bound but the heavier one does, this is a configuration
with k stable states and we are in the presence of the mild version of the weak
gravity conjecture. For a potential given by the sum of N copies of (13.45) and
taking the lowest-order contribution l D 1, we have f2 D f1 =k < f1 and

4 m1 =MPl 1 k 2
V. 1 ; 2 / ' 1 e cos C 42 em2 =MPl cos : (13.46)
f1 f1
The heavier particle has a smaller decay constant and cannot enjoy the super-
Planckian enhancement in axion inflation. Its contribution to (13.46) is suppressed
by an exponential factor exp.m2 =MPl / exp.m1 =MPl / and the dominant
contribution is by 1 , which can play the role of the inflation [105].
However, there are no counter-examples to the strong weak gravity conjecture
in string theory and it is non-trivial to find realistic embeddings of the suppres-
sion mechanism leading to (13.46). Attempts to realize it in aligned/hierarchical
models [117, 118] face a series of obstacles [120] which, nevertheless, might be
circumvented; see the type-IIB construction of [118], where a potential different
from (13.39) is obtained.
To summarize, in Sect. 13.4 we have examined cosmological string scenarios
aiming to recover the periodic potential (5.90) in some limit and to explain inflation
with such potential. Starting from a population of axions, the continuous shift
symmetry (12.90) is explicitly broken by non-perturbative effects and many string
embeddings of the potential can be found in the literature. Examples of this class
of models are N-flation, aligned and hierarchical inflation. The majority of these
cases are constrained by the weak gravity conjecture and either fine tuning or
special hand-picked Calabi–Yau spaces are the price to pay in order to attain viable
inflation. Some models [112] avoid this theoretical constraint but not other issues
[116]. Other multi-axion models under construction might survive the screening of
the mild version of the weak gravity conjecture [105, 117, 118, 292] but not easily
[120]. We also mention that one can give up the alignment mechanism and consider
other types of mixing in type-II compactifications with intersecting D-branes, giving
rise to the natural-inflation potential (5.90) for a certain linear combination of
axions [293, 294]. The resulting effective decay constant f is different from the feff
of previous models and it may avoid the weak-gravity bound. A particular case
of natural inflation [295] can evade the weak gravity conjecture but not all its
versions [291].
Finally, a model-independent study of configurations with N axions shows that, at
large N, the effective enhanced decay constant feff converges to a finite value (hence
the scenario is consistent with the weak gravity conjecture) only if the number of
instantons in the inflationary potential grows fast with N, more than quadratically
and perhaps exponentially [296]. Since multi-axion large-field models such as N-
flation and aligned/hierarchical inflation predict, by their assumptions, a different
(much slower) scaling of feff .N/, one can reach two mutually exclusive conclusions:
either the weak gravity principle holds and these models are inconsistent at large
N, or we are witnessing a violation of the weak gravity principle. The latter
interpretation is suggested by the lack of such a huge number of unsuppressed
instantons in controlled string compactifications.
All in all, the debate on axion inflation with periodic potentials and the weak
gravity principle is fairly recent and wide open.
In axion monodromy inflation, the discrete shift symmetry is further broken,
explicitly (with branes) or spontaneously (by F-terms), by perturbative mechanisms
and inflation is driven by the symmetry breaking term, which dominates over the
sinusoidal term. The instantonic corrections to the periodic part of the potential
are therefore innocuous. However, the weak gravity principle may still pose a
problem, since it implies the existence of a UV cut-off smaller than the Planck
scale.
P The problem of large-field scenarios is to control the perturbative corrections
n . =m Pl / n
to the inflationary potential. The shift symmetry of axions protects
their potential from such corrections but this symmetry is broken by stringy non-
perturbative effects. In turn, these effects strongly depend on the details of the UV
model, so that the challenge is to balance realistic UV physics on one hand and
good inflation on the other hand. The weak gravity principle can constrain the UV
cut-off in axion monodromy models and, from that, the inflationary parameter space.
Large-field displacements are allowed but there is a lower bound for the axion decay
constant [142].
So far, axion monodromy inflation seems safe from weak-gravity assaults.
However, we can appreciate other difficulties in obtaining a viable UV model of
monodromy inflation by recalling that even F-term monodromy, proposed to bypass
the problems of brane monodromy, has its own issues. Single-field inflation from
F-term axion monodromy in type-IIB orientifold compactifications is possible but
subject to several theoretical constraints [134, 135, 138]. In F-term models, the
inflaton does not appear in the Kähler potential and there is no danger to have
an -problem, provided the inflaton be a linear combination of axions only. Also,
identifying the inflaton with the universal axion C0 [130] may be problematic
due to the guaranteed presence of other light axions, which cannot be stabilized
beforehand. Moreover, the fine tuning of the parameters may be severe or even
impossible in certain cases, such as in the weak-coupling regime gs 1 of
type-IIB theory on an orientifold. The conclusion from various negative and
positive examples is that not all regions in the string landscape can support F-term
monodromy inflation and care must be exercised in model building and in the choice
of the Calabi–Yau space.
13.5 Slow-Roll D-Brane Inflation
13.5.1 Early Brane-Inflation Models
In old Kaluza–Klein scenarios, compactification of a higher-dimensional space

down to 3 C 1 directions is the most direct way to obtain a world with the observed
number of dimensions. Essentially the same could be told about string theory but
with a notable difference: the extra dimensions can be much larger than the Planck
scale [237–239] and the observer may live in a D3-brane.
13.5 Slow-Roll D-Brane Inflation 743
An interesting situation is that of a stack of D-branes at a non-minimal energy

configuration, that is to say, the branes are displaced from one another with respect
to the transverse extra dimensions [297]. The displacement is governed by the scalar
mode ' of an open string connecting two separate branes. For an observer in a brane,
this field plays the role of the inflaton. As the branes approach one another but while
still being at a distance, the weakly-coupled scalar rolls slowly down its potential
and causes the non-compact spatial directions in the branes to expand exponentially.
At some critical inter-brane distance, the potential V.'/ becomes too steep and
inflation ends. At small distances compared to the string scale, the tachyonic mode
switches on and triggers the brane decay. Reheating consists in the collision of the
branes and the consequent release of energy as radiation.
Consider one extra direction and assume, for the moment, that all moduli but
the brane separation have been stabilized. For Hubble patches H 1 R much
larger than the size R of the extra dimension, the dynamics on one brane is
described by the first Friedmann equation H 2 / eff , where the effective four-
dimensional energy density eff ' T3 'P 2 C V.'/ (from the DBI action; T3 is the
3-brane tension) encodes the kinetic energy of the relative brane motion and the
inter-brane interaction potential V.'/. At zero separation, the potential vanishes,
V.0/ D 0. At distances ' ¤ 0, the coupling between modes on different branes is
suppressed at least exponentially, so that there is a short-range attractive contribution
T3 Œ1 exp.j'='0 j/ (or a more strongly damped profile) to V.'/. This adds
to other terms including a potential exp.mj'j/j'j2NDp which describes the
exchange of massive (or massless, when m D 0) bulk modes, where NDp is the
number of branes. Due to the strong fall-off of brane interactions, no mass term for
' is generated in the potential and a slow-roll regime is easily achieved.
From the total V.'/, one can work out the details of the inflationary era. Brane–
anti-brane interactions are under better control than those of branes of the same
orientation and one can realize brane inflation [297] in string theory [298–303]. In
this case, the effective potential is of the form [299–301]

ˇ
VDp-Dp .'/ D V0 1 : (13.47)
' d? 2
. pC1/
and ˇ / msd? 2 are dimensionful constants and d? D
pC1
where V0 / ms D ls
9p is the number of large dimensions transverse to the Dp-branes. One has d? D 6
if all extra dimensions are large; if d? D 2, the potential becomes logarithmic. For
1=6
d? D 6 and assuming the branes are at the maximal possible distance R ' V6 (the
size of the compact space), inflation lasts about Ne 80 e-foldings and generates
a scalar spectrum with index ns 1 0:97. The typical compactification scale is
about R1 1012 GeV, while the string scale is at ms 1015 – 1016 GeV. However,
1=6
non-extremal but more natural cases with R V6 do not give enough e-folds
6 1
Ne / R =V6 / jj and the perturbation spectrum strongly deviates from scale
invariance. This is nothing but the -problem.
Other configurations with branes intersecting at an angle can solve this

problem and appear in many constructions of the Standard Model (Sect. 12.3.9).
The coefficients V0 and ˇ in (13.47) depend on and the brane–anti-brane case
corresponds to D . More general potentials V D V0 .1 ˇ' n / are also possible.
Apart from the specific form of the potential and a limited range of realistic
couplings, brane inflation has other characteristic features that distinguish it from a
standard inflationary mechanism. For instance, cosmic strings are produced at the
end of acceleration; their fate depends on the details of the model [304–306].
These possibilities were recognized before the advent of KLT scenarios
[297–299, 304, 305] and promptly embedded in them soon afterwards. In concrete
compactification schemes of moduli stabilization, pairs of D3-branes and D3-
branes are located at the tip of a deformed Klebanov–Strassler throat, where the
warped geometry naturally leads to acceleration. The Randall–Sundrum braneworld
scenario we will see in Sect. 13.7.1 [307] can nest in the full theory via the KLT
mechanism of moduli stabilization: the braneworld then coincides with a stack of
D3-branes (for the Standard Model) or D3-branes (for supersymmetric extensions
of the Standard Model) at the tip of the throat [6]. Then, the scalar ' corresponds to
the radial direction along the throat.
13.5.2 Warped D-Brane Inflation and KLMT Model
A crucial assumption of brane inflation is that all moduli have been stabilized,
so that V0 and ˇ in (13.47) are constant. If the internal space were not static,
the potential V.'; V6 / / V62 would be too steep along the direction of the
Kähler modulus and brane inflation would end quickly. The challenge of obtaining
viable inflation well illustrates the huge gap between cosmological models where
moduli stabilization is assumed and top-down scenarios where such stabilization
is carried out explicitly. To differentiate modern cosmological models with respect
to early proposals, we call the former warped D-brane inflation. We begin with
models where the slow-roll conditions are satisfied, slow-roll D-brane inflation
[144–167, 193, 194] and their prototype KLMT inflation (pronounced “KKLMMT”
or “KLMT,” from the acronym of the six authors who first studied this scenario
[144]). The KLMT construction is fairly simple: in the KLT setting, one adds a
mobile D3-brane to the Klebanov–Strassler throat which makes the de Sitter vacuum
dynamical. The model must be engineered carefully in order to make the inter-brane
potential flat and the inflaton sufficiently light. First, warping of the background
space flattens the potential much more than brane-inflation models on flat spaces.
Second, gaugino condensation gives a heavy mass to the inflaton and one should
consider stabilization mechanisms alternative to that of Sect. 12.3.9. ˛ 0 -corrections
to the Kähler potential (although not of the form computed in [308]) can stabilize
the Kähler modulus while avoiding this problem. Third, while the anti-brane is
automatically fixed at the tip of the throat by the dynamics, the position of the mobile
brane must be subject to a suitable non-perturbative superpotential. Let us see these

ingredients in detail in type-IIB flux compactification, starting from the background
geometry.
13.5.2.1 Step 1: Klebanov–Strassler Geometry
The Klebanov–Strassler throat in six dimensions can be parametrized by a conical

metric
ds2 D e2!.r/ ds24 C e2!.r/ ds26 ; (13.48)
where the internal line element ds26 depends on the radial coordinate r (spanning
the length of the throat) and on angular coordinates. The conifold singularity
corresponds to r D 0. When deforming the throat into a smooth compact space,
the singularity is replaced by a smooth tip isomorphic to S3 and located at some
r D rN 1. !.r/ increases with r monotonically and tends to a constant at small r.
For r > rN but smaller than the gluing point rmax of the conifold with the rest
of the Calabi–Yau space, the cross section of the throat is the coset space T 1;1 D
ŒSU.2/ ˝ SU.2/=U.1/, with topology S2 S3 , while away from the tip the throat
is isomorphic to AdS5 S2 S3 spacetime with characteristic scale RAdS . More
generically, the cross section of a warped space can be an Einstein manifold X5 .
Along the throat, there is a large gravitational redshift e!.r/ 1 in the warped
metric (13.48). The tip and the gluing point act as, respectively, an IR and a UV
cut-off for AdS5 . For this reason, the tip is sometimes called the infrared end of the
throat, while the gluing point is the ultraviolet one.
The scale RAdS is related to the location of the tip at rN . As a matter of fact, the
deformation of the singular Klebanov–Strassler conifold into a compact space is
achieved thanks to the presence of the background 3-forms F3 and H3 , which carry
the quantized fluxes
Z Z
TQ3 TQ3
F 3 D kF ; H3 D kH : (13.49)
2 S3 2 3
As in (12.87), TQ3 D Œ.2/3 ˛ 0 2 1 , while S3 and 3 are, respectively, the sphere S3
at the tip and its dual 3-cycle; kF ; kH 1 are positive integers. Then, the minimum
warp factor in (13.48) (i.e., the ratio between the tip position and the AdS radius) is
finite [204]:
2
rN 'N 2 4kH
e2!.Nr/ D D p D exp ; (13.50)
RAdS T3 3gs kF
p
where, for later
pconvenience, we have defined 'N :D T3 rN and a dimensionless
coupling :D T3 R2AdS . The radius RAdS D .4gs N˛ 02 /1=4 depends on the number
N D kF kH 1 (13.51)
of D-brane charges of the background. This numerology gets a fresh insight in the
AdS/CFT correspondence and 1 is interpreted as the strong-coupling limit of
the dual CFT (see (12.103) and [196]).
An exact metric describing the whole deformed throat exists but for our purposes
the above rough decomposition into tip, throat and gluing point will be enough.
13.5.2.2 Step 2: D3-Brane
Consider now a D3-brane with tension T3 (see (12.42)) placed in the throat at a point
r D rD3 , with rN < rD3 < rmax . Its action on AdS5 S5 [309, 310] well approximates
the brane dynamics away from tip and gluing point:
R p 2
SDBI D d4 x gLDBI ; f .'/ D ;
'4
(13.52)
p
LDBI D f 1 .'/ 1 C f .'/ g @ '@ ' 1 ;
p
where ' :D T3 rD3 and we have neglected fermionic and gauge fields. Notice the
non-perturbative dependence 1=gs / 1=2 of the action from the string coupling,
when written in terms of rD3 . The strong-coupling regime corresponds to 1,
but (13.52) is valid also Rat weak coupling [311].
p
The action SNDBI D d4 x gLN DBI for a dynamical D3-brane is the same as
(13.52) except for the last term, which is C1:
q
LN DBI D f 1 .'/
Q Q g @ '@
1 C f .'/ Q 'Q C 1 ; (13.53)
p
where 'Q :D T3 rD3 . Expanding this action for a small derivative term, one finds
a quartic potential with a global minimum at the origin, i.e., near the tip of the
Klebanov–Strassler throat:
2 2'Q 4
LN DBI D Q 2 D 2 C OŒ.@'/
C OŒ.@'/ Q 2 : (13.54)
f .'/
Q
This means that any D3-brane placed in the deformed throat, and actually onto
any spot on the Calabi–Yau space [150], will be eventually drawn towards the
region 'Q ! 'N with smaller warp factor, i.e., down the throat. This is an intuitive
justification of an ingredient of the KLT stabilization of Sect. 13.1.1, where a
stack of anti-branes was put in the throat to cancel tadpole anomalies and to break
supersymmetry. Similarly, in the KLMT construction one adds a D3-brane at the tip
r D rN of the Klebanov–Strassler throat, assuming that it was fixed there dynamically.
Then, rD3 D rN and 'Q D '.
N
13.5.2.3 Step 3: D3-Brane
In type-IIB theory, D3-branes are stable and there is no potential term in the DBI
action (13.52). The presence of a D3-brane trapped at the tip, parallel to the D3-
brane, induces an attractive potential V.'/ which pulls the D3-brane towards the
tip. This situation is depicted in Fig. 13.10a.
If we truncate the action (13.52) to second order in the derivatives, we obtain a
canonical kinetic term for the inflaton ':
1
L' ' @ '@ ' VD3-D3 .'/ ; (13.55)
2
where the potential is (13.47) with d? D 6:

'N 4 1 'N 4
VD3-D3 .'/ D 2 2 1 : (13.56)
N '4
Note that the constant term is nothing but LN DBI for a fixed anti-brane position.
To the brane–anti-brane potential (13.56), one should also add the contribution
VCY .'/ from all corrections coming from the specific embedding in a Calabi–Yau
Fig. 13.10 Warped D-brane inflation (a) in the K LM T model [144], (b) with a Z 2 symmetry
[151] and (c) with only D3-branes [150]
space, populated by fluxes, branes of different kind (in particular, D7-branes) at

different positions and more moduli that are stabilized eventually. The total potential
then reads
V.'/ D VD3-D3 .'/ C VCY .'/ : (13.57)
13.5.3 Cosmological KLMT Dynamics
Thanks to (13.50), for certain choices of fluxes V0 can be very small and the potential
extremely flat. The negative sign in (13.56) indicates an attractive force between the
two branes. When the latter are too close and rN rD3 , the potential (13.56) breaks
down; this stage corresponds to reheating.
Assuming that the compactification is stable, viable inflation follows through.
The first slow-roll parameters of Sect. 5.4.2.2 are V ' .8 2 'N 8 =N 2 /' 10 and V '
.20 2 'N 4 =N/' 6 ' 5=.6Nk /, which are both small if ' is sufficiently large.
Since N 1, V V and the tensor spectrum is unobservable. In fact, it is easy to
see that for Nk D 60 e-foldings both the tensor-to-scalar ratio (4.66) and the tensor
index nt D 2 are extremely small,
Pt
rD D O. / 1011 ; (13.58)
Ps
while the inflaton potential at horizon crossing is lower than the GUT scale by a
couple of orders of magnitude, V 1=4 1014 GeV. The scalar spectral index ns D
1 C 2V 6V ' 1 C 2V is well within the observed range,
5
ns ' 1 0:97 (13.59)
3Nk
for 60 e-folds.
The potentials (13.56) (which appears in old brane inflation [299] and in the
KLMT model [144]) and (13.47) with d? D 4 (typical of old brane inflation with
brane at angles [303]) are both compatible with PLANCK 2015 data [285]. However
pleasing this result may be, the real challenge of string cosmology is not to fit data
per se but, rather, to do so in the natural parameter space of the theory. The remainder
of this section is dedicated to this goal.
To ensure that inflation actually takes place, the moduli must be stabilized.
However, the KLT scheme of Sects. 12.3.9 and 13.1.1 receives corrections that
risk to spoil the cosmological scenario just detailed. The Kähler potential (12.99)
is augmented non-trivially by the contribution of the brane moduli,
Q
K.%; / D 3 lnŒ2Re% c. ;
/ D 2 ln V ; (13.60)
where is a multiplet of fields consisting in the position of the D-branes on the

Calabi–Yau space. The superpotential W D W.%/ can be taken for the moment to
be dependent only on the Kähler modulus, as in (12.111) (but it will be generic in
what follows). Then, one can compute the F-term (12.112) with KQ given by (13.60).
To this F-term, one must add the D-term (13.3) with the replacement Re% ! Re%
c. ; /=2. In a neighborhood in the moduli space where c. ; / ' j j2 and
assuming that Im% D 0 and that W.%/ D W./ is real, the total potential reads
2 V0 ./ 1
V.; / D ; V0 ./ D W;2 3W W; C ˇ1 :
. j j2 =2/2 44 2
(13.61)
In the putative de Sitter minimum D min

p, the potential can be expanded to obtain
the effective mass for the inflaton ' :D 3=.2Re%/=4 :

2 2
VCY .'/ D V.min ; '/ ' V0 .min / 1 C 4 j'j2 ; (13.62)
3
which yields m2' D 242 V0 .min /=3 ' 2H 2 and a large slow-roll parameter V '
m2' =.3H 2 / ' 2=3. This relation does not agree with the slow-roll condition (5.59),
m2' H 2 : the -problem of old brane inflation and of F-term SUGRA inflation is
brought to the surface also in slow-roll D-brane inflation. Ultimately, the problem
is due to the fact that the inflaton couples non-minimally with gravity in the total
effective action in four dimension, via a j'j2 R term [144, 312] (the relation m2' D
2H 2 is an indication of this conformal coupling).
13.5.3.1 Gravitino Mass
In the KLMT model, the scale of inflation is related to the gravitino mass by
(12.118). General considerations show that V . .m3=2 MPl /2 , hence H . m3=2
[11]. The same relation holds in monodromy inflation [121]. Thus, to have a large
inflationary scale (and observable tensor modes), one should develop phenomeno-
logical models where supersymmetry is broken at high energies. This is not an
issue by itself but it does not produce interesting phenomenology at accelerators,
apart in specific scenarios (for instance, the split supersymmetry mentioned later
in Chap. 14). Conversely, for m3=2 D O.TeV/ there is the problem of obtaining
long slow-roll inflation with a very low scale. Although one can engineer low-scale
inflation [313–315], its typical parameter space cannot be reproduced in a string-
theory embedding. The gravitino-mass problem can be circumvented in racetrack
inflation (13.31) [11, 86, 87], where one can have a low gravitino mass and a high
inflationary scale at the same time; a discussion of these issues in large-volume
inflation can be found in [82].
13.5.3.2 Cosmic Strings
At the end of inflation, lower-dimensional wrapped branes and metastable funda-

mental and Dirichlet strings are produced during brane collision [199]. These long
strings interact in a complex network of cosmological size and they eventually
decay into shorter strings and gravitational radiation [305, 306, 316–321]. The
evolution and characteristic tension of cosmic strings produced inside the throat
are constrained by CMB and large-scale-structure observations [322, 323]. No
signature of a string network has been detected so far. The prediction
G 1010 – 109 (13.63)
of the KLMT model [317] is compatible with the experimental bound
G < 107 – 106 (13.64)
and within the detection threshold of future experiments dedicated to gravitational

waves and B-mode polarization [154, 199, 324, 325] (see Chap. 14).
13.5.4 Refinements and Related Models
So far, the form of the total potential (13.57) has been determined only for the
SUGRA terms (13.62). Taking into account more curvature, flux and brane effects,
in the conifold approximation of the warped throat, the potential for the D3-brane
radial position is [155, 157, 160, 166]
3 5
p 5
VCY .'/ D V0 C c1 ' C c 3 ' 2 C c2 ' 2 C c 5 ' 2 C b' 27 2
C ::: ; (13.65)
2 2
where the coefficients ci and b depend on the angular position on the conifold
and the ellipsis denotes higher-order terms negligible in a small-' expansion.
The -problem arises from the supergravity coupling of the inflaton to the four-
dimensional curvature. In fact, the leading contribution to c2 is the O.H 2 / mass
of the KLMT model from curvature effects, which produce terms p / ' n with
n D 2; 3; 7=2; 4; : : : . The last term in (13.65), with irrational exponent 2 75=2
2:79, is a flux-related contribution stemming from the compactification of the non-
compact throat. Terms of order less than 2 help to alleviate the -problem and, in
general, the coefficients can be adjusted to give viable slow-roll inflation, possibly
without much fine tuning [158, 159, 167].
The KLMT example is quite instructive of the importance of constructing, rather
then assuming, a robust moduli stabilization scheme. To have a frozen internal
volume, a dynamical field ' and viable inflation at the same time, one can relax the
assumption that the superpotential depends only on the Kähler moduli and consider
the more general case
W D W.%; '/ : (13.66)
A suitable dependence on the brane moduli can compensate the large inflaton mass
coming from the Kähler modulus and lower it to an acceptable level. Since viable
models have V D O.0:01/, the tuning required on fluxes and brane positions
to solve the -problem may be modest, of order of percent. Constructions with
specific choices of (13.66) confirm this expectation [144]. Furthermore, symmetry
arguments [145, 147, 148] and one-loop corrections to the superpotential from the
open-string channel [152] and the closed-string channel [155, 157, 326] can lower
the inflaton mass to a viable value. However, at least for models with closed-string
corrections, a relaxation of the -problem is possible only in a limited portion of
the landscape, for certain embeddings of D3/D3-branes and wrapped D7-branes
[155–157, 327]. In certain cases, one can skip Step 2 (Sect. 13.5.2.2) and sustain
inflation only with D3-branes: a running dilaton can suffice to uplift the minimum
of the potential to de Sitter or Minkowski spacetime without D3-branes in the
throat [327].
Extensions of the KLMT model (Fig. 13.10a) to a multi-moduli dynamics
usually entail a solid Calabi–Yau construction motivated from the bottom, for
instance requiring to have Standard-Model-like chiral families on the anti-brane.
The type-IIB flux compactification of [6], mentioned in Sect. 12.3.9.2, is of this
sort. The fixed points on the throat are associated with additional moduli such as
chiral scalars, chiral matter fields and gauge multiplets. These fields i contribute
to the total potential V.'; i / by an F-term (with a racetrack superpotential W), a
D-term and a supersymmetry-breaking term coming from anti-branes. For suitable
initial conditions, the trajectories of the fields collectively drive the inflaton ' along
a sufficiently flat trough. To obtain enough e-foldings, all parameters must be tuned
by one part in 103 [149], about the same order of magnitude of the tuning in volume-
modulus inflation (Sect. 13.3.2). The observed amplitude of the scalar spectrum is
recovered if the string scale is close to the GUT scale, ms 1015 GeV, while the
scalar spectral index ns turns out to be slightly blue-tilted. The modest fine tuning
and the wrong sign of the typical ns 1 can be overcome by a modification of the
physical picture. In particular, the presence of many fields suggest the possibility to
realize inflation not by the slow rolling of the inter-brane separation ' but through
a sequence of short stages of acceleration driven by different fields, as in the model
of chain inflation [253–257] (Sect. 13.2).
A similar alleviation of the -problem via fixed points has been explored in [151].
In that case, one considers a Calabi–Yau space with a Z2 symmetry and a D3-
brane located at the fixed point. Symmetric with respect to this fixed point, there
are two Klebanov–Strassler throats and a D3-brane at the bottom (i.e., tip) of each
(Fig. 13.10b). The D3-brane feels an attractive force towards the anti-branes, which
results in a total hilltop potential
V.'/ D V0 12 m2 ' 2 ; (13.67)

where ' is the DD-brane separation (identical for both throats) and m2 H 2
is a small effective mass. The inflaton slowly rolls down the local maximum of
the potential and 102 . The scalar index ns ' 1 C 2 is red-tilted,
the inflationary scale is V 1=4 1014 GeV and the gravitational-wave spectrum is
suppressed below any future detection level (remember the discussion about the
Lyth bound (5.161)). Also in this case, the string scale is ms 1015 GeV.
A third possibility, based on the brane dynamics developed in [206], is to start
with D3-branes only and to identify the inflaton with their collective motion near
the tip [150]. When the anti-branes approach the tip of the throat and stack together,
due to the presence of the background R-R 3-form flux they coalesce into an NS5-
brane (this is known as Myers effect [328]). This NS5-brane has topology R4
S2 : it is wrapped on an S2 sphere inside S3 , leaving four directions non-compact.
Parametrizing the tip S3 with the metric d˝32 D d 2 C .sin /2 d˝22 , the NS5-brane
action is [206]
Z " s #
T3 kF ˛04 4 p g s kF 1
SNS5 D d x g V2 . / 1 C 2 @ @ C U. / ;
gs ˛0
(13.68)
1p N3N 1
V2 . / D .sin /4 C U 2 . / ; U. / D C sin.2 /;
kF 2
(13.69)
where kF are the R-R flux units in (13.49) associated with the F3 form, ˛0 D
exp !.Nr / is the warp factor of (13.48) at the tip and N3N is the number of D3-branes
in the condensate. The effective potential

T3 kF ˛04 1
Veff . / D V2 . / C U. / (13.70)
gs
has a metastable vacuum at some min for N3N =kF . 0:08, while for N3N =kF > 0:08
it is monotonic and the potential develops a plateau around min . In both cases, the
local vacuum or the plateau correspond to the NS5-brane state. When enough D3-
branes have condensed into the metastable NS5-brane, the latter acquires enough
energy to unwrap itself and pass through S3 , eventually decaying (via quantum
tunneling or classically) into a cluster of N3 D kF N3N D3-branes (Fig. 13.10c).
This final configuration is supersymmetric.
The flatness of the potential near min is a promising region where the inflation
can roll slowly enough to sustain a prolonged era of acceleration. In the case of a
local minimum, from the point of view of a four-dimensional observer, a metastable
bubble is created by the brane condensation. If the potential is flat enough, the
bubble inflates. The typical inflationary scale is rather high, of order of the string
energy ms . This renders moduli stabilization more difficult, as one should fix the
dilaton and the Kähler modulus already at these high scales. Some fine tuning may
be in action here: viable inflation occurs for choices of the parameters which lie just
beyond the range allowed by the approximations of the model.
Brane inflation has also been constructed in the low-energy limit of M-theory
compactified on an S1 =Z2 orbifold (a configuration described more in detail in
Sect. 13.7.1), using the non-perturbative dynamics of M5-branes in the presence
of fluxes [329–334]. If a single M5-brane approaches the brane at the boundary of
the orbifold, the inflaton is the distance / y between the boundary and the M5-
brane. More generally, one can distribute a stack of M5-branes along the orbifold;
equally-separated branes are a dynamical attractor solution and the inflaton is the
distance / y D y1 D y2 D : : : . Therefore, this is a case of assisted inflation
[331] (Sect. 5.5.3). Corrections to the superpotential can destabilize the inflationary
trajectory unless one fine tunes the parameters of the model [332], in particular the
number of M5-branes [334].
13.5.5 Why the Tensor Spectrum Is Small
All warped D-brane inflationary models in string theory predict a negligible

amplitude for the tensor spectrum. This general result can be explained with
disarming simplicity by the Lyth bound (5.161) [193]. After calculating the scalar
and tensor spectra, one realizes that, as a function of the number of e-foldings,
2
the tensor-to-scalar ratio r.N / D 8Œ'=.HMP Pl / D 8.';N =MPl /2 is the same as
in standard inflation. Thus, the inequality (5.161) holds. On the other hand, the
field excursion ' during inflation has an upper bound determined by the throat
geometry. It is not difficult to show that ' is limited by the number of background
charges (13.51) in a generic flux compactification, . '=MPl/2 < 4=N [193]. From
(5.161),
32 0:036
r< 2
: (13.71)
30 N N
Since N is large in configurations where the AdS/CFT correspondence holds and
the SUGRA approximation is under control, the field variation during inflation is
usually much smaller than the reduced Planck mass and r 103 . On the other
hand, a positive detection of gravitational waves at the level r & 0:01 would paint
warped D-brane inflation into a very tight corner because it would imply N & 4,
an almost unwarped Calabi–Yau space which cannot be described consistently with
the tools employed so far.
In contrast, multi-brane inflation in M-theory can give rise to a detectable r
because the bound on the tensor-to-scalar ratio applies not to the inflation ' but to
the inter-brane separation y. ' and y are related to each other by a proportionality
factor that weakens the Lyth bound [333].
13.6 DBI Inflation
13.6.1 Setting
In the previous section, we introduced a model of inflation with the canonical

Lagrangian (13.55) and discussed the challenges it has to face in order to sustain
a sufficiently long period of acceleration. An assumption we have kept so far, and
that we will abandon now, is to neglect higher-order derivatives in the DBI action.
The dynamics governed by (13.52) with an effective potential,
Z
p
S D Sg C S' ; S' D d4 x g ŒLDBI V.'/ ; (13.72)
is capable of addressing the technical issues of the KLMT scenario related to

moduli stabilization.
The warp factor f .'/ in (13.52), valid for an exact AdS5 S5 background, receives
corrections from the actual shape of the Klebanov–Strassler geometry. In particular,
the effect of the IR cut-off of AdS5 can be modeled by
2
f .'/ D ; '02 > 0 : (13.73)
.' 2 C '02 /2
The effective potential V.'/ is generated by the interaction of the D3-brane with
the matter and Kaluza–Klein modes of the theory. Since conformal invariance of
AdS5 X5 is broken by the cut-offs, a mass term for ' can be generated.
We will call the cosmological model of the early universe based on the action
(13.72) DBI inflation [168–194]. (The NS5-brane action (13.68) is a special model
of DBI inflation.) It has several points of contact with other scenarios mentioned
throughout the book, mainly k-essence (Sect. 7.5.1, characterized by Lagrangians
L. ; r / which are higher order in .r /2 ), k-inflation [335, 336] (the same type
of model as k-essence but applied to the early universe) and early models based on
the DBI action for an unstable brane (which will be discussed in Sect. 13.7.2).10
On a FLRW background, L' D .1 1 /f 1 V, where the factor
1
:D p (13.74)
1 f 'P 2
10
DBI inflation is not different, conceptually, from slow-roll D-brane inflation: the setting is about
the same (branes and anti-branes moving on a warped Calabi–Yau space) and, just like in the
comparison between k- and standard inflation, all the changes stem from the kinetic term.
13.6 DBI Inflation 755
places an upper limit on the kinetic energy of the field. Slow-roll models correspond
to D O.1/; models with & 1 and 1 are called, respectively, relativistic and
ultra-relativistic.
Since the field momentum is ˘' D @L' =@'P D ', P the energy density and
pressure of the scalar are
1 1
D ˘' 'P L' D CV; P D L' D V: (13.75)
f f
13.6.2 UV Model
There are mainly two versions of DBI inflation. In the first (UV model) [168, 169,
173–175, 177, 178, 191–194], the D3-brane travels from the UV region of AdS5
towards the tip of the throat. For a quadratic potential
V.'/ D 12 m2 ' 2 ; (13.76)
one obtains a sustainable era of inflation along the AdS part of the throat. The mass
in (13.76) is not very small, in general, due to the relative position of the anti-branes
with respect to the moving brane. The warp factor f 1 at the IR end of the throat
takes its minimum value, so that, in the absence of a potential, the energy of the
D3-branes at the tip is lowered to levels that cannot sustain inflation. Therefore, the
extra potential term V.'/ in (13.72) must be sufficiently steep. This will limit the
freedom in the choice of parameters in V, a fact that we will better appreciate when
considering the IR model of Sect. 13.6.3.
Fortunately, a steep potential does not spoil inflation. While in the KLMT
model the D3-brane must travel slowly down the throat, in DBI inflation the D3-
brane moves fast. The inflaton potential can be steep without spoiling the slow-roll
condition, due to the upper bound 'P < f 1=2 on the speed of '. When the brane
reaches the tip (' ' '0 ), it is slowed down by the production of virtual light
particles. Subsequently, the scalar field decays into dust matter and radiation and
inflation ends.
These features can help to alleviate the technical difficulties met by the models
of Sect. 13.5, where we saw that, in general, moduli stabilization induces a large
mass for the inflaton (-problem). This is no longer an issue if the kinetic term
takes the DBI form. Therefore, both the suppression of the conformal coupling ' 2 R
and the fast-roll phase conspire to give viable inflation without conflicting with the
stabilization of the moduli fields.
Let us look at the model in detail. In the AdS part of the Klebanov–Strassler
geometry, away from the cut-offs,
'0 ' MPl : (13.77)

The upper bound (sub-Planckian vacuum expectation values) is necessary to avoid

a proliferation of quantum corrections to the gravitational Lagrangian Lg in (13.72)
[168]. The latter is of the form 2Lg D .MPl2 C' 2 /RCO.R=' 2 /C: : : (bars omitted).
In the regime (13.77), the conformal coupling is sub-dominant with respect to the
Einstein–Hilbert term, provided D O.1/, while terms of the form R=' 2 are of
order of H 2 =' 2 / .m=MPl /2 , which are negligible if the inflaton mass is sufficiently
small. This is indeed the case, as we shall see shortly.
With an Einstein–Hilbert gravitational sector, the cosmological dynamics is
given by the standard Friedmann equations with the energy density and pressure
(13.75). The scalar field approaches zero as ' ' =t, while the scale factor obeys
an approximate power law,
1 m
a.t/ / tp ; pD ' p : (13.78)
6 MPl
In the strong-coupling regime 1, one can obtain enough acceleration (p 1

and Ne D O. p/ D O.100/) for a mass m lower than the reduced Planck mass.
However, and contrary to standard inflation, m does not have to be much lower
than MPl .
The inflationary observables are not difficult to compute [169, 336]. Linear scalar
perturbations uk D zRk obey the Mukhanov–Sasaki equation

z00 1 3=2 a'
P
u00k 2 2
C cs k uk D 0 ; cs D ; zD : (13.79)
z H
Note that the propagation speed of the fluctuations is much smaller than the speed
of light, cs 1. In the limit ! 1, one recovers the usual result (5.118). On
the adiabatic vacuum and in the slow-roll approximation, the power spectrum is
Ps / H 2 =.cs /. To obtain a small-enough amplitude, for 0:05 < < 0:2 the
parameter .RAdS =ls /4 must be quite large, D 1010 – 1012 gs , implying for
gs D O.1/ that the curvature radius of the throat is about 100–1000 times larger
than the string length scale ls .
The deviation from a pure de Sitter expansion is compensated by a shrinking in
the sound horizon, which entails a freezing of perturbation modes at progressively
smaller scales. The scalar tilt is
ns 1 D 2 4 2s ; (13.80)
where s :D cP s =.Hcs /.
As in KLMT models, DBI inflation produces a tensor spectrum with a small
index nt D 2 and a small amplitude,
r D 16 cs ; (13.81)
which is suppressed by a factor cs D 1 . However, there is a lower bound on r

which makes the tensor signal potentially observable. In fact, another characteristic
effect of the model is a large non-Gaussianity [169, 337]. When expanding the action
(13.72) in the fluctuations ı', higher-order terms acquire higher-order coefficients
in 1. This enhances non-linear effects and deviations from Gaussian statistics.
The equilateral non-linear parameter is

equil 35 1
fNL ' 1 : (13.82)
108 c2s
equil
The 1-level PLANCK 2015 constraint fNL D 11 ˙ 69 on this model gives a lower
bound
cs > 0:07 (95 % CL) (13.83)
on the sound speed [338] which, for D O.102 /, implies r > 0:1. For a constant
non-linear parameter, there is a more precise lower bound on the tensor-to-scalar
equil
ratio, written in terms of fNL and the scalar index [177]:
4.1 ns /
r> q : (13.84)
equil
1C 3j fNL j
For 1 ns D O.102 / and j fNL j . O.10/, one has

equil
r > 102 ; (13.85)
not far from the upper limits (4.73) and (4.74).

However, there also exists an upper bound on the observed r which depends on
the volume Vol.X5 / (in string length units) of the Einstein-manifold cross-section of
the Calabi–Yau space C6 [177]:
106
r< : (13.86)
Vol.X5 /
The volume of the cross section of the Klebanov–Strassler throat is Vol.T 1;1 / D
.16 3 =27/l5s , which fixes a typical order of magnitude for Vol.X5 / D O.10/ and
the upper bound (13.86): r < 107 , incompatible with (13.85) by several orders
of magnitude. This means that the UV DBI model can accommodate observations
only if the compactification volume of the X5 cross section of C6 is unnaturally
small, Vol.X5 / 105 .
In this sense, the UV DBI model is under very strong observational pressure.
Together, non-Gaussianity and the Lyth bound (valid not only for standard and
D-brane inflation but also for k-inflation and, in particular, DBI inflation [193])
limit the number of background fluxes. The quadratic potential (13.76) is already
excluded, since N < r 2 =8 ' 27rj fNL j=70 < 3 [193]. Similarly, for large but not
equil
too large values of one obtains a relativistic scenario situated between the KLMT
slow-roll scenario and the UV DBI model [192]. This case is also unfeasible because
a scalar index compatible with observations is usually associated with a large tensor-
to-scalar ratio. Also inflation too close to the tip of the Klebanov–Strassler throat is
ruled out, as it produces too large non-Gaussianities [174].
Generalizations and modifications of basic DBI inflation do not alleviate these
problems in a substantial way. Relaxing the assumption that the D3-brane moves
towards the throat tip along the radial direction and allowing for a more generic
spiral motion, one obtains a multi-field model where the dynamics of the fields is
governed by the DBI Lagrangian [181, 182, 184–190]
0v
u 1
u 6
X
LDBI Œ'i D f 1 @t1 C f @ 'i @ 'i 1A V.f'i g/ :
iD1
The non-zero angular momentum of the brane turns out to provide a few extra e-
folds of inflation but the level of non-Gaussianity is still large, since the leading
contribution is the single-field one. A large fNL is also obtained when
P one considers
many non-interacting D-branes, which give rise to several copies n LDBI Œ'n of the
single-field DBI Lagrangian (13.52) [339, 340].
On the other hand, replacing the D3-brane with a D5-brane wrapped on a 2-cycle
or a D7-brane wrapped on a 3-cycle increases the upper bound (13.86) but there are
other issues to solve, first of all how to obtain a sufficient number of background
charges (limited by a maximum value in Calabi–Yau compactifications) [179, 180].
13.6.3 IR Model
Generically, Calabi–Yau spaces are multi-throat and many D3-branes live in the
low-curvature regions. Depending on the number and position of these branes and of
the background fluxes, one can imagine a qualitative scenario combining bits of the
information we already gathered. Due to the attraction (13.54) towards the warped
regions, some of these anti-branes settle down some throats (Fig. 13.11a), while
others decay into D3-branes passing through the NS5-brane condensate (13.68).
The lifetime of the anti-branes and their condensates depend on the ratio N=kF . After
some time, the typical snapshot will show stacks of D3-branes in some throats and
stacks of D3-branes in other throats with different shallowness (warping factors).
These D3-branes may exit their throat attracted by the anti-branes in the region
with the largest warp factor, which we call A-throat (A stems from anti-brane; see
Fig. 13.11b).
The surviving D3-branes lift the AdS vacuum to de Sitter and pave the ground
for inflation. In the so-called IR model of DBI inflation [170–174, 176, 183, 191,
Fig. 13.11 IR DBI inflation as described in the text
194], D3-branes are placed in a throat in a stabilized Calabi–Yau space (dubbed

brane-throat or B-throat) and the motion is from the IR to the UV end. The inflaton
is the AdS radial coordinate of the D3-stack along the B-throat and inflation occurs
when the branes exit the IR region. Once got away from the B-throat, the branes
spread towards different throats (not necessarily the A-throat) and inflation ends
when they annihilate or collide with the D-branes therein. The Standard Model lives
in one of these throats (S-throat).
From the KLT moduli stabilization, we saw that, in the absence of special
precautions the inflaton tends to acquire a mass m2 D O.H 2 / [144]. The mass of
the UV DBI model may be not very small, either, due to the suppression of the
D3-branes energy by the warp factor f 1 . On the other hand, by placing the anti-
branes in or past the UV part of the B-throat one avoids such suppression and their
energy is enough to pull the D3-branes at the IR end efficiently. As a consequence,
there is more freedom in the shape of the potential V.'/, which we will encode in
a phenomenological parameter ˇ. The sign of the mass term changes if we switch
the direction of motion from UV!IR to IR!UV, so that we can describe the force
sucking N3 D3-branes from the B-throat by the potential
V.'/ D V0 12 ˇH 2 ' 2 ; (13.87)
where ˇ > 0 and the Hubble parameter H is approximately constant. This potential
covers several scenarios with quadratic potentials: for 1 ˇ < 0 and 1, the
slow-roll (small mass) KLMT model; for ˇ 1 and & 1, the intermediate
model of [192]; for ˇ 1 and 1, ultra-relativistic UV DBI inflation; for
ˇ 1 and 1, the slow-roll model (13.67) of [151].
For the potential (13.87) with ˇ > 0 and the dynamics (13.72), the attractor
solution is ' ' =t (here the initial conditions are set at t D 1) and H const,
p
followed by a non-relativistic phase with / expŒ. 9 C 4ˇ 3/Ht=2 (which is
of slow rolling only if ˇ 1).
Inflation lasts about t NB =m D NB =.ˇH/, where NB (estimated to be &
gs N32 to keep the warp factor (13.50) at the tip small) is the number of background
charges in the B-throat. The DBI action is valid as long as one can ignore the back-
reaction of the background with respect to the brane energy, which translates into
the inequality NB . Since NB D O.106 / or higher is easily achieved in flux
compactifications, it is not difficult to have enough e-foldings,
NB
Ne H t ; (13.88)
ˇ
even if ˇ is large. The propagation speed of cosmological perturbations is
3 3
cs ' 1; (13.89)
ˇNe NB
while the scalar spectral index is
4 2
ns 1 ' C p : (13.90)
Ne Ne C 27N3 NB =8
The first term in (13.90) always dominates and the spectrum is red-tilted. For 50 <
3=2
Ne < 70, N3 NB NB .ˇNe /3=2 and ˇ < 106 , we obtain
0:92 < ns 6 0:97 ; (13.91)
consistent with PLANCK 2015 data at the 3-level. The scalar running ˛s is negative.
Since the cosmic expansion is exponential during inflation (H const), the first
slow-roll parameter is negligibly small and so are the tensor-to-scalar ratio r and
the tensor index nt . Again, the Lyth bound provides an enlightening explanation of
the suppression of primordial gravitational waves [193].
Finally, non-Gaussianity constrains the range of ˇ through the non-linear
parameter (13.82). Using (13.89),

equil 35 ˇ 2 Ne2
fNL ' 1 : (13.92)
108 9
When combined with PLANCK data, this formula enforces a strong restriction on
the parameter space [338]:
ˇ 6 0:7 (95 % CL) : (13.93)
This is the same constraint as (13.83) for Ne D 60. As in the UV case, non-
Gaussianities are excessive if inflation takes place close to the tip of the throat [174],
since their magnitude increases with the decrease of the warp factor f 1 . At the UV
13.7 Other Models 761
end of the throat, the warp factor increases and non-Gaussianities are lowered down
to acceptable levels.
The details of the reheating phase depend on how the D3-branes end their life,
by collision or annihilation with anti-branes. In general, cosmic strings will be
produced after inflation in the A-throat with N3N anti-branes, with a string tension
in the range
r
p gs
3 1012 gs < G < 9 106 ; (13.94)
N3N
compatible with the experimental bound (13.64).
13.7 Other Models
We continue with a brief mention of models which appeared during the years
and tackled inflation and the cosmological constant problem independently of the
landscape picture.11 The interested reader should look into the literature for more
details. In this section, we do not attempt to provide an exhaustive list of references.
13.7.1 Braneworld
Before modern models of moduli stabilization came to prominence, an alternative

to the brane scenarios of Sect. 13.5 attracted the attention of the community for
several years. Regarded as the D D 11 strong-coupling limit of the E8 E8 heterotic
string [341, 342], M-theory admits a consistent compactification on a Calabi–Yau
space, resulting in a five-dimensional non-compact spacetime [343, 344]. This
model, called heterotic M-theory [345], consists in two 9-branes at the boundary
of an eleven-dimensional bulk, each carrying an N D 1 super-Yang–Mills theory
with group E8 . The inter-brane direction has the structure of an orbifold S1 =Z2 .
Upon compactification, the dynamics of this spacetime admits a solution with two
parallel D3-branes at the end-points of the fifth dimension [346, 347], on which
11
The research on some of these proposals (such as braneworld or tachyon cosmology) has been
discontinued for several reasons, mainly for observational or theoretical difficulties. Experimental
problems arose when WMAP, PLANCK and supernovæ data begun to rule out cosmological
models of inflation or dark energy, making the parameter space of rigidly string-motivated
approaches unviable. Theoretical difficulties include the presence of undesired collateral effects or
an incomplete control on moduli stabilization, the latter being improved in the K LT construction.
Concerning this point, one might differentiate between string approaches to cosmology and models
which are only string-inspired, i.e., that contain certain ingredients borrowed from or similar
to some features of string theory but that do not meet the highest standard of rigor in their
assumptions, their justification of their starting point or their controllability within the fundamental
theory. Such a discrimination [198] would be unavoidably subjective and we will not attempt it.
one can construct a cosmology [348, 349]. The distance between the boundary
branes is larger, or even much larger, than the size of the Calabi–Yau manifold
[343, 344, 350, 351].
This set-up inspired a plethora of braneworld scenarios where the observed
universe with the Standard Model is assumed to live in a brane embedded in
a non-compact AdS5 five-dimensional bulk (see [352, 353] for reviews). One of
the first phenomenological models proposed is the Randall–Sundrum braneworld
[307]. The invisible brane with negative tension is sent away to infinity, while
the observed universe is on the brane with positive tension > 0.12 The strong
curvature of AdS5 localizes gravity mainly on the visible brane but there is some
leakage of information between brane and bulk. The extra non-compact dimension
can communicate with the matter confined in the brane and the cosmological
evolution thereon can change substantially from the standard four-dimensional case
[355–362]. If the brane energy density is comparable with its tension, = & 1,
then the bulk back-reacts for the presence of the brane matter and, for an FLRW
metric with a perfect fluid, quadratic corrections to the Friedmann equation arise
[349, 356, 357]:
42 E
H2 D .2 C / C 4 ; (13.95)
6 a
where E D const is a “dark radiation” term arising from the brane-bulk gravitational
interaction. Gravity experiments impose the bulk curvature scale to be .1 mm, that
is, 51 m5 & 108 GeV and 1=4 & 103 GeV.
In another braneworld scenario [363], one includes also the ˛ 0 -leading-order
quantum corrections to the heterotic low-energy effective action. These corrections
take the form of the Gauss–Bonnet term (7.78) [364]. Up to a boundary term, the
five-dimensional bulk Lagrangian with a negative cosmological constant 5 < 0
reads ŒR 25 C .8g2s /1 LGB =.252 /. Gauss–Bonnet braneworld cosmology differs
from (13.95) in the modifications at high energy density [365–373]. A cosmological
solution of the theory can be found via a five-dimensional warped metric such that
its projection on the 3-brane is FLRW-like. The effective Friedmann equation on the
brane is [368, 370, 371]
2s 3 23
3=2 2
5 ı ı
H 2 D g2s .cC C c 2/; c˙ D 4 1C 2 C ˙ 5 ;
6gs ı0 ı0
(13.96)
12
In an earlier version where the Standard Model is on the brane with negative tension, the extra
direction is compact [354]; however, this model is not realistic since the visible brane turns out to
be an anti-gravity world [355].
where ı01 :D 52 =.4gs/ and ı D C . Expanding (13.96) to quadratic order

in ı, one recovers (13.95) provided some relations between 5 , 4 , gs and are
satisfied. When ı=ı0 1, one has a high-energy non-standard cosmology H 2 '
2 2 2=3 2=3
.gs 5 =2/ . When the energy density is far below the string scale (=ı0
ı=ı0 1) but , we have H 2 ' Œ42 =.6/2 , the high-energy regime of
(13.95). Finally, when the brane grows stiff with respect to its matter content
and the standard evolution is recovered, H 2 ' .42 =3/.
13.7.1.1 Inflation
These results from Randall–Sundrum and Gauss–Bonnet braneworlds indicate that,

in general, the physics in the bulk can influence the effective dynamics induced on
the brane, giving rise to modifications of standard cosmology. The dynamics can be
studied with a deformed Friedmann equation [374–376]
H 2 / q ; q > 0; (13.97)
where different values of q D 1; 2; 2=3; : : : describe high-energy patches of the

early universe valid in certain epochs. Many models of braneworld inflation have
been studied in the background dynamics (13.95), (13.96) or (13.97) and including
inhomogeneous cosmological perturbations [377–382]. The problems of the hot-
big-bang models are solved via the traditional inflationary mechanism, the only
difference being in the effective dynamics the scalar field obeys.
13.7.1.2 Big-Bang Problem
A resident on a brane may or may not experience a singular cosmological evolution,

depending on the immense wealth of details that characterize brane cosmolog-
ical models. The cosmological solutions of braneworld scenarios with dynamics
(13.95), (13.96) and (13.97) have, in general, a big bang, which is not approached
with the mixmaster behaviour discussed in Sects. 6.3.4 and 13.9.2 (at least in
the Randall–Sundrum case [383]). Nevertheless, when considering the effects of
the bulk, there are hints about a singularity resolution. In the simplified model of
(13.48), a stable D3-brane travels towards the tip of the Klebanov–Strassler throat
of a type-IIB Calabi–Yau compactification and bounces back. An observer on the
brane would start with a contraction period, see a bounce instead of a bang and then
go through an epoch of expansion. The presence of a bounce can be understood
by looking at the metric (13.48), where ds24 is chosen to be the four-dimensional
Minkowski metric. Parametrizing the brane trajectory with the proper time t of the
in-falling brane, the brane observer lives in a flat FLRW universe with scale factor
a.t/ D exp !Œr.t/. Since ! is finite at the tip of the throat, a.t/ acquires a finite
minimum there [384]. However, issues appear when trying to connect non-singular
solutions with a realistic cosmological evolution [385–387].
13.7.1.3 Self-Tuning Cosmological Constant
In certain solutions of five-dimensional Randall–Sundrum brane models [307, 354],

the zero mode of the warp factor ! in (12.107) is moderately adjusted to suppress
the four-dimensional cosmological constant to zero. Poincaré invariance is therefore
preserved in the brane.
This result, which is the stringy version of an earlier proposal in generic Kaluza–
Klein compactifications [388], can be obtained in two ways. One is to consider
the renormalization-group flow of the couplings of the theory and to recognize the
existence of solutions such that k D 0 throughout the trajectory [389, 390].13 The
other is to check the stability of D 0 against radiative corrections in perturbative
quantum field theory on the brane [391, 392]. In both cases, however, these solutions
are in general not preferred, since their consistency requires some fine tuning
on other parameters of the model [393, 394] and they possess singularities if
the brane is flat [395]. Also, the cosmological evolution dictated by the effective
Friedmann equations on the brane may require some extra tuning, at least in some
toy models [396]. A generalization of this setting to six-dimensional supergravity,
where supersymmetry is broken by the branes, can circumvent the fine tuning; its
chance of success depends on the details of the construction [397–399].
13.7.1.4 Open Problems
Nowadays, braneworld scenarios have partly lost their appeal mainly because they
are not as phenomenologically robust as the warped D-brane models of Sect. 13.5.
To begin with, the majority of braneworld models are not based on concrete schemes
of moduli stabilization, as they predate the KLT construction. Also, contrary to
its compact but yet more unrealistic version [307], the non-compact Randall–
Sundrum scenario [354] does not solve the hierarchy problem. Moreover, for the
values q D 2; 2=3, the inflationary dynamics associated with (13.97) is limited by
the most recent data even more severely than standard general-relativity models
[400]. Despite its drastic change with respect to (2.81), a modified Friedmann
equation such as (13.97) does not lead, once all parameter constraints are met,
to characteristic features in the cosmic microwave background that could strongly
differentiate them from Einstein gravity.
13
We met a similar type of trajectory in the RG flow of asymptotic safety (Sect. 11.2.2).
13.7.2 Cosmological Tachyon
In Sect. 12.2.8, we mentioned that unstable Dp-branes (p even and odd in, respec-
tively, type-IIB and type-IIA theory) can decay into lower-dimensional branes or
into the closed-string vacuum via tachyon condensation. Tachyon condensation on
an unstable Dp-brane with metric g is described, up to fermionic fields, by the
DBI action [401–403]
Z
SDBI ;T D dpC1 x LDBI ;T ; (13.98a)
q
LDBI ;T D V.T/ det.g C 2l2s Fab @ T@ T/ : (13.98b)
Here, the scalar field T has dimension ŒT D 1 and V is its potential, which is
calculated exactly to all orders in the Regge slope ˛ 0 but at the tree level in gs .
On the other hand, the SFT tachyonic action (12.52) is non-perturbative also in gs .
To recapitulate, in this book we have seen four different DBI actions: the actions
(12.19) and (12.42) for a stable Dp-brane in flat or curved spacetime, the action
(13.52) for a stable D3-brane on AdS5 S5 and the action (13.98) for a unstable
Dp-brane on a generic curved background.
Expression (13.98) can be easily compared with the action (12.42) for a stable
brane: the tension Tp has been replaced by the tachyon potential and a kinetic term
has been introduced. On the other hand, the difference between (13.98) and (13.52)
is the regime of the string coupling: strong in the model of DBI inflation of Sect. 13.6
(where the D3-D3 pair is unstable but has no tachyonic mode), weak in the present
case. Note that (13.98) can be mapped to the square-root part of (13.52) with the
field redefinition T D =' and the identification V.T/ D 2 =T 4 .
The tachyon potential has some universal features. There is a maximum at
Tmax D 0, near which V.T/ D 1 M 2 T 2 =2 C O.T 4 /, where M 2 < 0 is the
negative tachyon mass. A local minimum V.Tmin / D 0 is located at infinity,
Tmin D ˙1. Potentials interpolating between the maximum and the minimum are
V.T/ D exp.M 2 T 2 =2/ [404–409] and V.T/ D 1= cosh.MT/ [410, 411].14
13.7.2.1 Inflation
In the brane inflation models of Sect. 13.5, the inflationary era takes place before
the onset of the tachyon instability, i.e., when the colliding branes are still at a
distance larger than the string length. In warped D-brane inflation, the inflaton is
14
Another but rather different formalism where one can analyze tachyon condensation is string
field theory (Sects. 12.1.6 and 12.2.9), in which case the kinetic term for the tachyon is canonical
and the potential is dressed with exponential non-local operators. The local minimum is at some
finite Tmin in this case.
the distance between the D3-brane of the KLT construction and a mobile D3-brane
is added on the warped throat. However, one can try to postpone inflation to tachyon
condensation in a setting where the individual branes are unstable. The first example
of this scenario is a type-IIA brane scenario where one considers the annihilation of
a D5- and a D5-brane into an unstable 3-brane [298]. On the 3-brane, the tachyon
scalar field vanishes, T D 0. This is the location of the local maximum of its
potential. At this point, the tachyon rolls down and drives or helps inflation on the
3-brane [298, 412].
The tachyon field as an inflationary agent has been studied in a more cosmolog-
ical fashion, often quite independently from string theory, both in four dimensions
and on a braneworld [413–436] (see also [374–376]). The idea is simply to take
(13.98) with p D 3 as the matter source for the Friedmann equations.
13.7.2.2 Dark Energy
Tachyonic models with inverse-power-law and exponential potentials have also

been proposed as a source of dark energy [417, 418, 432, 437–443]. They can
be viewed as a special case of k-essence (Sect. 7.5.1), they have quite similar
phenomenology to canonical quintessence and suffer from the same issues of fine
tuning [443]. Therefore, there is no apparent gain in choosing a DBI scalar field
instead of a Klein–Gordon one. Also, future singularities can appear in the late-time
evolution [431].
Cosmology based upon a rolling tachyon suffers from a number of problems,

including a small number of e-foldings (insufficient slow rolling), a large amplitude
ı= 1 for density perturbations and the formation of caustics [418, 419, 421,
424, 426, 428, 433]. A longer inflationary period can be obtained but only in a
configuration with 106 –1013 D-branes.15 Moreover, if the local minimum of V.T/ is
at infinity, there are no oscillations at the end of inflation and a reheating mechanism
appears difficult [421, 427].
These issues arise if one does not embed the model in KLT scenarios, where the
compactification scheme and moduli stabilization follow a certain procedure. In this
case, the parameters of the tachyon potential and the number of branes must be fine
tuned. Attempts to realize tachyon inflation in KLT scenarios is attractive because,
instead of adding a mobile D3-brane and facing the stabilization issues of standard
15
Although a tachyonic inflationary period with not many branes is too short, it can be a viable
means to provide natural initial conditions for a standard scalar inflationary period, similarly to
what happens in fast-roll inflation [444]. Then, this standard inflation lasts a sufficient number of
e-foldings and dilutes the non-linear perturbation structure generated by the tachyonic phase.
KLT inflation, acceleration is driven by the open-string tachyon of the D3-brane and
no new moduli are introduced. In practice, in the KLT construction of Sect. 13.1.1,
the constant brane tension in (13.3) is replaced by (13.98) [432],
ˇ1 / T3 ! LDBI ;T : (13.99)
Then, the problem of large density perturbations is solved by considering a small

warp factor in a warped metric, while the problem of reheating is overcome by
accounting for the negative cosmological constant arising from gaugino condensa-
tion. However, the modification (13.99) can spoil the local minimum of the total
SUGRA potential and jeopardize the viability of the model as a competitive string-
theory candidate.16 A more effective use of the DBI action, where the scalar field
is the inter-brane separation rather than a tachyonic mode, has been introduced in
Sect. 13.6.
If, on the other hand, one gives complete freedom on the choice of the potential
V.T/ and ignores any other stringy detail such as the compactification scheme and
the moduli problem, the tachyon is not distinguishable phenomenologically from
a canonical inflaton due to a field redefinition [417, 422]. Conversely, given the
same potential, one can distinguish the tachyon from a standard canonical scalar
but it does not give any appreciable advantage over the latter, even on a braneworld
[435]. Overall, the connection with string theory becomes progressively weaker as
cosmological applications of the tachyon advance.
13.7.3 Modified Gravity
String theory can give a motivation to some (but very few) of the modified gravity
models considered in Sect. 7.5.
13.7.3.1 Gauss–Bonnet Gravity
In order to be of use for phenomenology, the low-energy limit of string theory should
be free from instabilities. This is not guaranteed in principle, since the OŒ.˛ 0 R/n
corrections to the beta functions in the universal SUGRA sector introduce higher-
order derivatives and, hence, possible Ostrogradski instabilities (Sect. 11.8.1). These
corrections are known up to O.R4 / [446–451]. In particular, the correction at
quadratic order is the Gauss–Bonnet term (7.78), which is non-topological in
D > 4 and ghost-free on a Minkowski R target spacetime [447]. When compactifying
p
the low-energy action S D ˛ 0 1 d10 x gŒO.˛ 0 /LEH C O.˛ 0 2 /LGB to four
dimensions, the couplings of the Einstein–Hilbert term LEH / R and of the Gauss–
16
On top of that, it is not clear whether the promotion of the classical field T to a quantum
object correctly describes quantum string theory [445]. This makes the quantization of the tachyon
Lagrangian (13.98) and the interpretation of cosmological perturbations a delicate subject.
Bonnet invariant LGB acquire a dependence on the moduli fields. For only one such
field , the gravitational action is of the form [450]
Z
p R
SD d4 x g f1 . / 2 C f2 . /ŒLGB C a4 .r /4 C L C Lmat ;
24
(13.100)
where L D !. /.r /2 =2 V. / includes a quadratic kinetic term and a

potential for and Lmat is the -dependent matter Lagrangian. If the field 4 is
the dilaton ˚4D arising in the loop expansion of the low-energy effective action,
then
4
4 D ˚4D ; f1 D ! D e4 ; f2 D e (13.101)
2
in the string (Jordan) frame and at the tree level, where D 1=4, 1=8 or 0 for
the bosonic, heterotic and type-II string, respectively.17 The coefficient a4 is fixed
to recover the three-point scattering amplitude for the graviton [364, 450]. For the
heterotic string with stabilized dilaton ReS, stabilized axions and only one radion
modulus ReT D exp.24 /, one has f1 D 1, ! D 3=2, f2 / lnŒ2e4 4 .ie4 /
and a4 D 0, where is the Dedekind function [452]. At large j j, f2 cosh.4 /
[453]. The potential V. / in L is non-zero by the instantonic and condensation
effects described in Sect. 12.3.9.
The main problem of (13.100) is its loose connection with detailed flux-
compactification scenarios of moduli stabilization, so crucial for phenomenology.
The use of this action in modern string cosmology suffers from various limitations.
In Sect. 7.5.1, we have discussed the inadequacy of (13.100) to explain the fine
tuning of dark energy in the parameter space allowed by the no-ghost, classical-
stability and sub-luminal conditions which arise on a perturbed FLRW background
[454–456]. As a model of the early universe, (13.100) has been employed to study
the dynamics of perturbations across the bounce (Sect. 13.7.5). A class of bouncing
cosmological solutions with a modulus give circumstantial evidence that the big-
bang singularity may be resolved in string theory [452]. However, as repeated time
and again, the existence of bouncing solutions in any system which also has singular
attractor solutions is insufficient to address the big-bang problem. Moreover, the
curvature expansion in low-energy string actions is expected to break down at
such high densities and other stringy effects, of which we will see a sample in
Sects. 13.7.4, 13.7.5, 13.7.6 and 13.7.7, may become predominant. Issues related
to inflation will be discussed in Sect. 13.7.5.
17
In type-II theories, the Gauss–Bonnet term is removed after a field redefinition. Therefore, the
first higher-curvature correction to the Einstein–Hilbert action is O.˛ 0 3 R4 /.
13.7.3.2 Inverse-Power-Law Gravity
The compactifications used in the string cosmological models of this chapter rely
on Calabi–Yau spaces. Two alternatives are: (i) hyperbolic compactifications, where
a .4 C d/-dimensional spacetime is reduced to M 4 Hd and Hd is a d-dimensional
compact hyperbolic manifold with d > 2 and time-dependent volume Vd ; (ii)
product-space compactifications, where the internal space is the product of flat,
spherical and hyperbolic spaces. Hyperbolic and product-space compactifications
produce exponential potentials of the type (7.42), which have been used in models
of cosmic acceleration (inflation or dark energy) [457–463]. Presently, it is not clear
whether a viable parameter space is generated in a natural way in these scenarios,
which suffer from some yet unresolved issues [461, 464].
A curious application of hyperbolic compactifications concerns the model
ˇ4
f .R/ D R ; (13.102)
R
equation (7.87) with cn D ˇ 4 < 0 and n D 1 [465]. Compactifying the D D
10 Einstein–Hilbert action on a six-dimensional hyperbolic manifold, after a few
field redefinitions one obtains an effective action / 1=R [466]. Presumably, the
Einstein–Hilbert term can be recovered in the complete SUGRA action.
As we remarked in Sect. 7.5.4, Minkowski spacetime is not a solution of systems
with inverse powers of the Ricci scalar, an aspect that would worry most particle
physicists. However, a p conformal transformation recasts (13.102) into 242 LN D RN
2 2 2 =3
.r / =2 2ˇ e 1 e =3 , where D 3 ln.1 C ˇ 4 =R2 / and Minkowski is
a solution in the Einstein frame. Playing around with the compactification scheme
can yield more general inverse powers of R. This may be a tenuous indication that
special compactifications in string theory can give rise to specific classes of f .R/
modified gravity models in four dimensions, similar to (7.91). However, the bound
(7.92) is typically violated in these cases.
13.7.4 Non-local Models
The theories of non-polynomial gravity described in Sect. 11.8 have, by construc-

tion, the same type of non-locality as the one realized in the effective spacetime
limit (12.52) of string field theory. To see if they can be derived as effective
models stemming from SFT, one should face the yet-unsolved difficult problem
of constructing a well-defined string field theory from a conformal field theory
representation. From that, one should derive an effective non-linear action for the
graviton while retaining all non-local effects. As we mentioned in Sect. 12.1.6,
the gauge symmetry group of string field theory is wide. It includes spacetime
diffeomorphisms and supersymmetry transformations [467] but the way in which
such symmetries arise in the low-energy field-theory limit is not straightforward.
For instance, in order to identify general coordinate transformations in closed string
theory one has to make field redefinitions order by order in the string coupling gs
[468]; the same holds for Abelian transformations in the open case [469]. Therefore,
it is not even obvious whether one can obtain, directly from closed SFT, an effective
action of gravity which is simultaneously non-local (with the correct type of form
factors), covariant and non-linear.18
The action (11.118) has these three properties, with the form factor (11.119)
chosen to match with the graviton propagator of (12.51) in the linear limit. In D D
11 and with the inclusion of the 3-form A3 and of fermionic super-partners, it also
has other features compatible with string theory, including S-duality and the correct
local SUGRA limit when compactified to D D 10 [472]. Since (11.118) and the
other spacetime actions discussed in Sect. 11.8 with exponential non-locality are not
derived from SFT, they are commonly referred to as string-inspired models [473].
When properly motivated, they can help us to understand the effect of exponential
operators on the evolution of the early universe. In particular, we have seen that the
form factors (11.106) and (11.119) can remove the classical singularities of local
gravity. It may therefore be possible to address the big-bang problem in string field
theory. The cosmological dynamics of non-local gravity is under study. p
We conclude this sub-section by quoting some bounds on the scale M 1= ˛ 0
appearing in the exponential operator (11.106) using simplified particle-physics
models and the non-local dispersion relation characteristic of these scenarios. LHC
data bound the scale of non-locality as [474]
M > 2:5 103 GeV ; (13.103)

p
roughly corresponding to ˛ 0 D ls < 1019 m. Observations on opto-mechanical
heavy quantum oscillators are less constraining, ls < 1015 m [475]. The range
(13.103) is not surprising since it is close to the lowest energy scale for super-
symmetry breaking. However, table-top high-precision experiments already under
construction will be able to place much stronger bounds ls < 1022 1029 m
[475], thus providing independent information on the string scale.
13.7.5 Pre-Big-Bang and Dilaton Cosmology
The T-duality (12.83) is strongly suggestive

p of the existence of an effective minimal
length scale, the self-dual radius ls D ˛ 0 . This critical scale hints that singularities
may be resolved in string theory: a big-bang initial configuration would be as much
unlikely as an infinitely large newborn Universe.
If the effective dynamics obeys T-duality, the evolution of the universe is
symmetric under the mapping [476, 477]
18
The novel formulation of closed super-SFT advanced in [470, 471] may open up new
possibilities.
ln a.t/ ! ln a.t/ (13.104)
and the big bang of classical general relativity can be replaced by a bounce at a.0/.
Imposing the scale-factor duality (13.104) to the solutions of effective equations of
motion leads to generic pre-big-bang scenarios where the dynamics is well defined
before and after a cosmic bounce [478–501]. The bounce can take place only in
the presence of inhomogeneous perturbations, since the junction of the solutions at
t D 0 is singular in purely homogeneous backgrounds [481, 485–487]. The trans-
Planckian problem is also solved, since the perturbation modes we observe today
started with an initial wave-length much larger than the Planck length.
The low-energy dilaton-gravity action (12.40) respects the scale-factor duality
(13.104) and is often used in these scenarios, including in many early studies on
string-gas cosmology (Sect. 13.7.6). CMB anisotropies are generated during the
deflationary contracting phase and, although this might be considered an alternative
to inflation, the ingredients producing inhomogeneities are the same: a fluctuating
scalar field (in this case, the dilaton) and the back-reaction of the metric.
The regime of applicability of perturbation theory is reduced by the extreme
conditions near the bounce and, in fact, dilaton deflation generates scale-dependent
spectra. In models where the gravitational action is linear in the Ricci scalar R, the
typical prediction is ns 1 ' nt 3 [484, 491]. Inclusion of the Gauss–Bonnet
curvature correction and of the quartic term for the dilaton [488, 490, 492–496, 498]
(equations (13.100) and (13.101)) flattens the spectra slightly but not enough [496]:
2 6 ns 1 ' nt 6 3 : (13.105)
These results will be discussed more in detail in the ekpyrotic scenario of

Sect. 13.7.7.
We take this opportunity to stress the most formidable obstacle dilaton-based
scenarios must face to explain the cosmic acceleration, at either early or late
times: the overshooting problem cursorily mentioned in Sect. 13.3 [84, 269, 283].
For definiteness, consider the low-energy limit of the E8 E8 superstring and
recall, from (12.92), that ˚4D D ln.ReS/ and that (12.43) holds.19 The non-
perturbative potential V.˚4D / of the dilaton has an asymptotic absolute minimum
at ˚4D ! 1, corresponding to the weak-coupling limit of string theory at low
energy [502]. Unless V.˚4D / has other minima, the field will run away to infinity
where V.˚4D ! 1/ D 0, either by classical motion (for a high kinetic energy) or
by thermal effects [503]. Usually the runaway is classical, since the typical potential
is very steep due to gaugino condensation; see the racetrack superpotential (12.110).
If the dilaton is identified with the inflaton, it is difficult to find a regime of slow
rolling and the perturbation spectrum strongly deviates from scale invariance. If
the dilaton is not the inflaton , it couples to it via a non-minimal interaction. The
19
Note that a relation analogous to (12.108) holds in the SU.3/ super-Yang–Mills sector, so that
(12.108) and (12.103) give (12.43).
kinetic energy of ˚4D dominates the evolution of and, again, any slow-roll regime
is quickly disrupted.
In both cases, a dynamical dilaton during or after inflation may give rise
to variations of the fundamental couplings [504] and violations of the weak
equivalence principle larger than experimental constraints [505],20 unless the dilaton
couples universally to all matter sectors [506, 507] (i.e., for all matter species i
the couplings bi are the same, bi .˚4D / D b.˚4D /) or the chameleon mechanism is
enforced (Sect. 7.4.5). In the scenarios of moduli and D-brane inflation, we have
seen that the above problems are avoided: the dilaton can be trapped in a local
minimum while inflation is successfully driven by other fields.
The dilaton has also been considered as a quintessence candidate [508, 509]. The
full form of the couplings bi is obtained at any given perturbative order by the string
loop expansion in Riemann surfaces of higher genus (Sect. 12.1.4). In the weak-
coupling limit gs 1, this expansion is bi .˚4D / D e˚4D C ci0 C ci1 e˚4D C ci2 e2˚4D C
: : : , for some constants cin . Under certain assumptions on the quantum corrections
to the string effective action [510], one may expect the opposite gs 1 limit to take
the form
bi .˚4D / ' Ci C O.e˚4D / ;
so that these couplings tend to a constant Ci when ˚4D ! C1. In particular,

the dilaton can evolve to strong coupling at late times and all its field couplings
except the one with dark matter become trivial in this limit [508]. The equivalence
principle is still violated but by effects below or near present constraints [511–
513]. The resulting model has all the properties of tracking solutions, including
their shortcomings.
13.7.6 String-Gas Cosmology
The resolution of the big bang in string theory could make use of the thermodynam-
ical properties of superstring winding modes in the weak-coupling regime gs 1
[514–516]. In a compact space, the excitation modes of a thermal ensemble of
strings are momentum modes and winding modes. According to (12.82), the energy
of winding modes decreases with the size of the space, so that, for an adiabatic
process, they will dominate the thermal bath in small boxes. The temperature of
this bath of light modes cannot rise indefinitely and, in fact, the total energy and
the partition function diverge as the system approaches a critical temperature TH
called Hagedorn temperature [517]. This is the maximum attainable temperature
physically.
20
The bound (7.73) from big-bang nucleosynthesis and post-Newtonian solar-system tests is ! >
4 104 , while for the dilaton ! D 1 (apply the field redefinition ' D e4 =2 to (13.101)). The
actual check of violations is much more involved than this back-of-the-envelope observation.
One can exploit this phenomenon to study the very early Universe at the string
scale, a scenario dubbed string-gas cosmology [514–516, 518–542] (reviews are
[543, 544]).21 In a non-compact space, the specific heat is not positive definite; a
realistic scenario thus requires space to be compact. A nine-dimensional torus is
the simplest choice but orbifold and Calabi–Yau compactifications are also possible
which do not alter the main features of the proposal [521, 522].
As in pre-big-bang cosmology, T-duality is invoked to replace the big bang with
a bounce where the temperature is very close to TH [514]. The radius at which
the bounce phase is triggered increases with the entropy of the Universe. Three of
the spatial directions then inflate enough to push the radius of the compact space
beyond the particle horizon. Intuitively, this may be achieved by the tendency of
the system to attain thermal equilibrium, which is reached only if the world-sheets
of winding strings interact. However, world-sheets can intersect effectively only
in a spacetime with four or less dimensions. Therefore, it may be possible for a
compact Universe to have three large spatial directions, while the remaining six
remain compact [514, 516, 518]. The same mechanism can be realized also in the
more general case of a gas of D-branes [519] and in the low-energy limit of M-
theory [524, 526].
It has been verified that, for reasonable initial conditions, three spatial directions
can grow large (and become isotropic), while the six-dimensional compact space is
stabilized around the string scale ls by a mechanism involving the massless modes
of the string spectrum [523, 525, 527, 528, 530, 531]. This mechanism takes care
of the Kähler moduli, while fluxes stabilize the shape moduli [530]. Due to the
absence of R-R fluxes in the simplified treatment of these models, the dilaton
must be stabilized separately. In toroidal compactification, the dilaton sits at the
minimum of a non-perturbative potential generated by gaugino condensation [540].
The same mechanism breaks supersymmetry in a way compatible both with late-
time cosmology and with particle physics [541].
13.7.6.2 Alternative to Inflation
String-gas cosmology offers an alternative to inflation based on string thermody-

namics. By T-duality, both the big-bang and the trans-Planckian problem can be
solved in principle, since scales smaller than the string scale are not physically
within reach. The horizon problem is solved by an early stage of contraction when
the Hubble horizon is larger than the physical Universe [520].
For three large spatial dimensions, one can also obtain an almost scale-invariant
primordial scalar spectrum from thermal fluctuations. The mean square energy-
21
String-gas cosmology has been one of the very first applications of string theory to the early
universe [545–548].
density fluctuations in a 3-torus of radius % and volume V are hı2 i / T 2 CV =%6 ,

where CV D ıE=ıTjVDconst is the specific heat capacity of the string gas in the
volume V. From the string partition function in the Hagedorn phase, one finds that
the heat capacity is CV / %2 =ŒT.1 T=TH / [515]. Intuitively, it scales as a surface
rather than a volume because winding modes (which dominate the Hagedorn phase)
appear with one dimension less to a bulk observer. On the other hand, the Poisson
equation r 2 ˚ / ı (the 00 component of Einstein’s equations, which are assumed)
reads ˚k / k2 ık in momentum space. Therefore, Ps / k3 hk2 i / k3 h˚k2 i /
k1 hık2 i / k4 hı2 i and, replacing % k1 , one has [533, 535]
T.k/ 1
Ps / ; (13.106)
TH 1 T.k/=TH
where the temperature T is evaluated at the time when the mode with comoving
wave-number k exits the horizon. Since T is almost constant during the Hagedorn
phase, the spectrum is almost constant. The spectral index is
1 d.T=TH /
ns 1 ' < 0; (13.107)
1 T.k/=TH d ln k
where the approximation considers that T.k/ is very close to the Hagedorn
temperature. A small red tilt is produced because scalar modes are generated by
the energy density, which increases with T; as dT=dk < 0, an increase of power is
observed for small k.
Contrary to the standard inflationary paradigm, the tensor spectrum is blue tilted,
since [534, 535, 542]

T.k/ T.k/ 2 1 T.k/
Pt / 1 ln 1 (13.108)
TH TH l2s k2 TH
and
nt ' 1 ns > 0 : (13.109)
Intuitively, tensor modes are generated by anisotropic pressure terms in the energy-
momentum tensor, but near the Hagedorn temperature the thermal bath is dominated
by winding modes and the pressure decreases. There is thus a decrease of power
at low k. A blue-tilted tensor spectrum is one of the characteristic predictions of
string-gas cosmology that could be tested if a primordial gravitational signal was
discovered. The tensor-to-scalar ratio can be computed from (13.108) and (13.106)
and its magnitude depends on the string scale ls . The latter can be tuned (to about
ls 103 lPl ) to obtain an observable r D O.0:1/.
The level of non-Gaussianity of the model depends on the string scale. The local
non-linear parameter is [539]
2 2
ls H 30 ls
local
fNL ' ' 10 k 0 : (13.110)
lPl T lPl
At scales comparable with the horizon today, fNL is negligible if the string length is
close to the Planck scale, while fNL D O.1/ if the string energy is O.TeV/.
The lack of an analytic treatment of the Hagedorn phase forbids a direct deter-
mination of the effective dynamical equations in the very early universe. Dilaton
gravity, given by the low-energy effective actions (12.40) and (12.41), is unviable.
A background with a non-stabilized dilaton puts the whole string-gas scenario
in jeopardy due to several issues, including cosmic spectra with a strong scale
dependence and a difficulty in smoothly connecting the Hagedorn phase with a late-
time sensible evolution of the universe [536, 538]. Einstein gravity is excluded per
se because it does not exhibit T-duality, but it can be regarded as a dynamics with
frozen dilaton. However, higher-order curvature terms are expected to dominate at
such high-density regimes. In other words, the assumption of weak coupling gs 1
may break down near the singularity or the bounce. Therefore, expressions such
as (13.106) and (13.108), obtained in Einstein gravity (ideally, in dilaton gravity
in a quasi-static configuration where the dilaton is constant), should be taken cum
grano salis until a more fundamental setting, with a robust mechanism for dilaton
stabilization, becomes available. The discovery of a blue-tilted tensor spectrum
would stimulate further research in this direction.
The non-local models of Sect. 13.7.4 can provide bouncing scenarios wherein
to embed the Hagedorn phase [537]. However, in this case the scalar spectrum
is very nearly Harrison–Zel’dovich, unless the string scale were O.TeV/. Also, a
sufficiently long Hagedorn phase (necessary to maintain thermal equilibrium over
the observed scales) requires some fine tuning.
The understanding of moduli stabilization is limited by an incomplete knowledge
of string and brane dynamics at strong coupling. Some results in D D 11 SUGRA
(regarded as the low-energy limit of M-theory) confirm that three spatial directions
decompactify but the compact directions grow slowly in time instead of being
stabilized [524]. The number of decompactified directions and their evolution is
also sensitive to the initial conditions, both in eleven and ten dimensions, which
points towards some fine tuning [526, 529]. Typical initial conditions in D D 10 are
such that either there are too few strings to wrap around space and all dimensions
grow (strong coupling), or string interactions switch off too rapidly to allow the
system to thermalize and all directions remain wrapped by a large number of strings
(weak coupling). Massless string modes can indeed help to stabilize the moduli
but, again, the fate of the dilaton is a separate and important question, as we just
remarked. Although several ingredients have already been proposed to get viable
stabilization mechanisms, they are still to be harmonized rigorously in a realistic
effective evolution of the early universe.
13.7.7 Cyclic Ekpyrotic Universe
A stringy version of cyclic scenarios (Sect. 6.2.4.1) and an alternative to inflation

is the cyclic ekpyrotic universe [549–582], strongly based on an older non-cyclic
scenario [583–587] (see [588, 589] for reviews).22 For the greater richness of the
details of its embedding in string theory, this setting may be considered as an
evolution of the pre-big-bang scenario of Sect. 13.7.5 [498] and, in fact, the analyses
of that bounce apply also to the ekpyrotic case [497, 499–501]. We will first describe
the original proposal and then introduce some modifications necessary to make it
compatible with observations.
Imagine two flat 3-branes constituting the boundary of a five-dimensional
spacetime and interacting with an attractive potential V.'/ along a compact fifth
dimension parametrized by the radion '. These branes can be the orbifold planes of
heterotic M-theory (M-theory compactified on C3 S1 =Z2 ) [346–349] or the .3C1/-
dimensional manifolds of a Randall–Sundrum setting. The radion potential takes the
general form

V.'/ D V0 e2'='1 e2'='0 F.'/ ; (13.111)
where 1 < ' 6 'max is dimensionless, '='1 1 and '='0 1. The function
F.'/ represents non-perturbative effects and scales as exp.1=gs / or exp.1=g2s /.
In M-theory, ' is the 11th dimension and gs / exp. '/, with > 0. The potential
(13.111) has a local negative minimum and is very flat at the sides, tending to V0 > 0
for ' ! 'max 1 and to 0 for ' ! 1 (Fig. 13.12).
As the branes get closer, the gravitational energy in the bulk is converted into
brane kinetic energy. Since the branes are boundary ones, instead of collapsing via
tachyon condensation they collide and oscillate back and forth their center of mass
along the extra direction. During the collision at coincident branes (' D 1), part
of the brane kinetic energy is converted into matter and radiation. An observer on
one of the branes experiences the brane collision as a big bang (vanishing scale
factor aE in the Einstein frame) after a period of contraction. Even if the fifth
dimension does experience a big crunch and it collapses to a point, and even if
aE .t/ D 0 for the brane observer, the brane metric in the Jordan frame, the local
temperature and the energy density on the brane remain finite at the event. This
22
The term “ekpyrotic” is inspired by a cosmogonic model, attributed to Greek Stoicism, of cyclic
destruction (ekpyrosis) and creation (palingenesis) of the world in and from a great fire.
Fig. 13.12 Typical single-field potential in ekpyrotic scenarios and dynamical stages in a cycle
(Reprinted figure with permission from [562]. ©2013 by the American Physical Society)
bounce is not symmetric, an essential feature for realizing the difference in scales
between primordial and late-time acceleration.
13.7.7.1 Alternative to Inflation
The evolution of the universe can be described purely in terms of four-dimensional

variables (Fig. 13.12). Just after the brane collision and the onset of a new
separation, the observable universe expands in a sequence of eras dominated by
radiation. The radion ' turns around and rolls back along its potential during
radiation domination, at which point w' D 1 (zero kinetic energy) but ' is
still sub-dominant. There follows matter domination and, when the matter energy
density is red-shifted away, ' dominates and a quintessence era begins, with a
barotropic index w' 1 increasing to less negative values. After several trillion
years of acceleration, the kinetic energy of ' starts to dominate over the potential,
the observed universe decelerates and then contracts slowly, entering a fast-roll
ekpyrotic phase with V.'/ ' V0 e2'='0 and D 3.w' C 1/=2 1. The horizon
and flatness problems are solved at this stage. Eventually, at the end of the ekpyrotic
phase the contraction becomes faster (w' 1) until the system collapses into what
the observer would perceive as a big crunch (but not with BKL oscillations, which
are suppressed by the w' 1 ekpyrotic phase [558]). A big bang and a new era of
expansion follow.
During the contraction phase, a pattern of inhomogeneities is developed. In fact,

due to quantum fluctuations the branes are not parallel at all points and they collide
at slightly different times in different places. These patches begin their evolution and
cooling down from the bounce out of sync, which causes the anisotropies observed
in the sky. The spectral properties of this pattern depend on the specific model of
ekpyrosis. We discuss three models: (i) perturbations generated during the ekpyrotic
phase by a single field, (ii) perturbations generated during the ekpyrotic phase by
two or more scalar fields and (iii) perturbations generated before ekpyrosis by a
single field.
(i) The original single-field model does not work. For large ' > 0, the
potential (13.111) can be approximated by the exponential potential (5.78) with
negative-definite normalization, which is a solution of the Friedmann equations
for a power-law expansion a .t/p with p D '02 =2 (equation (5.79)) and
spectral indices (5.143) and (5.154). Since in this phase gravitational effects
are weak, the tensor spectrum is suppressed. However, the universe is quasi
static during contraction and 0 . p < 1, so that both spectra are strongly blue-
tilted, ns 1 D nt & 2 (in the Einstein frame). This blue-tilted deviation from
scale invariance is a typical issue of steep exponential potentials and it also
excludes dilaton inflation in pre-big-bang models, for which (13.105) holds
[552]. The problem of all these single-field models is that linear perturbation
theory breaks down near the bounce in most gauge choices and the power
spectra, which are produced in the contracting phase, acquire a strong scale
dependence [551–553, 560, 584, 586] unless specific potentials and matching
conditions at the bounce are chosen [499, 500, 554, 556].
(ii) Apart from extending the choice of matching conditions at the bounce, one can
relax the assumption that all moduli are frozen except the radion. Including
also a dynamical axion or the Kähler modulus of the Calabi–Yau space, one
is capable to generate red-tilted, almost scale-invariant scalar perturbations
through the bounce [554, 555, 563, 566, 567]. The effective potential V.'1 ; '2 /
during this ekpyrotic phase can take a variety of forms. Typically, it can be
parametrized by a double exponential
R R
V.'1 ; '2 / D V1 e d'1 c1 .'1 /
V 2 e d'2 c2 .'1 /
; (13.112)
where V1;2 > 0 and c1;2 are generic functions which become almost constant
during ekpyrosis. Defining :D .'P1 '1 C 'P2 '2 /=P and s :D .'P1 '2 '1 'P2 /=,
P
up to third order in s, the potential during ekpyrosis reads [567]
p 3 3=2 3
2
V.; s/ ' V0 e 1 C s2 C s ;
3Š
where 3 is a constant. Entropy perturbations produced at a first stage are
subsequently converted into curvature perturbations by the interaction of the
two fields. The typical range of the scalar spectral index ns 1 D 2=
d ln =dN (where N is the number of e-folds before the end of the ekpyrotic
phase) is
0:97 < ns < 1:02 I (13.113)
the observed value lies in this interval within the experimental uncertainty.
While scale invariance in standard inflation is a consequence of the slow-roll
approximation, the cyclic ekpyrotic model achieves this goal when 1.
The typical level of non-Gaussianity of this scenario is at least one order
of magnitude larger than the one of inflation, j fNLlocal
j 5 – 100 [567–569,
571, 572]. The actual range of values depends on when entropy perturbations
were converted into curvature perturbations. A conversion during the ekpyrotic
local
phase is ruled out, since fNL < 40. If the conversion took place during
kinetic-energy domination after the ekpyrotic phase, then [570]
p
local
fNL ' 5 C 32 3 : (13.114)
Since D O.102 /, the parameter space giving rise to acceptable local non-
Gaussianity is severely constrained, 0:8 < 3 < 0:5 at the 95 % CL [338].
The model is not ruled out but specific potentials must be chosen [582].
In an alternative model where '1 drives ekpyrosis and '2 generates entropy
perturbations, V1 > 0 and V2 D 0 in (13.112) and '2 has a non-canonical
kinetic term / f .'1 /.r'2 /2 [577, 578]. This configuration suppresses any non-
Gaussian signal down to acceptable levels (in particular, fNL local
5) and reduces
the level of fine tuning [580, 581].
(iii) An alternative which employs only one scalar field is to assume that the
observed perturbations were created before the ekpyrotic phase, during the
transition phase when the kinetic energy 'P 2 =2 of the field almost cancels the
second exponential term in (13.111) [573, 575]. The potential is simplified
to V.'/ D V0 .1 ec0 ' /, where c0 :D 2='0 , and the nearly constant term
V V0 fuels the cosmic evolution. At this stage (which is a dynamical
attractor), both the scale factor and the Hubble parameter are almost constant.
The resulting scalar spectrum is adiabatic and scale invariant. In order to obtain
a reheating phase at a scale between electroweak and GUT, c0 D 1028 – 1040 ,
but for a constant c0 scale invariance is almost exact. To get an observable red
tilt, it is sufficient to generalize the potential in the transition phase as
h R i
V.'/ ' V0 1 e d' c.'/ ; (13.115)
where c.'/ is slowly varying around c c0 . Then,
2
ns 1 ' D O.102 / : (13.116)
ln.c20 =2/
Unfortunately, a simple exponential potential strongly suffers from a trans-

Planckian problem [574]. It also gives rise to momentum-dependent large non-
Gaussianities, with an equilateral non-linear parameter fNL ' 5k2 =.6c20 V0 /
equil
[573]. To overcome these issues, one has to ensure that the transition phase
does not last too long, so that to avoid the late-time creation of modes
with strong non-Gaussian statistics. For example, one can consider a c.'/
which decreases to some value c0 just after nearly-Gaussian scale-invariant
fluctuations have been generated [575].
To summarize, all models (i)–(iii) are characterized by a tensor spectrum with
negligible amplitude and a strongly blue tilt:
nt & 2 ; r 0: (13.117)
On the other hand, in those models which achieve near scale invariance in the scalar
sector a conspicuous non-Gaussian signal is typically produced, unless the potential
is carefully tailored or the kinetic term of the entropy-generating field modified.
After the ekpyrotic phase, the qualitative features of the evolution of the universe
are the same for all models. The transition between a big crunch and the next big
bang is governed by the smooth brane dynamics. At the bounce, matter and radiation
are created on the brane and the universe begins with high density again. One should
take into account the effect of non-linearities at this moment [554, 556, 576, 579,
585, 587].
When the branes are sufficiently far apart, the inter-brane potential V.'/ becomes
positive and fuels a dark-energy dominated era. The late-time accelerated expansion
washes away any relic produced since the preceding big bang and drives the universe
to a nearly empty state, thus restoring the local conditions in existence exactly one
cycle before. The total entropy increases at each new cycle but the entropy density
is cyclically diluted from a bang to the following crunch. The horizon problem
is automatically solved in this scenario and inflation is not needed. The flatness
problem is also nullified, since the brane universe remains flat all the time either
because it began on an almost stable (hence almost flat) brane or, more generally,
due to the curvature dilution by the preceding cycles of expansion.
The cycles of the ekpyrotic universe are asymmetric and they spend most of the time
expanding. Therefore, the averaged expansion condition (6.14) is satisfied both in a
cycle and when averaging over an arbitrary number of identical cycles, Hav > 0.
Thus, the BGV theorem of Sect. 6.2.3 applies: the ekpyrotic universe is geodesically
past-incomplete and there must have been a primordial Big Bang where and when
everything begun.
However, the cyclic solution is a dynamical attractor and the system becomes
insensitive to the initial conditions. Our cycle is only the last of a very long sequence
which will continue in the future indefinitely. The probability that the observed
particles have been created at the beginning of a past cycle is argued to become
exponentially smaller the longer the time interval between that cycle and ours.
Therefore, the issue of the past-incompleteness of geodesics becomes physically
irrelevant in this scenario. In this sense, but modulo a measure problem yet to be
addressed, this may be regarded as a solution to the big-bang problem [550].
13.7.7.3 Cosmological Constant Problem
The ekpyrotic scenario embeds a relaxation mechanism for the cosmological

constant problem [561] via an adaptation of an older proposal by Abbott [590].
Consider an axion field coupled with a non-Abelian gauge sector (e.g., the strong
Q where f is a mass scale. Integrating out the
force) through an interaction / . =f /FF,
gauge fields, one obtains a cosine potential similar to (5.90). It is also assumed that
the discrete symmetry ! C 2fn is softly broken by a term with no minimum
in the range of interest in . For simplicity, one can take a linear term:

U. / D A cos CB ; (13.118)
f f
where B A. Since radiative corrections to B are proportional to B, a very small

value of B would be protected against quantum effects. This potential is depicted in
Fig. 13.13 and has a typical “washboard” form with gradually decreasing periodic
minima Vnmin . As the universe evolves from some vacuum with large D Vnmin1 , the
axion undergoes a sequence of quantum tunneling events Vnmin ! Vn1 min
! Vn2
min
!
: : : describing de Sitter bubbles, eventually relaxing the effective cosmological
Fig. 13.13 Relaxation of the cosmological constant in the ekpyrotic cyclic universe. The differ-
ence Vnmin Vn1
min
between two adjacent vacua (dashed lines, n decreasing from right to left) must
be small enough to hit the observed value of within the experimental uncertainty
constant to a very small value 0 < V0min < B. A transition to negative values V1min
,
V2 ; : : : would make a bubble collapse in about one Hubble time, while bubbles
min
in other regions would continue to expand with > 0. The time spent in each local
vacuum becomes large according to an exponential-of-exponential progression, and
configurations with a small are most favored.
Abbott’s model was proposed within a non-cyclic, standard big-bang cosmology.
The problem is that the tunneling sequence occurs too slowly compared with the
Hubble expansion and the universe becomes “empty” by over-acceleration well
before the formation of structures and the reaching of the desired vacuum V0min .
However, by including the non-Abelian sector in the ekpyrotic scenario the empty-
universe problem is easily solved, since matter and radiation are created at every
bounce event. Also, plausibly the Universe is much older than the age of the present
cycle and the relaxation mechanism can take place on a time scale far greater than
H 1 . Moreover, since the brane does not become singular at the bounce, the kinetic
energy of the axion never diverges and the tunneling events down the washboard
take place from one minimum to the adjacent one, without erratic jumps. Cycle
after cycle, the universe spends an exponentially longer time in a state with smaller
and smaller cosmological constant.
Like most braneworld scenarios, the ekpyrotic proposal does not include the details
of the compactification scheme and of the stabilization of non-dynamical moduli. A
posteriori, this turns out to be a disadvantage because it introduces free parameters
(types of potentials, number of dynamical moduli, type of matching conditions, and
so on) and a loss of contact with string theory, as we discussed in Sect. 13.7.1. The
choice of potential is especially crucial for the avoidance of trans-Planckian issues
and unacceptably large non-Gaussianities. By itself, V.'/ does not entail any fine
tuning of the initial conditions [582] but, without any input from the full theory, it
remains purely phenomenological. Also, moduli stabilization would become even
more challenging in more realistic scenarios where the bulk is populated by branes,
the latter being necessary to get viable particle physics.
Moreover, the fact that one of the compact directions becomes singular cyclically
may not be attractive in a theory supposed to be finite, even if the type of singularity
involved is much milder than the outright disappearance of the whole spacetime as
in general relativity.
Regarding the cosmological constant problem, the relaxation mechanism
described above does not provide a complete solution because the parameter B
must be fine tuned to a very small value to guarantee that the interval 0 < V0min < B
is compatible with the experimental uncertainty on . At present, a theoretical
justification for this tuning within string theory is still missing. This gap problem
and further criticism are discussed in [591].
13.8 Inflation and Alternatives: Compact Summary 783
13.8 Inflation and Alternatives: Compact Summary
Table 13.1 is a grand summary of the strongest contenders for a scenario of the early
universe motivated by string theory. We have left out models which are in contrast
with experiments for string-motivated values of their parameters (KLMT models
with large , tachyon inflation and several versions of the ekpyrotic scenario),
phenomenological models which do not have full control on moduli stabilization
(the braneworld, among others) and models which have not been developed to the
point of giving predictions for the cosmological observables (such as non-local
gravity). The level of fine tuning of all these scenarios is similar and modest.
• Qualitatively, models of moduli inflation all share the same predictions of single-
field standard inflation: a nearly scale-invariant scalar spectrum, a negligible
tensor spectrum and a low level of non-Gaussianity. Contrary to the brane-
based models of Sects. 13.5 and 13.6, these scenarios do not involve complicated
brane dynamics. Since they are completely described by low-energy D D 4
supergravity, after writing down the total potential they can be treated with the
field-theory formalism of Chap. 5, at least during inflation and perhaps also in
the reheating phase. The main difference with respect to the models of Chap. 5
is that both the form and the parameters of the inflaton potential are motivated
by string theory, since the details of the super- and Kähler potentials depend on
the compactification scheme and on the geometry of the underlying Calabi–Yau
space.
• N-flation is not among the most favoured models due to the low scalar index
(13.37), a possible fNL D O.1/ non-Gaussianity and its uncertain embedding
in string theory. It is not excluded either, since it is not in clear tension with
observations.
• D-brane inflation has three characteristics which immediately tell it apart from
moduli inflation: (a) it is more involved due to the fundamental role played by
branes and anti-branes, (b) it takes more efforts to obtain a flat potential and
to solve the -problem and (c) the typical inter-brane potential cannot sustain
eternal inflation (this property has some consequences on the choice of initial
conditions but it does not worsen the level of fine tuning of the model). The
richness of this proposal makes it a tantalizing example of how string theory can
get in touch with observations.
• String-gas cosmology exploits string thermodynamics to manufacture a cosmo-
logical model of density fluctuations, while the ekpyrotic universe exploits the
dynamics of brane collisions to sustain cyclic fast-rolling eras of acceleration and
contraction. With respect to moduli, D-brane and DBI inflation, both proposals
rely much less heavily on the details of flux compactification and moduli
stabilization, they focus more on a resolution of the big-bang problem and they
offer an alternative to inflation.
All scenarios entail a few simplifications and many open problems. The details of
moduli stabilization are taken care of but only partially. In fact, there are not many
784
Table 13.1 Typical values of the cosmological observables and level of fine tuning of some early-universe string models compatible with observations.
The interval 0:95 < ns < 0:98 roughly corresponds to the 3 -level range allowed by observations. “Moduli inflation” includes racetrack axion inflation,
fluxless inflation, large-volume blow-up inflation (fibre inflation yields r 103 ) and volume-modulus inflation. “Axion inflation” includes N-flation and
aligned/hierarchical multi-axion models. “Warped D-brane inflation” includes the K LM T model and its most promising modifications. “Ekpyrotic universe
(ii)” refers to models of type (ii) with entropy-to-curvature conversion of perturbations after the ekpyrotic phase. Very low values of (the first slow-roll
parameter at horizon crossing) or of the tensor-to-scalar ratio, undetectable by present or near-future experiments, have been approximated to zero. “Small”
and “large” indicate the typical level of non-Gaussianity. Models with potentially dangerous non-Gaussianities are tuned to be compatible with observations.
UV DBI inflation is almost ruled out due to some problems in its theoretical consistency
Model j j 0:95 < ns < 0:98 nt r fNL Fine tuning

Moduli inflation 0 102 X .0 0 Small 100 – 103
Axion inflation 102 &102 X 102 .0:001 Small? 102 – 104
Monodromy inflation 102 &102 X 102 0:01 – 0:1 Small 102
Warped D-brane inflation 0 102 X .0 0 Small 102 – 103
UV DBI inflation 101 101 X 101 &0:01 Large 102
IR DBI inflation 0 102 X .0 0 Large 101 – 102
String gas – – X C 102 .0:1 Small 102
Ekpyrotic universe (ii) 102 102 X &2 0 Large 101 – 102
13 String Cosmology
13.8 Inflation and Alternatives: Compact Summary 785
examples based on concrete Calabi–Yau spaces and, in the absence of an explicit

construction, an unspecified number of moduli is usually assumed to be frozen.
For the same reason, there is some margin of freedom in the choice of the super-
and Kähler potentials. The making of Calabi–Yau spaces populated by Klebanov–
Strassler throats with a suitable brane population must pass through a series of
technical steps and consistency checks, including the stability of supergravity
solutions.23 The multi-throat, multi-brane scenario of IR DBI inflation is only
qualitative and still under study. One of the consequences is that the potential
attracting the D-branes out of their throats is phenomenological.
Despite all their limitations, models of string cosmology are rigid enough to
be falsifiable by current experiments. As a matter of fact, a number of them have
already been ruled out in the last decade, not long after their birth.
While all the viable models pass the test of near scale invariance of curvature
perturbations, they greatly differ in the tensor sector. Moduli inflation, warped D-
brane inflation, IR DBI inflation and the ekpyrotic universe predict a negligible
tensor-to-scalar ratio, either because the amplitude of primordial gravitational waves
is too low or due to a strong blue tilt. A detection of a non-zero r D O.102 /
would rule out these models, while it would restrict the parameter space of string-
gas cosmology. The level of non-Gaussianity of DBI inflation and of the ekpyrotic
universe is typically large and current estimates of the non-linear parameter fNL
place tight constraints on the parameters.
These predictions can be assessed within the bigger picture of the string
landscape. Hoping to confront string theory with experiments soon, one may wonder
whether large-r scenarios are “more common” than the others. We do not know the
answer yet but a preliminary counting of the relative frequency of large-field and
small-field models indicates that string theory has no preference towards a large or
small tensor-to-scalar ratio r [56]. However, this estimate has been made locally in a
mini-landscape of controllable compactification models which does not encompass
all the cases in the table.
The issue of fine tuning is also open. There are indications that the initial
conditions chosen for inflation may be not so special, at least in the mini-landscape
of one model of volume-modulus inflation [57]. Counting the number of viable
23
The inclusion of D-branes in the throat is a delicate issue, since the stability of the de Sitter
vacuum may be compromised by the brane back-reaction [592–600]. Supergravity solutions
describing branes in a throat have, in general, a divergent flux density. If the singularity is not
physical, it indicates that the configuration is not static: the anti-branes annihilate with the fluxes.
In this case, one cannot uplift the potential with the K LT mechanism and the viability of K LM T
inflation, based exquisitely on the supergravity approximation of string theory, is put to question.
This problem is not settled in stone for several reasons. On one hand, it is possible (but very difficult
to prove, technically) that stable branes-in-a-throat configurations could be realized beyond the
SUGRA low-energy limit. On the other hand, we have seen that one can realize D-brane inflation
with widely different brane configurations (even without D-branes), some of which may be free
from singularities even if the original K LM T set-up were in danger. Third, there are arguments
which regard these flux singularities as physical and, hence, resolvable [601–603]. In this case, the
de Sitter vacuum would be metastable and D-brane inflation could safely take place.
configurations for type-IIB F-term monodromy inflation shows that these models
should be realized with enough frequency in the landscape [135]. The mapping of
the landscape of multi-axion potentials similar to (13.39) is in progress [604]. The
numbers in the last column of Table 13.1 should then be interpreted with care.
To conclude, cosmological string models have enough predictive power to go
beyond simple compatibility checks with present bounds. The challenge they have
to face, also in the light of upcoming or near-future data on the tensor spectrum and
on non-Gaussianity, is to give robust theoretical motivations to the ever-shrinking
viable parameter space.
13.9 Big-Bang Problem
Having discussed how inflation is realized in string theory and string-inspired

scenarios, and how string theory can address the cosmological constant problem,
we turn to the big-bang problem. The subject is surrounded by a perhaps inevitable
halo of incompleteness due to the very fragmentary information collected from the
theory during the years. Still, it is important to assess this information and place
it side by side, either as a follow-up or as an alternative path, with the results on
classical singularities in Chap. 6 and with the big-bang resolution proposed by loop
quantum cosmology in Sect. 10.3. After summarizing the study of singularities in
string theory, we divide the rest of the section
p in two parts, which cover scales
larger or smaller than the string scale ls D ˛ 0 : a classical chaotic approach to the
putative singularity in the low-energy limit, valid for scales > ls (Sect. 13.9.2), and
a quantum approach at scales . ls (Sect. 13.9.3).
13.9.1 Big Bang in String Theory
In this chapter, we have seen three mechanisms that could remove the cosmic
singularity:
• Non-locality. Non-locality is implemented as exponential differential operators
exp.˛ 0 / in an effective low-energy dynamics (Sect. 13.7.4, non-perturbative in
˛ 0 and gs ).
• T-duality. T-duality is implemented in the FLRW solutions of pre-big-bang
cosmology (Sect. 13.7.5, perturbative in ˛ 0 and gs ) and in string-gas cosmology
(Sect. 13.7.6);
• Ekpyrotic mechanism. The colliding branes of the ekpyrotic paradigm
(Sect. 13.7.7, perturbative in ˛ 0 and non-perturbative in gs ).
All these frameworks have some drawbacks. Realistic non-local cosmological
models have not yet been derived from full string field theory. Both the pre-big-
bang and the ekpyrotic scenarios have been developed much more extensively than
13.9 Big-Bang Problem 787
non-local models but they are not quite embedded in a flux compactification scheme
of moduli stabilization.
Other approaches probe different corners of the string parameter space and open
widely different views on the big-bang problem (see [605–608] for reviews).
• Orbifold singularities. For instance, in contrast with non-local scenarios and the
ekpyrotic solution of [559], non-perturbative in ˛ 0 , time-dependent orbifolds with
a space-like or light-like singularity have been used as models of cosmological
singularities24 in perturbative string theory (in ˛ 0 and gs ) [609–619]. In a class of
cases, divergences in scattering amplitudes and various instabilities can arise and
signal a fundamental singularity in the geometry. At the intuitive level of general
relativity, near the big-bang singularity the kinetic terms in the non-linear sigma
model (12.37) are suppressed (some coefficients in gN and B go to zero),
leading to non-suppressed terms in the world-sheet path integral (i.e., violent
fluctuations of the fields) and to divergent amplitudes in the genus expansion.
However, ingredients such as orientifold planes [616] or the contribution of
twisted states [617, 619] can resolve the singularity of specific orbifold models
into a smooth bounce.
• Tachyon condensation. A positive outcome is also achieved in another type-
II or heterotic setting perturbative in ˛ 0 and gs [620]. Closed-string tachyon
condensation introduces action terms in (12.37) that grow towards the space-
like singularity, thus balancing the suppression of the non-linear sigma model
and controlling field fluctuations. As a consequence, the big bang is replaced by
a phase characterized by a thermal distribution of closed-string modes and where
spacetime simply ends through a topology change [621, 622]. This “Nothing
state” [620, 623, 624] is a string realization of the Hartle–Hawking no boundary
proposal (Sect. 9.2.3). The same mechanism has been utilized also to resolve the
singularity inside black holes [625].
• AdS/CFT correspondence. Going beyond perturbation theory and using non-
perturbative (in gs ) techniques such as the AdS/CFT correspondence, it was
found that black-hole space-like singularities are resolved in D D 3 [626], while
the D > 3 case is more complicated [627, 628]. Contrary to the cosmological
singularity, this one is hidden beyond an event horizon. To obviate this problem,
solutions in N D 8 SUGRA compactified to D D 4 and D D 5 have been
constructed such that smooth, asymptotically AdS initial data evolve, without
fine tuning, to a space-like big-crunch singularity [629]. This can open up the
study of the naked big-bang singularity in these regimes [630, 631]. Examples
of light-like singularities in a type-IIB AdS bulk show hints of a resolution in
the super-Yang–Mills sector of the theory, which is particularly well behaved
[632, 633].
24
In particular, the two-dimensional Misner space R 1;1 =boost has been a popular subject of study,
where R 1;1 D M2 is D D 2 Minkowski spacetime and “boost” is a finite boost transformation of
the light-cone coordinates.
• Matrix theory. In the matrix incarnation of M-theory, as gs increases one

promotes a light-like linear-dilaton background of type-IIA string theory to a
solution of M-theory [634, 635]. Both solutions have a light-like singularity but,
in the D D 11 extension, near the singularity the notion of spacetime is replaced
by that of non-commuting matrices. The physics at the big bang turns out to be
under control because the Yang–Mills gauge sector is weakly coupled there. Also
in this case, the analysis of the singularity is facilitated.
There is another context, perturbative in ˛ 0 and non-perturbative in gs , indicating
that cosmological singularities may be resolved. It is the study of early-time homo-
geneous solutions (including the Kasner metric (6.20)) to the Einstein equations in
the low-energy limit of M-theory [636, 637]. Compactifying D D 11 gravity on
hypertori of various dimensions, one notices that the ensuing cosmology has a big-
bang singularity which, however, disappears by U-duality in vacuum. (Recall from
Sect. 12.4 that U-duality combines S- and T-duality to map D D 11 supergravity
into weakly-coupled type-II string theories.) While vacuum solutions are U-dual
invariant, the matter sector is not. The physical interpretation is that, near the
singularities found in the D D 11 low-energy bosonic sector, matter decays into
new states that cannot be captured by the SUGRA field theory and that require a
dual, more efficient description in terms of type-II string modes. This is the subject
of the rest of the chapter.
13.9.2 Classical Billiards: Living with the Singularity
In Sect. 6.3.5, we had occasion to remark that the BKL oscillatory behaviour in
general relativity is typical for 9 or fewer spatial dimensions, while for D D 10 C 1
the approach to the singularity is monotonic. In Sect. 12.4, we met again the magic
value D D 11, this time in the context of supergravity and M-theory. Before
looking for a singularity resolution in string theory, it is interesting to explore this
coincidence of numbers and check whether the BKL behaviour detailed in Sect. 6.3
persists in supergravity, i.e., in the low-energy limit of string theories and M-theory.
Such is the subject of this section.
A central part of the discussion will be based on a characterization of the BKL
singularity we have not explicitly used in Sects. 6.3.4–6.3.5, but that is hidden in the
formulæ therein. Instead as “oscillations,” the chaotic sequence of Kasner epochs
can be described as collisions on the walls of a billiard [638]. The term billiard in
cosmological applications was introduced in the case of D D 4 general relativity
in vacuum [639] and, more extensively, in D dimensions either without matter or
with several perfect fluids [640–642] (other early works are cited in Sect. 6.3.4.1).
The most interesting case for string theory is when matter is constituted by several
p-forms [643–650] and fermions [651–656]. Reference [657] is a comprehensive
introduction and summary to general billiards and their applications to string and
M-theory.
To check whether space-like singularities are approached chaotically, consider

first the D-dimensional toy model [644]
Z " #
1 X
D p R n D 2
SD d x g @ @ e FnC1 ; (13.119)
2 D2 n
where n is dimensionless and FnC1 is the field strength of an n-form An (we use n
instead of p to avoid confusion in what follows). The action (13.119) is a schematic
representation of several cases. For D D 11, one has the low-energy M-theory
(SUGRA) action (12.128), with D 0 and n D 3. For D D 10 and / ˚ the
dilaton, one gets the massless bosonic sector of the low-energy string-theory actions
in the Einstein frame, including type-I theory (n D 2), the heterotic theories (n D 2)
and the transformed versions of both the type-IIA action (12.74) (n D 1; 2; 3) and
the type-IIB action (12.75) (n D 0; 2; 2; 4). Other terms such as the Chern–Simons
action and interactions among the n-forms do not change the results below [649].
The values and signs of the couplings n are given in [644] (or can be derived by
the reader for type-II theories).
In vacuum, the Kasner metric (6.20)–(6.21) and the profile D D '.x/ C
p ln.t=t0 / are solutions to the Einstein equations stemming from (13.119), provided
the generalization of conditions (6.25) holds:
X
D1 X
D1
pi D 1 ; p2 C p2i D 1 : (13.120)
iD1 iD1
The singularity at t D 0 is approached

P monotonically if the Kasner solution is stable
against the contributions / n;I t2wn of the n-forms to the Einstein equations and
I
P J
those / J t2wr of the spatial derivatives of the metric, where I and J are generic
labels. This happens if such contributions drop faster than the gR =g C.@ /2 / t2
leading terms in the limit t ! 0, which is guaranteed if all the exponents wIn and
wJr are positive. Adopting a generalized Kasner metric (6.19) and the same Ansätze
above for ai and (where now, approximately, the exponents pi .x/ and p .x/ depend
on spatial coordinates), the exponents wJr have been computed in [658–660] and
actually have a triple index:
ijk
wr D 1 C pi pj pk > 0 : (13.121)
For every n in the spectrum, the n-form contribution is split into two parts (one
“electric” and the other “magnetic”) which are sub-dominant if [644]
win1 :::in D pi1 C C pin 12 n p > 0 ; (13.122a)

wnj1 :::jD2n D pj1 C C pjD2n C 12 n p > 0 ; (13.122b)
where the indices in each expression are all different. Overall, one can define the set
of exponents
wK D fwr ; win1 :::in ; wnj1 :::jD2n g ;

ijk
K D 1; : : : ; Ktot : (13.123)
Their number is typically large. In M-theory, Ktot D 690.

In the absence of n-forms (Einstein gravity plus a scalar field), the Kasner
solution is asymptotically stable for any D [658], while in the absence also of the
dilaton the solution is asymptotically stable for D > 11 [659–661]. Turning on just
a vector field is sufficient to make the BKL dynamics appear [658]. Adding several
n-forms generates a behaviour more complex than the usual BKL approach.
In M-theory, there is only a 3-form and no dilaton and it is possible to show that
the electric exponent ˛ni1 i2 i3 D pi1 C pi2 C pi3 is always negative semi-definite. With
S-duality (D D 11 SUGRA compactified on a circle), one can map this condition
to the type-IIA case, which is in turn mapped to type IIB by T-duality. A direct
calculation in heterotic theory and its S-dualization to type-I theory close the web
of relations among the theories and eventually lead to the conclusion that none of
them admits an asymptotically stable Kasner solution [644]. Monotonic solutions
do exist but they are non-generic [645].
Having excluded a monotonic solution in the presence of inhomogeneities and
n-forms, one may apply the analysis of the mixmaster dynamics of Sect. 6.3.4
to the string case. Kasner epochs with different exponents pQ :D .Qp ; pQ i / are
joined together at the points where the conditions (13.121)–(13.122) are violated,
to produce a pattern of oscillations towards the singularity. Each Kasner epoch is
analogous to the free motion of a billiard ball interrupted by collisions against the
Ktot walls wK D 0 of the pool table, hence the nickname billiards to the model of
approach to the singularity.
The law p ! pQ . p / can be written down in a covariant form and governs
iteratively a chaotic motion on the sphere (13.120). To describe relativistic billiards,
we define the ten Einstein-frame variables
ˇ i :D ln ai (13.124)
in M-theory (i D 1; : : : ; 10) or ˇ i D ln ai and ˇ 10 :D 10 in string theories

(i D 1; : : : ; 9). In each Kasner epoch, ˇ i ' pi . In the string frame, ˇ 10 ! ˇ 0 :D
p
ln. gs e2˚ /. Then, in the limit t ! 0C the action (13.119) evaluated on the
generalized Kasner metric (6.19) with scale factors (13.124) can be approximated
by [646, 649]
Z
dˇ dˇ
SŒˇ D d G V.ˇ/ ; (13.125)
d d
p
where D 1; : : : ; 10 or D 0; : : : ; 9 depending on the frame, d D dt= gD1
and G is a metric with Lorentzian signature P .; C; P; C/ determined by the
model. In D D 11 SUGRA, G dˇ dˇ D .ˇ /2 . ˇ /2 , while in string
theories G D in the string frame. The exponents (13.123) can be rewritten as

the scalar product wK D wK .ˇ/ :D G wK ˇ between the model-dependent vectors

ˇ and wK . All the walls have space-like gradients, G wK wK > 0. The effective

25
potential V reads
X
Ktot
V.ˇ/ ' cK e2wK .ˇ/ ; (13.126)
KD1
where cK > 0 are approximately constant coefficients all positive in the models of
interest; therefore, all the walls are repulsive.
Fortunately, most of the walls are not involved in the billiard motion when
approaching the singularity and they can be neglected to a first approximation. It
turns out that only 10 walls out of O.700/ play a relevant role in this otherwise

tremendously complicated game. We will denote these walls by ˛I .ˇ/ D G ˛I ˇ ,
I D 1; : : : ; 10. For the block of N D 2 supergravities (M-theory, type-IIA and
type-IIB superstrings), they are
˛i .ˇ/ D ˇ iC1 ˇ i ; i D 1; : : : ; 9 ; (13.127a)

˛10 .ˇ/ D ˇ 1 C ˇ 2 C ˇ 3 ; (13.127b)
while for the block of N D 1 supergravities (type-I, heterotic SO.32/ and heterotic
E8 E8 superstrings) the relevant walls in the string frame are
˛1 .ˇ/ D ˇ 1 ; ˛i .ˇ/ D ˇ i ˇ i1 ; i D 2; : : : ; 9 ; (13.128a)

0 7 8 9
˛10 .ˇ/ D ˇ .ˇ C ˇ C ˇ / : (13.128b)
When normalized to 1, the relevant walls (13.127) and (13.128) define unit vectors
normal to the faces of a simplex in the 9-dimensional hyperbolic space H 9 (loosely
speaking, a maximally symmetric Riemannian manifold with constant negative
curvature). The angles between faces are given by the scalar products ˛I ˛I 0 D

G ˛I ˛I0 . The set f˛I g forms a basis of simple roots for a rank-10 hyperbolic
Kac–Moody algebra, while the vectors ˇ parametrize the associated Cartan sub-
algebra.26 For the N D 2 and N D 1 SUGRA blocks, this algebra is, respectively,
E10 and BE10 [646]. The reflections in the walls of the cosmic billiard form a
group corresponding to the Weyl group of E10 or BE10 , i.e., the sub-group of the
isometry group of the root system f˛I g generated by reflections with respect to the
25
The dynamics (13.125) with an exponential potential of the form (13.126) is said to be Toda-like
and has been considered in classical and quantum cosmology since the first descriptions of billiard
systems [638]; see also [662–664].
26
This sudden escalation in the terminology would deserve a digression for which we lack space
here. Instead, we refer to [649, 657] for a brief introduction to Kac–Moody algebras and to [665]
for an in-depth study, in particular Sects. 6.2, 6.4–6.7 and 7.8 therein.
hyperplanes orthogonal to the root vectors. From these properties, it is possible to

show that the evolution towards the singularity is chaotic in a precise mathematical
sense [646]. The emergence of chaos relies on the facts that the billiard H 9 is
hyperbolic and has a finite volume.
More generic billiards and their underlying algebras can be classified according
to the content of the action and to the dimensionality of spacetime [647, 649, 657].
Dimensional reduction of any given model preserves its chaotic properties. The
bosonic sector of D D 4, N D 1; 2; 3; 4; 5; 6; 8 SUGRA has been studied in [650].
Fermions have been included in [651–656].
The physical interpretation of the billiard in M-theory is that of an ensemble of
interacting branes [666–671]. This result relies on an extension of the dynamics of
the above simple billiard model with infinite-potential dominant walls. Including
also the walls neglected at first and, going beyond the Cartan sub-algebra, all the
roots of E10 , one finds a correspondence between the roots of E10 and the content of
M-theory. Arguments in D D 11 SUGRA compactified on hypertori and various
orbifolds suggest the following picture. The walls of the billiard correspond to
background fluxes, with the orientation and position of each wall determined by
the type and magnitude of the associated flux. Fluxes are changed by discrete units
via instantons, among which we can find Kaluza–Klein particles and Euclidean M2-
and M5-branes. Real roots of E10 (i.e., those such that ˛ ˛ > 0; they are all mapped
into simple roots by a transformation ˛ ! ˛I in the Weyl group) represent the
instantons changing the fluxes of the model or, in orbifold compactifications, extra
D-branes contained in the twisted sector. Imaginary roots (˛ ˛ 6 0; they are not
mappable to simple roots through the Weyl group) correspond to physical branes
with Lorentzian world-volume. Inner products ˛ ˛ 0 encode brane interactions.27
13.9.3 Quantum Billiards: Avoiding the Singularity?
The analysis of Sect. 13.9.2 relies on low-energy actions and is, therefore, classical
and perturbative in ˛ 0 . It is thus expected
p to be modified, or to break down
altogether, at time scales t . ts D ˛ 0 of order of the string scale. Assuming
ts D tGUT 1040 s, the number of collisions (the analogue of the BKL oscillations
in Sect. 6.3.4) from today t0 1018 s until ts is about

t0
Ncoll ' ln ' ln ln 5: (13.129)
ts
27
In a complementary approach, real and imaginary roots have been identified with, respectively,
the spatial gradients of SUGRA fields and the higher-order corrections O.Rm / to the leading-˛ 0 -
order low-energy action [648, 657, 672, 673].
Although Ncoll is quite small, the mixing properties of the system are, to quote the
suggestive words of Damour and Henneaux [646], “enough for churning up the
fabric of spacetime and transforming any [. . . ] patch of space into a turbulent foam
at ts .” Below the string scale, non-perturbative and quantum corrections to the low-
energy dynamics could spoil the chaotic model of billiards and even remove the
singularity.
A first and still preliminary inspection of the problem has been conducted on a
model of canonical quantum cosmology on a mini-superspace [674–679]. Consider
pure bosonic gravity in D dimensions, no dilatons and the generalized Kasner metric
(6.19) with scale factors (13.124) (the role of fermionic partners is discussed in
[676]). Take also the decomposition ˇ i D ! i , where is the radial direction in the
future light cone and the ! i .z/ depend on the coordinates z on the hyperbolic space
H D2 [649]. Since the unit hyperboloid is !i ! i D 1, one has 2 D ˇi ˇ i > 0. The
classical Hamiltonian HSUGRA in the canonical variables and ! i and their momenta
can be found readily [649]. The Wheeler–DeWitt equation stems from a canonical
quantization of HSUGRA :

O Œ; z D 0 ;
H HO ' G ij @i @j D 2D @ D2 @ C 2 rH2 D2 ; (13.130)
where rH2 D2 is the Laplacian operator in H D2 . The effect of matter fields is
approximated by sharp potential walls, which we do not see in (13.130) because the
exponential potential terms are negligible in the BKL limit ˇ i ! C1. As in early
studies of cosmological billiards, the walls are encoded in boundary conditions on
the wave-function [680]. Separating variables as Œ; z D r./ f .z/ and imposing
the wave-function to vanish at the boundary, one finds that
q
2
E. D3
2 / ln
e˙i
D3
r./ D 2 ; (13.131)
where E is the eigenvalue in rH2 D2 f .z/ D E f .z/ and is bounded from below by
E > Œ.D 3/=22.
The wave-function is complex and oscillating. In general, wave-packets
travelling towards the singularity at ! C1 spread across the whole billiard
domain and, noting that lim!C1 ReŒr./ D 0C for D > 3, the probability density
j j2 vanishes at the singularity. This result looks similar to the familiar case of
WDW quantum cosmology with three spatial directions, presented in Sect. 10.2. In
particular, in Sect. 10.2.3 we remarked that having D 0 at the singularity is not
enough to solve the big-bang problem, mainly because such wave-functions are not
typical and the lowest eigenvalue of the volume operator is usually zero. Here the
situation is somewhat different because the boundary conditions are fixed by the
billiard problem and having D 0 at the singularity seems to be a robust feature
across several modifications of (13.130) [676–679].
However, this is not the end of the story. In D D 11 SUGRA and its
compactifications, one further requires that the scalar product in H D2 is invariant
under the action of the Weyl group of E10 , which translates into a symmetry
condition on the wave-function,
Œ; z D ˙ Œ; ˛I z : (13.132)
With this condition imposed on the eigenfunction f .z/, becomes a Maass wave-
form, whose actual precise behaviour near the singularity is still shrouded in
mystery. What emerges from this datum and the results below is a purely algebraic
structure near the singularity, which suggests a novel scenario for the birth of the
Universe: the big bang is neither smoothed out nor skipped, it simply cannot be
reached in an ordinary sense [681].
Let us explain. One of the most characteristic features of the BKL conjecture is
that spatial points are completely decoupled near the singularity. When approaching
the big bang, spacetime becomes nearly homogeneous and it experiences a sort of
simplification in the type and number of degrees of freedom. This phase transition
is brought to an extreme in higher-dimensional supergravity models. In [648, 657,
672, 673] and the fermionic extension [651–656], evidence has been gathered that,
near the singularity (t . ts ) and in the strong-coupling limit GD ! 1, all the D D
11 SUGRA dynamical fields (including the n-forms and the spatial metric gij ) are
replaced by just one degree of freedom in one dimension, a homogeneous dynamical
variable v.t/ 2 E10 . Denoting with E10 the group R related to the algebra E10 , the
group element associated with v is .t/ D expŒ dt v.t/ 2 E10 . The one-parameter
dynamics at scales t . ts is governed by the action
Z
dt 1 d ln .t/ d ln .t/ t
SŒ D hP.t/jP.t/i ; P.t/ :D C ;
n.t/ 2 dt dt
(13.133)
where n.t/ is the lapse function, hji is the standard bilinear form on E10 =K.E10 / and
the suffix t denotes the transpose. The solution v.t/ to the dynamics parametrizes a
null-geodesic motion in the infinite-dimensional quotient space E10 =K.E10 /, where
K.E10 / is the maximal compact sub-group of E10 . Elements ek of E10 =K.E10 / are
group elements of E10 3 e left-invariant under K.E10 / 3 k. Equation (13.133) is
known as the E10 =K.E10 / coset model. The classical Hamiltonian Hcoset associated
with (13.133) is formally identical to that of D D 11 SUGRA with a 3-form,
although written with different variables. By studying the fundamental billiard
model with simple roots and its extension to dozens of non-fundamental extra walls,
and including quantum corrections to the SUGRA action, it is possible to establish
a dynamics-preserving map between the SUGRA variables and the coset-space
canonical variables. In particular, one can write .t/ D ˇ i hi .t/ for some specific
hi .t/. A notable feature of the dynamics is that Hcoset vanishes only if the geodesics
are future-oriented, which implies that trajectories cannot go back in ˇ-space.
From the properties of the E10 =K.E10 / coset model, a new picture of the cosmic
singularity emerges [681]. As one gets close to the big bang, space dissolves into a
time-dependent description in terms of a Lie algebra. The notion of intrinsic time
References 795
disappears upon quantization, when the system (13.130) (and its generalizations)
deparametrizes and the role of evolution variable is played by one of the canonical
fields.
In WDW quantum cosmology, the classical trajectory in the space of 3-
geometries passing through the singularity is smeared at D C1 and the
distribution vanishes and is continuous therein (Sect. 10.2.3). In LQC, the
point D C1 is altogether removed from this trajectory, since evolves
discretely and jumps over the singular point D C1 through a cosmic bounce
(Sect. 10.3.7). Here, in contrast, the wave-function neither hits nor jumps through
the singularity: spacetime melts away to be replaced by a purely algebraic harmony.
This would be an original way to realize background independence in quantum
gravity: at the string scale, the Universe would become just symmetry.
In particular, it is possible that E10 is a key symmetry not only of supergravity
(as first conjectured in [682]) but also of full M-theory, both being characterized
by a very rich structure in addition to supersymmetry [683–691]. Once again, the
study of the very early Universe in extreme regimes is inextricably entangled with
the quest for a fundamental theory.
References
1. S. Kachru, R. Kallosh, A.D. Linde, S.P. Trivedi, de Sitter vacua in string theory. Phys. Rev. D
68, 046005 (2003). [arXiv:hep-th/0301240]
2. A.R. Frey, M. Lippert, B. Williams, Fall of stringy de Sitter spacetime. Phys. Rev. D 68,
046008 (2003). [arXiv:hep-th/0305018]
3. C. Escoda, M. Gómez-Reino, F. Quevedo, Saltatory de Sitter string vacua. JHEP 0311, 065
(2003). [arXiv:hep-th/0307160]
4. C.P. Burgess, R. Kallosh, F. Quevedo, de Sitter string vacua from supersymmetric D-terms.
JHEP 0310, 056 (2003). [arXiv:hep-th/0309187]
5. E.I. Buchbinder, B.A. Ovrut, Vacuum stability in heterotic M theory. Phys. Rev. D 69, 086010
(2004). [arXiv:hep-th/0310112]
6. J.F.G. Cascales, M.P. García del Moral, F. Quevedo, A.M. Uranga, Realistic D-brane models
on warped throats: fluxes, hierarchies and moduli stabilization. JHEP 0402, 031 (2004).
7. R. Brustein, S.P. de Alwis, Moduli potentials in string compactifications with fluxes: mapping
the discretuum. Phys. Rev. D 69, 126006 (2004). [arXiv:hep-th/0402088]
8. M. Becker, G. Curio, A. Krause, de Sitter vacua from heterotic M-theory. Nucl. Phys. B 693,
223 (2004). [arXiv:hep-th/0403027]
9. F. Denef, M.R. Douglas, B. Florea, Building a better racetrack. JHEP 0406, 034 (2004).
10. V. Balasubramanian, P. Berglund, Stringy corrections to Kähler potentials, SUSY breaking,
and the cosmological constant problem. JHEP 0411, 085 (2004). [arXiv:hep-th/0408054]
11. R. Kallosh, A.D. Linde, Landscape, the scale of SUSY breaking, and inflation. JHEP 0412,
004 (2004). [arXiv:hep-th/0411011]
12. K. Choi, A. Falkowski, H.P. Nilles, M. Olechowski, S. Pokorski, Stability of flux
compactifications and the pattern of supersymmetry breaking. JHEP 0411, 076 (2004).
13. V. Balasubramanian, P. Berglund, J.P. Conlon, F. Quevedo, Systematics of moduli stabilisa-
tion in Calabi–Yau flux compactifications. JHEP 0503, 007 (2005). [arXiv:hep-th/0502058]
14. K. Choi, A. Falkowski, H.P. Nilles, M. Olechowski, Soft supersymmetry breaking in KKLT
flux compactification. Nucl. Phys. B 718, 113 (2005). [arXiv:hep-th/0503216]
15. J.P. Conlon, F. Quevedo, K. Suruliz, Large-volume flux compactifications: moduli spectrum
and D3/D7 soft supersymmetry breaking. JHEP 0508, 007 (2005). [arXiv:hep-th/0505076]
16. S.B. Giddings, A. Maharana, Dynamics of warped compactifications and the shape of the
warped landscape. Phys. Rev. D 73, 126003 (2006). [arXiv:hep-th/0507158]
17. G. Villadoro, F. Zwirner, de Sitter vacua via consistent D terms. Phys. Rev. Lett. 95, 231602
(2005). [arXiv:hep-th/0508167]
18. J.J. Blanco-Pillado, R. Kallosh, A.D. Linde, Supersymmetry and stability of flux vacua. JHEP
0605, 053 (2006). [arXiv:hep-th/0511042]
19. A. Achúcarro, B. de Carlos, J.A. Casas, L. Doplicher, de Sitter vacua from uplift-
ing D-terms in effective supergravities from realistic strings. JHEP 0606, 014 (2006).
20. O. Lebedev, H.P. Nilles, M. Ratz, de Sitter vacua from matter superpotentials. Phys. Lett. B
636, 126 (2006). [arXiv:hep-th/0603047]
21. J.P. Conlon, F. Quevedo, Gaugino and scalar masses in the landscape. JHEP 0606, 029 (2006).
22. J.P. Conlon, Moduli stabilisation and applications in IIB string theory. Fortsch. Phys. 55, 287
(2007). [arXiv:hep-th/0611039]
23. A. Westphal, de Sitter string vacua from Kähler uplifting. JHEP 0703, 102 (2007).
24. M. Berg, M. Haack, E. Pajer, Jumping through loops: on soft terms from large volume
compactifications. JHEP 0709, 031 (2007). [arXiv:0704.0737]
25. J.P. Conlon, C.H. Kom, K. Suruliz, B.C. Allanach, F. Quevedo, Sparticle spectra and LHC sig-
natures for large volume string compactifications. JHEP 0708, 061 (2007). [arXiv:0704.3403]
26. M. Cicoli, J.P. Conlon, F. Quevedo, General analysis of LARGE volume scenarios with string
loop moduli stabilisation. JHEP 0810, 105 (2008). [arXiv:0805.1029]
27. M. Cicoli, D. Klevers, S. Krippendorf, C. Mayrhofer, F. Quevedo, R. Valandro, Explicit
de Sitter flux vacua for global string models with chiral matter. JHEP 1405, 001 (2014).
[arXiv:1312.0014]
28. D. Ciupke, J. Louis, A. Westphal, Higher-derivative supergravity and moduli stabilization.
JHEP 1510, 094 (2015). [arXiv:1505.03092]
29. L. Susskind, The anthropic landscape of string theory, in Universe or Multiverse? ed. by B.
Carr (Cambridge University Press, Cambridge, 2007). [arXiv:hep-th/0302219]
30. M.R. Douglas, The statistics of string/M theory vacua. JHEP 0305, 046 (2003).
31. S. Ashok, M.R. Douglas, Counting flux vacua. JHEP 0401, 060 (2004).
32. T. Banks, M. Dine, E. Gorbatov, Is there a string theory landscape? JHEP 0408, 058 (2004).
33. M.R. Douglas, B. Shiffman, S. Zelditch, Critical points and supersymmetric vacua. Commun.
Math. Phys. 252, 325 (2004). [arXiv:math/0402326]
34. F. Denef, M.R. Douglas, Distributions of flux vacua. JHEP 0405, 072 (2004).
35. A. Giryavets, S. Kachru, P.K. Tripathy, On the taxonomy of flux vacua. JHEP 0408, 002
(2004). [arXiv:hep-th/0404243]
36. L. Susskind, Supersymmetry breaking in the anthropic landscape, in From Fields to Strings,
ed. by M. Shifman (World Scientific, Singapore, 2005). [arXiv:hep-th/0405189]
37. M.R. Douglas, Statistical analysis of the supersymmetry breaking scale.
38. M. Dine, E. Gorbatov, S.D. Thomas, Low energy supersymmetry from the landscape. JHEP
0808, 098 (2008). [arXiv:hep-th/0407043]
39. B. Freivogel, L. Susskind, Framework for the string theory landscape. Phys. Rev. D 70,
126007 (2004). [arXiv:hep-th/0408133]
References 797
40. M.R. Douglas, Basic results in vacuum statistics. C. R. Phys. 5, 965 (2004).
41. J.P. Conlon, F. Quevedo, On the explicit construction and statistics of Calabi–Yau flux vacua.
JHEP 0410, 039 (2004). [arXiv:hep-th/0409215]
42. R. Blumenhagen, F. Gmeiner, G. Honecker, D. Lüst, T. Weigand, The statistics of supersym-
metric D-brane models. Nucl. Phys. B 713, 83 (2005). [arXiv:hep-th/0411173]
43. F. Denef, M.R. Douglas, Distributions of nonsupersymmetric flux vacua. JHEP 0503, 061
(2005). [arXiv:hep-th/0411183]
44. T. Banks, Landskepticism or why effective potentials don’t count string models.
45. N. Arkani-Hamed, S. Dimopoulos, S. Kachru, Predictive landscapes and new physics at a
TeV. arXiv:hep-th/0501082
46. B.S. Acharya, F. Denef, R. Valandro, Statistics of M theory vacua. JHEP 0506, 056 (2005).
47. M.R. Douglas, B. Shiffman, S. Zelditch, Critical points and supersymmetric vacua, III:
string/M models. Commun. Math. Phys. 265, 617 (2006). [arXiv:math-ph/0506015]
48. J. Gomis, F. Marchesano, D. Mateos, An open string landscape. JHEP 0511, 021 (2005).
49. F. Gmeiner, R. Blumenhagen, G. Honecker, D. Lüst, T. Weigand, One in a billion: MSSM-like
D-brane statistics. JHEP 0601, 004 (2006). [arXiv:hep-th/0510170]
50. K.R. Dienes, Statistics on the heterotic landscape: gauge groups and cosmologi-
cal constants of four-dimensional heterotic strings. Phys. Rev. D 73, 106010 (2006).
51. A. Ceresole, G. Dall’Agata, A. Giryavets, R. Kallosh, A.D. Linde, Domain walls, near-
BPS bubbles, and probabilities in the landscape. Phys. Rev. D 74, 086010 (2006).
52. M.R. Douglas, W. Taylor, The landscape of intersecting brane models. JHEP 0701, 031
(2007). [arXiv:hep-th/0606109]
53. F. Gmeiner, G. Honecker, Millions of standard models on Z 06 ? JHEP 0807, 052 (2008).
[arXiv:0806.3039]
54. H.P. Nilles, S. Ramos-Sánchez, M. Ratz, P.K.S. Vaudrevange, From strings to the MSSM.
Eur. Phys. J. C 59, 249 (2009). [arXiv:0806.3905]
55. C. Asensio, A. Seguí, Applications of an exact counting formula in the Bousso–Polchinski
landscape. Phys. Rev. D 82, 123532 (2010). [arXiv:1003.6011]
56. A. Westphal, Tensor modes on the string theory landscape. JHEP 1304, 054 (2013).
[arXiv:1206.4034]
57. J.J. Blanco-Pillado, M. Gómez-Reino, K. Metallinos, Accidental inflation in the landscape.
JCAP 1302, 034 (2013). [arXiv:1209.0796]
58. A.P. Braun, T. Watari, Distribution of the number of generations in flux compactifications.
Phys. Rev. D 90, 121901 (2014). [arXiv:1408.6156]
59. Y.-H. He, V. Jejjala, L. Pontiggia, Patterns in Calabi–Yau distributions. arXiv:1512.01579
60. M. Graña, Flux compactifications in string theory: a comprehensive review. Phys. Rep. 423,
91 (2006). [arXiv:hep-th/0509003]
61. M.R. Douglas, S. Kachru, Flux compactification. Rev. Mod. Phys. 79, 733 (2007).
63. F. Denef, M.R. Douglas, S. Kachru, Physics of string flux compactifications. Ann. Rev. Nucl.
Part. Sci. 57, 119 (2007). [arXiv:hep-th/0701050]
64. F. Denef, Course 12 – Lectures on constructing string vacua. Les Houches 87, 483 (2008).
[arXiv:0803.1194]
65. R. Bousso, J. Polchinski, Quantization of four-form fluxes and dynamical neutralization of
the cosmological constant. JHEP 0006, 006 (2000). [arXiv:hep-th/0004134]
66. J.L. Feng, J. March-Russell, S. Sethi, F. Wilczek, Saltatory relaxation of the cosmological
constant. Nucl. Phys. B 602, 307 (2001). [arXiv:hep-th/0005276]
67. N. Kaloper, L. Sorbo, Where in the string landscape is quintessence? Phys. Rev. D 79, 043528
(2009). [arXiv:0810.5346]
68. S. Panda, Y. Sumitomo, S.P. Trivedi, Axions as quintessence in string theory. Phys. Rev. D
83, 083506 (2011). [arXiv:1011.5877]
69. G. Gupta, S. Panda, A.A. Sen, Observational constraints on axions as quintessence in string
70. J.P. Conlon, F. Quevedo, Kähler moduli inflation. JHEP 0601, 146 (2006).
71. J.R. Bond, L. Kofman, S. Prokushkin, P.M. Vaudrevange, Roulette inflation with Kähler
moduli and their axions. Phys. Rev. D 75, 123511 (2007). [arXiv:hep-th/0612197]
72. Z. Lalak, D. Langlois, S. Pokorski, K. Turzynski, Curvature and isocurvature perturbations in
two-field inflation. JCAP 0707, 014 (2007). [arXiv:0704.0212]
73. M. Cicoli, C.P. Burgess, F. Quevedo, Fibre inflation: observable gravity waves from IIB string
compactifications. JCAP 0903, 013 (2009). [arXiv:0808.0691]
74. J.J. Blanco-Pillado, D. Buck, E.J. Copeland, M. Gómez-Reino, N.J. Nunes, Kähler moduli
inflation revisited. JHEP 1001, 081 (2010). [arXiv:0906.3711]
75. C.P. Burgess, M. Cicoli, M. Gómez-Reino, F. Quevedo, G. Tasinato, I. Zavala, Non-standard
primordial fluctuations and nongaussianity in string inflation. JHEP 1008, 045 (2010).
[arXiv:1005.4840]
76. M. Cicoli, A. Mazumdar, Reheating for closed string inflation. JCAP 1009, 025 (2010).
[arXiv:1005.5076]
77. M. Cicoli, F.G. Pedro, G. Tasinato, Poly-instanton inflation. JCAP 1112, 022 (2011).
[arXiv:1110.6182]
78. B.J. Broy, D. Ciupke, F.G. Pedro, A. Westphal, Starobinsky-type inflation from ˛ 0 -
corrections. JCAP 1601, 001 (2016). [arXiv:1509.00024]
79. C.P. Burgess, M. Cicoli, S. de Alwis, F. Quevedo, Robust inflation from fibrous strings. JCAP
1605, 032 (2016). [arXiv:1603.06789]
80. M. Cicoli, F. Muia, P. Shukla, Global embedding of fibre inflation models. arXiv:1611.04612
81. A.D. Linde, A. Westphal, Accidental inflation in string theory. JCAP 0803, 005 (2008).
[arXiv:0712.1610]
82. J.P. Conlon, R. Kallosh, A.D. Linde, F. Quevedo, Volume modulus inflation and the gravitino
mass problem. JCAP 0809, 011 (2008). [arXiv:0806.0809]
83. M. Cicoli, F. Muia, F.G. Pedro, Microscopic origin of volume modulus inflation. JCAP 1512,
040 (2015). [arXiv:1509.07748]
84. Z. Lalak, G.G. Ross, S. Sarkar, Racetrack inflation and assisted moduli stabilisation. Nucl.
85. L. Alabidi, D.H. Lyth, Inflation models and observation. JCAP 0605, 016 (2006).
86. J.J. Blanco-Pillado, C.P. Burgess, J.M. Cline, C. Escoda, M. Gómez-Reino, R. Kallosh,
A.D. Linde, F. Quevedo, Racetrack inflation. JHEP 0411, 063 (2004). [arXiv:hep-th/0406230]
87. J.J. Blanco-Pillado, C.P. Burgess, J.M. Cline, C. Escoda, M. Gómez-Reino, R. Kallosh,
A.D. Linde, F. Quevedo, Inflating in a better racetrack. JHEP 0609, 002 (2006).
88. R. Kallosh, On inflation in string theory. Lect. Notes Phys. 738, 119 (2008).
89. R. Kallosh, N. Sivanandam, M. Soroush, Axion inflation and gravity waves in string theory.
Phys. Rev. D 77, 043501 (2008). [arXiv:0710.3429]
90. S. Dimopoulos, S. Kachru, J. McGreevy, J.G. Wacker, N-flation. JCAP 0808, 003 (2008).
91. R. Easther, L. McAllister, Random matrices and the spectrum of N-flation. JCAP 0605, 018
(2006). [arXiv:hep-th/0512102]
92. S.A Kim, A.R. Liddle, N-flation: multi-field inflationary dynamics and perturbations. Phys.
References 799
93. Y.-S. Piao, Perturbation spectra of “N-flation”. Phys. Rev. D 74, 047302 (2006).
94. S.A Kim, A.R. Liddle, N-flation: non-Gaussianity in the horizon-crossing approximation.
95. T. Battefeld, R. Easther, Non-Gaussianities in multi-field inflation. JCAP 0703, 020 (2007).
96. D. Battefeld, T. Battefeld, Non-Gaussianities in N-flation. JCAP 0705, 012 (2007).
97. S.A Kim, A.R. Liddle, N-flation: observable predictions from the random matrix mass
spectrum. Phys. Rev. D 76, 063515 (2007). [arXiv:0707.1982]
98. D.R. Green, Reheating closed string inflation. Phys. Rev. D 76, 103504 (2007).
[arXiv:0707.3832]
99. T.W. Grimm, Axion inflation in type II string theory. Phys. Rev. D 77, 126007 (2008).
[arXiv:0710.3883]
100. S.A Kim, A.R. Liddle, D. Seery, Non-Gaussianity in axion N-flation models. Phys. Rev. Lett.
105, 181302 (2010). [arXiv:1005.4410]
101. S.A Kim, A.R. Liddle, D. Seery, Non-Gaussianity in axion N-flation models: detailed
predictions and mass spectra. Phys. Rev. D 85, 023532 (2012). [arXiv:1108.2944]
102. M. Cicoli, K. Dutta, A. Maharana, N-flation with hierarchically light axions in string
compactifications. JCAP 1408, 012 (2014). [arXiv:1401.2579]
103. T.W. Grimm, Axion inflation in F-theory. Phys. Lett. B 739, 201 (2014). [arXiv:1404.4268]
104. T. Rudelius, On the possibility of large axion moduli spaces. JCAP 1504, 049 (2015).
[arXiv:1409.5793]
105. J. Brown, W. Cottrell, G. Shiu, P. Soler, Fencing in the swampland: quantum gravity
constraints on large field inflation. JHEP 1510, 023 (2015). [arXiv:1503.04783]
106. J.E. Kim, H.P. Nilles, M. Peloso, Completing natural inflation. JCAP 0501, 005 (2005).
107. M. Berg, E. Pajer, S. Sjors, Dante’s Inferno. Phys. Rev. D 81, 103535 (2010).
[arXiv:0912.1341]
108. K. Choi, H. Kim, S. Yun, Natural inflation with multiple sub-Planckian axions. Phys. Rev. D
90, 023545 (2014). [arXiv:1404.6209]
109. T. Higaki, F. Takahashi, Natural and multi-natural inflation in axion landscape. JHEP 1407,
074 (2014). [arXiv:1404.6923]
110. R. Kappl, S. Krippendorf, H.P. Nilles, Aligned natural inflation: monodromies of two axions.
Phys. Lett. B 737, 124 (2014). [arXiv:1404.7127]
111. I. Ben-Dayan, F.G. Pedro, A. Westphal, Hierarchical axion inflation. Phys. Rev. Lett. 113,
261301 (2014). [arXiv:1404.7773]
112. C. Long, L. McAllister, P. McGuirk, Aligned natural inflation in string theory. Phys. Rev. D
90, 023501 (2014). [arXiv:1404.7852]
113. X. Gao, T. Li, P. Shukla, Combining universal and odd RR axions for aligned natural inflation.
JCAP 1410, 048 (2014). [arXiv:1406.0341]
114. I. Ben-Dayan, F.G. Pedro, A. Westphal, Towards natural inflation in string theory. Phys. Rev.
D 92, 023515 (2015). [arXiv:1407.2562]
115. T.C. Bachlechner, C. Long, L. McAllister, Planckian axions in string theory. JHEP 1512, 042
(2015). [arXiv:1412.1093]
116. T. Rudelius, Constraints on axion inflation from the weak gravity conjecture. JCAP 1509, 020
(2015). [arXiv:1503.00795]
117. M. Montero, A.M. Uranga, I. Valenzuela, Transplanckian axions!? JHEP 1508, 032 (2015).
[arXiv:1503.03886]
118. A. Hebecker, P. Mangat, F. Rompineve, L.T. Witkowski, Winding out of the swamp: evading
the weak gravity conjecture with F-term winding inflation? Phys. Lett. B 748, 455 (2015).
[arXiv:1503.07912]
119. E. Palti, On natural inflation and moduli stabilisation in string theory. JHEP 1510, 188 (2015).
[arXiv:1508.00009]
120. J. Brown, W. Cottrell, G. Shiu, P. Soler, On axionic field ranges, loopholes and the weak
gravity conjecture. JHEP 1604, 017 (2016). [arXiv:1504.00659]
121. E. Silverstein, A. Westphal, Monodromy in the CMB: gravity waves and string inflation. Phys.
Rev. D 78, 106003 (2008). [arXiv:0803.3085]
122. L. McAllister, E. Silverstein, A. Westphal, Gravity waves and linear inflation from axion
monodromy. Phys. Rev. D 82, 046003 (2010). [arXiv:0808.0706]
123. R. Flauger, L. McAllister, E. Pajer, A. Westphal, G. Xu, Oscillations in the CMB from axion
monodromy inflation. JCAP 1006, 009 (2010). [arXiv:0907.2916]
124. S. Hannestad, T. Haugbølle, P.R. Jarnhus, M.S. Sloth, Non-Gaussianity from axion mon-
odromy inflation. JCAP 1006, 001 (2010). [arXiv:0912.3527]
125. R. Flauger, E. Pajer, Resonant non-Gaussianity. JCAP 1101, 017 (2011). [arXiv:1002.0833]
126. X. Dong, B. Horn, E. Silverstein, A. Westphal, Simple exercises to flatten your potential.
Phys. Rev. D 84, 026011 (2011). [arXiv:1011.4521]
127. J.P. Conlon, Brane-antibrane backreaction in axion monodromy inflation. JCAP 1201, 033
(2012). [arXiv:1110.6454]
128. H. Peiris, R. Easther, R. Flauger, Constraining monodromy inflation. JCAP 1309, 018 (2013).
[arXiv:1303.2616]
129. F. Marchesano, G. Shiu, A.M. Uranga, F-term axion monodromy inflation. JHEP 1409, 184
(2014). [arXiv:1404.3040]
130. R. Blumenhagen, E. Plauschinn, Towards universal axion inflation and reheating in string
theory. Phys. Lett. B 736, 482 (2014). [arXiv:1404.3542]
131. A. Hebecker, S.C. Kraus, L.T. Witkowski, D7-brane chaotic inflation. Phys. Lett. B 737, 16
(2014). [arXiv:1404.3711]
132. M. Arends, A. Hebecker, K. Heimpel, S.C. Kraus, D. Lüst, C. Mayrhofer, C. Schick, T.
Weigand, D7-brane moduli space in axion monodromy and fluxbrane inflation. Fortsch. Phys.
62, 647 (2014). [arXiv:1405.0283]
133. L. McAllister, E. Silverstein, A. Westphal, T. Wrase, The powers of monodromy. JHEP 1409,
123 (2014). [arXiv:1405.3652]
134. R. Blumenhagen, D. Herschmann, E. Plauschinn, The challenge of realizing F-term axion
monodromy inflation in string theory. JHEP 1501, 007 (2015). [arXiv:1409.7075]
135. A. Hebecker, P. Mangat, F. Rompineve, L.T. Witkowski, Tuning and backreaction in F-term
axion monodromy inflation. Nucl. Phys. B 894, 456 (2015). [arXiv:1411.2032]
136. R. Flauger, L. McAllister, E. Silverstein, A. Westphal, Drifting oscillations in axion mon-
odromy. arXiv:1412.1814
137. R. Blumenhagen, A. Font, M. Fuchs, D. Herschmann, E. Plauschinn, Towards axionic
Starobinsky-like inflation in string theory. Phys. Lett. B 746, 217 (2015). [arXiv:1503.01607]
138. R. Blumenhagen, A. Font, M. Fuchs, D. Herschmann, E. Plauschinn, Y. Sekiguchi, F. Wolf, A
flux-scaling scenario for high-scale moduli stabilization in string theory. Nucl. Phys. B 897,
500 (2015). [arXiv:1503.07634]
139. D. Escobar, A. Landete, F. Marchesano, D. Regalado, Large field inflation from D-branes.
Phys. Rev. D 93, 081301 (2016). [arXiv:1505.07871]
140. D. Andriot, A no-go theorem for monodromy inflation. JCAP 1603, 025 (2016).
[arXiv:1510.02005]
141. D. Escobar, A. Landete, F. Marchesano, D. Regalado, D6-branes and axion monodromy
inflation. JHEP 1603, 113 (2016). [arXiv:1511.08820]
142. A. Hebecker, F. Rompineve, A. Westphal, Axion monodromy and the weak gravity conjec-
ture. JHEP 1604, 157 (2016). [arXiv:1512.03768]
143. A. Hebecker, J. Moritz, A. Westphal, L.T. Witkowski, Axion monodromy inflation with
warped KK-modes. Phys. Lett. B 754, 328 (2016). [arXiv:1512.04463]
144. S. Kachru, R. Kallosh, A.D. Linde, J.M. Maldacena, L.P. McAllister, S.P. Trivedi, Towards
inflation in string theory. JCAP 0310, 013 (2003). [arXiv:hep-th/0308055]
145. J.P. Hsu, R. Kallosh, S. Prokushkin, On brane inflation with volume stabilization. JCAP 0312,
009 (2003). [arXiv:hep-th/0311077]
References 801
146. A. Buchel, R. Roiban, Inflation in warped geometries. Phys. Lett. B 590, 284 (2004).
147. H. Firouzjahi, S.-H.H. Tye, Closer towards inflation in string theory. Phys. Lett. B 584, 147
(2004). [arXiv:hep-th/0312020]
148. J.P. Hsu, R. Kallosh, Volume stabilization and the origin of the inflaton shift symmetry in
string theory. JHEP 0404, 042 (2004). [arXiv:hep-th/0402047]
149. C.P. Burgess, J.M. Cline, H. Stoica, F. Quevedo, Inflation in realistic D-brane models. JHEP
0409, 033 (2004). [arXiv:hep-th/0403119]
150. O. DeWolfe, S. Kachru, H.L. Verlinde, The giant inflaton. JHEP 0405, 017 (2004).
151. N. Iizuka, S.P. Trivedi, An inflationary model in string theory. Phys. Rev. D 70, 043519
(2004). [arXiv:hep-th/0403203]
152. M. Berg, M. Haack, B. Körs, Loop corrections to volume moduli and inflation in string theory.
153. K. Dasgupta, J.P. Hsu, R. Kallosh, A.D. Linde, M. Zagermann, D3/D7 brane inflation and
semilocal strings. JHEP 0408, 030 (2004). [arXiv:hep-th/0405247]
154. U. Seljak, A. Slosar, B polarization of cosmic microwave background as a tracer of strings.
155. D. Baumann, A. Dymarsky, I.R. Klebanov, L. McAllister, P.J. Steinhardt, A delicate
universe: compactification obstacles to D-brane inflation. Phys. Rev. Lett. 99, 141601 (2007).
[arXiv:0705.3837]
156. A. Krause, E. Pajer, Chasing brane inflation in string theory. JCAP 0807, 023 (2008).
[arXiv:0705.4682]
157. D. Baumann, A. Dymarsky, I.R. Klebanov, L. McAllister, Towards an explicit model of D-
brane inflation. JCAP 0801, 024 (2008). [arXiv:0706.0360]
158. S. Panda, M. Sami, S. Tsujikawa, Prospects of inflation in delicate D-brane cosmology. Phys.
Rev. D 76, 103512 (2007). [arXiv:0707.2848]
159. L. Hoi, J.M. Cline, How delicate is brane-antibrane inflation? Phys. Rev. D 79, 083537 (2009).
[arXiv:0810.1303]
160. D. Baumann, A. Dymarsky, S. Kachru, I.R. Klebanov, L. McAllister, Holographic systematics
of D-brane inflation. JHEP 0903, 093 (2009). [arXiv:0808.2811]
161. N. Agarwal, R. Bean, L. McAllister, G. Xu, Universality in D-brane inflation. JCAP 1109,
002 (2011). [arXiv:1103.2775]
162. M. Dias, J. Frazer, A.R. Liddle, Multifield consequences for D-brane inflation. JCAP 1206,
020 (2012); Erratum-ibid. 1303, E01 (2013). [arXiv:1203.3792]
163. D. Battefeld, T. Battefeld, S. Schulz, On the unlikeliness of multi-field inflation: bounded
random potentials and our vacuum. JCAP 1206, 034 (2012). [arXiv:1203.3941]
164. L. McAllister, S. Renaux-Petel, G. Xu, A statistical approach to multifield inflation: many-
field perturbations beyond slow roll. JCAP 1210, 046 (2012). [arXiv:1207.0317]
165. D. Battefeld, T. Battefeld, A smooth landscape: ending saddle point inflation requires features
to be shallow. JCAP 1307, 038 (2013). [arXiv:1304.0461]
166. D. Baumann, A. Dymarsky, S. Kachru, I.R. Klebanov, L. McAllister, D3-brane potentials
from fluxes in AdS/CFT. JHEP 1006, 072 (2010). [arXiv:1001.5028]
167. A. Ali, A. Deshamukhya, S. Panda, M. Sami, Inflation with improved D3-brane potential and
the fine tunings associated with the model. Eur. Phys. J. C 71, 1672 (2011). [arXiv:1010.1407]
168. E. Silverstein, D. Tong, Scalar speed limits and cosmology: acceleration from D-cceleration.
169. M. Alishahiha, E. Silverstein, D. Tong, DBI in the sky: non-Gaussianity from inflation with a
speed limit. Phys. Rev. D 70, 123505 (2004). [arXiv:hep-th/0404084]
170. X. Chen, Multithroat brane inflation. Phys. Rev. D 71, 063506 (2005).
171. X. Chen, Inflation from warped space. JHEP 0508, 045 (2005). [arXiv:hep-th/0501184]
172. X. Chen, Running non-Gaussianities in Dirac–Born–Infeld inflation. Phys. Rev. D 72, 123518
(2005). [arXiv:astro-ph/0507053]
173. X. Chen, M.-x. Huang, S. Kachru, G. Shiu, Observational signatures and non-Gaussianities
of general single field inflation. JCAP 0701, 002 (2007). [arXiv:hep-th/0605045]
174. S. Kecskemeti, J. Maiden, G. Shiu, B. Underwood, DBI inflation in the tip region of a warped
throat. JHEP 0609, 076 (2006). [arXiv:hep-th/0605189]
175. G. Shiu, B. Underwood, Observing the geometry of warped compactification via cosmic
inflation. Phys. Rev. Lett. 98, 051301 (2007). [arXiv:hep-th/0610151]
176. S. Thomas, J. Ward, IR inflation from multiple branes. Phys. Rev. D 76, 023509 (2007).
177. J.E. Lidsey, I. Huston, Gravitational wave constraints on Dirac–Born–Infeld inflation. JCAP
0707, 002 (2007). [arXiv:0705.0240]
178. H.V. Peiris, D. Baumann, B. Friedman, A. Cooray, Phenomenology of D-brane inflation with
general speed of sound. Phys. Rev. D 76, 103517 (2007). [arXiv:0706.1240]
179. T. Kobayashi, S. Mukohyama, S. Kinoshita, Constraints on wrapped DBI inflation in a warped
throat. JCAP 0801, 028 (2008). [arXiv:0708.4285]
180. M. Becker, L. Leblond, S.E. Shandera, Inflation from wrapped branes. Phys. Rev. D 76,
123516 (2007). [arXiv:0709.1170]
181. D.A. Easson, R. Gregory, D.F. Mota, G. Tasinato, I. Zavala, Spinflation. JCAP 0802, 010
(2008). [arXiv:0709.2666]
182. M.-x. Huang, G. Shiu, B. Underwood, Multifield Dirac–Born–Infeld inflation and non-
Gaussianities. Phys. Rev. D 77, 023511 (2008). [arXiv:0709.3299]
183. R. Bean, X. Chen, H. Peiris, J. Xu, Comparing infrared Dirac–Born–Infeld brane inflation to
observations. Phys. Rev. D 77, 023527 (2008). [arXiv:0710.1812]
184. D. Langlois, S. Renaux-Petel, D.A. Steer, T. Tanaka, Primordial fluctuations and non-
Gaussianities in multifield Dirac–Born–Infeld inflation. Phys. Rev. Lett. 101, 061301 (2008).
[arXiv:0804.3139]
185. D. Langlois, S. Renaux-Petel, D.A. Steer, T. Tanaka, Primordial perturbations and non-
Gaussianities in DBI and general multifield inflation. Phys. Rev. D 78, 063523 (2008).
[arXiv:0806.0336]
186. F. Arroja, S. Mizuno, K. Koyama, Non-Gaussianity from the bispectrum in general multiple
field inflation. JCAP 0808, 015 (2008). [arXiv:0806.0619]
187. D. Langlois, S. Renaux-Petel, D.A. Steer, Multi-field DBI inflation: introducing bulk forms
and revisiting the gravitational wave constraints. JCAP 0904, 021 (2009). [arXiv:0902.2941]
188. S. Mizuno, F. Arroja, K. Koyama, T. Tanaka, Lorentz boost and non-Gaussianity in multifield
DBI inflation. Phys. Rev. D 80, 023530 (2009). [arXiv:0905.4557]
189. S. Mizuno, F. Arroja, K. Koyama, Full quantum trispectrum in multifield DBI inflation. Phys.
Rev. D 80, 083517 (2009). [arXiv:0907.2439]
190. S. Renaux-Petel, Combined local and equilateral non-Gaussianities from multifield DBI
191. L. Lorenz, J. Martin, J. Yokoyama, Geometrically consistent approach to stochastic DBI
192. S.E. Shandera, S.-H.H. Tye, Observing brane inflation. JCAP 0605, 007 (2006).
193. D. Baumann, L. McAllister, A microscopic limit on gravitational waves from D-brane
194. R. Bean, S.E. Shandera, S.-H.H. Tye, J. Xu, Comparing brane inflation to WMAP. JCAP
0705, 004 (2007). [arXiv:hep-th/0702107]
195. S.-H.H. Tye, Brane inflation: string theory viewed from the cosmos. Lect. Notes Phys. 737,
949 (2008). [arXiv:hep-th/0610221]
196. L. McAllister, E. Silverstein, String cosmology: a review. Gen. Relat. Grav. 40, 565 (2008).
[arXiv:0710.2951]
197. M. Cicoli, F. Quevedo, String moduli inflation: an overview. Class. Quantum Grav. 28,
204001 (2011). [arXiv:1108.2659]
198. C.P. Burgess, L. McAllister, Challenges for string cosmology. Class. Quantum Grav. 28,
204002 (2011). [arXiv:1108.2660]
References 803
199. E.J. Copeland, L. Pogosian, T. Vachaspati, Seeking string theory in the cosmos. Class.
200. E. Pajer, M. Peloso, A review of axion inflation in the era of Planck. Class. Quantum Grav.
30, 214002 (2013). [arXiv:1305.3557]
201. C.P. Burgess, M. Cicoli, F. Quevedo, String inflation after Planck 2013. JCAP 1311, 003
(2013). [arXiv:1306.3512]
203. O. DeWolfe, A. Giryavets, S. Kachru, W. Taylor, Type IIA moduli stabilization. JHEP 0507,
066 (2005). [arXiv:hep-th/0505160]
204. S.B. Giddings, S. Kachru, J. Polchinski, Hierarchies from fluxes in string compactifications.
205. S. Sethi, C. Vafa, E. Witten, Constraints on low-dimensional string compactifications. Nucl.
206. S. Kachru, J. Pearson, H.L. Verlinde, Brane/flux annihilation and the string dual of a non-
supersymmetric field theory. JHEP 0206, 021 (2002). [arXiv:hep-th/0112197]
207. J.M. Maldacena, H.S. Nastase, The supergravity dual of a theory with dynamical supersym-
metry breaking. JHEP 0109, 024 (2001). [arXiv:hep-th/0105049]
208. A. Saltman, E. Silverstein, The scaling of the no-scale potential and de-Sitter model building.
JHEP 0411, 066 (2004). [arXiv:hep-th/0402135]
209. S.R. Coleman, Fate of the false vacuum: semiclassical theory. Phys. Rev. D 15, 2929 (1977);
Erratum-ibid. D 16, 1248 (1977)
210. S.R. Coleman, F. De Luccia, Gravitational effects on and of vacuum decay. Phys. Rev. D 21,
3305 (1980)
211. N. Goheer, M. Kleban, L. Susskind, The trouble with de Sitter space. JHEP 0307, 056 (2003).
212. D. Robbins, S. Sethi, A barren landscape?—Metastable de Sitter vacua are nongeneric in
string theory. Phys. Rev. D 71, 046008 (2005). [arXiv:hep-th/0405011]
213. F. Saueressig, U. Theis, S. Vandoren, On de Sitter vacua in type IIA orientifold compactifica-
tions. Phys. Lett. B 633, 125 (2006). [arXiv:hep-th/0506181]
214. V. Braun, B.A. Ovrut, Stabilizing moduli with a positive cosmological constant in heterotic
M-theory. JHEP 0607, 035 (2006). [arXiv:hep-th/0603088]
215. E.I. Buchbinder, Raising anti-de Sitter vacua to de Sitter vacua in heterotic M theory. Phys.
Rev. D 70, 066008 (2004). [arXiv:hep-th/0406101]
216. R. Blumenhagen, S. Moster, E. Plauschinn, Moduli stabilisation versus chirality for MSSM
like type IIB orientifolds. JHEP 0801, 058 (2008). [arXiv:arXiv:0711.3389]
217. E. Dudas, S.K. Vempati, Large D-terms, hierarchical soft spectra and moduli stabilisation.
218. M.J. Duff, P. van Nieuwenhuizen, Quantum inequivalence of different field representations.
Phys. Lett. B 94, 179 (1980)
219. A. Aurilia, H. Nicolai, P.K. Townsend, Hidden constants: the parameter of QCD and the
cosmological constant of N D 8 supergravity. Nucl. Phys. B 176, 509 (1980)
220. M. Henneaux, C. Teitelboim, The cosmological constant as a canonical variable. Phys. Lett.
B 143, 415 (1984)
221. J.D. Brown, C. Teitelboim, Dynamical neutralization of the cosmological constant. Phys. Lett.
B 195, 177 (1987)
222. J.D. Brown, C. Teitelboim, Neutralization of the cosmological constant by membrane
creation. Nucl. Phys. B 297, 787 (1988)
223. J. Polchinski, A. Strominger, New vacua for type II string theory. Phys. Lett. B 388, 736
(1996). [arXiv:hep-th/9510227]
224. M.J. Duff, R.R. Khuri, J.X. Lu, String solitons. Phys. Rep. 259, 213 (1995).
225. B. Carter, Large number coincidences and the anthropic principle in cosmology, in IAU
Symposium 63: Confrontation of Cosmological Theories with Observational Data, ed. by
M.S. Longair (Reidel, Dordrecht, 1974)
226. S. Weinberg, Anthropic bound on the cosmological constant. Phys. Rev. Lett. 59, 2607 (1987)
228. H. Martel, P.R. Shapiro, S. Weinberg, Likely values of the cosmological constant. Astrophys.
J. 492, 29 (1998). [arXiv:astro-ph/9701099]
229. J. Garriga, A. Vilenkin, On likely values of the cosmological constant. Phys. Rev. D 61,
083502 (2000). [arXiv:astro-ph/9908115]
230. V. Agrawal, S.M. Barr, J.F. Donoghue, D. Seckel, Viable range of the mass scale of the
standard model. Phys. Rev. D 57, 5480 (1998). [arXiv:hep-ph/9707380]
232. W.R. Stoeger, G.F.R. Ellis, U. Kirchner, Multiverses and cosmology: philosophical issues.
233. B. Freivogel, Making predictions in the multiverse. Class. Quantum Grav. 28, 204007 (2011).
[arXiv:1105.0244]
234. T. Banks, M. Dine, L. Motl, On anthropic solutions of the cosmological constant problem.
JHEP 0101, 031 (2001). [arXiv:hep-th/0007206]
235. M.L. Graesser, S.D.H. Hsu, A. Jenkins, M.B. Wise, Anthropic distribution for cosmo-
logical constant and primordial density perturbations. Phys. Lett. B 600, 15 (2004).
236. H. Firouzjahi, S. Sarangi, S.-H.H. Tye, Spontaneous creation of inflationary universes and the
cosmic landscape. JHEP 0409, 060 (2004). [arXiv:hep-th/0406107]
237. N. Arkani-Hamed, S. Dimopoulos, G.R. Dvali, The hierarchy problem and new dimensions
at a millimeter. Phys. Lett. B 429, 263 (1998). [arXiv:hep-ph/9803315]
238. N. Arkani-Hamed, S. Dimopoulos, G.R. Dvali, Phenomenology, astrophysics and cosmology
of theories with submillimeter dimensions and TeV scale quantum gravity. Phys. Rev. D 59,
086004 (1999). [arXiv:hep-ph/9807344]
239. I. Antoniadis, N. Arkani-Hamed, S. Dimopoulos, G.R. Dvali, New dimensions at a millimeter
to a Fermi and superstrings at a TeV. Phys. Lett. B 436, 257 (1998). [arXiv:hep-ph/9804398]
240. B. Freivogel, M. Kleban, M. Rodríguez Martínez, L. Susskind, Observational consequences
of a landscape. JHEP 0603, 039 (2006). [arXiv:hep-th/0505232]
241. D. Yamauchi, A. Linde, A. Naruko, M. Sasaki, T. Tanaka, Open inflation in the landscape.
Phys. Rev. D 84, 043513 (2011). [arXiv:1105.2674]
242. A. De Simone, M.P. Salem, Distribution of ˝k from the scale-factor cutoff measure. Phys.
Rev. D 81, 083527 (2010). [arXiv:0912.3783]
243. R. Bousso, D. Harlow, L. Senatore, Inflation after false vacuum decay: observational
prospects after Planck. Phys. Rev. D 91, 083527 (2015). [arXiv:1309.4060]
244. M. Cicoli, S. Downes, B. Dutta, F.G. Pedro, A. Westphal, Just enough inflation: power
spectrum modifications at large scales. JCAP 1412, 030 (2014). [arXiv:1407.1048]
245. C.R. Contaldi, M. Peloso, L. Kofman, A.D. Linde, Suppressing the lower multipoles in the
CMB anisotropies. JCAP 0307, 002 (2003). [arXiv:astro-ph/0303636]
246. N. Arkani-Hamed, L. Motl, A. Nicolis, C. Vafa, The string landscape, black holes and gravity
as the weakest force. JHEP 0706, 060 (2007). [arXiv:hep-th/0601001]
247. T. Banks, M. Johnson, A. Shomer, A note on gauge theories coupled to gravity. JHEP 0609,
049 (2006). [arXiv:hep-th/0606277]
248. G. Dvali, Black holes and large N species solution to the hierarchy problem. Fortsch. Phys.
58, 528 (2010). [arXiv:0706.2050]
249. C. Cheung, G.N. Remmen, Naturalness and the weak gravity conjecture. Phys. Rev. Lett. 113,
051601 (2014). [arXiv:1402.2287]
250. C. Cheung, G.N. Remmen, Infrared consistency and the weak gravity conjecture. JHEP 1412,
087 (2014). [arXiv:1407.7865]
251. F.G. Pedro, A. Westphal, The scale of inflation in the landscape. Phys. Lett. B 739, 439 (2014).
[arXiv:1303.3224]
252. C.P. Burgess, R. Easther, A. Mazumdar, D.F. Mota, T. Multamäki, Multiple inflation, cosmic
string networks and the string landscape. JHEP 0505, 067 (2005). [arXiv:hep-th/0501125]
References 805
253. K. Freese, D. Spolyar, Chain inflation in the landscape: ‘Bubble bubble toil and trouble’.
JCAP 0507, 007 (2005). [arXiv:hep-ph/0412145]
254. K. Freese, J.T. Liu, D. Spolyar, Chain inflation via rapid tunneling in the landscape.
255. Q.-G. Huang, Simplified chain inflation. JCAP 0705, 009 (2007). [arXiv:0704.2835]
256. D. Chialva, U.H. Danielsson, Chain inflation revisited. JCAP 0810, 012 (2008).
[arXiv:0804.2846]
257. J.M. Cline, G.D. Moore, Y. Wang, Chain inflation reconsidered. JCAP 1108, 032 (2011).
[arXiv:1106.2188]
258. R. Easther, Folded inflation, primordial tensors, and the running of the scalar spectral index.
259. J. Frazer, A.R. Liddle, Exploring a string-like landscape. JCAP 1102, 026 (2011).
[arXiv:1101.1619]
260. J. Frazer, A.R. Liddle, Multi-field inflation with random potentials: field dimension, feature
scale and non-Gaussianity. JCAP 1202, 039 (2012). [arXiv:1111.6646]
261. G.D. Coughlan, W. Fischler, E.W. Kolb, S. Raby, G.G. Ross, Cosmological problems for the
Polonyi potential. Phys. Lett. B 131, 59 (1983)
262. J.R. Ellis, D.V. Nanopoulos, M. Quirós, On the axion, dilaton, Polonyi, gravitino and shadow
matter problems in supergravity and superstring models. Phys. Lett. B 174, 176 (1986)
263. T. Banks, D.B. Kaplan, A.E. Nelson, Cosmological implications of dynamical supersymmetry
breaking. Phys. Rev. D 49, 779 (1994). [arXiv:hep-ph/9308292]
264. B. de Carlos, J.A. Casas, F. Quevedo, E. Roulet, Model-independent properties and cosmo-
logical implications of the dilaton and moduli sectors of 4D strings. Phys. Lett. B 318, 447
(1993). [arXiv:hep-ph/9308325]
265. M. Endo, M. Yamaguchi, K. Yoshioka, Bottom-up approach to moduli dynamics in heavy
gravitino scenario: superpotential, soft terms, and sparticle mass spectrum. Phys. Rev. D 72,
015004 (2005). [arXiv:hep-ph/0504036]
266. K.-i. Maeda, M.D. Pollock, On inflation in the heterotic superstring model. Phys. Lett. B 173,
251 (1986)
267. J.R. Ellis, K. Enqvist, D.V. Nanopoulos, M. Quiros, Evolution with temperature and the
possibility of inflation from the superstring in four dimensions. Nucl. Phys. B 277, 231 (1986)
268. P. Binétruy, M.K. Gaillard, Candidates for the inflaton field in superstring models. Phys. Rev.
D 34, 3069 (1986)
269. R. Brustein, P.J. Steinhardt, Challenges for superstring cosmology. Phys. Lett. B 302, 196
(1993). [arXiv:hep-th/9212049]
270. T. Banks, M. Berkooz, S.H. Shenker, G.W. Moore, P.J. Steinhardt, Modular cosmology. Phys.
Rev. D 52, 3548 (1995). [arXiv:hep-th/9503114]
271. M. Cicoli, F.G. Pedro, G. Tasinato, Natural quintessence in string theory. JCAP 1207, 044
(2012). [arXiv:1203.6655]
272. J.A. Casas, Baryogenesis, inflation and superstrings, in International Europhysics Conference
on High Energy Physics, ed. by D. Lellouch, G. Mikenberg, E. Rabinovici (Springer, Berlin,
1999). [arXiv:hep-ph/9802210]
273. J. Ellis, M.A.G. García, D.V. Nanopoulos, K.A. Olive, A no-scale inflationary model to fit
them all. JCAP 1408, 044 (2014). [arXiv:1405.0271]
274. J. Ellis, M.A.G. García, D.V. Nanopoulos, K.A. Olive, Two-field analysis of no-scale
supergravity inflation. JCAP 1501, 010 (2015). [arXiv:1409.8197]
275. J. Ellis, M.A.G. García, D.V. Nanopoulos, K.A. Olive, Phenomenological aspects of no-scale
inflation models. JCAP 1510, 003 (2015). [arXiv:1503.08867]
276. E. Witten, Dimensional reduction of superstring models. Phys. Lett. B 155, 151 (1985)
277. S. Ferrara, C. Kounnas, M. Porrati, General dimensional reduction of ten-dimensional
supergravity and superstring. Phys. Lett. B 181, 263 (1986)
278. M. Cvetič, J. Louis, B.A. Ovrut, A string calculation of the Kähler potentials for moduli of
ZN orbifolds. Phys. Lett. B 206, 227 (1988)
279. L.J. Dixon, V. Kaplunovsky, J. Louis, On effective field theories describing .2; 2/ vacua of
the heterotic string. Nucl. Phys. B 329, 27 (1990)
280. J.P. Derendinger, S. Ferrara, C. Kounnas, F. Zwirner, On loop corrections to string effective
field theories: field-dependent gauge couplings and -model anomalies. Nucl. Phys. B 372,
145 (1992)
281. A. Brignole, L.E. Ibáñez, C. Muñoz, Towards a theory of soft terms for the supersymmetric
Standard Model. Nucl. Phys. B 422, 125 (1994); Erratum-ibid. B 436, 747 (1995)
282. M. Cicoli, S. Krippendorf, C. Mayrhofer, F. Quevedo, R. Valandro, D-branes at del
Pezzo singularities: global embedding and moduli stabilisation. JHEP 1209, 019 (2012).
[arXiv:1206.5237]
284. T. Banks, M. Dine, P.J. Fox, E. Gorbatov, On the possibility of large axion decay constants.
JCAP 0306, 001 (2003). [arXiv:hep-th/0303252]
286. N. Barnaby, M. Peloso, Large non-Gaussianity in axion inflation. Phys. Rev. Lett. 106, 181301
(2011). [arXiv:1011.1500]
287. M.P. Hertzberg, M. Tegmark, S. Kachru, J. Shelton, O. Özcan, Searching for inflation in
simple string theory models: an astrophysical perspective. Phys. Rev. D 76, 103521 (2007).
[arXiv:0709.0002]
288. G. Villadoro, F. Zwirner, N D 1 effective potential from dual type-IIA D6/O6 orientifolds
with general fluxes. JHEP 0506, 047 (2005). [arXiv:hep-th/0503169]
289. M. Ihl, T. Wrase, Towards a realistic type IIA T 6 =Z 4 orientifold model with background
fluxes, part 1. Moduli stabilization. JHEP 0607, 027 (2006). [arXiv:hep-th/0604087]
290. E. Silverstein, Simple de Sitter solutions. Phys. Rev. D 77, 106006 (2008). [arXiv:0712.1196]
291. B. Heidenreich, M. Reece, T. Rudelius, Weak gravity strongly constrains large-field axion
inflation. JHEP 1512, 108 (2015). [arXiv:1506.03447]
292. T.C. Bachlechner, C. Long, L. McAllister, Planckian axions and the weak gravity conjecture.
JHEP 1601, 091 (2016). [arXiv:1503.07853]
293. G. Shiu, W. Staessens, F. Ye, Widening the axion window via kinetic and Stückelberg mixings.
294. G. Shiu, W. Staessens, F. Ye, Large field inflation from axion mixing. JHEP 1506, 026 (2015).
[arXiv:1503.02965]
295. A. de la Fuente, P. Saraswat, R. Sundrum, Natural inflation and quantum gravity. Phys. Rev.
Lett. 114, 151303 (2015). [arXiv:1412.3457]
296. D. Junghans, Large-field inflation with multiple axions and the weak gravity conjecture. JHEP
1602, 128 (2016). [arXiv:1504.03566]
297. G.R. Dvali, S.-H.H. Tye, Brane inflation. Phys. Lett. B 450, 72 (1999).
298. S.H.S. Alexander, Inflation from D – D brane annihilation. Phys. Rev. D 65, 023507 (2002).
299. G.R. Dvali, Q. Shafi, S. Solganik, D-brane inflation. arXiv:hep-th/0105203
300. C.P. Burgess, M. Majumdar, D. Nolte, F. Quevedo, G. Rajesh, R.J. Zhang, The inflationary
brane-antibrane universe. JHEP 0107, 047 (2001). [arXiv:hep-th/0105204]
301. G. Shiu, S.-H.H. Tye, Some aspects of brane inflation. Phys. Lett. B 516, 421 (2001).
302. B. Kyae, Q. Shafi, Branes and inflationary cosmology. Phys. Lett. B 526, 379 (2002).
303. J. García-Bellido, R. Rabadán, F. Zamora, Inflationary scenarios from branes at angles. JHEP
0201, 036 (2002). [arXiv:hep-th/0112147]
304. N.T. Jones, H. Stoica, S.-H.H. Tye, Brane interaction as the origin of inflation. JHEP 0207,
051 (2002). [arXiv:hep-th/0203163]
References 807
305. S. Sarangi, S.-H.H. Tye, Cosmic string production towards the end of brane inflation. Phys.
306. N.T. Jones, H. Stoica, S.-H.H. Tye, The production, spectrum and evolution of cosmic strings
in brane inflation. Phys. Lett. B 563, 6 (2003). [arXiv:hep-th/0303269]
307. L. Randall, R. Sundrum, An alternative to compactification. Phys. Rev. Lett. 83, 4690 (1999).
308. K. Becker, M. Becker, M. Haack, J. Louis, Supersymmetry breaking and ˛ 0 corrections to
flux induced potentials. JHEP 0206, 060 (2002). [arXiv:hep-th/0204254]
309. R.G. Leigh, Dirac–Born–Infeld action from Dirichlet -model. Mod. Phys. Lett. A 4, 2767
(1989)
310. O. Aharony, S.S. Gubser, J.M. Maldacena, H. Ooguri, Y. Oz, Large N field theories, string
theory and gravity. Phys. Rep. 323, 183 (2000). [arXiv:hep-th/9905111]
311. J.M. Maldacena, The large-N limit of superconformal field theories and supergrav-
ity. Adv. Theor. Math. Phys. 2, 231 (1998) [Int. J. Theor. Phys. 38, 1113 (1999)].
312. N. Seiberg, E. Witten, The D1/D5 system and singular CFT. JHEP 9904, 017 (1999).
313. G.G. Ross, S. Sarkar, Successful supersymmetric inflation. Nucl. Phys. B 461, 597 (1996).
314. G. German, G.G. Ross, S. Sarkar, Low-scale inflation. Nucl. Phys. B 608, 423 (2001).
315. R. Allahverdi, K. Enqvist, J. García-Bellido, A. Jokinen, A. Mazumdar, MSSM flat direc-
tion inflation: slow roll, stability, fine tunning and reheating. JCAP 0706, 019 (2007).
316. G. Dvali, A. Vilenkin, Formation and evolution of cosmic D strings. JCAP 0403, 010 (2004).
317. E.J. Copeland, R.C. Myers, J. Polchinski, Cosmic F- and D-strings. JHEP 0406, 013 (2004).
318. M.G. Jackson, N.T. Jones, J. Polchinski, Collisions of cosmic F- and D-strings. JHEP 0510,
013 (2005). [arXiv:hep-th/0405229]
319. T. Damour, A. Vilenkin, Gravitational radiation from cosmic (super)strings: bursts,
stochastic background, and observational windows. Phys. Rev. D 71, 063510 (2005).
320. T. Damour, A. Vilenkin, Gravitational wave bursts from cosmic strings. Phys. Rev. Lett. 85,
3761 (2000). [arXiv:gr-qc/0004075]
321. E.J. Copeland, T.W.B. Kibble, Cosmic strings and superstrings. Proc. R. Soc. Lond. A 466,
623 (2010). [arXiv:0911.1345]
322. P.A.R. Ade et al. [Planck Collaboration], Planck 2013 results. XXV. Searches for cosmic
strings and other topological defects. Astron. Astrophys. 571, A25 (2014). [arXiv:1303.5085]
324. S. Kuroyanagi, K. Miyamoto, T. Sekiguchi, K. Takahashi, J. Silk, Forecast constraints on
cosmic string parameters from gravitational wave direct detection experiments. Phys. Rev. D
86, 023503 (2012). [arXiv:1202.3032]
cosmic strings from future CMB, pulsar timing and gravitational wave direct detection
experiments. Phys. Rev. D 87, 023522 (2013); Erratum-ibid. D 87, 069903(E) (2013).
[arXiv:1210.2829]
326. D. Baumann, A. Dymarsky, I.R. Klebanov, J.M. Maldacena, L.P. McAllister, A. Murugan, On
D3-brane potentials in compactifications with fluxes and wrapped D-branes. JHEP 0611, 031
(2006). [arXiv:hep-th/0607050]
327. C.P. Burgess, J.M. Cline, K. Dasgupta, H. Firouzjahi, Uplifting and inflation with D3 branes.
JHEP 0703, 027 (2007). [arXiv:hep-th/0610320]
328. R.C. Myers, Dielectric branes. JHEP 9912, 022 (1999). [arXiv:hep-th/9910053]
329. B. de Carlos, J. Roberts, Y. Schmohe, Moving five-branes and membrane instantons in low
energy heterotic M-theory. Phys. Rev. D 71, 026004 (2005). [arXiv:hep-th/0406171]
330. E.I. Buchbinder, Five-brane dynamics and inflation in heterotic M-theory. Nucl. Phys. B 711,
314 (2005). [arXiv:hep-th/0411062]
331. K. Becker, M. Becker, A. Krause, M-theory inflation from multi M5-brane dynamics. Nucl.
332. J. Ward, Instantons, assisted inflation and heterotic M-theory. Phys. Rev. D 73, 026004
(2006). [arXiv:hep-th/0511079]
333. A. Krause, Large gravitational waves and the Lyth bound in multi-brane inflation. JCAP 0807,
001 (2008). [arXiv:0708.4414]
334. P. Vargas Moniz, S. Panda, J. Ward, Higher order corrections to heterotic M-theory inflation.
335. C. Armendáriz-Picón, T. Damour, V.F. Mukhanov, k-inflation. Phys. Lett. B 458, 209 (1999).
336. J. Garriga, V.F. Mukhanov, Perturbations in k-inflation. Phys. Lett. B 458, 219 (1999).
(2004). [arXiv:astro-ph/0405356]
338. P.A.R. Ade et al. [Planck Collaboration], Planck 2013 Results. XXIV. Constraints on
primordial non-Gaussianity. Astron. Astrophys. 571, A24 (2014). [arXiv:1303.5084]
339. Y.-F. Cai, W. Xue, N-flation from multiple DBI type actions. Phys. Lett. B 680, 395 (2009).
[arXiv:0809.4134]
340. Y.-F. Cai, H.-Y. Xia, Inflation with multiple sound speeds: a model of multiple DBI type
actions and non-Gaussianities. Phys. Lett. B 677, 226 (2009). [arXiv:0904.0062]
341. P. Hořava, E. Witten, Heterotic and type I string dynamics from eleven dimensions. Nucl.
342. P. Hořava, E. Witten, Eleven-dimensional supergravity on a manifold with boundary. Nucl.
343. E. Witten, Strong coupling expansion of Calabi–Yau compactification. Nucl. Phys. B 471,
135 (1996). [arXiv:hep-th/9602070]
344. T. Banks, M. Dine, Couplings and scales in strongly coupled heterotic string theory. Nucl.
345. B.A. Ovrut, Lectures on heterotic M-theory, in Strings, Branes and Extra Dimensions, ed. by
S.S. Gubser, J.D. Lykken (World Scientific, Singapore, 2004). [arXiv:hep-th/0201032]
346. A. Lukas, B.A. Ovrut, K.S. Stelle, D. Waldram, Universe as a domain wall. Phys. Rev. D 59,
086001 (1999). [arXiv:hep-th/9803235]
347. A. Lukas, B.A. Ovrut, K.S. Stelle, D. Waldram, Heterotic M-theory in five dimensions. Nucl.
348. A. Lukas, B.A. Ovrut, D. Waldram, Cosmological solutions of Hořava–Witten theory. Phys.
Rev. D 60, 086001 (1999). [arXiv:hep-th/9806022]
349. A. Lukas, B.A. Ovrut, D. Waldram, Boundary inflation. Phys. Rev. D 61, 023506 (2000).
350. I. Antoniadis, M. Quirós, Large radii and string unification. Phys. Lett. B 392, 61 (1997).
351. K. Benakli, Phenomenology of low quantum gravity scale models. Phys. Rev. D 60, 104002
(1999). [arXiv:hep-ph/9809582]
352. P. Brax, C. van de Bruck, A.C. Davis, Brane world cosmology. Rep. Prog. Phys. 67, 2183
(2004). [arXiv:hep-th/0404011]
353. R. Maartens, K. Koyama, Brane-world gravity. Living Rev. Relat. 13, 5 (2010)
354. L. Randall, R. Sundrum, Large mass hierarchy from a small extra dimension. Phys. Rev. Lett.
83, 3370 (1999). [arXiv:hep-ph/9905221]
355. T. Shiromizu, K.-i. Maeda, M. Sasaki, The Einstein equation on the 3-brane world. Phys. Rev.
D 62, 024012 (2000). [arXiv:gr-qc/9910076]
References 809
356. P. Binétruy, C. Deffayet, D. Langlois, Non-conventional cosmology from a brane universe.

357. P. Binétruy, C. Deffayet, U. Ellwanger, D. Langlois, Brane cosmological evolution in a bulk
with cosmological constant. Phys. Lett. B 477, 285 (2000). [arXiv:hep-th/9910219]
358. C. Csáki, M. Graesser, C. Kolda, J. Terning, Cosmology of one extra dimension with localized
gravity. Phys. Lett. B 462, 34 (1999). [arXiv:hep-ph/9906513]
359. J.M. Cline, C. Grojean, G. Servant, Cosmological expansion in the presence of extra
dimensions. Phys. Rev. Lett. 83, 4245 (1999). [arXiv:hep-ph/9906523]
360. É.É. Flanagan, S.-H.H. Tye, I. Wasserman, Cosmological expansion in the Randall–Sundrum
brane world scenario. Phys. Rev. D 62, 044039 (2000). [arXiv:hep-ph/9910498]
361. R. Maartens, D. Wands, B.A. Bassett, I.P.C. Heard, Chaotic inflation on the brane. Phys. Rev.
D 62, 041301 (2000). [arXiv:hep-ph/9912464]
362. E.J. Copeland, A.R. Liddle, J.E. Lidsey, Steep inflation: ending brane world inflation by
gravitational particle production. Phys. Rev. D 64, 023509 (2001). [arXiv:astro-ph/0006421]
363. J.E. Kim, B. Kyae, H.M. Lee, Effective Gauss–Bonnet interaction in Randall–Sundrum
compactification. Phys. Rev. D 62, 045013 (2000). [arXiv:hep-ph/9912344]
364. D.J. Gross, J.H. Sloan, The quartic effective action for the heterotic string. Nucl. Phys. B 291,
41 (1987)
365. J.E. Kim, B. Kyae, H.M. Lee, Various modified solutions of the Randall–Sundrum model
with the Gauss–Bonnet interaction. Nucl. Phys. B 582, 296 (2000); Erratum-ibid. B 591, 587
(2000). [arXiv:hep-th/0004005]
366. S. Nojiri, S.D. Odintsov, Brane world cosmology in higher derivative gravity or warped
compactification in the next-to-leading order of AdS/CFT correspondence. JHEP 0007, 049
(2000). [arXiv:hep-th/0006232]
367. I.P. Neupane, Consistency of higher derivative gravity in the brane background. JHEP 0009,
040 (2000). [arXiv:hep-th/0008190]
368. C. Charmousis, J.-F. Dufaux, General Gauss–Bonnet brane cosmology. Class. Quantum Grav.
19, 4671 (2002). [arXiv:hep-th/0202107]
369. S. Nojiri, S.D. Odintsov, S. Ogushi, Friedmann–Robertson–Walker brane cosmological
equations from the five-dimensional bulk (A)dS black hole. Int. J. Mod. Phys. A 17, 4809
(2002). [arXiv:hep-th/0205187]
370. S.C. Davis, Generalized Israel junction conditions for a Gauss–Bonnet brane world. Phys.
Rev. D 67, 024030 (2003). [arXiv:hep-th/0208205]
371. E. Gravanis, S. Willison, Israel conditions for the Gauss–Bonnet theory and the Friedmann
equation on the brane universe. Phys. Lett. B 562, 118 (2003). [arXiv:hep-th/0209076]
372. J.E. Lidsey, N.J. Nunes, Inflation in Gauss–Bonnet brane cosmology. Phys. Rev. D 67, 103510
(2003). [arXiv:astro-ph/0303168]
373. J.-F. Dufaux, J.E. Lidsey, R. Maartens, M. Sami, Cosmological perturbations from brane
inflation with a Gauss–Bonnet term. Phys. Rev. D 70, 083525 (2004). [arXiv:hep-th/0404161]
374. G. Calcagni, Slow-roll parameters in braneworld cosmologies. Phys. Rev. D 69, 103508
(2004). [arXiv:hep-ph/0402126]
375. G. Calcagni, Consistency relations and degeneracies in (non)commutative patch inflation.
Phys. Lett. B 606, 177 (2005). [arXiv:hep-ph/0406057]
376. G. Calcagni, Braneworld Cosmology and Noncommutative Inflation. Ph.D. thesis, Parma
University, Parma (2005). [arXiv:hep-ph/0503044]
377. R. Maartens, Cosmological dynamics on the brane. Phys. Rev. D 62, 084023 (2000).
378. C. van de Bruck, M. Dorca, R.H. Brandenberger, A. Lukas, Cosmological perturbations in
brane-world theories: formalism. Phys. Rev. D 62, 123515 (2000). [arXiv:hep-th/0005032]
379. K. Koyama, J. Soda, Evolution of cosmological perturbations in the brane world. Phys. Rev.
D 62, 123502 (2000). [arXiv:hep-th/0005239]
380. D. Langlois, Evolution of cosmological perturbations in a brane universe. Phys. Rev. Lett. 86,
2212 (2001). [arXiv:hep-th/0010063]
381. D. Langlois, R. Maartens, M. Sasaki, D. Wands, Large-scale cosmological perturbations on

the brane. Phys. Rev. D 63, 084009 (2001). [arXiv:hep-th/0012044]
382. D. Langlois, R. Maartens, D. Wands, Gravitational waves from inflation on the brane. Phys.
383. A. Coley, No chaos in brane-world cosmology. Class. Quantum Grav. 19, L45 (2002).
384. S. Kachru, L. McAllister, Bouncing brane cosmologies from warped string compactifications.
JHEP 0303, 018 (2003). [arXiv:hep-th/0205209]
385. S. Mukherji, M. Peloso, Bouncing and cyclic universes from brane models. Phys. Lett. B 547,
297 (2002). [arXiv:hep-th/0205180]
386. P. Kanti, K. Tamvakis, Challenges and obstacles for a bouncing universe in brane models.
387. J.L. Hovdebo, R.C. Myers, Bouncing brane worlds go crunch! JCAP 0311, 012 (2003).
388. V.A. Rubakov, M.E. Shaposhnikov, Extra space-time dimensions: towards a solution to the
cosmological constant problem. Phys. Lett. B 125, 139 (1983)
389. E.P. Verlinde, H.L. Verlinde, RG flow, gravity and the cosmological constant. JHEP 0005,
034 (2000). [arXiv:hep-th/9912018]
390. S.P. de Alwis, Brane world scenarios and the cosmological constant. Nucl. Phys. B 597, 263
(2001). [arXiv:hep-th/0002174]
391. N. Arkani-Hamed, S. Dimopoulos, N. Kaloper, R. Sundrum, A small cosmological constant
from a large extra dimension. Phys. Lett. B 480, 193 (2000). [arXiv:hep-th/0001197]
392. S. Kachru, M.B. Schulz, E. Silverstein, Self-tuning flat domain walls in 5D gravity and string
theory. Phys. Rev. D 62, 045021 (2000). [arXiv:hep-th/0001206]
393. S. Förste, Z. Lalak, S. Lavignac, H.P. Nilles, A comment on self-tuning and vanishing cosmo-
logical constant in the brane world. Phys. Lett. B 481, 360 (2000). [arXiv:hep-th/0002164]
394. S. Förste, Z. Lalak, S. Lavignac, H.P. Nilles, The cosmological constant problem from a
brane-world perspective. JHEP 0009, 034 (2000). [arXiv:hep-th/0006139]
395. I. Antoniadis, S. Cotsakis, I. Klaoudatou, Brane singularities and their avoidance. Class.
396. S.M. Carroll, L. Mersini-Houghton, Can we live in a self-tuning universe? Phys. Rev. D 64,
124008 (2001). [arXiv:hep-th/0105007]
397. Y. Aghababaie, C.P. Burgess, S.L. Parameswaran, F. Quevedo, Towards a naturally small
cosmological constant from branes in 6D supergravity. Nucl. Phys. B 680, 389 (2004).
398. Y. Aghababaie, C.P. Burgess, J.M. Cline, H. Firouzjahi, S.L. Parameswaran, F. Quevedo,
G. Tasinato, I. Zavala, Warped brane worlds in six dimensional supergravity. JHEP 0309, 037
(2003). [arXiv:hep-th/0308064]
399. C.P. Burgess, Supersymmetric large extra dimensions and the cosmological constant: an
update. Ann. Phys. (N.Y.) 313, 283 (2004). [arXiv:hep-th/0402200]
400. G. Calcagni, S. Kuroyanagi, J. Ohashi, S. Tsujikawa, Strong Planck constraints on braneworld
and non-commutative inflation. JCAP 1403, 052 (2014). [arXiv:1310.5186]
401. A. Sen, Universality of the tachyon potential. JHEP 9912, 027 (1999).
402. M.R. Garousi, Tachyon couplings on non-BPS D-branes and Dirac–Born–Infeld action. Nucl.
403. E.A. Bergshoeff, M. de Roo, T.C. de Wit, E. Eyras, S. Panda, T-duality and actions for non-
BPS D-branes. JHEP 0005, 009 (2000). [arXiv:hep-th/0003221]
404. D. Kutasov, M. Mariño, G.W. Moore, Some exact results on tachyon condensation in string
field theory. JHEP 0010, 045 (2000). [arXiv:hep-th/0009148]
405. D. Kutasov, M. Mariño, G.W. Moore, Remarks on tachyon condensation in superstring field
theory. arXiv:hep-th/0010108
406. P. Kraus, F. Larsen, Boundary string field theory of the DDN system. Phys. Rev. D 63, 106004
(2001). [arXiv:hep-th/0012198]
References 811
407. T. Takayanagi, S. Terashima, T. Uesugi, Brane-antibrane action from boundary string field
theory. JHEP 0103, 019 (2001). [arXiv:hep-th/0012210]
408. M.R. Garousi, Off-shell extension of S-matrix elements and tachyonic effective actions. JHEP
0304, 027 (2003). [arXiv:hep-th/0303239]
409. M.R. Garousi, Slowly varying tachyon and tachyon potential. JHEP 0305, 058 (2003).
410. F. Leblond, A.W. Peet, SD-brane gravity fields and rolling tachyons. JHEP 0304, 048 (2003).
411. N.D. Lambert, H. Liu, J.M. Maldacena, Closed strings from decaying D-branes. JHEP 0703,
014 (2007). [arXiv:hep-th/0303139]
412. A. Mazumdar, S. Panda, A. Pérez-Lorenzana, Assisted inflation via tachyon condensation.
Nucl. Phys. B 614, 101 (2001). [arXiv:hep-ph/0107058]
413. G.W. Gibbons, Cosmological evolution of the rolling tachyon. Phys. Lett. B 537, 1 (2002).
414. M. Fairbairn, M.H.G. Tytgat, Inflation from a tachyon fluid? Phys. Lett. B 546, 1 (2002).
415. S. Mukohyama, Brane cosmology driven by the rolling tachyon. Phys. Rev. D 66, 024009
(2002). [arXiv:hep-th/0204084]
416. A. Feinstein, Power-law inflation from the rolling tachyon. Phys. Rev. D 66, 063511 (2002).
417. T. Padmanabhan, Accelerated expansion of the universe driven by tachyonic matter. Phys.
Rev. D 66, 021301 (2002). [arXiv:hep-th/0204150]
418. D. Choudhury, D. Ghoshal, D.P. Jatkar, S. Panda, On the cosmological relevance of the
tachyon. Phys. Lett. B 544, 231 (2002). [arXiv:hep-th/0204204]
419. A. Frolov, L. Kofman, A. Starobinsky, Prospects and problems of tachyon matter cosmology.
420. G. Shiu, I. Wasserman, Cosmological constraints on tachyon matter. Phys. Lett. B 541, 6
(2002). [arXiv:hep-th/0205003]
421. L. Kofman, A. Linde, Problems with tachyon inflation. JHEP 0207, 004 (2002).
422. H.B. Benaoum, Accelerated universe from modified Chaplygin gas and tachyonic fluid.
423. M. Sami, Implementing power law inflation with rolling tachyon on the brane. Mod. Phys.
Lett. A 18, 691 (2003). [arXiv:hep-th/0205146]
424. M. Sami, P. Chingangbam, T. Qureshi, Aspects of tachyonic inflation with exponential
potential. Phys. Rev. D 66, 043530 (2002). [arXiv:hep-th/0205179]
425. G. Shiu, S.-H.H. Tye, I. Wasserman, Rolling tachyon in brane world cosmology from
superstring field theory. Phys. Rev. D 67, 083517 (2003). [arXiv:hep-th/0207119]
426. Y.-S. Piao, R.-G. Cai, X. Zhang, Y.-Z. Zhang, Assisted tachyonic inflation. Phys. Rev. D 66,
121301(R) (2002). [arXiv:hep-ph/0207143]
427. J.M. Cline, H. Firouzjahi, P. Martineau, Reheating from tachyon condensation. JHEP 0211,
041 (2002). [arXiv:hep-th/0207156]
428. M.C. Bento, O. Bertolami, A.A. Sen, Tachyonic inflation in the braneworld scenario. Phys.
Rev. D 67, 063511 (2003). [arXiv:hep-th/0208124]
429. G.W. Gibbons, Thoughts on tachyon cosmology. Class. Quantum Grav. 20, S321 (2003).
430. D.A. Steer, F. Vernizzi, Tachyon inflation: tests and comparison with single scalar field
431. V. Gorini, A. Kamenshchik, U. Moschella, V. Pasquier, Tachyons, scalar fields and cosmol-
ogy. Phys. Rev. D 69, 123512 (2004). [arXiv:hep-th/0311111]
432. M.R. Garousi, M. Sami, S. Tsujikawa, Cosmology from a rolling massive scalar field on the
anti-D3 brane of de Sitter vacua. Phys. Rev. D 70, 043536 (2004). [arXiv:hep-th/0402075]
433. J. Raeymaekers, Tachyonic inflation in a warped string background. JHEP 0410, 057 (2004).
434. H. Yavartanoo, Cosmological solution from D-brane motion in NS5-branes background. Int.
436. A. Ghodsi, A.E. Mosaffa, D-brane dynamics in RR deformation of NS5-branes background
and tachyon cosmology. Nucl. Phys. B 714, 30 (2005). [arXiv:hep-th/0408015]
437. J.-g. Hao, X.-z. Li, Reconstructing the equation of state of the tachyon. Phys. Rev. D 66,
087301 (2002). [arXiv:hep-th/0209041]
438. J.S. Bagla, H.K. Jassal, T. Padmanabhan, Cosmology with tachyon field as dark energy. Phys.
439. L.R.W. Abramo, F. Finelli, Cosmological dynamics of the tachyon with an inverse power-law
potential. Phys. Lett. B 575, 165 (2003). [arXiv:astro-ph/0307208]
440. L.P. Chimento, Extended tachyon field, Chaplygin gas and solvable k-essence cosmologies.
441. J.M. Aguirregabiria, R. Lazkoz, Tracking solutions in tachyon cosmology. Phys. Rev. D 69,
123502 (2004). [arXiv:hep-th/0402190]
442. E.J. Copeland, M.R. Garousi, M. Sami, S. Tsujikawa, What is needed of a tachyon if it is to
be the dark energy? Phys. Rev. D 71, 043003 (2005). [arXiv:hep-th/0411192]
443. G. Calcagni, A.R. Liddle, Tachyon dark energy models: dynamics and constraints. Phys. Rev.
D 74, 043528 (2006). [arXiv:astro-ph/0606003]
444. A.D. Linde, Fast-roll inflation. JHEP 0111, 052 (2001). [arXiv:hep-th/0110195]
445. A. Sen, Time and tachyon. Int. J. Mod. Phys. A 18, 4869 (2003). [arXiv:hep-th/0209122]
446. E.S. Fradkin, A.A. Tseytlin, Effective field theory from quantized strings. Phys. Lett. B 158,
316 (1985)
447. B. Zwiebach, Curvature squared terms and string theories. Phys. Lett. B 156, 315 (1985)
448. C.G. Callan, D. Friedan, E.J. Martinec, M.J. Perry, Strings in background fields. Nucl. Phys.
B 262, 593 (1985)
449. D.J. Gross, J.H. Sloan, The quartic effective action for the heterotic string. Nucl. Phys. B 291,
41 (1987)
450. R.R. Metsaev, A.A. Tseytlin, Order ˛ 0 (two-loop) equivalence of the string equations of
motion and the -model Weyl invariance conditions: dependence on the dilaton and the
antisymmetric tensor. Nucl. Phys. B 293, 385 (1987)
451. I. Jack, D.R.T. Jones, D.A. Ross, On the relationship between string low-energy effective
actions and O.˛ 0 3 / -model ˇ-functions. Nucl. Phys. B 307, 130 (1988)
452. I. Antoniadis, J. Rizos, K. Tamvakis, Singularity-free cosmological solutions of the super-
string effective action. Nucl. Phys. B 415, 497 (1994). [arXiv:hep-th/9305025]
453. R. Easther, K.-i. Maeda, One loop superstring cosmology and the nonsingular universe. Phys.
Rev. D 54, 7252 (1996). [arXiv:hep-th/9605173]
454. S. Kawai, M.-a. Sakagami, J. Soda, Instability of 1-loop superstring cosmology. Phys. Lett.
B 437, 284 (1998). [arXiv:gr-qc/9802033]
455. S. Kawai, J. Soda, Evolution of fluctuations during graceful exit in string cosmology. Phys.
Lett. B 460, 41 (1999). [arXiv:gr-qc/9903017]
457. N. Kaloper, J. March-Russell, G.D. Starkman, M. Trodden, Compact hyperbolic extra
dimensions: branes, Kaluza–Klein modes, and cosmology. Phys. Rev. Lett. 85, 928 (2000).
458. G.D. Starkman, D. Stojkovic, M. Trodden, Large extra dimensions and cosmological
problems. Phys. Rev. D 63, 103511 (2001). [arXiv:hep-th/0012226]
459. G.D. Starkman, D. Stojkovic, M. Trodden, Homogeneity, flatness, and “large” extra dimen-
sions. Phys. Rev. Lett. 87, 231303 (2001). [arXiv:hep-th/0106143]
460. P.K. Townsend, M.N.R. Wohlfarth, Accelerating cosmologies from compactification. Phys.
References 813
461. R. Emparan, J. Garriga, A note on accelerating cosmologies from compactifications and S-

branes. JHEP 0305, 028 (2003). [arXiv:hep-th/0304124]
462. C.-M. Chen, P.-M. Ho, I.P. Neupane, J.E. Wang, A note on acceleration from product space
compactification. JHEP 0307, 017 (2003). [arXiv:hep-th/0304177]
463. C.-M. Chen, P.-M. Ho, I.P. Neupane, N. Ohta, J.E. Wang, Hyperbolic space cosmologies.
JHEP 0310, 058 (2003). [arXiv:hep-th/0306291]
464. M. Gutperle, R. Kallosh, A.D. Linde, M/string theory, S-branes and accelerating universe.
JCAP 0307, 001 (2003). [arXiv:hep-th/0304225]
465. S.M. Carroll, V. Duvvuri, M. Trodden, M.S. Turner, Is cosmic speed-up due to new
gravitational physics? Phys. Rev. D 70, 043528 (2004). [arXiv:astro-ph/0306438]
466. S. Nojiri, S.D. Odintsov, Where new gravitational physics comes from: M-theory? Phys. Lett.
B 576, 5 (2003). [arXiv:hep-th/0307071]
467. Z.-a. Qiu, A. Strominger, Gauge symmetries in (super)string field theory. Phys. Rev. D 36,
1794 (1987)
468. D. Ghoshal, A. Sen, Gauge and general coordinate invariance in nonpolynomial closed string
theory. Nucl. Phys. B 380, 103 (1992). [arXiv:hep-th/9110038]
469. J.R. David, U.1/ gauge invariance from open string field theory. JHEP 0010, 017 (2000).
470. A. Sen, BV master action for heterotic and type II string field theories. JHEP 1602, 087
(2016). [arXiv:1508.05387]
471. A. Sen, Covariant action for type IIB supergravity. JHEP 1607, 017 (2016).
[arXiv:1511.08220]
472. G. Calcagni, L. Modesto, Nonlocal quantum gravity and M-theory. Phys. Rev. D 91, 124059
(2015). [arXiv:1404.2137]
473. T. Biswas, A. Mazumdar, W. Siegel, Bouncing universes in string-inspired gravity. JCAP
0603, 009 (2006). [arXiv:hep-th/0508194]
474. T. Biswas, N. Okada, Towards LHC physics with nonlocal Standard Model. Nucl. Phys. B
898, 113 (2015). [arXiv:1407.3331]
475. A. Belenchia, D.M.T. Benincasa, S. Liberati, F. Marin, F. Marino, A. Ortolan, Tests of
quantum gravity induced nonlocality via optomechanical quantum oscillators. Phys. Rev. Lett.
116, 161303 (2016). [arXiv:1512.02083]
476. G. Veneziano, Scale factor duality for classical and quantum strings. Phys. Lett. B 265, 287
(1991)
477. K.A. Meissner, G. Veneziano, Symmetries of cosmological superstring vacua. Phys. Lett. B
267, 33 (1991)
478. M. Gasperini, G. Veneziano, Pre-big-bang in string cosmology. Astropart. Phys. 1, 317
(1993). [arXiv:hep-th/9211021]
479. M. Gasperini, G. Veneziano, O.d; d/-covariant string cosmology. Phys. Lett. B 277, 256
(1992). [arXiv:hep-th/9112044]
480. M. Gasperini, G. Veneziano, Inflation, deflation, and frame independence in string cosmology.
Mod. Phys. Lett. A 8, 3701 (1993). [arXiv:hep-th/9309023]
481. R. Brustein, G. Veneziano, The graceful exit problem in string cosmology. Phys. Lett. B 329,
429 (1994). [arXiv:hep-th/9403060]
482. E.J. Copeland, A. Lahiri, D. Wands, Low-energy effective string cosmology. Phys. Rev. D 50,
4868 (1994). [arXiv:hep-th/9406216]
483. E.J. Copeland, A. Lahiri, D. Wands, String cosmology with a time-dependent antisymmetric
tensor potential. Phys. Rev. D 51, 1569 (1995). [arXiv:hep-th/9410136]
484. R. Brustein, M. Gasperini, M. Giovannini, V.F. Mukhanov, G. Veneziano, Metric perturba-
tions in dilaton driven inflation. Phys. Rev. D 51, 6744 (1995). [arXiv:hep-th/9501066]
485. N. Kaloper, R. Madden, K.A. Olive, Towards a singularity-free inflationary universe? Nucl.
486. R. Easther, K.-i. Maeda, D. Wands, Tree level string cosmology. Phys. Rev. D 53, 4247 (1996).
487. N. Kaloper, R. Madden, K.A. Olive, Axions and the graceful exit problem in string
cosmology. Phys. Lett. B 371, 34 (1996). [arXiv:hep-th/9510117]
488. M. Gasperini, M. Maggiore, G. Veneziano, Towards a non-singular pre-big-bang cosmology.
489. E.J. Copeland, R. Easther, D. Wands, Vacuum fluctuations in axion-dilaton cosmologies.
490. R. Brustein, R. Madden, Model of graceful exit in string cosmology. Phys. Rev. D 57, 712
(1998). [arXiv:hep-th/9708046]
491. J.-c. Hwang, Gravitational wave spectra from pole-like inflations based on generalized gravity
theories. Class. Quantum Grav. 15, 1401 (1998). [arXiv:gr-qc/9710061]
492. S. Foffa, M. Maggiore, R. Sturani, Loop corrections and graceful exit in string cosmology.
493. M. Gasperini, Tensor perturbations in high-curvature string backgrounds. Phys. Rev. D 56,
4815 (1997). [arXiv:gr-qc/9704045]
494. J.-c. Hwang, H. Noh, Conserved cosmological structures in the one-loop superstring effective
action. Phys. Rev. D 61, 043511 (2000). [arXiv:astro-ph/9909480]
495. C. Cartier, E.J. Copeland, R. Madden, The graceful exit in string cosmology. JHEP 0001, 035
(2000). [arXiv:hep-th/9910169]
496. C. Cartier, J.-c. Hwang, E.J. Copeland, Evolution of cosmological perturbations in nonsingu-
lar string cosmologies. Phys. Rev. D 64, 103504 (2001). [arXiv:astro-ph/0106197]
497. F. Finelli, Assisted contraction. Phys. Lett. B 545, 1 (2002). [arXiv:hep-th/0206112]
498. M. Gasperini, G. Veneziano, The pre-big bang scenario in string cosmology. Phys. Rep. 373,
1 (2003). [arXiv:hep-th/0207130]
499. M. Gasperini, M. Giovannini, G. Veneziano, Perturbations in a non-singular bouncing
Universe. Phys. Lett. B 569, 113 (2003). [arXiv:hep-th/0306113]
500. M. Gasperini, M. Giovannini, G. Veneziano, Cosmological perturbations across a curvature
bounce. Nucl. Phys. B 694, 206 (2004). [arXiv:hep-th/0401112]
501. L.E. Allen, D. Wands, Cosmological perturbations through a simple bounce. Phys. Rev. D 70,
063515 (2004). [arXiv:astro-ph/0404441]
502. M. Dine, N. Seiberg, Is the superstring weakly coupled? Phys. Lett. B 162, 299 (1985)
503. W. Buchmüller, K. Hamaguchi, O. Lebedev, M. Ratz, Dilaton destabilization at high
temperature. Nucl. Phys. B 699, 292 (2004). [arXiv:hep-th/0404168]
504. E. Witten, Some properties of O.32/ superstrings. Phys. Lett. B 149, 351 (1984)
505. T.R. Taylor, G. Veneziano, Dilaton couplings at large distances. Phys. Lett. B 213, 450 (1988)
506. T. Damour, A.M. Polyakov, The string dilaton and a least coupling principle. Nucl. Phys. B
423, 532 (1994). [arXiv:hep-th/9401069]
507. T. Damour, A.M. Polyakov, String theory and gravity. Gen. Relat. Grav. 26, 1171 (1994).
508. M. Gasperini, F. Piazza, G. Veneziano, Quintessence as a runaway dilaton. Phys. Rev. D 65,
023508 (2002). [arXiv:gr-qc/0108016]
509. F. Piazza, S. Tsujikawa, Dilatonic ghost condensate as dark energy. JCAP 0407, 004 (2004).
510. G. Veneziano, Large-N bounds on, and compositeness limit of, gauge and gravitational
interactions. JHEP 0206, 051 (2002). [arXiv:hep-th/0110129]
511. T. Damour, F. Piazza, G. Veneziano, Runaway dilaton and equivalence principle violations.
512. T. Damour, F. Piazza, G. Veneziano, Violations of the equivalence principle in a dilaton
runaway scenario. Phys. Rev. D 66, 046007 (2002). [arXiv:hep-th/0205111]
513. C.J.A.P. Martins, P.E. Vielzeuf, M. Martinelli, E. Calabrese, S. Pandolfi, Evolution of
the fine-structure constant in runaway dilaton models. Phys. Lett. B 743, 377 (2015).
[arXiv:1503.05068]
514. R. Brandenberger, C. Vafa, Superstrings in the early universe. Nucl. Phys. B 316, 391 (1989)
515. N. Deo, S. Jain, O. Narayan, C.-I Tan, Effect of topology on the thermodynamic limit for a
string gas. Phys. Rev. D 45, 3641 (1992)
References 815
516. A.A. Tseytlin, C. Vafa, Elements of string cosmology. Nucl. Phys. B 372, 443 (1992).
517. R. Hagedorn, Statistical thermodynamics of strong interactions at high energies. Nuovo Cim.
Suppl. 3, 147 (1965)
518. M. Sakellariadou, Numerical experiments on string cosmology. Nucl. Phys. B 468, 319
(1996). [arXiv:hep-th/9511075]
519. S. Alexander, R.H. Brandenberger, D.A. Easson, Brane gases in the early universe. Phys. Rev.
D 62, 103509 (2000). [arXiv:hep-th/0005212]
520. R. Brandenberger, D.A. Easson, D. Kimberly, Loitering phase in brane gas cosmology. Nucl.
521. D.A. Easson, Brane gases on K3 and Calabi–Yau manifolds. Int. J. Mod. Phys. A 18, 4295
(2003). [arXiv:hep-th/0110225]
522. R. Easther, B.R. Greene, M.G. Jackson, Cosmological string gas on orbifolds. Phys. Rev. D
66, 023502 (2002). [arXiv:hep-th/0204099]
523. S. Watson, R.H. Brandenberger, Isotropization in brane gas cosmology. Phys. Rev. D 67,
043510 (2003). [arXiv:hep-th/0207168]
524. R. Easther, B.R. Greene, M.G. Jackson, D.N. Kabat, Brane gas cosmology in M theory: late
time behavior. Phys. Rev. D 67, 123501 (2003). [arXiv:hep-th/0211124]
525. S. Watson, R.H. Brandenberger, Stabilization of extra dimensions at tree level. JCAP 0311,
008 (2003). [arXiv:hep-th/0307044]
526. R. Easther, B.R. Greene, M.G. Jackson, D.N. Kabat, Brane gases in the early universe:
thermodynamics and cosmology. JCAP 0401, 006 (2004). [arXiv:hep-th/0307233]
527. S.P. Patil, R. Brandenberger, Radion stabilization by stringy effects in general relativity. Phys.
Rev. D 71, 103522 (2005). [arXiv:hep-th/0401037]
528. S. Watson, Moduli stabilization with the string Higgs effect. Phys. Rev. D 70, 066005 (2004).
529. R. Easther, B.R. Greene, M.G. Jackson, D.N. Kabat, String windings in the early universe.
JCAP 0502, 009 (2005). [arXiv:hep-th/0409121]
530. Y.-K.E. Cheung, S. Watson, R. Brandenberger, Moduli stabilization with string gas and fluxes.
JHEP 0605, 025 (2006). [arXiv:hep-th/0501032]
531. S.P. Patil, R. Brandenberger, The cosmology of massless string modes. JCAP 0601, 005
(2006). [arXiv:hep-th/0502069]
532. T. Battefeld, S. Watson, String gas cosmology. Rev. Mod. Phys. 78, 435 (2006).
533. A. Nayeri, R.H. Brandenberger, C. Vafa, Producing a scale-invariant spectrum of pertur-
bations in a Hagedorn phase of string cosmology. Phys. Rev. Lett. 97, 021302 (2006).
534. R.H. Brandenberger, A. Nayeri, S.P. Patil, C. Vafa, Tensor modes from a primordial Hagedorn
phase of string cosmology. Phys. Rev. Lett. 98, 231302 (2007). [arXiv:hep-th/0604126]
535. R.H. Brandenberger, A. Nayeri, S.P. Patil, C. Vafa, String gas cosmology and structure
formation. Int. J. Mod. Phys. A 22, 3621 (2007). [arXiv:hep-th/0608121]
536. N. Kaloper, L. Kofman, A.D. Linde, V. Mukhanov, On the new string theory inspired
mechanism of generation of cosmological perturbations. JCAP 0610, 006 (2006).
537. T. Biswas, R. Brandenberger, A. Mazumdar, W. Siegel, Non-perturbative gravity, the
Hagedorn bounce and the cosmic microwave background. JCAP 0712, 011 (2007).
538. N. Kaloper, S. Watson, Geometric precipices in string cosmology. Phys. Rev. D 77, 066002
(2008). [arXiv:0712.1820]
539. B. Chen, Y. Wang, W. Xue, R. Brandenberger, String gas cosmology and non-Gaussianities.
arXiv:0712.2477
540. R.J. Danos, A.R. Frey, R.H. Brandenberger, Stabilizing moduli with thermal matter and
nonperturbative effects. Phys. Rev. D 77, 126009 (2008). [arXiv:0802.1557]
541. S. Mishra, W. Xue, R. Brandenberger, U. Yajnik, Supersymmetry breaking and dilaton

stabilization in string gas cosmology. JCAP 1209, 015 (2012). [arXiv:1103.1389]
542. R.H. Brandenberger, A. Nayeri, S.P. Patil, Closed string thermodynamics and a blue tensor
spectrum. Phys. Rev. D 90, 067301 (2014). [arXiv:1403.4927]
543. R.H. Brandenberger, String gas cosmology: progress and problems. Class. Quantum Grav. 28,
204005 (2011). [arXiv:1105.3247]
544. R.H. Brandenberger, Unconventional cosmology. Lect. Notes Phys. 863, 333 (2013).
[arXiv:1203.6698]
545. E. Alvarez, Superstring cosmology. Phys. Rev. D 31, 418 (1985); Erratum-ibid. D 33, 1206
(1986)
546. M.J. Bowick, L.C.R. Wijewardhana, Superstrings at high temperature. Phys. Rev. Lett. 54,
2485 (1985)
547. E. Alvarez, Strings at finite temperature. Nucl. Phys. B 269, 596 (1986)
548. D. Mitchell, N. Turok, Statistical properties of cosmic strings. Nucl. Phys. B 294, 1138 (1987)
549. P.J. Steinhardt, N. Turok, A cyclic model of the universe. Science 296, 1436 (2002).
550. P.J. Steinhardt, N. Turok, Cosmic evolution in a cyclic universe. Phys. Rev. D 65, 126003
(2002). [arXiv:hep-th/0111098]
551. D.H. Lyth, The failure of cosmological perturbation theory in the new ekpyrotic and cyclic
ekpyrotic scenarios. Phys. Lett. B 526, 173 (2002). [arXiv:hep-ph/0110007]
552. S. Tsujikawa, Density perturbations in the ekpyrotic universe and string inspired generaliza-
tions. Phys. Lett. B 526, 179 (2002). [arXiv:gr-qc/0110124]
553. J. Martin, P. Peter, N. Pinto Neto, D.J. Schwarz, Passing through the bounce in the ekpyrotic
models. Phys. Rev. D 65, 123513 (2002). [arXiv:hep-th/0112128]
554. F. Finelli, R. Brandenberger, Generation of a scale-invariant spectrum of adiabatic fluctu-
ations in cosmological models with a contracting phase. Phys. Rev. D 65, 103522 (2002).
555. S. Gratton, J. Khoury, P.J. Steinhardt, N. Turok, Conditions for generating scale-invariant
density perturbations. Phys. Rev. D 69, 103505 (2004). [arXiv:astro-ph/0301395]
556. A.J. Tolley, N. Turok, P.J. Steinhardt, Cosmological perturbations in a big-crunch–big-bang
space-time. Phys. Rev. D 69, 106005 (2004). [arXiv:hep-th/0306109]
557. J. Khoury, P.J. Steinhardt, N. Turok, Designing cyclic universe models. Phys. Rev. Lett. 92,
031302 (2004). [arXiv:hep-th/0307132]
558. J.K. Erickson, D.H. Wesley, P.J. Steinhardt, N. Turok, Kasner and mixmaster behavior in uni-
verses with equation of state w & 1. Phys. Rev. D 69, 063514 (2004). [arXiv:hep-th/0312009]
559. N. Turok, M. Perry, P.J. Steinhardt, M theory model of a big crunch/big bang transition. Phys.
Rev. D 70, 106004 (2004); Erratum-Ibid. D 71, 029901 (2005). [arXiv:hep-th/0408083]
560. P. Creminelli, A. Nicolis, M. Zaldarriaga, Perturbations in bouncing cosmologies: dynamical
attractor versus scale invariance. Phys. Rev. D 71, 063505 (2005). [arXiv:hep-th/0411270]
561. P.J. Steinhardt, N. Turok, Why the cosmological constant is small and positive. Science 312,
1180 (2006). [arXiv:astro-ph/0605173]
562. J.K. Erickson, S. Gratton, P.J. Steinhardt, N. Turok, Cosmic perturbations through the cyclic
ages. Phys. Rev. D 75, 123507 (2007). [arXiv:hep-th/0607164]
563. J.-L. Lehners, P. McFadden, N. Turok, P.J. Steinhardt, Generating ekpyrotic curvature
perturbations before the big bang. Phys. Rev. D 76, 103501 (2007). [arXiv:hep-th/0702153]
564. E.I. Buchbinder, J. Khoury, B.A. Ovrut, New ekpyrotic cosmology. Phys. Rev. D 76, 123503
(2007). [arXiv:hep-th/0702154]
565. K. Koyama, D. Wands, Ekpyrotic collapse with multiple fields. JCAP 0704, 008 (2007).
566. A.J. Tolley, D.H. Wesley, Scale-invariance in expanding and contracting universes from two-
field models. JCAP 0705, 006 (2007). [arXiv:hep-th/0703101]
567. E.I. Buchbinder, J. Khoury, B.A. Ovrut, On the initial conditions in new ekpyrotic cosmology.
JHEP 0711, 076 (2007). [arXiv:0706.3903]
568. K. Koyama, S. Mizuno, F. Vernizzi, D. Wands, Non-Gaussianities from ekpyrotic collapse
with multiple fields. JCAP 0711, 024 (2007). [arXiv:0708.4321]
References 817
569. E.I. Buchbinder, J. Khoury, B.A. Ovrut, Non-Gaussianities in new ekpyrotic cosmology. Phys.
Rev. Lett. 100, 171302 (2008). [arXiv:0710.5172]
570. J.-L. Lehners, P.J. Steinhardt, Non-Gaussian density fluctuations from entropically generated
curvature perturbations in ekpyrotic models. Phys. Rev. D 77, 063533 (2008); Erratum-ibid.
D 79, 129903(E) (2009). [arXiv:0712.3779]
571. J.-L. Lehners, P.J. Steinhardt, Intuitive understanding of non-Gaussianity in ekpyrotic and
cyclic models. Phys. Rev. D 78, 023506 (2008); Erratum-ibid. D 79, 129902(E) (2009).
[arXiv:0804.1293]
572. J. Khoury, F. Piazza, Rapidly-varying speed of sound, scale invariance and non-Gaussian
signatures. JCAP 0907, 026 (2009). [arXiv:0811.3633]
573. J. Khoury, P.J. Steinhardt, Adiabatic ekpyrosis: scale-invariant curvature perturbations from
a single scalar field in a contracting universe. Phys. Rev. Lett. 104, 091301 (2010).
[arXiv:0910.2230]
574. A. Linde, V. Mukhanov, A. Vikman, On adiabatic perturbations in the ekpyrotic scenario.
JCAP 1002, 006 (2010). [arXiv:0912.0944]
575. J. Khoury, P.J. Steinhardt, Generating scale-invariant perturbations from rapidly-evolving
equation of state. Phys. Rev. D 83, 123502 (2011). [arXiv:1101.3548]
576. Y.-F. Cai, D.A. Easson, R. Brandenberger, Towards a nonsingular bouncing cosmology. JCAP
1208, 020 (2012). [arXiv:1206.2382]
577. T. Qiu, X. Gao, E.N. Saridakis, Towards anisotropy-free and nonsingular bounce cosmology
with scale-invariant perturbations. Phys. Rev. D 88, 043525 (2013). [arXiv:1303.2372]
578. M. Li, Note on the production of scale-invariant entropy perturbation in the ekpyrotic
universe. Phys. Lett. B 724, 192 (2013). [arXiv:1306.0191]
579. B. Xue, D. Garfinkle, F. Pretorius, P.J. Steinhardt, Nonperturbative analysis of the evolution
of cosmological perturbations through a nonsingular bounce. Phys. Rev. D 88, 083509 (2013).
[arXiv:1308.3044]
580. A. Fertig, J.-L. Lehners, E. Mallwitz, Ekpyrotic perturbations with small non-Gaussian
corrections. Phys. Rev. D 89, 103537 (2014). [arXiv:1310.8133]
581. A. Ijjas, J.-L. Lehners, P.J. Steinhardt, General mechanism for producing scale-invariant
perturbations and small non-Gaussianity in ekpyrotic models. Phys. Rev. D 89, 123520
(2014). [arXiv:1404.1265]
582. A.M. Levy, A. Ijjas, P.J. Steinhardt, Scale-invariant perturbations in ekpyrotic cosmologies
without fine-tuning of initial conditions. Phys. Rev. D 92, 063524 (2015). [arXiv:1506.01011]
583. J. Khoury, B.A. Ovrut, P.J. Steinhardt, N. Turok, Ekpyrotic universe: colliding branes and the
origin of the hot big bang. Phys. Rev. D 64, 123522 (2001). [arXiv:hep-th/0103239]
584. D.H. Lyth, The primordial curvature perturbation in the ekpyrotic universe. Phys. Lett. B 524,
1 (2002). [arXiv:hep-ph/0106153]
585. J. Khoury, B.A. Ovrut, N. Seiberg, P.J. Steinhardt, N. Turok, From big crunch to big bang.
586. R. Brandenberger, F. Finelli, On the spectrum of fluctuations in an effective field theory of the
ekpyrotic universe. JHEP 0111, 056 (2001). [arXiv:hep-th/0109004]
587. J. Khoury, B.A. Ovrut, P.J. Steinhardt, N. Turok, Density perturbations in the ekpyrotic
scenario. Phys. Rev. D 66, 046005 (2002). [arXiv:hep-th/0109050]
588. J.-L. Lehners, Ekpyrotic and cyclic cosmology. Phys. Rep. 465, 223 (2008).
[arXiv:0806.1245]
589. J.-L. Lehners, Ekpyrotic nongaussianity: a review. Adv. Astron. 2010, 903907 (2010).
[arXiv:1001.3125]
590. L.F. Abbott, A mechanism for reducing the value of the cosmological constant. Phys. Lett. B
150, 427 (1985)
591. R. Bousso, The cosmological constant. Gen. Relat. Grav. 40, 607 (2008). [arXiv:0708.4231]
592. I. Bena, M. Graña, N. Halmagyi, On the existence of meta-stable vacua in Klebanov–Strassler.
JHEP 1009, 087 (2010). [arXiv:0912.3519]
593. A. Dymarsky, On gravity dual of a metastable vacuum in Klebanov–Strassler theory. JHEP

1105, 053 (2011). [arXiv:1102.1734]
594. I. Bena, G. Giecold, M. Graña, N. Halmagyi, S. Massai, On metastable vacua and the
warped deformed conifold: analytic results. Class. Quantum Grav. 30, 015003 (2013).
[arXiv:1102.2403]
595. J. Bläbäck, U.H. Danielsson, T. Van Riet, Resolving anti-brane singularities through time-
dependence. JHEP 1302, 061 (2013). [arXiv:1202.1132]
596. I. Bena, M. Graña, S. Kuperstein, S. Massai, Anti-D3 branes: singular to the bitter end. Phys.
Rev. D 87, 106010 (2013). [arXiv:1206.6369]
597. D. Junghans, Dynamics of warped flux compactifications with backreacting antibranes. Phys.
Rev. D 89, 126007 (2014). [arXiv:1402.4571]
598. D. Junghans, D. Schmidt, M. Zagermann, Curvature-induced resolution of anti-brane singu-
larities. JHEP 1410, 34 (2014). [arXiv:1402.6040]
599. J. Bläbäck, U.H. Danielsson, D. Junghans, T. Van Riet, S.C. Vargas, Localised anti-branes in
non-compact throats at zero and finite T. JHEP 1502, 018 (2015). [arXiv:1409.0534]
600. U.H. Danielsson, T. Van Riet, Fatal attraction: more on decaying anti-branes. JHEP 1503, 087
(2015). [arXiv:1410.8476]
601. B. Michel, E. Mintun, J. Polchinski, A. Puhm, P. Saad, Remarks on brane and antibrane
dynamics. JHEP 1509, 021 (2015). [arXiv:1412.5702]
602. G.S. Hartnett, Localised anti-branes in flux backgrounds. JHEP 1506, 007 (2015).
[arXiv:1501.06568]
603. J. Polchinski, Brane/antibrane dynamics and KKLT stability. arXiv:1509.05710
604. G. Wang, T. Battefeld, Vacuum selection on axionic landscapes. JCAP 1604, 025 (2016).
[arXiv:1512.04224]
605. H. Liu, G.W. Moore, N. Seiberg, The challenging cosmic singularity. arXiv:gr-qc/0301001
606. L. Cornalba, M.S. Costa, Time-dependent orbifolds and string cosmology. Fortsch. Phys. 52,
145 (2004). [arXiv:hep-th/0310099]
607. B. Craps, Big bang models in string theory. Class. Quantum Grav. 23, S849 (2006).
608. M. Berkooz, D. Reichmann, A short review of time dependent solutions and space-like
singularities in string theory. Nucl. Phys. Proc. Suppl. 171, 69 (2007). [arXiv:0705.2146]
609. H. Liu, G.W. Moore, N. Seiberg, Strings in a time-dependent orbifold. JHEP 0206, 045
(2002). [arXiv:hep-th/0204168]
610. B. Craps, D. Kutasov, G. Rajesh, String propagation in the presence of cosmological
singularities. JHEP 0206, 053 (2002). [arXiv:hep-th/0205101]
611. A. Lawrence, On the instability of 3d null singularities. JHEP 0211, 019 (2002).
612. H. Liu, G.W. Moore, N. Seiberg, Strings in time-dependent orbifolds. JHEP 0210, 031 (2002).
613. M. Fabinger, J. McGreevy, On smooth time-dependent orbifolds and null singularities. JHEP
0306, 042 (2003). [arXiv:hep-th/0206196]
614. G.T. Horowitz, J. Polchinski, Instability of spacelike and null orbifold singularities. Phys.
Rev.D 66, 103512 (2002). [arXiv:hep-th/0206228]
615. M. Berkooz, B. Craps, D. Kutasov, G. Rajesh, Comments on cosmological singularities in
string theory. JHEP 0303, 031 (2003). [arXiv:hep-th/0212215]
616. L. Cornalba, M.S. Costa, On the classical stability of orientifold cosmologies. Class. Quantum
Grav. 20, 3969 (2003) .[arXiv:hep-th/0302137]
617. M. Berkooz, B. Pioline, Strings in an electric field, and the Milne universe. JCAP 0311, 007
(2003). [arXiv:hep-th/0307280]
618. M. Berkooz, B. Pioline, M. Rozali, Closed strings in Misner space: cosmological production
of winding strings. JCAP 0408, 004 (2004). [arXiv:hep-th/0405126]
References 819
619. B. Durin, B. Pioline, Closed strings in Misner space: a toy model for a big bounce? in
String Theory: From Gauge Interactions to Cosmology, ed. by L. Baulieu, J. de Boer, B.
Pioline, E. Rabinovici (Springer, Berlin/Germany, 2006); NATO Sci. Ser. II 208, 177 (2006).
[arxiv:hep-th/0501145]
620. J. McGreevy, E. Silverstein, The tachyon at the end of the universe. JHEP 0508, 090 (2005).
621. A. Adams, X. Liu, J. McGreevy, A. Saltman, E. Silverstein, Things fall apart: topology change
from winding tachyons. JHEP 0510, 033 (2005). [arXiv:hep-th/0502021]
622. G.T. Horowitz, Tachyon condensation and black strings. JHEP 0508, 091 (2005).
623. A. Strominger, T. Takayanagi, Correlators in timelike bulk Liouville theory. Adv. Theor.
Math. Phys. 7, 369 (2003). [arXiv:hep-th/0303221]
624. Y. Nakayama, S.J. Rey, Y. Sugawara, The Nothing at the beginning of the universe made
precise. arXiv:hep-th/0606127
625. G.T. Horowitz, E. Silverstein, The inside story: quasilocal tachyons and black holes. Phys.
Rev. D 73, 064016 (2006). [arXiv:hep-th/0601032]
626. P. Kraus, H. Ooguri, S. Shenker, Inside the horizon with AdS/CFT. Phys. Rev. D 67, 124022
(2003). [arXiv:hep-th/0212277]
627. L. Fidkowski, V. Hubeny, M. Kleban, S. Shenker, The black hole singularity in AdS/CFT.
JHEP 0402, 014 (2004). [arXiv:hep-th/0306170]
628. G. Festuccia, H. Liu, Excursions beyond the horizon: black hole singularities in Yang–Mills
theories (I). JHEP 0604, 044 (2006). [arXiv:hep-th/0506202]
629. T. Hertog, G.T. Horowitz, Towards a big crunch dual. JHEP 0407, 073 (2004).
630. T. Hertog, G.T. Horowitz, Holographic description of AdS cosmologies. JHEP 0504, 005
(2005). [arXiv:hep-th/0503071]
631. B. Craps, T. Hertog, N. Turok, Quantum resolution of cosmological singularities using
AdS/CFT correspondence. Phys. Rev. D 86, 043513 (2012). [arxiv:hep-th/0712.4180]
632. C.-S. Chu, P.-M. Ho, Time-dependent AdS/CFT duality and null singularity. JHEP 0604, 013
(2006). [arXiv:hep-th/0602054]
633. S.R. Das, J. Michelson, K. Narayan, S.P. Trivedi, Time-dependent cosmologies and their
duals. Phys. Rev. D 74, 026002 (2006). [arXiv:hep-th/0602107]
634. B. Craps, S. Sethi, E.P. Verlinde, A matrix big bang. JHEP 0510, 005 (2005).
635. B. Craps, A. Rajaraman, S. Sethi, Effective dynamics of the matrix big bang. Phys. Rev. D
73, 106005 (2006). [arXiv:hep-th/0601062]
636. T. Banks, W. Fischler, L. Motl, Dualities versus singularities. JHEP 9901, 019 (1999).
637. A. Feinstein, M.A. Vázquez-Mozo, M-theory resolution of four-dimensional cosmological
singularities via U-duality. Nucl. Phys. B 568, 405 (2000). [arXiv:hep-th/9906006]
638. C.W. Misner, Mixmaster universe. Phys. Rev. Lett. 22, 1071 (1969)
639. A.A. Kirillov, On the nature of the spatial distribution of metric inhomogeneities in the general
solution of the Einstein equations near a cosmological singularity. Zh. Eksp. Teor. Fiz. 103,
721 (1993) [Sov. Phys. JETP 76, 355 (1993)]
640. V.D. Ivashchuk, V.N. Melnikov, A.A. Kirillov, Stochastic properties of multidimensional
cosmological models near a singular point. Pis’ma Zh. Eksp. Teor. Fiz. 60, 225 (1994) [JETP
Lett. 60, 235 (1994)]
641. V.D. Ivashchuk, V.N. Melnikov, Billiard representation for multidimensional cosmology with
multicomponent perfect fluid near the singularity. Class. Quantum Grav. 12, 809 (1995)
642. A.A. Kirillov, V.N. Melnikov, Dynamics of inhomogeneities of metric in the vicinity of a sin-
gularity in multidimensional cosmology. Phys. Rev. D 52, 723 (1995). [arXiv:gr-qc/9408004]
643. V.D. Ivashchuk, V.N. Melnikov, Billiard representation for multidimensional cosmol-
ogy with intersecting p-branes near the singularity. J. Math. Phys. 41, 6341 (2000).
644. T. Damour, M. Henneaux, Chaos in superstring cosmology. Phys. Rev. Lett. 85, 920 (2000).
645. T. Damour, M. Henneaux, Oscillatory behaviour in homogeneous string cosmology models.
Phys. Lett. B 488, 108 (2000); Erratum-ibid. B 491, 377 (2000). [arXiv:hep-th/0006171]
646. T. Damour, M. Henneaux, E10 , BE10 and arithmetical chaos in superstring cosmology. Phys.
647. T. Damour, S. de Buyl, M. Henneaux, C. Schomblond, Einstein billiards and overextensions
of finite-dimensional simple Lie algebras. JHEP 0208, 030 (2002). [arXiv:hep-th/0206125]
648. T. Damour, M. Henneaux, H. Nicolai, E10 and a small tension expansion of M theory. Phys.
649. T. Damour, M. Henneaux, H. Nicolai, Cosmological billiards. Class. Quantum Grav. 20, R145
(2003). [arXiv:hep-th/0212256]
650. M. Henneaux, B. Julia, Hyperbolic billiards of pure D D 4 supergravities. JHEP 0305, 047
(2003). [arXiv:hep-th/0304233]
651. T. Damour, A. Kleinschmidt, H. Nicolai, Hidden symmetries and the fermionic sector of
eleven-dimensional supergravity. Phys. Lett. B 634, 319 (2006). [arXiv:hep-th/0512163]
652. S. de Buyl, M. Henneaux, L. Paulot, Extended E8 invariance of 11-dimensional supergravity.
JHEP 0602, 056 (2006). [ arXiv:hep-th/0512292]
653. T. Damour, A. Kleinschmidt, H. Nicolai, K.E10 /, supergravity and fermions. JHEP 0608, 046
(2006). [arXiv:hep-th/0606105]
654. T. Damour, A. Kleinschmidt, H. Nicolai, Constraints and the E10 coset model. Class. Quantum
Grav. 24, 6097 (2007). [arXiv:0709.2691]
655. T. Damour, C. Hillmann, Fermionic Kac–Moody billiards and supergravity. JHEP 0908, 100
(2009). [arXiv:0906.3116]
656. T. Damour, A. Kleinschmidt, H. Nicolai, Sugawara-type constraints in hyperbolic coset
models. Commun. Math. Phys. 302, 755 (2011). [arXiv:0912.3491]
658. V.A. Belinskiı̆, I.M. Khalatnikov, Effect of scalar and vector fields on the nature of the
cosmological singularity. Zh. Ehsp. Teor. Fiz. 63, 1121 (1972) [Sov. Phys. JETP 36, 591
(1973)]
J. Theor. Phys. 28, 1067 (1989)
662. V.D. Ivashchuk, V.N. Melnikov, Perfect-fluid type solution in multidimensional cosmology.
Phys. Lett. A 136, 465 (1989)
663. V.D. Ivashchuk, V.N. Melnikov, A.I. Zhuk, On Wheeler–DeWitt equation in multidimensional
cosmology. Nuovo Cim. B 104, 575 (1989)
664. V.D. Ivashchuk, V.N. Melnikov, Multidimensional cosmology with m-component perfect
fluid. Int. J. Mod. Phys. D 3, 795 (1994). [arXiv:gr-qc/9403064]
665. J. Fuchs, C. Schweigert, Symmetries, Lie Algebras and Representations (Cambridge Univer-
sity Press, Cambridge, 2003)
666. N.A. Obers, B. Pioline, E. Rabinovici, M-theory and U-duality on T d with gauge back-
grounds. Nucl. Phys. B 525, 163 (1998). [ arXiv:hep-th/9712084]
667. N.A. Obers, B. Pioline, U-duality and M-theory. Phys. Rep. 318, 113 (1999).
668. J. Brown, O.J. Ganor, C. Helfgott, M theory and E10 : billiards, branes, and imaginary roots.
JHEP 0408, 063 (2004). [arXiv:hep-th/0401053]
669. J. Brown, S. Ganguli, O.J. Ganor, C. Helfgott, E10 orbifolds. JHEP 0506, 057 (2005).
References 821
670. M. Henneaux, M. Leston, D. Persson, P. Spindel, Geometric configurations, regular subalge-

bras of E10 and M-theory cosmology. JHEP 0610, 021 (2006). [arXiv:hep-th/0606123]
671. M. Bagnoud, L. Carlevaro, Hidden Borcherds symmetries in Z n orbifolds of M-theory and
magnetized D-branes in type 00 orientifolds. JHEP 0611, 003 (2006). [arXiv:hep-th/0607136]
672. T. Damour, H. Nicolai, Eleven dimensional supergravity and the E10 =K.E10 / -model at low
A9 levels. arXiv:hep-th/0410245
673. T. Damour, H. Nicolai, Higher-order M-theory corrections and the Kac–Moody algebra E10 .
674. V.D. Ivashchuk, V.N. Melnikov, Multidimensional classical and quantum cosmology with
intersecting p-branes. J. Math. Phys. 39, 2866 (1998). [arXiv:hep-th/9708157]
675. L.A. Forte, Arithmetical chaos and quantum cosmology. Class. Quantum Grav. 26, 045001
(2009). [arXiv:0812.4382]
676. A. Kleinschmidt, M. Koehn, H. Nicolai, Supersymmetric quantum cosmological billiards.
Phys. Rev. D 80, 061701 (2009). [arXiv:0907.3048]
677. M. Koehn, Relativistic wavepackets in classically chaotic quantum cosmological billiards.
Phys. Rev. D 85, 063501 (2012). [arXiv:1107.6023]
678. V.D. Ivashchuk, V.N. Melnikov, Quantum billiards in multidimensional models with fields of
forms. Grav. Cosmol. 19, 171 (2013). [arXiv:1306.6521]
679. V.D. Ivashchuk, V.N. Melnikov, Quantum billiards in multidimensional models with branes.
Eur. Phys. J. C 74, 2805 (2014). [arXiv:1310.4451]
680. C.W. Misner, Minisuperspace, in Magic Without Magic, ed. by J.R. Klauder (Freeman, San
Francisco, 1972)
681. T. Damour, H. Nicolai, Symmetries, singularities and the de-emergence of space. Int. J. Mod.
Phys. D 17, 525 (2008). [arXiv:0705.2643]
682. B. Julia, Kac–Moody symmetry of gravitation and supergravity theories, in Applications of
Group Theory in Physics and Mathematical Physics, ed. by M. Flato, P. Sally, G. Zuckerman.
Lectures in Applied Mathematics, vol. 21 (AMS, Providence, 1985)
683. G.W. Moore, Finite in all directions. arXiv:hep-th/9305139
684. S. Mizoguchi, E10 symmetry in one-dimensional supergravity. Nucl. Phys. B 528, 238 (1998).
685. S. Elitzur, A. Giveon, D. Kutasov, E. Rabinovici, Algebraic aspects of matrix theory on T d .
686. O.J. Ganor, Two conjectures on gauge theories, gravity, and infinite dimensional Kac–Moody
groups. arXiv:hep-th/9903110
687. P.C. West, E11 and M theory. Class. Quantum Grav. 18, 4443 (2001). [arXiv:hep-th/0104081]
688. I. Schnakenburg, P.C. West, Kac–Moody symmetries of 2B supergravity. Phys. Lett. B 517,
421 (2001). [arXiv:hep-th/0107181]
689. F. Englert, L. Houart, A. Taormina, P.C. West, The symmetry of M-theories. JHEP 0309, 020
(2003). [arXiv:hep-th/0304206]
690. P.C. West, E11 , SL.32/ and central charges. Phys. Lett. B 575, 333 (2003).
691. F. Englert, L. Houart, G +++ invariant formulation of gravity and M-theories: exact BPS
solutions. JHEP 0401, 002 (2004). [arXiv:hep-th/0311255]
Chapter 14
Perspective
And I said to my spirit, When we become the enfolders of those

orbs and the pleasure and knowledge of every thing in them,
shall we be filled and satisfied then?
And my spirit said No, we level that lift to pass and continue
beyond.
— Walt Whitman, Leaves of Grass (1855 edition)
Observations have verified the cosmic concordance model to a high degree of

accuracy. We have detailed information about the accelerating phase of the early
universe and the formation of cosmic structures since then, from super-clusters
of galaxies down to our solar system. Still, the theoretical framework is largely
incomplete and many question marks remain. We do not know whether the universe
begun with a bang or passed through a state of minimal size. We do not know what
triggered primordial inflation, whether a scalar field or gravity itself or some other
mechanism. We do not know what constitutes two thirds of the energy density of
the universe and is responsible for late-time acceleration. The impressive number of
quantum-gravity and string scenarios proposed in the last years is a manifestation of
this remarkable juncture of questions. Any such model should be able to address
a large set of issues: Does it solve the big-bang problem? Does it explain the
cosmological constant? Does it solve the coincidence problem? What is the nature
of the agent responsible for primordial inflation? Do the trans-Planckian and
problems naturally disappear? Does the model leave characteristic signatures in the
sky, something different from the standard cosmological theory?
Most scenarios answer only a minimal part of these and other questions. For
instance, while the problem of the initial singularity has been attacked in many
different ways and found possible resolutions, the cosmological constant problem
is still alive and in good health. We do not observe a big bang; we do observe a
cosmological constant. Getting the measured value of in a robust theory without
fine tuning is proving to be a formidable task.
Eventually, gathering more data will help to disentangle those We Don’t Knows.
The increased precision of our knowledge of the primordial CMB correlation
functions thanks to WMAP and PLANCK have already ruled out whole classes
of inflationary models while putting others under tight pressure, most of which
had been thriving for years. Further progress in the same direction via a better

DOI 10.1007/978-3-319-41127-9_14
824 14 Perspective
determination by PLANCK of the polarization spectra and of the level of non-

Gaussianity, as well as via other experiments such as BICEP3 [1, 2] (in operation)
and LiteBIRD [3] (to be launched in 2022) will stimulate the study of the most
favoured scenarios of inflation [4, 5] and their origin from a fundamental theory.
The detection of a B-mode polarization signal in the CMB is particularly important
because it would make contact with the physics of gravitational waves and vorticity
(vector) perturbations and, perhaps, also with quantum gravity. In fact, a large
tensor-to-scalar ratio r can be produced not only by specific models of slow-roll
inflation (such as those with quadratic and cosine potential), but also in scenarios
beyond the standard big-bang model, for instance in the presence of a cosmological
bounce [6] or provided the dispersion relation of perturbation modes be modified
as in the large class of quantum-gravity models of Sect. 11.9. Also, non-zero TB
and EB spectra would also announce parity violation and call for new, possibly
gravity-related phenomenology (Sects. 5.9.2 and 11.6.2). The forecast sensitivity of
near-future experiments, both ground-based and air-borne, can saturate the lower
bound (5.162) and permit a detection of a tensor-to-scalar ratio r D O.103 / [7, 8],
up to two orders of magnitude smaller than the current bounds (4.71) and (4.73).
Together with polarization, the search for gravitational waves will receive the
lion’s share of the community attention in the next few years, especially after the
observation of the black-hole mergers GW150914 and GW151226 by Advanced
LIGO [9, 10]. A battery of experiments will scan the sky for the direct detection of
signals originated from various sources, including primordial inflation and cosmic
strings [11, 12]. Observations will be carried on by ground-based interferometers
(Advanced LIGO [13, 14], in operation; Advanced Virgo [15], to be in operation
by 2017; KAGRA [16], to be in operation by 2018; LIGO-India [17, 18], under
construction), space-borne laser interferometer antennæ (eLISA [19, 20], to be in
operation by 2034; DECIGO [21, 22], proposed) and pulsar timing array projects
(PPTA [23], in operation; EPTA [24], in operation; NANOGrav [25], in operation;
SKA [26], to be in operation by 2020). Some of these experiments will be sensitive
to a range of frequencies overlapping with the inflationary spectrum (e.g., DECIGO
[27]) and can give valuable information on the post-inflationary reheating epoch
[28]. As GW150914 showed, it is possible to use gravitational waves to constrain
different kinds of deviation from general relativity, including violations of Lorentz
symmetry from quantum-gravity or exotic geometry effects [29–32]. The indepen-
dent discovery of gravitational waves, or the determination of strong upper bounds
on the signal, will allow us to test several models beyond standard cosmology,
in primis string theory. In particular, the cosmic strings produced in D-brane
inflationary models (equation (13.63)) will be within the range of detectability.
In parallel, the composition of the universe and the behaviour of the gravitational
interaction at different curvature and energy scales will be clarified by different
experiments. Multi-purpose missions such as the Fermi Gamma-ray Space Tele-
scope [33] will study dark matter and highly energetic astrophysical processes.
The nature of dark energy is expected to be pinned down by detailed surveys of
cosmic structures, baryon acoustic oscillations and galaxies at a wide range of
redshift spanning 10 billion years, such as the one to be performed by the E UCLID
satellite to be launched in 2020 [34, 35]. And, although we celebrated the CDM
14 Perspective 825
model from page 1, it must also be said that not all observations are in that much
concordance, since the estimates of some PLANCK 2015 parameters have not been
reproduced by other data sets. This discrepancy can disappear if a dynamical dark-
energy component is assumed instead of a cosmological constant [36]. Near-future
data will settle this interesting issue.
Probably, cosmology alone cannot provide the smoking gun of quantum gravity
or string theory. As a matter of fact, as the concept of falsifiability evolves with the
challenges offered by modern theories, a smoking gun of such theories may assume
rather subtle shapes. Consider for instance the issue of whether and how one can test
the string landscape. The Standard Model of electroweak and strong interactions
works extremely well also at scales above the electroweak scale mW , but it
cannot be extended arbitrarily: if E was the energy scale below which the Standard
Model is valid, then the Higgs mass should be fine tuned to an accuracy of order
.mW =E/2 . To avoid such a fine tuning, a low-energy scale E MSUSY for the onset
of supersymmetry has been traditionally invoked. However, the string landscape
has revolutionized the concept of naturalness (Sect. 5.10.4) and our expectations
about how small the coupling constants of Nature can be. Supersymmetry with
a large energy scale MSUSY is no longer excluded on account of an unwanted
fine tuning and there may be sufficiently many vacua with a large supersymmetry
scale where the observed Higgs mass and cosmological constant are realized. A
supersymmetric extension of the Standard Model with large MSUSY , such as the
scenario of split supersymmetry [37–42], can have measurable consequences which
include a prediction range for the Higgs mass and the observability of the gluino at
the LHC. The observed Higgs mass falls within the predicted range; the appearance
of long-lived gluinos in LHC particle collisions can be the next signature of a
high supersymmetry scale at 10 – 109 TeV. Indirectly, the string landscape would
gain support from this observation: the large number of vacua of the landscape,
coupled with weak anthropic arguments of super-selection [43], can accommodate
the overt “unnaturalness” of split supersymmetry in a natural albeit peculiar way.
At the time of completion of this book, no significant evidence for new particles has
been observed and previous hints about a resonance at 750 GeV [44–46] have been
recognized as statistical fluctuations.
Information on quantum gravity is leaking also from experiments with analogue-
gravity systems (Sect. 7.6.4). In one of the most famous semi-classical approxi-
mations to quantum gravity, Hawking predicted the evaporation of a black hole
via the production of virtual quantum particles at the horizon and the subsequent
splitting into two real particles: one falling back into the black hole and the other
escaping away [47, 48]. Soon after, it was realized that the propagation of particles
travelling at the speed of light near a black-hole horizon is described by the same
equations governing the motion of sound waves in a convergent quantum-fluid flow
(in other words, phonons play the role of photons) [49], in particular a Bose–
Einstein condensate [50, 51]. The possibility to create an analogue black hole in the
laboratory has been stimulating experimental research and, recently, the analogue
of the thermal spectrum of Hawking radiation has been finally observed in a low-
temperature Bose–Einstein condensate [52, 53]. This step forward in the empirical
determination of simulated quantum-gravity effects in controlled conditions might
826 14 Perspective
just as well be the first of a long series, hopefully leading us to uncharted territory
away from semi-classicality and giving us a valuable insight into new gravitational
physics.
String and quantum-gravity cosmologies have started to produce characteristic
and sophisticated predictions. A non-ambiguous physical evidence for exotic states
of matter or geometry in the universe would open up a new season for our view of the
high-energy and gravitational structure of spacetime and would dramatically boost
the research for a viable, completely consistent theory of fundamental interactions.
Whatever the final answer (if any) turns out to be, our journey through modern
cosmology has just begun.
References
1. Z. Ahmed et al. [BICEP3 Collaboration], BICEP3: a 95GHz refracting telescope for

degree-scale CMB polarization. Proc. SPIE Int. Soc. Opt. Eng. 9153, 91531N (2014).
[arXiv:1407.5928]
2. W.L.K. Wu et al., Initial performance of BICEP3: a degree angular scale 95 GHz band
polarimeter. J. Low Temp. Phys. 184, 765 (2016). [arXiv:1601.00125]
3. T. Matsumura et al., Mission design of LiteBIRD. J. Low. Temp. Phys. 176, 733 (2014).
[arXiv:1311.2847]
4. J. Martin, C. Ringeval, V. Vennin, Encyclopædia inflationaris. Phys. Dark Univ. 5–6, 75 (2014).
[arXiv:1303.3787]
5. J. Martin, C. Ringeval, R. Trotta, V. Vennin, The best inflationary models after Planck. JCAP
1403, 039 (2014). [arXiv:1312.3529]
6. J.-Q. Xia, Y.-F. Cai, H. Li, X. Zhang, Evidence for bouncing evolution before inflation after
BICEP2. Phys. Rev. Lett. 112, 251301 (2014). [arXiv:1403.7623]
7. P. Creminelli, D. López Nacir, M. Simonović, G. Trevisan, M. Zaldarriaga, Detecting
primordial B-modes after Planck. JCAP 1511, 031 (2015). [arXiv:1502.01983]
8. M. Remazeilles, C. Dickinson, H.K.K. Eriksen, I.K. Wehus, Sensitivity and foreground mod-
elling for large-scale cosmic microwave background B-mode polarization satellite missions.
Mon. Not. R. Astron. Soc.458, 2032 (2016). [arXiv:1509.04714]
9. B.P. Abbott et al. [LIGO Scientific and Virgo Collaborations], Observation of gravita-
tional waves from a binary black hole merger. Phys. Rev. Lett. 116, 061102 (2016).
[arXiv:1602.03837]
10. B.P. Abbott et al. [LIGO Scientific and Virgo Collaborations], GW151226: observation of
gravitational waves from a 22-solar-mass binary black hole coalescence. Phys. Rev. Lett. 116,
241103 (2016). [arXiv:1606.04855]
cosmic string parameters from gravitational wave direct detection experiments. Phys. Rev. D
86, 023503 (2012). [arXiv:1202.3032]
cosmic strings from future CMB, pulsar timing and gravitational wave direct detection
experiments. Phys. Rev. D 87, 023522 (2013); Erratum-ibid. D 87, 069903(E) (2013).
[arXiv:1210.2829]
13. J. Aasi et al. [The LIGO Scientific Collaboration], Advanced LIGO. Class. Quantum Grav. 32,
074001 (2015). [arXiv:1411.4547]
14. https://www.advancedligo.mit.edu
15. T. Accadia et al., Status of the Virgo project. Class. Quantum Grav. 28, 114002 (2011)
16. K. Kuroda [LCGT Collaboration], Status of LCGT. Class. Quantum Grav. 27, 084004 (2010)
References 827
17. C.S. Unnikrishnan, IndIGO and LIGO-India: scope and plans for gravitational wave research
and precision metrology in India. Int. J. Mod. Phys. D 22, 1341010 (2013). [arXiv:1510.06059]
18. https://www.ligo.caltech.edu/news/ligo20160217
19. https://www.elisascience.org
20. P. Amaro-Seoane et al., eLISA/NGO: astrophysics and cosmology in the gravitational-wave
millihertz regime. GW Notes 6, 4 (2013). [arXiv:1201.3621]
21. N. Seto, S. Kawamura, T. Nakamura, Possibility of direct measurement of the acceleration of
the universe using 0.1-Hz band laser interferometer gravitational wave antenna in space. Phys.
Rev. Lett. 87, 221103 (2001). [arXiv:astro-ph/0108011]
22. S. Kawamura et al., The Japanese space gravitational wave antenna: DECIGO. Class. Quantum
Grav. 28, 094011 (2011)
23. http://www.atnf.csiro.au/research/pulsar/ppta
24. R.D. Ferdman et al., The European Pulsar Timing Array: current efforts and a LEAP toward
the future. Class. Quantum Grav. 27, 084014 (2010). [arXiv:1003.3405]
25. P.B. Demorest et al., Limits on the stochastic gravitational wave background from the North
American Nanohertz Observatory for Gravitational Waves. Astrophys. J. 762, 94 (2013).
[arXiv:1201.6641]
26. http://www.skatelescope.org
27. S. Kuroyanagi, S. Tsujikawa, T. Chiba, N. Sugiyama, Implications of the B-mode polarization
measurement for direct detection of inflationary gravitational waves. Phys. Rev. D 90, 063513
(2014). [arXiv:1406.1369]
28. S. Kuroyanagi, K. Nakayama, S. Saito, Prospects for determination of thermal history
after inflation with future gravitational wave detectors. Phys. Rev. D 84, 123513 (2011).
[arXiv:1110.4169]
29. S.B. Giddings, Gravitational wave tests of quantum modifications to black hole structure.
30. J. Ellis, N.E. Mavromatos, D.V. Nanopoulos, Comments on graviton propagation in light of
GW150914. Mod. Phys. Lett. A 31, 1675001 (2016). [arXiv:1602.04764]
31. M. Arzano, G. Calcagni, What gravity waves are telling about quantum spacetime. Phys. Rev.
D 93, 124065 (2016). [arXiv:1604.00541]
32. G. Calcagni, Lorentz violations in multifractal spacetimes. arXiv:1603.03046
33. http://www.nasa.gov/mission_pages/GLAST/main/index.html
34. http://www.euclid-ec.org
35. L. Amendola et al. [Euclid Theory Working Group Collaboration], Cosmology and fundamen-
tal physics with the Euclid satellite. Living Rev. Relat. 16, 6 (2013). [arXiv:1206.1225]
36. S. Joudaki et al., KiDS-450: testing extensions to the standard cosmological model.
arXiv:1610.04606
37. N. Arkani-Hamed, S. Dimopoulos, Supersymmetric unification without low energy
supersymmetry and signatures for fine-tuning at the LHC. JHEP 0506, 073 (2005).
38. G.F. Giudice, A. Romanino, Split supersymmetry. Nucl. Phys. B 699, 65 (2004); Erratum-ibid.
B 706, 65 (2005). [arXiv:hep-ph/0406088]
39. N. Arkani-Hamed, S. Dimopoulos, G.F. Giudice, A. Romanino, Aspects of split supersymme-
try. Nucl. Phys. B 709, 3 (2005). [arXiv:hep-ph/0409232]
40. I. Antoniadis, S. Dimopoulos, Splitting supersymmetry in string theory. Nucl. Phys. B 715,
120 (2005). [arXiv:hep-th/0411032]
41. N. Arkani-Hamed, S. Dimopoulos, S. Kachru, Predictive landscapes and new physics at a TeV.
42. N. Arkani-Hamed, A. Gupta, D.E. Kaplan, N. Weiner, T. Zorawski, Simply unnatural
supersymmetry. arXiv:1212.6971
43. V. Agrawal, S.M. Barr, J.F. Donoghue, D. Seckel, Viable range of the mass scale of the standard
model. Phys. Rev. D 57, 5480 (1998). [arXiv:hep-ph/9707380]
44. G. Aad et al.p[ATLAS Collaboration], Search for high-mass diphoton resonances in pp
collisions at s D 8 TeV with the ATLAS detector. Phys. Rev. D 92, 032004 (2015).
[arXiv:1504.05511]
828 14 Perspective
45. V. Khachatryan et al. [CMS Collaboration], Search

p for diphoton resonances in the mass range
from 150 to 850 GeV in pp collisions at s D 8 TeV. Phys. Lett. B 750, 494 (2015).
[arXiv:1506.02301]
46. A. Strumia, Interpreting the 750 GeV digamma excess: a review. arXiv:1605.09401
47. S.W. Hawking, Black hole explosions. Nature 248, 30 (1974)
48. S.W. Hawking, Particle creation by black holes. Commun. Math. Phys. 43, 199 (1975);
Erratum-Ibid. 46, 206 (1976)
49. W.G. Unruh, Experimental black hole evaporation. Phys. Rev. Lett. 46, 1351 (1981)
50. L.J. Garay, J.R. Anglin, J.I. Cirac, P. Zoller, Black holes in Bose–Einstein condensates. Phys.
51. C. Barceló, S. Liberati, M. Visser, Analog gravity from Bose–Einstein condensates. Class.
52. J. Steinhauer, Observation of self-amplifying Hawking radiation in an analog black hole laser.
Nat. Phys. 10, 864 (2014). [arXiv:1409.6550]
53. J. Steinhauer, Observation of thermal Hawking radiation and its entanglement in an analogue
black hole. Nature Phys. 12, 959 (2016). [arXiv:1510.00621]
Index
A Asymptotic safety, 224, 401, 544, 547, 599,

˛-attractors, 234 603, 604
˛-vacua. See Vacuum state Atomic principle, 713
a`m . See Cosmic microwave background Attractor. See Inflation
(CMB) Automorphism group, 557, 572
Acoustic oscillations. See Cosmic microwave Autonomous systems, 420
background (CMB) Averaged expansion condition, 275
Adiabatic perturbations, 111 Axion, 184, 186, 216, 222, 446, 648
ADM. See Arnowitt–Deser–Misner (ADM) dark matter, 36
variables in string theory, 321, 664, 728
AdS/CFT correspondence, 668 valley, 730
Affine parameter, 263, 354 Axion inflation, 217, 702, 728
Age of the universe, 23, 46 aligned, 702, 734
Airy functions, 474 hierarchical, 702, 734
Alternatives to inflation, 552, 594, 773, 777, monodromy, 702, 735, 749
783
Analogue gravity, 348, 825
Angular resolution, 104 B
Angular scale, 106 B-modes. See Cosmic microwave background
Anisotropic stress tensor, 65, 144 (CMB) polarization
Anisotropies Back-reaction of a brane, 737, 760, 785
CMB, 64, 97 Background field method, 64
primary, 110 Barbero–Immirzi
secondary, 110, 118 field, 451, 453, 456
Anomalies in constraint algebras, 517 parameter, 413, 442, 571
Anthropic principle, 483, 712, 825 Bardeen–Cooper–Schrieffer (BCS) model, 449
Anti-branes, 675, 705 Bardeen potential, 79
Anti-de Sitter spacetime, 270, 341, 668 Barotropic index, 29
Area operator, 441 Baryon acoustic oscillations, 40
Arnowitt–Deser–Misner (ADM) variables, Baryon drag, 116
408, 426 BBN. See Nucleosynthesis
Ashtekar variables, 443 BCS model. See Bardeen–Cooper–Schrieffer
Ashtekar–Barbero variables, 408, 424 (BCS) model
Assisted inflation. See Inflation Becchi–Rouet–Stora–Tyutin (BRST)
Asymptotic freedom, 599, 603 quantization, 632

DOI 10.1007/978-3-319-41127-9
830 Index
Bekenstein–Hawking entropy-area law, 313, Borde–Guth–Vilenkin theorem, 275, 338

357, 360, 442 Borde–Vilenkin theorem, 272
Bessel functions, 73 Born–Oppenheimer approximation, 485,
spherical, 99 525
BGV theorem. See Borde–Guth–Vilenkin Bose–Einstein condensate, 450, 575, 825
theorem Bose–Einstein distribution, 37, 96
Bianchi identities, 26, 412, 413 Bounce, 277, 824
Bianchi models, 508 Boundary term. See York–Gibbons–Hawking
Bianchi spacetimes, 279 boundary term
I, 279, 284, 291, 508, 509 Brane inflation, 743
II, 291, 508 Branes, 400, 631, 662
IX, 279, 284, 290, 292, 508 gas, 773
V, 279 tension, 632
VI0 , 291 Braneworld, 703
VIII, 284, 290 Gauss–Bonnet, 762
BICEP2, 125, 134 Randall–Sundrum, 762, 764
BICEP3, 824 Brans–Dicke theory. See Jordan–Brans–Dicke
Big bang, 3, 38, 267 theory
shape of singularity, 293 Bremsstrahlung, 36, 95, 96
Big crunch, 268, 481, 507 BRST quantization. See Becchi–Rouet–Stora–
Big-bang nucleosynthesis (BBN). See Tyutin (BRST) quantization
Nucleosynthesis Bubble diagrams, 305, 347
Big-bang problem, 3, 4, 38, 261 Bubble nucleation, 437, 710, 782
and the braneworld, 763
and the ekpyrotic universe, 780, 786
in LQC, 494, 498, 508 C
and modified gravity, 338 Calabi–Yau spaces, 660, 681
in non-local gravity, 603, 786 CAMB , 134
in string theory, 768, 770, 771, 773, 786, Canonical dimension. See Scaling dimension
792 Canonical formalism, 401
in WDW quantum cosmology, 480 Cartan structure equation, 413
Billiards, 290, 482, 679, 703, 788, 792 Cartan sub-algebra, 791
Bimetric gravity, 5 Cartan–Maurer invariant, 445
Bispectrum, 83, 135 Casimir invariant, 440, 441
Belinsky–Khalatnikov–Lifshitz (BKL) Cauchy hypersurface, 264
conjecture, 292, 679 Causal curve, 263
Belinsky–Khalatnikov–Lifshitz (BKL) Causal dynamical triangulations (CDT), 401,
singularity, 290 555, 604
in LQC, 508 bifurcation phase, 561
quantum, 482 branched-polymer phase, 559
in string theory, 777, 788 crumpled phase, 559
time-like, 291 phase-space diagram, 559
BKL. See Belinsky–Khalatnikov–Lifshitz Causal sets, 401, 545, 584
Black holes, 6, 74, 270, 271, 277, 313, 352, CDT. See Causal dynamical triangulations
357, 361, 442, 606, 825 (CDT)
Black-body spectrum, 37, 92 Central charge, 397
Blow-up inflation, 721 CFT. See Conformal field theory (CFT)
Bogoliubov transformations, 191 Chain inflation, 719, 751
Bohm interpretation, 188 Chameleon mechanism, 335, 342, 772
Boltzmann Chan–Paton factors, 637
constant, 8, 357 Chaos, 289, 482, 792
equation, 94, 131 Chaplygin gas, 318
BOOMERanG, 119 Characteristic equation, 177, 429
Borde theorem, 273 Chemical potential, 37, 96
Index 831
Chern–Simons Ricci, 412

action, 653, 654, 678 so.3/ spin, 422
state, 443, 584 spin, 411
Chromodynamics. See Quantum Consistency relations, 204
chromodynamics (QCD) Constraints, 414
CL. See Confidence level (CL) first-class, 423
CLASS, 135 primary, 419
Clifford–Klein space, 43 second-class, 423
Closed Lorentzian manifold, 265 secondary, 420, 422
Closed universe, 108, 164 Continuity equation, 26
in LQC, 500, 507 Continuum limit, 556
Closure constraint, 570 Contortion, 412
CMB. See Cosmic microwave background Convergence condition
(CMB) null, 27, 271, 272
Coarse graining, 68, 139, 207, 548 time-like, 27, 265, 269
COBE, 92, 95, 106, 141 Copenhagen interpretation, 188, 390
Coherent states, 574 Copernican principle, 18, 344
Coincidence problem, 42, 310, 314, 317, 739 Correlation functions, 82, 101, 142
Cold big bang model, 3, 163 Cosmic confusion, 123
Coleman mechanism, 483 Cosmic microwave background (CMB), 4, 15,
Comoving 16, 91
coordinates, 21 acoustic oscillations, 114
distance, 21 cold spots, 110, 118, 132
Compactification, 322, 651, 656, 659 damping tail, 117
flux, 322, 667, 669, 704 dipole, 107
hyperbolic, 769 frame, 52, 107
product space, 769 monopole, 107
spherical, 322 multipole coefficients, 100
spontaneous, 658, 662, 678 multipoles, 98
toroidal, 322, 649 non-Gaussianity (see Non-Gaussianity)
Compatibility equation, 414 polarization, 126, 216, 588, 717, 739, 824
Complex. See Simplicial complex quadrupole, 108
Complex projective plane, 660 spectrum enhancement (see Spectrum
Compton (cosmological))
distortion, 96 spectrum suppression (see Spectrum
drag, 118 (cosmological))
scattering, 36, 95, 96 temperature, 16
Concordance model. See CDM concordance Cosmic strings, 158, 729, 744, 750, 761
model Cosmic variance, 103
Condensates, 349, 442, 449, 574, 672 Cosmological constant, 24, 39
Confidence level (CL), 23 and the CMB, 119
Conformal cyclic cosmology, 278 in LQC, 501
Conformal field theory (CFT), 450, 629, 638 old problem, 39
Conformal time, 20 Cosmological constant problem, 3, 4, 41, 301,
Conformal transformation, 70, 328, 366–368, 442, 589
402, 628 and the braneworld, 764
Congruence and the ekpyrotic universe, 781
null, 266, 354 new, 44, 310
null geodesic, 354 old, 44, 310
time-like, 265 in quantum cosmology, 482
Connection and quantum fluctuations, 304
Ashtekar, 426 self-tuning , 764
Ashtekar–Barbero, 570 in string theory, 709
Levi-Civita, 24 and supersymmetry, 308
832 Index
Cosmological principle, 19, 64, 198, 344 de Sitter

Cosmology, 14 attractor, 218, 581
COSMO MC, 134, 214 in canonical quantum cosmology, 473
Covariant formalism, 28, 76, 262 cosmology, 39, 56, 67, 73, 159, 164, 195,
Covariant path-integral quantization. See 200, 208, 237, 358, 438
String theory horizon (see Horizon)
Coxeter polyhedron, 679 instanton (see Instanton)
CPnC1 . See Complex projective plane in quantum gravity, 561, 562
CPT symmetry, 444, 446 metastability, 708
Creation and annihilation operators, 188, 190, spacetime, 270, 444
191, 305, 449, 570, 632 temperature, 196, 358
Creation of particles, 191 vacua (see Vacuum state)
Creeping solutions, 318, 320, 363 DEC. See Energy condition
Critical density, 33 DECIGO, 824
Critical dimension, 635 Decoherence, 188, 437, 473, 484
Curvaton, 186, 213, 221, 326, 724 Decoupling, 46, 93
Curvature Decoupling theorem, 393
extrinsic, 416 Dedekind function, 768
intrinsic, 19, 43 Deformed algebra, 517
linear comoving perturbation R, 81 Degenerate sector, 439, 447
linear perturbation on uniform density Densitized triad, 418
slices , 80, 332 Density matrix, 437
non-linear comoving perturbation R , 77, Density parameter, 33
210 Deparametrization, 564
non-linear perturbation RNL , 78, 332 Descent relations, 655
non-linear perturbation on uniform density DeWitt metric, 427, 429
slices , 76, 137 DGP scenario. See Dvali–Gabadadze–Porrati
non-linear Salopek–Bond perturbation NL , (DGP) scenario
78, 138 Diffeomorphisms, 66, 346, 401, 423, 429, 431,
second-order comoving perturbation R.2/ , 626, 637
212 transverse, 346
Cycle. See String theory Dilatino, 648, 650
Cyclic cosmology, 277, 776 Dilaton, 231, 328, 637, 638, 657, 683, 768
and inflation, 771, 778
runaway problem (see Overshooting
D problem)
ıN Dimension of spacetime, 635
formalism, 76 Dimensional flow, 401, 606
formula, 78, 206 Dimensional reduction. See Compactification
d’Alembertian. See Laplace–Beltrami operator Dimensional regularization, 306
D-branes. See Branes Dipole. See Cosmic microwave background
D-strings, 654 (CMB)
D-term inflation, 236 Dirac
D-terms, 228, 236, 706, 708, 735 bracket, 423
Dangerous operators, 394 conjecture, 423
Dark energy, 3, 39, 42, 44, 221 Hamiltonian, 419
parametrizations, 318 matrices, 225, 448
tachyonic, 766 observables, 480
Dark matter, 3, 18, 35, 333, 335, 734 procedure, 415
Dark radiation, 762 Dirac–Born–Infeld (DBI) action
Davies–Unruh temperature, 354 of a stable brane on AdS5 X5 , 746, 754,
DBI action. See Dirac–Born–Infeld (DBI) 765
action of a stable brane on curved space, 640, 735,
DBI inflation. See Warped D-brane inflation 765
Index 833
of a stable brane on flat space, 632, 765 Equality

of an unstable brane, 765 strong, 419
Dirichlet boundary conditions, 631 weak, 419
Dispersion relations, 195, 220, 592, 594, 606, Equality (matter-radiation), 34
824 Equivalence principle
Domain walls, 158 strong, 331, 396
Dominant energy condition (DEC). See Energy weak, 331, 772
condition Ergodic field, 81
Doppler effect, 117 Ergodicity, 103
Dressed metric approach, 517 ESR. See Extreme slow roll (ESR)
Dust. See Matter Essential couplings, 547
Dvali–Gabadadze–Porrati (DGP) scenario, Eternal inflation. See Inflation
5 EUCLID , 824
Dynkin indices, 323 Euclidean action, 435, 475, 483, 555
Dyons, 654 Euclidean dynamical triangulations (EDT),
Dyson resummation, 398 556
Euler theorem, 395, 602
Euler–Mascheroni constant, 73, 307
E Everett interpretation, 188, 197, 391
e-foldings, 28, 68, 78, 160, 167, 169, 175, 237, Expansion parameter. See Scale factor
311, 361 Extended supersymmetry. See Supergravity
anisotropic, 287 Exterior derivative, 411
E10 =K.E10 / coset model, 794 Extreme slow roll (ESR), 164, 316
-problem, 175, 229, 326, 720, 729, 743, 750,
755
EDT. See Euclidean dynamical triangulations F
(EDT) F-strings, 654
Effective action, 548 F-term, 228, 325, 670, 738
Effective field theory, 221, 397 f .R/ models, 232, 338
Einstein equations, 24 in LQG quantization, 490
linearized, 65, 84 in string theory, 769
semi-classical, 40, 390 Fast roll, 165, 316, 326, 718
Einstein frame, 328, 331 Fermi constant, 304
Einstein–Hilbert action, 24 Fermi frame, 310, 409
in ADM variables, 427 Fermi–Dirac distribution, 37
Ekpyrotic universe, 703, 776, 786 Fermions
eLISA, 126, 824 and gravity, 412
Emergent gravity, 2, 351, 400, 572, 627 Majorana, 449, 643, 647
Energy condition Weyl, 647
dominant (DEC), 27, 268 Fibre inflation, 723
null (NEC), 26, 271 Fiducial volume, 470, 490
relations among DEC, WEC, NEC and Fierz identities, 226
SEC, 30 Fierz–Pauli Lagrangian, 347
strong (SEC), 27, 270 Fine graining, 68, 139, 207
weak (WEC), 27 Fine-structure constant, 343
Energy-momentum tensor, 25 FIRAS. See COBE
Engineering dimension. See Scaling dimension Fixed point
Engle–Pereira–Rovelli–Livine (EPRL) model. Gaussian, 548
See Spin foams non-Gaussian, 547
Ensemble of skies, 100 Flatness condition, 573
Enthalpy density. See Heat density Flatness problem, 2, 156, 160
Entropy problem. See Flatness problem Floor function, 286
Entropy-area law. See Bekenstein–Hawking FLRW. See Friedmann–Lemaître–Robertson–
entropy-area Walker (FLRW)
834 Index
Flux. See String theory transverse-traceless, 70, 85, 601

Flux (of the triad), 440 uniform density, 79
Flux compactification. See Compactification Gaugino, 650, 672
Fock space, 188, 570, 640, 645 condensation, 667, 672, 704, 705, 727, 767,
Focusing theorems, 265 773
Fokker–Planck equation, 208 Gaunt integral, 136
Foliation, 265 Gauss constraint, 424
Form factor, 598 Gauss–Bonnet
Forms, 222, 411, 648 braneworld (see Braneworld)
4 puzzle, 311, 360 gravity, 338, 352, 527, 767
4-flux. See String theory term, 336, 762
Fractal geometry, 278, 289, 548, 550, 559, 606, Gaussian
607 distribution, 81, 100
Freidel–Krasnov model. See Spin foams field, 81
Friedmann equations, 31, 32 wave-packets, 435
Friedmann–Lemaître–Robertson–Walker Gaussianity, 198
(FLRW), 19 Gell-Mann–Low equation, 547
curvature, 31 General relativity, 19
metric, 19 Generating functional. See Gauge
spacetime temperature, 196 Geodesic incompleteness, 267
FRW. See Friedmann–Lemaître–Robertson– Georgi–Glashow model. See Grand unification
Walker (FLRW) theories (GUT)
Functional renormalization approach, Geroch splitting theorem, 264
547 GFC. See Group field cosmology (GFC)
Fundamental form Ghosts, 334, 337, 396, 397, 527, 597, 635, 768
first, 28, 415 Faddeev–Popov, 636
second, 358 Giant vacua, 15
Future Gliozzi–Scherk–Olive (GSO) projection, 646
causal J C , 263 Global hyperbolicity, 19, 415, 557
chronological I C , 192, 263 Gluino, 672, 825
Fuzzy spacetime, 480, 606 Goldstone
boson, 184, 216, 303, 321, 351, 447
model, 303
G theorem, 303
Galaxy bias, 138 Goncharov–Linde inflation, 233
Galileon, 342 Gowdy model, 508, 517
Gap problem, 712, 782 Graceful entry problem, 180, 197, 218
Gauge, 19 Graceful exit problem, 182, 219
fixing, 66, 415, 422, 518, 635 Gradient expansion, 76
flat, 79 Grand unification theories (GUT), 155, 158,
generating functional, 457 674
harmonic (see Transverse-traceless) Georgi–Glashow model, 155
invariance, 66 scale, 155, 161, 162, 167, 169, 312, 719,
light-cone, 635 748
longitudinal, 79 Gravielectric field. See Densitized triad
modes, 66 Gravitational field. See Vielbein
Newtonian, 79 Gravitational potential, 79, 111
static, 630 Gravitational waves, 70, 190, 198, 216, 824
synchronous, 20, 79, 209 discovery, 6, 74
total-matter, 79 stochastic background, 125
transformations, 414 Gravitino, 648, 650, 677, 749
transformations (large), 444, 446 Graviton, 70, 86, 190, 307, 332, 337, 339, 346,
transformations (small), 66, 423, 444, 457, 351, 391, 392, 396, 397, 398
517 mass, 74
Index 835
and non-locality, 600 Hodge dual, 653, 664

vs. non-perturbative gravity, 402 Holographic equipartition, 358
polarization, 70 Holography, 356, 668
in string theory, 637 Holonomy, 439
Gravity as a gauge theory, 426 Holst term, 413
Green–Schwarz mechanism, 649 Homogeneity, 15, 18
Gross–Pitaevskii equation, 575 Homogeneous structure equation, 422
Group averaging, 478, 501, 563 Hořava–Lifshitz gravity, 5, 606
Group field cosmology, 483, 575 Horizon
Group field theory, 2, 400, 442, 545, 569, 605 crossing, 68, 115
and non-commutativity, 571, 592 and decoherence, 189
Gukov–Vafa–Witten superpotential, 669 de Sitter, 67
GUT. See Grand unification theories (GUT) Hubble, 22, 67, 162
particle, 22, 47, 67, 162, 294, 344
problem, 2, 161, 290
H sound, 115
Haar measure, 479, 570 Horndeski theory, 342, 597
Hagedorn temperature, 772 Hot big bang, 4, 18, 36, 154
Hamilton–Jacobi Hubble
formalism, 171, 581 distance (see Horizon)
momenta, 436, 474 law, 18, 21
Hamiltonian formalism, 414 parameter, 22, 23
constrained systems, 414 parameter (generalized), 274
and quantization, 408 volume, 311
Hankel functions, 194 Hybrid inflation. See Inflation
Harrison–Zel’dovich spectrum. See Spectrum Hybrid quantization, 517
(cosmological) Hyperbolic space H D2 , 791
Hartle–Hawking Hypergeometric function, 50, 306
no-boundary proposal, 437, 482, 787
wave-function, 474
Hausdorff dimension, 545, 606 I
Hawking radiation, 825 Immirzi parameter. See Barbero–Immirzi
Hawking–Penrose theorem, 269 parameter
Heat capacity, 774 Improved quantization. See Loop quantum
Heat density, 355 cosmology (LQC)
Heaviside step function, 478 Inextendible curve, 264
Heisenberg principle. See Uncertainty principle Inflation, 1, 4, 159
Higgs assisted, 185, 213, 753
boson, 222, 304 attractor, 176
discovery, 5 chaotic, 166
inflation, 222, 232 equation of state, 164
mass, 222 eternal, 163, 197, 201, 272, 276, 476, 483,
mechanism, 158, 304 725, 729, 783
and naturalness, 220 extended, 333
Higher-order gravity, 216, 277, 291, 336, 523, Higgs (see Higgs)
703, 767 hilltop, 183
quantum, 337 hybrid, 186, 219
quartic, 336, 397 in string theory, 783
Hilbert space large-field models, 179, 718, 735, 737
non-separable, 494 low-scale, 749
rigged, 497 multi-field, 185, 719, 727
separable, 498 natural, 184, 216, 222, 476, 665, 728, 730
Hilbert–Palatini action, 413, 451 non-commutative, 594
History of the universe, 52, 54, 170 observational constraints, 213
836 Index
old scenario, 165 KKLMMT inflation, 744

potentials (see Potential (scalar field)) KKLT
probability of, 473, 507 stabilization scenarios, 673, 676
pseudo-scalar, 217 uplifting scenarios, 702, 705
quadratic potential, 182, 738 Klebanov–Strassler throat, 661, 670, 675, 705,
small-field models, 183, 719, 721 709, 739, 745, 751, 755, 758, 763
stochastic, 206 Klein–Gordon equation, 31, 477, 503
Inflaton, 1, 164 Kodama state. See Chern–Simons state
mass, 167 Komar energy. See Tolman–Komar energy
natural, 732 density
Infrared, 83
Inhomogeneities, 64, 68, 97, 484, 509, 576, 591
Inner product. See also Two-point function L
non-relativistic, 480 CDM concordance model, 43, 106
relativistic, 479 Lanczos–Lovelock gravity, 352
Instanton, 427, 446, 672, 710, 737, 740, 792 Landscape, 702, 704, 825
Coleman–de Luccia, 708, 717 Langevin equation, 208
Hawking–Moss, 438 Laplace–Beltrami operator, 25, 31, 432, 546
Internal clock, 171, 477 Laplacian, 71
Intertwiners, 441, 509 on a group manifold, 577
Irrelevant operator, 393 Lapse function, 415
Isocurvature perturbations. See Perturbations Large Hadron Collider (LHC), 5, 54
Isotropy, 15, 16, 18, 81, 100 Large scales (cosmological), 68
Large-field inflation. See Inflation
Large-scale structure, 15, 125, 316, 345
J Large-volume
Jordan frame, 328, 331, 639 inflation, 721
Jordan–Brans–Dicke theory, 327, 330, 333, stabilization scenarios, 676, 708
334, 340. See also Scalar-tensor uplifting scenarios, 702, 705, 708, 721
theories Last-scattering surface, 51, 93, 97, 131
Lattice refinement, 510, 514, 578, 583, 591
Legendre polynomials, 98
K associated, 98, 129, 579
k-essence, 342, 638, 754, 766 Lemaître–Tolman–Bondi (LTB) model,
k-inflation, 754 345
-Minkowski, 592 Lensing (gravitational), 118, 138
K3 manifold, 659 Levi-Civita
Kac–Moody algebra, 791 connection (see Connection)
Kähler symbol, 397, 410, 416
manifold, 227, 660 tensor, 410
metric, 227 LHC. See Large Hadron Collider (LHC)
potential, 227, 665, 666, 669 Lie derivative, 26, 66, 76, 87, 358, 416
Kalb–Ramond field. See String theory Light cone
Kaluza–Klein scenarios, 180, 186, 314, 328, future, 263
742 localized, 272
Kasner past, 263
epoch, 285 Light-cone quantization. See String theory
era, 286 LIGO, 6, 74, 125
generalized metric, 282 discovery of gravitational waves (see
metric, 280 Gravitational waves)
Kibble LiteBIRD, 824
gradients, 158 Little group, 634
mechanism, 158 Local frame. See Vielbein
Kination. See Fast roll Local Group, 107
Index 837
Loop quantum cosmology (LQC), 3, 5, 413, in LQC, 508

489, 572 quantum, 482
critical density, 502, 503 in string theory, 790
deformed algebra, 509 Model-building problem, 220
dressed metric, 509 Modified gravity models, 336, 703, 826
and GFT, 580 in string theory, 767
hybrid quantization, 509 Moduli, 222, 230, 231, 325
improved quantization, 496 axio-dilaton, 665, 666
non-Gaussianity, 525 complex-structure (shape), 665
old quantization, 496 fibre, 723
scalar spectrum, 521 Kähler, 666
signature change, 509, 526 problem, 230, 673, 720
tensor spectrum, 521 size, 666
WDW equation, 497 stabilization, 667, 773
Loop quantum gravity (LQG), 3, 224, 401, in string theory, 638, 663, 667, 669
426, 439, 604 two-form scalars, 666
quantization of f .R/ models, 490 Moduli inflation, 702, 720
quantization of scalar-tensor theories, 490 Monodromy, 735
Low-energy action, 393, 397 Monodromy inflation. See Axion inflation
in string theory, 638, 651 Monopole. See Cosmic microwave background
LQC. See Loop quantum cosmology (LQC) (CMB)
LQG. See Loop quantum gravity (LQG) Monopole problem, 2, 158, 163
LTB model. See Lemaître–Tolman–Bondi M ONTE PYTHON , 135
(LTB) model Moyal product, 593
Luminosity distance, 52, 345 Mpc (megaparsec), 7
Lyth bound, 204, 752, 753, 757, 760 Mukhanov–Sasaki
equation, 71, 73, 193, 199, 201, 756
variable, 73, 520
M Multi-fractional spacetimes, 606, 607
M-theory, 186, 400, 672, 678, 773, 788 Multi-scale spacetimes, 592
flux compactification, 682, 753 Multiverse, 188, 189, 482, 569, 715
and non-commutativity, 592 Myers effect, 752
uplifting scenarios, 708
vacua, 704
Maass wave-form, 794 N
Marginal operator, 393 N-flation, 702, 719, 732
Marginalization, 123 Nambu–Goldstone boson. See Goldstone
Mass gap, 450 boson
Massive gravity, 5 Natural inflation. See Inflation
M ATHEMATICA , 364 Naturalness, 220, 704, 825
Matrix models, 400 NEC. See Energy condition
Matter, 33 Neumann boundary conditions, 629
dust, 33 Neveu–Schwarz field. See String theory
MAXIMA, 119 Newton coupling, 24
Mean-field approximation, 350, 449, 515, 575 varying, 399
Measure problem in cosmology, 198, 476, 484, Newtonian potential, 398
507, 715 Nieh–Yan invariant, 414, 451
Metastable vacua. See Vacuum state Nil manifold. See Twisted torus
Milne universe, 270 No-boundary proposal. See Hartle–Hawking
Mini-superspace, 3, 468 no-boundary proposal
Minimal coupling, 24, 223 No-hair theorem, 218
Misner space, 787 No-scale models, 230, 232, 309, 663, 670, 725
Mixmaster universe, 290 Non-adiabatic pressure, 77
and the braneworld, 763 Non-commutative
838 Index
cosmology, 486, 593 P

field theory, 592 p-adic string, 641
spacetimes, 401, 591 P 4Œ1;1;1;6;9 , 676, 729
Non-Gaussianity, 217, 594 Padmanabhan’s theory, 352
and the CMB, 135 Palatini formalism, 340, 341, 410, 411
equilateral form, 140 Parity, 408
experimental bounds, 141 conservation, 131, 134
and inflation, 205 in quantum cosmology, 477
in LQC, 525 transformation, 129
in string theory, 719, 724, 733, 757, 758, violation, 217, 446, 588, 651, 824
760, 775, 779, 783 Partially framed flag. See Big bang
local form, 138, 144 Particle
multi-field, 213 and non-commutativity, 592
non-linear parameter, 137, 206 non-relativistic, 414
orthogonal form, 141 relativistic, 415, 565
resonant form, 739 Particle horizon. See Horizon
squeezed limit, 139, 146 Partition function, 305, 348, 360, 555, 571
Non-linear parameter. See Non-Gaussianity discrete, 557
Non-linear sigma model, 227, 638 Past
Non-local causal J , 263, 585
cosmology, 703, 769 chronological I , 192, 263
theories, 401, 596, 775 Patch cosmology, 238
Non-locality, 449, 546, 587, 593, 596 Path integral, 350, 400, 401, 432, 433, 437,
in string theory, 641, 655, 769, 775, 786 475, 508, 555, 559, 564, 568, 571,
Non-minimal coupling, 223, 232, 315, 327, 575, 587, 605, 636
469 Pauli matrices, 225, 439
Non-polynomial theories. See Non-local Peccei–Quinn mechanism, 446
theories Penrose–Hawking theorem, 271
Non-relativistic matter. See Matter (dust) Perfect fluid, 27
NS5-brane, 654, 735, 752, 754, 758 Perturbations (cosmological), 4, 63
adiabatic, 80, 213
Nucleosynthesis, 39, 53, 93, 126, 154, 168,
curvature (see Curvature)
322, 495, 555, 720, 772
definition, 66
Null energy condition (NEC). See Energy
homogeneous, 177, 284
condition
isentropic, 80
Numerical solutions, 324, 363
isocurvature, 111, 115, 116, 213, 724
linear, 64, 198
multi-field, 185
O non-linear, 64, 69, 76, 198
O-planes, 670 scalar, 65, 76
On-shell condition, 26, 391 second-order, 198
One-particle-irreducible (1PI) diagrams, 395 stability, 177
Open Lorentzian manifold, 265 tensor, 65, 70, 144, 199
Open universe, 108, 164 vector, 65
in LQC, 501, 507 in WDW quantization, 484
Operator ordering, 430 Phantoms, 165, 334, 342
Optical depth, 97 Phase transition
Orbifold, 660, 787 first order, 561
Orientifold, 661, 671, 787 second order, 561
Oscillations in the CMB spectra. See Spectrum Phonons, 583
(cosmological) Pivot scale, 123
Ostriker–Vishniac effect, 118 PLANCK mission, 6, 15, 40, 104, 106
Ostrogradski instability, 597, 767 TT+lowP+lensing, 23
Overshooting problem, 728, 771 TT+lowP+lensing+BAO, 43
Index 839
TT+lowP+lensing+ext+BKP, 125 the problem of, 389, 398, 458

Planck scale renormalization, 24, 395
length, 7, 166 and the weak gravity conjecture, 719
mass, 7, 155 Quantum reduced loop gravity, 509
reduced mass, 8 Quintessence, 42, 315, 333, 361, 363
temperature, 8 coupled, 333
time, 7, 155 extended, 333
Poincaré dodecahedral space, 43 in string theory, 702, 724, 739, 772
Poisson bracket, 418 Quintessential inflation, 325, 715
Poisson distribution, 585
POLARBEAR , 134
Polarization (CMB). See Cosmic microwave R
background (CMB) R-boost, 334
Polarization (graviton). See Graviton Racetrack models, 673, 675
Poly-instanton inflation, 724 inflation, 729
Polyakov action, 627 Radiation, 33
Polymeric quantization, 497 Radiative transfer, 94
Pontryagin term, 217, 446 Radion, 666
Potential (scalar field) Ramond field. See String theory
cosine, 184 Ramond–Ramond potentials, 648
double-well, 183, 303 Randall–Sundrum model. See Braneworld
exponential, 180, 769 Random field, 81, 103
monomial, 181, 321 Raychaudhuri equation, 33, 265, 293
quadratic, 182, 216, 717, 733, 755 Recombination, 93
quartic, 183, 216 Redshift, 21
washboard, 781 Rees–Sciama effect, 118
Power counting. See Renormalization Regge
Power-law expansion, 35, 57, 67, 74, 159 action, 559
Pre-big-bang cosmology, 703, 771, 776, 778, calculus, 556
786 slope, 627
Preheating, 168, 217 Regularization, 393, 602
Probability of inflation. See Inflation of singularities, 661
Proper distance, 21 Reheating, 161, 167, 172, 219, 312
Proper time, 19, 263 in string theory, 724, 734, 761
Pseudo-scalar field, 409, 452 Reionization, 117
Pseudo-trace axial vector, 413 Relevant operator, 393
Renormalization
of gravity (see Quantum gravity)
Q group, 393, 401
Quadrupole. See Cosmic microwave perturbative, 307, 393
background (CMB) power counting, 393, 394, 402
Quantization, 428, 442, 493 of string theory, 648
Quantum chromodynamics (QCD), 349, 446 super-, 395, 599
super-, 323 Resonance at 750 GeV, 825
Quantum cosmology, 188 Resummed quantum gravity, 398, 550
Quantum field theory, 5 Ricci
effective, 41 scalar, 24
Quantum fluctuations, 2, 189 tensor, 24
Quantum gravity, 2, 4, 43, 155, 217, 314, Riemann tensor, 24
389 Rindler
approaches, 400, 604 frame, 352
canonical, 413 horizon, 353
higher-derivative, 338 observer, 353
perturbative, 337, 392 Runaway potential, 326
840 Index
S theorems, 269
-model. See Non-linear sigma model time-like, 267
S-duality, 680 weak, 267
S-matrix, 391, 392, 396, 626 6dFGS, 15
Sachs–Wolfe SKA, 824
effect, 110 Skewness, 136
integrated effect, 113 sl.2; C /. See Special linear algebra
plateau, 111 sl.2; R /. See Special linear algebra
Sample variance, 106 Sloan Digital Sky Survey (SDSS), 15, 52
Scalar constraint. See Super-Hamiltonian Sloan Great Wall, 15
constraint Slow roll
Scalar field, 30, 163, 395 first parameter, 55
Scalar-tensor theories, 164, 217, 277, towers, 172
291, 327, 346, 483. See also Slow-roll D-brane inflation. See Warped
Jordan–Brans–Dicke theory D-brane inflation
in LQG quantization, 490 Small scales (cosmological), 67
in WDW quantization, 469, 483 Small-field inflation. See Inflation
Scale factor, 19 Smeared functionals, 418, 421, 458
Scale invariance. See Spectrum (cosmological) Sound horizon. See Horizon
Scaling dimension, 30, 393 Spacetime, 262
Scaling solutions, 316, 317, 363 causal, 263
Schwarzschild length, 294 causally simple, 263
Schwinger singular, 267
gauge (see Time gauge) stably causal, 263
representation, 478 time-orientable, 263
variables, 418 Special linear algebra, 225, 633
SDSS. See Sloan Digital Sky Survey (SDSS) Spectral dimension, 546, 606
SEC. See Energy condition Spectral index
Seeds (primordial), 159, 163 scalar, 121, 203
Selection rules. See Super-selection rules running, 121, 203
Self-dual connection, 426 tensor, 122, 200
Semi-classical states, 434 Spectrum (cosmological)
Separate universe, 68, 76, 166, 511, 526, 577 blue tilt, 83
Sequestered models. See Volume-modulus CMB angular, 100
inflation CMB temperature, 101, 104
Shear, 28 de Sitter, 195, 200
Shift symmetry, 310, 356, 360 enhancement in quantum gravity, 486
Shift vector, 415 enhancement in string theory, 718, 774
Shooting method, 364 Harrison–Zel’dovich, 112, 124, 214, 216
Silk damping, 117, 164 inflationary, 195
Simplex, 556, 570 oscillations, 739
Simplicial complex, 400, 468, 572 power law, 83
Simplicity constraints, 573 power spectrum, 83
Singularity, 3, 267 red tilt, 83
big bang (see Big bang) scalar, 121, 203
big brake, 268 scalar amplitude, 122
big crunch (see Big crunch) scale invariant, 83
big freeze, 268 suppression in quantum gravity, 486,
big rip, 268 524–526, 595
conformal, 267 suppression in string theory, 718, 774
naked, 267 tensor, 112, 122, 200
space-like, 267 tensor amplitude, 122
strong, 267 Spherical harmonics, 98
sudden future, 268 Spin foams, 400, 401, 508, 545, 563, 605
Index 841
cosmology, 508 and non-commutativity, 592

EPRL model, 564, 583 null states, 633
Freidel–Krasnov model, 564 old covariant quantization, 632
vertex expansion, 564 open strings, 629
Spin networks, 401, 441, 493, 509, 510, 570 oriented strings, 637
quantum, 450 Ramond sector, 644
Spinors. See Fermions scale, 628, 743, 751, 752, 756, 792
Spontaneous symmetry breaking, 184, 222, spurious states, 633
302, 314 Standard Model, 659, 662, 671, 704
SPTPOL, 134 superstrings, 642
Squeezed limit. See Non-Gaussianity twist, 637
Stability. See Perturbations type-I, 646, 647, 661
Standard candles, 52, 318 type-II, 647
Standard Model, 18, 36, 222, 626, 825 unoriented strings, 637, 647
in string theory (see String theory) winding modes, 772
Starobinsky inflation, 232, 333, 341, 553 winding number, 657
in string theory, 724, 738 String-gas cosmology, 703, 771, 772, 786
Statistical selection, 713 Strong energy condition (SEC). See Energy
Stefan–Boltzmann law, 38 condition
Stelle theory, 337, 396, 597 Strong equivalence principle. See Equivalence
Sterile inflaton, 326 principle
Stiff matter, 39 Structure formation, 138, 322
Stochastic inflation. See Inflation Sum-over-histories. See Path integral
Stokes parameters, 128 Sunyaev–Zel’dovich effect, 96, 118, 138
Stress-energy tensor. See Energy-momentum Super-acceleration, 55
tensor Super-Hamiltonian constraint, 420, 423, 425,
String cosmology, 3, 5, 486 428
String field theory, 599, 640, 655, 765, 770 Super-inflation, 55
String frame. See Jordan frame Super-luminal modes, 338, 527, 768
String theory, 2, 5, 43, 182, 185, 186, 231, 291, Super-momentum constraint, 420, 423, 425,
309, 314, 315, 322, 325, 400, 408, 428
482, 606, 626 Super-selection rules, 480, 566, 712
admissible states, 633 Superconductivity. See Bardeen–Cooper–
anomalies, 646, 648, 649 Schrieffer (BCS) model
bosonic, 627 Superficial degree of divergence, 395
BRST quantization, 636, 646 Superfield, 226
closed strings, 630 Supergravity, 5, 185, 223, 224, 228, 338, 397,
coupling gs , 637 602, 651
covariant path-integral quantization, 636, compactification, 180, 322, 659
646 in D D 11, 678
critical dimension, 635, 646 extended, 321
cycle, 662, 721 and inflation, 180, 229, 230, 725
dualities, 678 and , 308, 325
flux, 662 in string theory, 649
gauge group, 641, 656, 769 universal sector, 648
heterotic, 649 Supernovæ, 18, 318
heterotic M-theory, 680, 761, 776 Superpotential, 226
interactions, 637, 648 Superspace (canonical gravity), 429
Kalb–Ramond field, 637–639, 662 Superspace (supersymmetry), 226
level matching, 634, 657 Supersymmetry, 5, 36, 42, 164, 223, 224, 642
light-cone quantization, 635 dynamical breaking, 323
low-energy limit, 638, 651 global, 225
Neveu–Schwarz sector, 592, 644 and Hamiltonian formalism, 408
no-ghost theorem, 636, 646 and , 308
842 Index
and quantum cosmology, 408, 792 Tunneling, 435. See also Instanton
scale MSUSY , 42, 308, 313, 677, 709, 713, Tunneling proposal. See Vilenkin tunneling
727, 825 proposal
split, 749, 825 Twisted torus, 736
Surface states, 656 2-complex, 564
Symmetry reduction, 467 2dFGRS, 15, 52
Symplectic structure, 418 Two-point function
Symplectomorphism, 424 Hadamard, 567
Synchronous time, 19 Newton–Wigner, 567
T U
T-duality, 657 U-duality, 680, 788
Tachyon, 634, 636, 641, 703 Ultra-local gravity, 459
condensation, 655, 765, 787 Ultraviolet, 83
cosmological, 765 Uncertainty principle, 468, 591
Target spacetime, 640 Uniform-density slices, 77
Temperature of the universe. See Cosmic Unimodular gravity, 346, 356, 360, 589
microwave background (CMB) Unitarity, 337, 396
Tensor models (quantum gravity), 569 Unitary gauge, 304
Tensor-to-scalar ratio, 123, 204 Universe, 23
Test field, 201, 484 Unruh temperature. See Davies–Unruh
Tetrad. See Vielbein temperature
Tetrahedron, 556, 570, 661
Textures, 158
Thawing solutions, 320 V
Theory of everything, 2, 400 Vacuum state, 188, 191
Thermal fluctuations, 594 anti-de Sitter, 671, 682, 705
Thermal history. See History of the universe Bunch–Davies, 192, 305
Thiemann identity, 440, 513 de Sitter, 705
Third quantization, 43, 188, 483, 569 de Sitter (˛-vacua), 192
Thomson scattering, 36, 117, 126 energy density (see Zero-point energy)
Tilt. See Spectrum equation of state, 40
3-flux. See String theory false, 302, 642
Time gauge, 415 instantaneous Minkowski, 192
Time reversal, 56 metastable, 708
Time-flow vector, 415 minimal-energy, 192
Toda dynamics, 791 topological, 445
Tolman–Komar energy density, 357 true, 302
Topological defects, 158, 205, 216 Varying couplings, 343, 772
Topology of the universe, 43 Varying-speed-of-light (VSL) theories,
change, 569, 583 343
Torsion, 412, 451, 453 Vector
Torus, 43 future-directed, 263
Trace vector, 413 inflation, 221
Tracking solutions, 317, 590, 772 null, 26
Trans-Planckian problem, 183, 219, 771 time-like, 26
Transfer function, 122 Vertex
Transition amplitude. See Path integral of a graph, 556
Transverse diffeomorphisms. See expansion (see Spin foams)
Diffeomorphisms Vielbein, 278, 408
Trapped surface, 271, 293 Vilenkin
Triangulation, 556 tunneling proposal, 437, 482
Trispectrum, 83, 137 wave-function, 475
Index 843
Virasoro Wess–Zumino model, 226, 308

algebra, 633 Weyl map, 592
operators, 632, 644 Weyl symmetry. See Conformal transformation
Virial theorem, 321 Weyl tensor, 337
Visibility function, 97 Wheeler–DeWitt
Voids, 344 equation, 431
Volume expansion, 28 quantum cosmology, 5, 43, 471, 593
Volume form, 410 White noise, 208
Volume operator, 490, 508 Wick rotation, 433, 475
Volume-modulus inflation, 725 Wiener–Khintchine theorem, 83, 102
Vorticity, 28 Wigner symbol, 3j, 136
Wilkinson Microwave Anisotropy Probe
(WMAP), 5, 15, 106, 141
W Wilson lines, 738
Warped D-brane inflation, 702, 744 Winding number, 445. See also String theory
DBI inflation, 702, 754 Window function, 105, 207
slow-roll D-brane inflation, 702, 742, 744 Witt algebra, 630
Warped spacetimes, 670 WKB approximation. See Wentzel–Kramers–
Wave-function of the universe, 431 Brillouin (WKB) approximation
Wave-length, 21 WMAP. See Wilkinson Microwave Anisotropy
Wave-number, 66 Probe (WMAP)
Weak equivalence principle. See Equivalence World-sheet, 448, 627
principle Wronskian, 190
Weak energy condition (WEC). See Energy
condition
Weak gravity conjecture, 718, 740 Y
Weakly interactive massive particles (WIMP), Yang–Mills theory, 415, 446, 483
36 super-, 647, 659, 668, 672, 678, 761
WEC. See Energy condition York–Gibbons–Hawking boundary term, 25,
Wedge product, 410 427
Wentzel–Kramers–Brillouin (WKB)
approximation, 433, 437, 473, 580
adiabatic expansion, 433 Z
states, 434, 444 Zero-point energy, 41, 305, 550
time, 436

Classical and Quantum Cosmology: Gianluca Calcagni

Uploaded by

Copyright:

Available Formats

Classical and Quantum Cosmology: Gianluca Calcagni

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Classical and Quantum Cosmology: Gianluca Calcagni

Uploaded by

Copyright:

Available Formats

Graduate Texts in Physics

Graduate Texts in Physics publishes core learning/teaching material for graduate-

More information about this series at http://www.springer.com/series/8431

Classical and Quantum

ISSN 1868-4513 ISSN 1868-4521 (electronic)

Library of Congress Control Number: 2016950570

© Springer International Publishing Switzerland 2017

Printed on acid-free paper

This Springer imprint is published by Springer Nature

3.1.3 Cosmological Horizons . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 66

4.6.3 Physical Origin . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 138

5.9 Observational Constraints on Inflation . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 213

7.2 Other Versions of the Problem and Strategies . .. . . . . . . . . . . . . . . . . . . . 310

8.4 Problems and Solutions . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 402

11 Cosmology of Quantum Gravities . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 543

12.3 Compactification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 656

13.7 Other Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 761

Nel suo profondo vidi che s’interna

1.1 Micro from Macro

© Springer International Publishing Switzerland 2017 1

Inflation was originally devised for solving a number of problems afflicting

1.2 Outline of the Topics

2 on constrained Hamiltonian systems), [44] (mathematical introduction to

1.3 About Citations

Whenever practical and up to exceptions indicated in the text, an exhaustive list of

function or a number to a number (e.g., 3:14, N 60 e-folds, or (1.2), (1.3)

1.5 Measure Units

A length unit commonly used in cosmology is the megaparsec (Mpc):

1 Mpc D 3:26156  106 light years 3:0857  1022 m :

TPl 1:4  1032 K :

Here we used the conversion formula

where kB 8:6173  105 eV K1 is Boltzmann constant.

These units will be occasionally restored.

1. S. Weinberg, Gravitation and Cosmology (Wiley, New York, 1972)

This to attain, whether heav’n move or earth,

© Springer International Publishing Switzerland 2017 13

2.1 Cosmic Expansion and Cosmological Principle

2.1.1 The Universe at Large Scales

Fig. 2.1 Cosmological scales (Source: adaptation of a figure by C. Schimd)

T0 D .2:7255 ˙ 0:0006/ K : (2.1)

The above ingredients can be formalized in Einstein’s theory of general relativity.

ds2 D g dx dx D dt2 C a2 .t/ ˛ˇ dx

where t D x0 is synchronous (or proper) time, a.t/ is the expansion parameter or

is the line element, in hyperspherical coordinates, of the maximally symmetric .D 

The line element (2.2) becomes

.proper distance/ D a.t/  .comoving distance/ : (2.4)

where  is the wave-length of radiation emitted by a source and 0 is the observed

H0 D 100 h km s1 Mpc1

The PLANCK 2015 value combining information on the temperature spectrum, on

h D 0:678 ˙ 0:009 .68 % CL/ ; (2.12)

t0 ' H01 D 9:786h1 Gyr ; (2.13)

t0 D 13:799 ˙ 0:038 Gyr .68 % CL/ : (2.14)

2.2 Einstein and Continuity Equations

The Ansatz of general relativity is the Einstein–Hilbert gravitational action

where g :D det.g / is the determinant of the dimensionless metric g ,  2 D

In a spacetime with D D 2 dimensions, the Newton constant is dimensionless,

G C g D  2 T ; (2.23)

Taking the trace of (2.23) gives

x0 D x C ıx ; ıx D a ; (2.28)

1 Mpc D 3:26156 106 light years 3:0857 1022 m :

TPl 1:4 1032 K :

where kB 8:6173 105 eV K1 is Boltzmann constant.

ds2 D g dx dx D dt2 C a2 .t/ ˛ˇ dx

is the line element, in hyperspherical coordinates, of the maximally symmetric .D

.proper distance/ D a.t/ .comoving distance/ : (2.4)

where is the wave-length of radiation emitted by a source and 0 is the observed

where g :D det.g / is the determinant of the dimensionless metric g , 2 D

G C g D 2 T ; (2.23)

x0 D x C ıx ; ıx D a ; (2.28)

ıg D g @ a C g @ a C a @ g ; (2.29)

T n n > 0 8 n null : (2.33)

R n n > 0 8 n null : (2.34)

T t t > 0 8 t time-like : (2.35)

T t D Au C Bn ; (2.37)

R t t > 0 8 t time-like : (2.38)

T D . C P/ u u C P g ; (2.40)

h D g C u u ; h D h h ; u h D 0 ; (2.41)

! :D rŒ u C uP Œ u D h Œ r u (2.45)

u PP C .Pu C u /. C P/ C r P D 0 : (2.52)

NEC: .1 C w/ > 0 ; (2.54)

WEC: > 0; w > 1 ; (2.55)

In D D 4, (2.58) reads 1=3 6 w 6 1.

R00 D .D 1/.H 2 C H/P D .D 1/ aR ; (2.70)

r D r0 .1 C z/4 ; (2.87)

m D m0 .1 C z/3 : (2.88)

˝r0 h2 D 2:469 105 ; (2.96)

G D 2 hT i ; (2.112)

hT i D vac g : (2.113)

This is a cosmological-constant contribution vac D = 2 D and the “vacuum

eq 2:4 10113 m4Pl ; (2.122)