Marshall

Complex Analysis
This user-friendly textbook introduces complex analysis at the beginning graduate or

advanced undergraduate level. Unlike other textbooks, it follows Weierstrass’s approach,
stressing the importance of power series expansions instead of starting with the Cauchy integ-
ral formula, an approach that illuminates many important concepts. This view allows readers
to quickly obtain and understand many fundamental results of complex analysis, such as the
maximum principle, Liouville’s theorem and Schwarz’s lemma.
The book covers all the essential material on complex analysis, and includes several elegant
proofs that were recently discovered. It includes the zipper algorithm for computing conformal
maps, a constructive proof of the Riemann mapping theorem, and culminates in a complete
proof of the uniformization theorem. Aimed at students with some undergraduate background
in real analysis, though not Lebesgue integration, this classroom-tested textbook will teach
the skills and intuition necessary to understand this important area of mathematics.
Donald E. Marshall is Professor of Mathematics at the University of Washington. He received

his PhD from UCLA in 1976. Professor Marshall is a leading complex analyst with a very
strong research record that has been continuously funded throughout his career. He has given
invited lectures in over a dozen countries. He is coauthor of the research-level monograph
Harmonic Measure, published by Cambridge University Press.
CAMBRIDGE MATHEMATICAL TEXTBOOKS
Cambridge Mathematical Textbooks is a program of undergraduate and beginning graduate-level text-

books for core courses, new courses, and interdisciplinary courses in pure and applied mathematics.
These texts provide motivation with plenty of exercises of varying difficulty, interesting examples,
modern applications, and unique approaches to the material.
Advisory Board
John B. Conway, George Washington University

Gregory F. Lawler, University of Chicago
John M. Lee, University of Washington
John Meier, Lafayette College
Lawrence C. Washington, University of Maryland, College Park
A complete list of books in the series can be found at www.cambridge.org/mathematics
Recent titles include the following:
Chance, Strategy, and Choice: An Introduction to the Mathematics of Games and Elections, S. B. Smith
Set Theory: A First Course, D. W. Cunningham
Chaotic Dynamics: Fractals, Tilings, and Substitutions, G. R. Goodson
A Second Course in Linear Algebra, S. R. Garcia & R. A. Horn
Introduction to Experimental Mathematics, S. Eilers & R. Johansen
Exploring Mathematics: An Engaging Introduction to Proof, J. Meier & D. Smith
A First Course in Analysis, J. B. Conway
Introduction to Probability, D. F. Anderson, T. Seppäläinen & B. Valkó
Linear Algebra, E. S. Meckes & M. W. Meckes
A Short Course in Differential Topology, B. I. Dundas
Abstract Algebra with Applications, A. Terras
Complex Analysis
Donald E. Marshall
University of Washington, Seattle, WA, USA
University Printing House, Cambridge CB2 8BS, United Kingdom
One Liberty Plaza, 20th Floor, New York, NY 10006, USA
477 Williamstown Road, Port Melbourne, VIC 3207, Australia
314–321, 3rd Floor, Plot 3, Splendor Forum, Jasola District Centre, New Delhi – 110025, India
79 Anson Road, #06–04/06, Singapore 079906
Cambridge University Press is part of the University of Cambridge.

It furthers the University’s mission by disseminating knowledge in the pursuit of
education, learning, and research at the highest international levels of excellence.
www.cambridge.org
Information on this title: www.cambridge.org/9781107134829
© Donald E. Marshall 2019
This publication is in copyright. Subject to statutory exception
and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without the written
permission of Cambridge University Press.
First published 2019
Printed in the United States of America by Sheridan Books, Inc.
A catalog record for this publication is available from the British Library.
Library of Congress Cataloging-in-Publication Data
Names: Marshall, Donald E. (Donald Eddy), 1947– author.
Title: Complex analysis / Donald E. Marshall.
Description: Cambridge, United Kingdom ; New York, NY :
Cambridge University Press, 2019.
Identifiers: LCCN 2018029851 | ISBN 9781107134829 (Hardback)
Subjects: LCSH: Functions of complex variables – Textbooks.
| Mathematical analysis – Textbooks.
Classification: LCC QA331.7 M365 2019 | DDC 515/.9–dc23
LC record available at https://lccn.loc.gov/2018029851
ISBN 978-1-107-13482-9 Hardback
Cambridge University Press has no responsibility for the persistence or accuracy of
URLs for external or third-party internet websites referred to in this publication
and does not guarantee that any content on such websites is, or will remain,
accurate or appropriate.
Contents
List of Figures page viii

Preface xi
Prerequisites xv
PART I
1 Preliminaries 3
1.1 Complex Numbers 3
1.2 Estimates 6
1.3 Stereographic Projection 8
1.4 Exercises 10
2 Analytic Functions 13
2.1 Polynomials 13
2.2 Fundamental Theorem of Algebra and Partial Fractions 15
2.3 Power Series 17
2.4 Analytic Functions 20
2.5 Elementary Operations 23
2.6 Exercises 27
3 The Maximum Principle 31

3.1 The Maximum Principle 31
3.2 Local Behavior 33
3.3 Growth on C and D 35
3.4 Boundary Behavior 38
3.5 Exercises 41
4 Integration and Approximation 43

4.1 Integration on Curves 43
4.2 Equivalence of Analytic and Holomorphic 47
4.3 Approximation by Rational Functions 53
4.4 Exercises 60
5 Cauchy’s Theorem 63
5.1 Cauchy’s Theorem 63
5.2 Winding Number 64
5.3 Removable Singularities 70
5.4 Laurent Series 72
vi Contents
5.5 The Argument Principle 76

5.6 Exercises 78
6 Elementary Maps 81
6.1 Linear Fractional Transformations 82
6.2 Exp and Log 86
6.3 Power Maps 87
6.4 The Joukovski Map 89
6.5 Trigonometric Functions 91
6.6 Constructing Conformal Maps 93
6.7 Exercises 98
PART II
7 Harmonic Functions 105

7.1 The Mean-Value Property and the Maximum Principle 105
7.2 Cauchy–Riemann and Laplace Equations 111
7.3 Hadamard, Lindelöf and Harnack 114
7.4 Exercises 118
8 Conformal Maps and Harmonic Functions 123

8.1 The Geodesic Zipper Algorithm 123
8.2 The Riemann Mapping Theorem 129
8.3 Symmetry and Conformal Maps 132
8.4 Conformal Maps to Polygonal Regions 135
8.5 Exercises 137
9 Calculus of Residues 141

9.1 Contour Integration and Residues 141
9.2 Some Examples 142
9.3 Fourier and Mellin Transforms 145
9.4 Series via Residues 149
9.5 Laplace and Inverse Laplace Transforms 150
9.6 Exercises 154
10 Normal Families 156

10.1 Normality and Equicontinuity 156
10.2 Riemann Mapping Theorem Revisited 162
10.3 Zalcman, Montel and Picard 164
10.4 Exercises 168
11 Series and Products 170

11.1 Mittag-Leffler’s Theorem 170
11.2 Weierstrass Products 175
11.3 Blaschke Products 181
Contents vii
11.4 The Gamma and Zeta Functions 184

11.5 Exercises 189
PART III
12 Conformal Maps to Jordan Regions 195

12.1 Some Badly Behaved Regions 195
12.2 Janiszewski’s Lemma 197
12.3 Jordan Curve Theorem 199
12.4 Carathéodory’s Theorem 201
12.5 Exercises 205
13 The Dirichlet Problem 207

13.1 Perron Process 207
13.2 Local Barriers 209
13.3 Riemann Mapping Theorem Again 212
13.4 Exercises 214
14 Riemann Surfaces 216

14.1 Analytic Continuation and Monodromy 216
14.2 Riemann Surfaces and Universal Covers 220
14.3 Deck Transformations 226
14.4 Exercises 227
15 The Uniformization Theorem 230

15.1 The Modular Function 230
15.2 Green’s Function 232
15.3 Simply-Connected Riemann Surfaces 237
15.4 Classification of All Riemann Surfaces 243
15.5 Exercises 243
16 Meromorphic Functions on a Riemann Surface 246

16.1 Existence of Meromorphic Functions 246
16.2 Properly Discontinuous Groups on C∗ and C 246
16.3 Elliptic Functions 248
16.4 Fuchsian Groups 251
16.5 Blaschke Products and Convergence Type 253
16.6 Exercises 257
Appendix 260
A.1 Fifteen Conditions Equivalent to Analytic 260
A.2 Program for Color Pictures 261
Bibliography 267
Index 269
Figures
1.1 Cartesian and polar representations of complex numbers. page 3

1.2 Addition. 4
1.3 Multiplication. 4
1.4 Triangle inequality. 6
1.5 Stereographic projection. 9
2.1 The power map. 14
2.2 p(z0 + ζ ) lies in a small disk of radius s = C |ζ |k+ 1 < r = |bk ||ζ | k . 14
2.3 Convergence of a power series. 19
2.4 Proof of Theorem 2.7. 21
3.1 Conformality. 34
3.2 Local behavior of an analytic function. 34
3.3 Stolz angle, ±α (ζ ). 38
4.1 A curve, γ . 43
4.2 A closed curve, γ = γ1 + γ2. 45
4.3 Integrals around squares. 45
4.4 Proof of Morera’s theorem. 52
4.5 If a is outside S. 53
4.6 The square S and its circumscribed circle C. 53
4.7 A union of closed squares covering K and contained in U. 55
5.1 Integration on a cycle. 64
5.2 Calculating the winding number. 66
5.3 Parity: which red point is not in the same component as the other two? 67
5.4 n(σ , 0) = n(σ , 1) = 0. 68
5.5 Proof of Rouché’s theorem. 77
6.1 A rational function. 81
6.2 The Cayley transform. 85
6.3 Maps ez and log z. 86
6.4 Maps z4 and z1/4 . 87
6.5 The map z1+i and its inverse z(1−i)/2 . 88
6.6 The map (z + 1/z)/2. 89
6.7 Some inverses of (z + 1/ z)/2. 91
6.8 The map cos(z). 92
6.9 Four intersections of disks meeting at angle π/ 4. 93
6.10 Mapping a circularly slit half-plane. 95
6.11 Below a parabola. 95
6.12 Exterior of an ellipse. 96
List of Figures ix
6.13 Region between the branches of a hyperbola. 96

6.14 “World’s greatest function,” exp ((z + 1)/(z − 1)). 98
6.15 Tangent circles. 101
7.1 The angle θ (z) as a harmonic function. 115
8.1 Construction of γ4 and ϕ4 . 124
8.2 Proof of Jørgensen’s theorem. 126
8.3 A disk-chain. 127
8.4 A smaller disk ² ⊂ Dj . 128
8.5 Conformal mapping using the geodesic zipper algorithm. 129
8.6 Turn angles at the vertices of a polygon. 135
9.1 Half-disk contour. 144
9.2 Rectangle contour. 145
9.3 Contour avoiding poles on R. 147
9.4 Keyhole contour. 148
9.5 Enlarged half-disk. 153
ζ ( 2 + iy), −40 ≤ y ≤ 40.
1
11.1 187
12.1 E2 and connecting segments. 196
12.2 Constructing a dense set of spirals. 196
12.3 Constructing the Lakes of Wada. 197
12.4 Janiszewksi’s lemma. 199
12.5 Boundary of a component. 200
12.6 A modification of J. 200
12.7 Crosscuts γδn and ϕ (γδn ). 202
12.8 Proof that ϕ is one-to-one. 203
13.1 Proof of Bouligand’s lemma. 211
14.1 Homotopy of γ0 to γ1. 218
14.2 Parking-lot ramp surface. 220
15.1 Reflections of a circular triangle T . 231
15.2 Harmonic measure of ∂ rU in W \ rU. 234
15.3 The surface Wt . 241
16.1 On C σ , ρ (z, 0) = ρ (z, σ (0)). 252
16.2 Normal fundamental domain. 253
Preface
This book provides a graduate-level introduction to complex analysis. There are four points
of view for this subject due primarily to Cauchy, Weierstrass, Riemann and Runge. Cauchy
thought of analytic functions in terms of a complex derivative and through his famous integral
formula. Weierstrass instead stressed the importance of power series expansions. Riemann
viewed analytic functions as locally rigid mappings from one region to another, a more geo-
metric point of view. Runge showed that analytic functions are nothing more than limits of
rational functions. The seminal modern text in this area was written by Ahlfors [1], which
stresses Cauchy’s point of view. Most subsequent texts have followed his lead. One aspect
of the first-year course in complex analysis is that the material has been around so long that
some very slick and elegant proofs have been discovered. The subject is quite beautiful as a
result, but some theorems then may seem mysterious.
I have decided instead to start with Weierstrass’s point of view for local behavior. Cartan [4]
has a similar approach. Power series are elementary and give you many non-trivial functions
immediately. In many cases it is a lot easier to see why certain theorems are true from this
point of view. For example, it is remarkable that a function which has a complex derivative
actually has derivatives of all orders. However, the derivative of a power series is just another
power series and hence has derivatives of all orders.
Cauchy’s theorem is a more global result concerned with integrals of analytic functions.
±
Why integrals of the form z −1 a dz are important in Cauchy’s theorem is very easy to under-
stand using partial fractions for rational functions. So we will use Runge’s point of view for
more global results: analytic functions are simply limits of rational functions.
As a pedagogic device we will use the term “analytic” for local power series expansion
and “holomorphic” for possessing a continuous complex derivative. We will of course prove
that these concepts (and several others) are equivalent eventually, but in the early chapters the
reader should be alert to the different definitions.
The emphasis in Chapters 1–6 is to view analytic functions as behaving like polynomials
or rational functions. Perhaps the most important elementary tool in this subject is the maxi-
mum principle, highlighted in Chapter 3. Runge’s theorem is proved in Chapter 4 and is used
to prove Cauchy’s theorem in Chapter 5. Chapter 6 uses color to visualize complex-valued
functions. Given a coloring of the complex plane, a function f can be illustrated by placing
the color of f (z) at the point z. See Section A.2 of the appendix for a computer program to
do this.
Chapters 7 and 8 introduce harmonic and subharmonic functions and highlight their appli-
cation to the study of analytic functions. Chapter 8 includes a method, called the geodesic
zipper algorithm, for numerically computing conformal maps, which is fast and simple to
program. Together with Harnack’s principle, it is used to give a somewhat constructive proof
xii Preface
of the Riemann mapping theorem in Chapter 8. Because it does not require the development of
normal families, it is possible to give a one-quarter course that includes this proof of the Rie-
mann mapping theorem. The standard proof based on normal families is given in Chapter 10.
In Chapter 10 we also give Zalcman’s remarkable characterization of non-normal families
in terms of an associated convergent sequence, then use it to prove Montel’s theorem and
Picard’s great theorem.
Complete and accessible proofs of Carathéodory’s theorem and the Jordan curve theorem
are included in Chapter 12. Local barriers instead of barriers are used to analyze regular points
for the Dirichlet problem in Chapter 13, so that it is easier to verify that every boundary point
of a simply-connected region is regular. This allows us to give another proof of the Riemann
mapping theorem. The uniformization theorem and the classification of all Riemann surfaces
in Chapters 14–16 tie together complex functions, topology, manifolds and groups. The proof
of the uniformization theorem here uses Green’s function, when available, and the dipole
Green’s function otherwise. This yields a very similar treatment of the two cases. The main
tool is simply the maximum principle, which allows a proof that avoids the “oil speck” method
of exhaustion by relatively compact surfaces, and avoids the need to prove triangulation or
Green’s theorem on Riemann surfaces. Another benefit of this approach is that it is then easy
to construct plenty of meromorphic functions on any Riemann surface in Chapter 16. The first
section in the appendix lists 15 ways developed in the text to determine whether a function is
analytic.
Each of the three parts of this book can be comfortably covered in a one-quarter course. A
one-semester course might include most of the material in Chapters 1–9. A list of prerequi-
sites follows this preface. Students should be encouraged to review this material as needed,
especially if they encounter difficulties in Chapter 1. Lebesgue integration is not needed in
this text because, by Theorem 4.32, we can integrate an analytic function on any continu-
ous curve using Riemann integration. The exercises at the end of each chapter are divided
by difficulty, though in some cases they can be solved in more than one way. Exercises A
are mostly straightforward, requiring little originality, and are designed for practice with the
material. The B exercises require a good idea or non-routine use of the results in the chapter.
Sometimes a creative idea or the right insight can lead to a simple solution. The C exercises
are usually much more difficult. You can think of “C” as “challenge.” I generally ask students
to do the A exercises while reading, but focus on the B exercises for homework. Class dis-
cussions are facilitated by asking the students to read as much as they can before we discuss
the material. Most of the B exercises come from the PhD qualifying exams in complex analy-
sis at the University of Washington [18]. It is entirely possible to find solutions to problems
by searching the internet. It is also possible to solve some problems using more advanced
techniques or theorems than have been covered in the text. Both will defeat the purpose of
developing the ability to solve problems, a goal of this book. For that reason, we also ask you
not to tempt others by posting solutions.
This book is not written as a novel that can be read passively. Active involvement will
increase your understanding as you read this material. You should have plenty of scratch
paper at hand so that you can check all details. The ideas in a proof are at least as important
as the statement of the corresponding theorem, if not more so. But the ideas are mean-
ingful only if you can fill in all the details. View this as practice for proving your own
theorems.
Preface xiii
I am grateful to many people for their assistance in preparing this text. First are all of
the students in my complex analysis classes who have pointed out errors, omissions and less
than stellar explanations. But I would particularly like to thank John Garnett, Pietro Poggi-
Corradini and Steffen Rohde, who have used the material in their classes and made numerous
excellent suggestions for improvement. Similar thanks go to Robert Burckel, who read the
first thirteen chapters with the eye of an eagle and a fine-toothed comb. I owe a great deal to
all my teachers, coauthors and the books I have read for the mathematics they have taught
me. Hopefully, some of the elegance, beauty and technique have been retained here. Several
excellent texts on this subject are also listed in the bibliography. As with any mathematics
text, errors still no doubt remain. I would appreciate receiving email at dmarshal@uw.edu
about any errors you encounter. I will list corrections on the web page:
www.cambridge.org/marshall
Wov’n through carefully chosen words

Lie the strands of key support
By my truelove Marianne
Mainstay of our life journey
Fifty years with more to come.
Prerequisites
You should be on friendly terms with the following concepts. If you have only seen the cor-
responding proofs for real numbers and real-valued functions, check to see whether the same
proofs also work when “real” is replaced by “complex,” after reading the first two sections of
Chapter 1. As you read the text, check all the details. If many of the concepts below are new
to you, then I would recommend that you first take a senior-level analysis class.
Let {an}∞ ∞
n=0, {b n} n=0 be sequences of real numbers and let { f n} be a sequence of real-
valued, continuous functions defined on some interval I ⊂ R.
1. {an } converges to a (notation: an → a) “± − δ ” version.
2. Cauchy sequence.
∑ ∑
3. a converges, converges absolutely (notation: |an| < ∞ ).
∑ n
4. an converges implies an → 0, but not conversely;
5. lim supn→∞ an , lim infn→∞ an.
6. Comparison test for convergence.
7. Rearranging absolutely convergent series gives the same sum, but a similar statement
does not hold for for conditionally convergent series.
∑ ∑∞
8. If ∞ n=0 an = A and n=0 b n = B then
∞
± ∞
±
A+ B = (an + bn ) and cA = ca n.
n=0 n=0
∑ ∑n
If an converges absolutely and cn = k=0 a k b n−k then
∞
±
AB = cn.
n=0
9.
∞±
± ∞ ∞±
± ∞
an,k = an,k
n=0 k=0 k=0 n=0
provided at least one sum converges absolutely. Absolute convergence of either of these
double sums is equivalent to the finiteness of
±
sup |an,k |.
S finite n,k ∈S
10. Continuous function, uniformly continuous function.

11. fn (x) → f (x) pointwise, fn (x) → f (x) uniformly.
12. Uniform limit of a sequence of continuous functions is continuous.
xvi Prerequisites
²³ ² ³
13. ² I f (x)dx ² ≤ I | f (x)|dx.
14. If f n → f uniformly on a bounded interval I then
´ ´ ´
lim fn (x)dx = lim fn (x)dx = f (x)dx.
I I I
15. Corollary:
∞´
± ∞
´ ±
fn (x)dx = f n(x)dx,
n=0 n=0
I I
∑
if the partial sums of fn converge uniformly on the bounded interval I.
16. Open set, closed set, connected set, compact set, metric space.
17. f continuous on a compact set X implies f is uniformly continuous on X.
18. X ⊂ Rn is compact if and only if it is closed and bounded.
19. A metric space X is compact if and only if every infinite sequence in X has a limit (cluster)
point in X. (This can fail if X is not a metric space.)
20. If f is continuous on a connected set U then f (U) is connected. If f is continuous on a
compact set K then f (K) is compact.
21. A continuous real-valued function on a compact set has a maximum and a minimum.
All of the above can be found in the undergraduate text Rudin [22], as well as many other
sources.
PART I
1 Preliminaries
1.1 Complex Numbers
The complex numbersC consist of pairs of real numbers: {(x, y) : x, y ∈ R}. The complex
number (x, y) can be represented geometrically as a point in the plane R2, or viewed as a vector
whose tip has coordinates (x, y) and whose tail has coordinates (0, 0). The complex number
(x, y) can be identified with another pair of real numbers (r, θ ), called the polar coordinate
representation. The line from (0, 0) to (x, y) has length r and forms an angle θ with the positive
x axis. The angle is measured by using the distance along the corresponding arc of the circle
of radius 1 (centered at (0, 0)). By similarity, the length of the subtended arc on the circle of
radius r is rθ . See Figure 1.1.
Conversion between these two representations is given by
x = r cos θ , y = r sin θ
and
±
r = x2 + y2 , tan θ = xy .
Care must be taken to find θ from the last equality since many angles can have the same tan-
gent. However, consideration of the quadrant containing (x, y) will give a unique θ ∈ [0, 2π ),
provided r > 0 (we do not define θ when r = 0).
Addition of complex numbers is defined coordinatewise:
(a, b) + (c, d) = (a + c, b + d),

and can be visualized by vector addition. See Figure 1.2.
Multiplication is given by
(a, b) · (c, d) = (ac − bd, bc + ad)

and can be visualized as follows. The points (0, 0), (1, 0), (a, b) form a triangle. Construct a
similar triangle with corresponding points (0, 0), (c, d), (x, y). Then it is an exercise in high-
(x, y)
y θ
r rθ
x 1
Figure 1.1 Cartesian and polar representations of complex numbers.

4 Preliminaries
( a + c, b + d)
(a, b)
( c, d)
Figure 1.2 Addition.
(a, b) · (c, d)
( c, d)
( a, b)
(0 , 0) (1 , 0)
Figure 1.3 Multiplication.
school geometry to show that (x, y) = (a, b) · (c, d). By similarity, the length of the product is
the product of the lengths and polar coordinate angles are added. See Figure 1.3.
The real number t is identified with the complex number (t, 0). With this identification,
complex addition and multiplication are extensions of the usual addition and multiplication of
real numbers. For conciseness, when t is real, t(x, y) means (t, 0) · (x, y) = (tx, ty). The additive
identity is 0 = (0, 0) and −(x, y) = (−x, −y). The multiplicative identity is 1 = (1, 0) and
the multiplicative inverse of (x, y) is (x/(x2 + y2), −y/ (x2 + y2 )). It is a tedious exercise to
check that the commutative and associative laws of addition and multiplication hold, as does
the distributive law.
The notation for complex numbers becomes much easier if we use a single letter instead
of a pair. It is traditional, at least among mathematicians, to use the letter i to denote the
complex number (0, 1). If z is the complex number given by (x, y), then, because (x, y) =
x(1, 0) + y(0, 1), we can write z = x + yi. If z = x + iy, then the real part of z is
Rez = x and the imaginary partis Imz = y. Note that i · i = −1. We can now just use
the usual algebraic rules for manipulating complex numbers together with the simplification
i2 = − 1. For example, z/w means multiplication of z by the multiplicative inverse of w.
To find the real and imaginary parts of the quotient, we use the analog of “rationalizing the
denominator”:
x + iy
a + ib
= (a(x ++ iy)(a − ib) = xa − i2 yb + iya − ixb
ib)(a − ib) a2 + b2
= xa + yb + ya − xb i.
a + b2
2 a2 + b2
Here is some additional notation: if z = x + iy is given in polar coordinates by the pair (r, θ )
then
±
|z| = r = x2 + y2
is called the modulus or absolute valueof z. Note that |z| is the distance from the complex
number z to the origin 0. The angle θ is called the argument of z and is written
θ = arg z.
1.1 Complex Numbers 5
The most common convention is that −π < arg z ≤ π , where positive angles are measured
counter-clockwise and negative angles are measured clockwise. The complex conjugate of z
is given by
z = x − iy.
The complex conjugate is the reflection of z about the real lineR.
It is an easy exercise to show the following:
|zw| = |z||w|,
|cz| = c|z| if c > 0,
z/| z| has absolute value 1,
zz = | z|2,
Rez = (z + z)/2,
Imz = (z − z)/(2i),
z + w = z + w,
zw = z · w,
z = z,
|z| = |z|,
arg zw = arg z + arg w modulo 2π ,
arg z = − arg z = 2π − arg z modulo 2π .
The statement modulo 2π means that the difference between the left- and right-hand sides of
the equality is an integer multiple of 2π .
The identity a + (z − a) = z expressed in vector form shows that z − a is (a translate of) the
vector from a to z. Thus |z − a| is the length of the complex number z − a but it is also equal
to the distance from a to z. The circle centered at a with radius r is given by {z : |z − a| = r}
and the disk centered at a of radius r is given by {z : |z − a| < r}. The open disks are the basic
open sets generating the standard topology on C. We will use D to denote the unit disk,
D = { z : | z| < 1 } ,
and use ∂ D to denote the unit circle,
∂ D = { z : | z| = 1 } .
Complex numbers were around for at least 250 years before good applications were found;
Cardano discussed them in his book Ars Magna (1545). Beginning in the 1800s, and continu-
ing today, there has been an explosive growth in their usage. Now complex numbers are very
important in the application of mathematics to engineering and physics.
It is a historical fiction that solutions to quadratic equations forced us to take complex
numbers seriously. How to solve x2 = mx + c has been known for 2000 years and can be
visualized as the points of intersection of the standard parabola y = x2 and the line y = mx+ c.
As the line is shifted up or down by changing c, it is easy to see there are two, one or no
(real) solutions. The solution to the cubic equation is where complex numbers really became
important. A cubic equation can be put in the standard form
6 Preliminaries
x3 = 3px + 2q
by scaling and translating. The solutions can be visualized as the intersection of the standard
cubic y = x3 and the line y = 3px + 2q. Every line meets the cubic, so there will always be a
solution. By formal manipulations, Cardano showed that a solution is given by
± ±
x = (q + + (q − q2 − p3 ) .
q 2 − p 3) 3
1 1
3
Bombelli pointed out 30 years later that if p = 5 and q = 2, then x = 4 is a solution, but
q2 −p3 < 0 so the above solution does not make sense. His “wild thought” was to use complex
numbers to understand the solution
x = (2 + 11i) 3 + (2 − 11i)
1 1
3 .
He found that (2 ± i)3 = 2 ± 11i, and so the above solution actually equals 4. In other words,
complex numbers were used to find a real solution. This is not just an oddity of Cardano’s
formula, because, for some cubics, complex numbers must be used in any rational formula
involving radicals by a theorem of O. Hölder [15]. See Exercises 1.9 and 1.10 for solutions of
cubic and quartic equations.
1.2 Estimates
Here are some elementary estimates which the reader should check:
− |z| ≤ Rez ≤ |z|,
− |z| ≤ Imz ≤ |z|
and
|z| ≤ |Rez| + |Imz|.
Perhaps the most useful inequality in analysis is the triangle inequality.
Theorem 1.1 (triangle inequality)

|z + w| ≤ |z| + |w|
and
|z + w| ≥ ||z| − |w|| .
The associated picture perhaps makes this result geometrically clear. See Figure 1.4. Analy-
sis is used to give a more rigorous proof of the triangle inequality (and it is good practice with
the notation we have introduced).
z+w w
z
Figure 1.4 Triangle inequality.

1.2 Estimates 7
Proof
|z + w|2 = (z + w)(z + w)
= zz + wz + zw + ww
= |z|2 + 2Re(wz) + |w|2
≤ |z|2 + 2|w||z| + |w|2
= ( |z| + |w|)2.
To obtain the second part of the triangle inequality we use
|z| = |z + w + (−w) | ≤ |z + w| + | − w| = |z + w| + |w|.

By subtracting | w|,
|z| − |w| ≤ |z + w|,
and switching z and w,
|w| − |z| ≤ |z + w|,
so that
||z| − |w|| ≤ |z + w|.
These estimates can be used to prove that {zn} converges if and only if both {Rezn } and
∑
{Imzn} converge. The series an is said to converge if the sequence of partial sums
m
²
Sm = an
n=1
∑
converges, and the series converges absolutelyif | an | converges. A series is said to diverge
if it does not converge. Absolute convergence implies convergence because Cauchy sequences
∑
converge. We sometimes write |an| < ∞ to∑denote absolute convergence because the
partial sums are increasing. It also follows that an is absolutely convergent if and only if
∑ ∑
both Rean and Iman are absolutely convergent. By comparing the nth partial sum and
∑
the (n − 1)st partial sum, if an converges then an → 0. The converse statement is false, for
example if an = 1/n.
Another useful estimate is the Cauchy–Schwarz inequality.
Theorem 1.2 (Cauchy–Schwarz inequality)

³ ³ ⎛ ⎞ 21 ⎛ ⎞ 21
³² ³
³ n ³ ² n ² n
³ aj bj ³ ≤ ⎝ |aj |2 ⎠ ⎝ | bj | 2⎠ .
³ ³
³ j=1 ³ j =1 j=1
If v and w are vectors in Cn , the Cauchy–Schwarz inequality says that |²v, w³| ≤ || v||||w||,
where the left-hand side is the absolute value of the inner product and the right-hand side is
the product of the lengths of the vectors.
8 Preliminaries
Proof The square of the right-hand side minus the square of the left-hand side in the
Cauchy–Schwarz inequality can be written as
n ´
n ²
² µ
|aj |2 |bi |2 − aj bj ai bi .
i=1 j=1
We can add another copy of this quantity, switching the index i and the index j to obtain
1 ²²´ 2 2 µ
n n
= 2
|aj | |bi | + |ai | 2| bj |2 − aj bj ai bi − ai bi aj bj .
i =1 j =1
Using the identity | A − B |2 = |A|2 + |B|2 − AB − AB, with A = aj bi and B = ai bj , we obtain

n ²
² n
= 21 |aj bi − ai bj |2.
i=1 j=1
The above proof also gives the error

n n
1 ²²
2
|aj bi − ai bj |2,
j =1 i=1
and so equality occurs if and only if aj = cbj for all j and some (complex) constant c, or
bj = 0 for all j.
The reader can use Riemann integration to deduce the following, which is also called the
Cauchy–Schwarz inequality. For a complex-valued function f defined on a real interval [a, b],
¶b ¶b ¶b
we define a f dx ≡ a Ref dx + i a Imf dx.
Corollary 1.3 If f and g are continuous complex-valued functions defined on [a, b] ⊂R

then
³· b ³ ¸· b ¹ 12 ¸· b ¹ 21
³ ³
³ a f (t)g(t)dt³≤ a | f (t)| dt |g(t)| dt .
³ ³ 2 2
a
This corollary can also be proved directly by expanding

· b· b
| f (x)g(y) − f (y)g(x)|2 dxdy (1.1)
a a
in a similar way, giving a proof for square integrable functions f , g. Moreover, the error term
is half of the integral (1.1) and equality occurs if and only if f = cg, for some constant c, or g
is identically zero.
1.3 Stereographic Projection
A component of Riemann’s point of view of functions as mappings is that ∞ is like any other
complex number. But we cannot extend the definition of complex numbers to include ∞ and
still have the usual laws of arithmetic hold. However, there is another “picture” of complex
1.3 Stereographic Projection 9
N =(0, 0,1)
(x1,x2,x3)
( x,y,0)
Figure 1.5 Stereographic projection.
numbers that can help us visualize this idea. The picture is called stereographic projection.
We identify the complex numbers with the plane {(x, y, 0) : x, y ∈ R} in R3 . If z = x + iy, let
z∗ be the unique point on the unit sphere in R3 which also lies on the line from the north pole
(0, 0, 1) to (x, y, 0). Thus
z∗ = (x1, x2, x3 ) = (0, 0, 1) + t[(x, y, 0) − (0, 0, 1)].

See Figure 1.5.
Then
±
|z∗ | = (tx)2 + (ty)2 + (1 − t)2 = 1,
which gives
2
t= ,
x2 + y2 + 1
where 0 < t ≤ 2, and
¸ ¹
2x 2y x2 + y2 − 1
z∗ = , , .
x2 + y2 + 1 x2 + y2 + 1 x2 + y2 + 1
The reader is invited to find z = x + iy from z∗ = (x1, x2 , x3 ). The sphere is sometimes called
the Riemann sphereand is denoted S2 . We can extend stereographic projection π : C → S2
to the extended planeC∗ = C ∪{∞} by defining π (∞) = (0, 0, 1). We give C∗ the topology
inherited from S 2. It is an explicit one-point compactification of the complex plane. The north
pole corresponds to “∞ .”
Theorem 1.4 Under stereographic projection, circles and straight lines in C correspond
precisely to circles on S 2.
Proof Every circle on the sphere is given by the intersection of a plane with the sphere, and
conversely the intersection of a plane with a sphere is a circle or a point. See Exercise 1.6. If
a plane is given by
Ax1 + Bx2 + Cx3 = D,
and if (x1 , x2, x3 ) corresponds to (x, y, 0) under stereographic projection, then
¸ ¹ ¸ ¹ ¸ ¹
+ C xx2 ++ yy2 +− 11
2 2
2x
A 2
x + y2 + 1
+ B x2 +2yy2 + 1 = D. (1.2)
10 Preliminaries
Equivalently,
(C − D)(x2 + y2) + 2Ax + 2By = C + D. (1.3)
If C = D, then this is the equation of a line, and all lines can be written this way. If C ´ = D,
then, by completing the square, we get the equation of a circle, and all circles can be put in
this form.
So we will consider a line in C as just a special kind of “circle.”

The sphere S2 inherits a topology from the usual topology on R3 generated by the balls
in R3 .
Corollary 1.5 The topology on S2 induces the standard topology on C via stereographic
projection, and moreover a basic neighborhood of ∞ is of the form {z : |z| > r}.
For later use, we note that the chordal distance between two points on the sphere induces a
metric, called the chordal metric, on C which is given by
2| z − w|
χ (z, w) = |z∗ − w∗ | = º º . (1.4)
1 + |z| 2 1 + |w |2
This metric is bounded (by 2). See Exercise 1.5.
1.4 Exercises
1.1 Check that item 9 of the prerequisites holds for complex an,k . Check that items 13, 14
and 15 of the prerequisites hold for complex-valued functions defined on an interval
I ⊂ R.
1.2 Check the details of the high-school geometry problem in the geometric version of
complex multiplication.
1.3 Prove the parallelogram equality:
|z + w|2 + |z − w|2 = 2(|z|2 + |w|2).
In geometric terms, the equality says that the sum of the squares of the lengths of the
diagonals of a parallelogram equals the sum of the squares of the lengths of the sides.
It is perhaps a bit easier to prove it using the complex notation of this chapter than to
prove it using high-school geometry.
1.4 Prove Corollary 1.3.
1.5 (a) Prove formula (1.4). An algebraic proof can be found in [1], p. 20. Alternatively,
use the law of cosines for the triangles with vertices N = (0, 0, 1), z, w and N, z∗,
w∗. Compute edge lengths of these two triangles using triangles that have N and
(0, 0, 0) as vertices.
(b) The chordal distance is bounded by 2, by the triangle inequality. Verify analytic-
ally that the formula for this distance given in the text is bounded by 2 using the
1.4 Exercises 11
Cauchy–Schwarz inequality, and also by directly multiplying it out using complex

notation and one of the estimates at the start of Section 1.2.
1.6 Show that the intersection of a plane with the unit sphere in R3 is a circle or a point
and conversely that every circle or point on the sphere is equal to the intersection of the
sphere with a plane. Hint: Rotate the plane and sphere so that the plane is parallel to the
(x, y, 0) plane.
(a) Suppose w is a non-zero complex number. Choose z so that |z| = |w| 2 and arg z =
1
1.7
1
2
arg w or arg z = 21 arg w + π . Show that z2 = w in both cases, and that these are
the only solutions to z2 = w.
(b) The quadratic formula gives two solutions to the equation az2 + bz + c = 0, when
a, b, c are complex numbers with a ´ = 0 because completing the square is a purely
algebraic manipulation of symbols, and there are two complex square roots of every
non-zero complex number by part (a). Check the details.
(c) If w is a non-zero complex number, find n solutions to zn = w using polar
coordinates.
B
1.8 Suppose that f is a continuous complex-valued function on a real interval [a, b]. Let
· b
1
A = b−a f (x)dx
a
be the average of f over the interval [a, b].
(a) Show¶that if | f (x)| ≤ |A| for all x ∈ [a, b], then f = A. Hint: Rotate f so that A > 0.
Then a (A − Ref )dx /(b − a) = 0, and A − Ref is continuous and non-negative.
b
¶b
(b) Show that if |A| = (1/ (b − a)) a | f (x)|dx, then arg f is constant modulo 2π on
{z : f (z) ´= 0}.
1.9 Formally solve the cubic equation ax3 + bx2 + cx + d = 0, where x, a, b, c, d ∈ C, a ´ = 0,
by the following reduction process:
(a) Set x = u + t and choose the constant t so that the coefficient of u2 is equal to zero.
(b) If the coefficient of u is also zero, then take a cube root to solve. If the coefficient
of u is non-zero, set u = kv and choose the constant k so that v3 = 3v + r, for some
constant r.
(c) Set v = z + 1/z and obtain a quadratic equation for z3 . The map z + 1/ z is important
for several reasons, including constructing what are called conformal maps. It will
be examined in more detail in Section 6.4.
(d) Use the quadratic formula to find two possible values for z3, and then take a cube
root to solve for z.
(e) In Section 2.2 we will show that the cubic equation has exactly three solutions,
counting multiplicity. But the process in this exercise appears to generate more solu-
tions, if we use two solutions to the quadratic and all three cube roots. Moreover,
there might be more than one valid choice for the constants used to reduce to a
simpler equation. Explain.
12 Preliminaries
1.10 The equation

¸ ¹2 ¸ ¹
1
a z+
z
+ b z + 1z + c = 0
has four solutions, which can be found by two applications of the quadratic formula. If
we multiply by z2 we obtain the quartic
az4 + bz 3 + (2a + c)z2 + bz + a = 0.
Which quartics Aw4 + Bw 3 + Cw 2 + Dw + E can be put in this form after a linear change
of variable w = αz + β?
1.11 Prove that stereographic projection preserves angles between curves. In other words, if
two curves γ1 , γ2 in the plane meet at an angle θ , then their lifts γ1∗ , γ2∗ to the sphere
meet at the same angle. Moreover, the direction of the angle from γ1 to γ2 corresponds to
the direction from γ1∗ to γ2∗ when viewed from inside the sphere. Orientation is reversed
when viewed from outside the sphere. Hint: This can be done without any calcula-
tions by considering intersecting planes. This exercise will be revisited in Exercise 6.16,
where it is used to find the Mercator projection, a map of tremendous economic impact.
1.12 Stereographic projection combined with rigid motions of the sphere can be used to
describe some transformations of the plane.
(a) Map a point z ∈ C to S2 , apply a rotation of the unit sphere, then map the resulting
point back to the plane. For a fixed rotation, find this map of the extended plane to
itself as an explicit function of z. Two cases are worth working out first: rotation
about the x3 axis and rotation about the x1 axis.
(b) Another map can be obtained by mapping a point z ∈ C to S2 , then translating the
sphere so that the origin is sent to (x0, y0 , z0 ), then projecting back to the plane. The
projection to the plane is given by drawing a line through the (translated) north pole
and a point on the (translated) sphere and finding the intersection with the plane
{(x, y, 0)}. For a fixed translation, find this map as an explicit function of z. In this
case it is worth working out a vertical translation and a translation in the plane
separately. Then view an arbitrary translation as a composition of these two maps.
Partial answer: the maps in parts (a) and (b) are of the form (az + b)/(cz + d) with
ad − bc ´ = 0.
For an award-winning movie of these maps, see
http://www-users.math.umn.edu/˜ arnold/moebius/
but do the exercise before viewing this link.
2 Analytic Functions
2.1 Polynomials
This course is about complex-valued functions of a complex variable. We could think of such
functions in terms of real variables as maps from R2 into R2 given by
f (x, y) = (u(x, y), v(x, y)),

and think of the graph of f as a subset of R4 . But the subject becomes more tractable if we
use a single letter z to denote the independent variable and write f (z) for the value at z, where
z = x + iy and f (z) = u(z) + iv(z). For example,
f (z) = zn
is much simpler to write (and understand) than its real equivalent. Here zn means the product
of n copies of z.
The simplest functions are the polynomials in z:
p(z) = a0 + a1 z + a2 z2 + . . . + an zn , (2.1)
where a0 , . . . , an are complex numbers. If an ± = 0, then we say that n is the degree of p. Note
that z is not a (complex) polynomial, and neither is Rez or Imz.
Let’s take a closer look at linear or degree 1 polynomials. For example, if b is a (fixed)
complex number, then
g(z) = z + b
translates, or shifts, the plane. If a is a (fixed) complex number then
h(z) = az
can be viewed as a dilation and rotation. To see this, recall that by Chapter 1 and Exercise 1.2,
|az| = |a||z| and arg(az) = arg(a) + arg(z) (up to a multiple of 2π ). So, h dilates z by a factor
of |a| and rotates the point z by the angle arg a. A linear function
f (z) = az + b
can then be viewed as a dilation and rotation followed by a translation. Equivalently, writing
f (z) = a(z + b/a) we can view f as a translation followed by a rotation and dilation.
Another instructive example is the function p(z) = zn. By Chapter 1 again,
|p(z)| = |z|n and arg p(z) = n arg z mod 2 π .

Each pie slice

± ² ² ³
² 2π k ²² π
Sk = z : ²²arg z −
n ²
<
n
∩ {z : |z| < r},
k = 0, . . . , n − 1 is mapped to a slit disk
{z : |z| < rn } \ ( −r n, 0).

Angles between straight-line segments issuing from the origin are multiplied by n, and, for
small r, the size of the image disk is much smaller than the “radius” of the pie slice. See
Figure 2.1.
The function k(z) = b(z − z0 )n can be viewed as a translation by −z0, followed by the power
function, and then a rotation and dilation. To put it another way, k translates a neighborhood
of z0 to the origin, then acts like the power function zn, followed by a dilation and rotation
by b.
To understand the local behavior of a polynomial (2.1) near a point z0 , write z = (z − z0 ) + z0
and expand (2.1) by multiplying out and collecting terms to obtain
p(z) = p(z0) + b1(z − z0) + b2(z − z0)2 + . . . + bn (z − z0)n . (2.2)
Another way to see this is to note that p(z) − an (z − z0 )n is a polynomial of degree at most
n − 1, so (2.2) follows by induction on the degree. If b1 ± = 0 then p(z) behaves like the linear
function p(z0) + b1(z − z0 ) for z near z0. If b1 = 0 then, near z0, p(z) is closely approximated
by p(z0 ) + bk (z − z0)k , where bk is the first non-zero coefficient in the expansion (2.2). Indeed,
for small ζ = z − z0 ,
|p(z0 + ζ ) − [p(z0) + bk ζ k ]| ≤ C|ζ |k +1,
for some constant C, by (2.2). Figure 2.2 is sometimes called “walking the dog,” where the
walking path has radius r = | bk ||ζ |k and the leash has length s = C |ζ | k+1 . As ζ traces a circle
centered at 0 of radius ε, the function p(z0) + bk ζ k winds k times around the circle centered
at p(z0 ) with radius r. For small ε, s is much smaller than r so the function p(z0 + ζ ) also then
traces a path which winds k times around p(z0 ).
So, for z near z0, p(z) behaves like a translation by −z0 , followed by a power function, a
rotation and dilation, and finally a translation by p(z0).
zn
0 r −r n 0
Figure 2.1 The power map.
p(z 0 ) + bk ζ k
r s p(z 0 + ζ )
p(z 0 )
Figure 2.2 p(z 0 + ζ ) lies in a small disk of radius s = C |ζ |k +1 < r = |bk ||ζ |k .
2.2 Fundamental Theorem of Algebra and Partial Fractions 15
2.2 Fundamental Theorem of Algebra and Partial Fractions
The local behavior of a polynomial described in Section 2.1 can be used to prove an important
result you will have seen in some form or another since high school. If p is a polynomial and
p(a) = 0 then a is called a zero of p.
Theorem 2.1 (fundamental theorem of algebra)Every non-constant polynomial has a

zero.
This remarkable result says that if we extend the real numbers to the complex numbers via
the solution to the equation z2 + 1 = 0 then every polynomial equation has a solution.
Proof Suppose p(z) = anzn + an−1zn−1 + . . . + a1 z + a0 , n ≥ 1, is a polynomial which

has no zeros and for which an ± = 0. We first prove that |p(z) | must have a non-zero minimum
value on C. Write
´ µ
an−1 a0
p(z) = zn an + + ... + .
z zn
Because an ± = 0, 1/zk → 0 and | zn | → ∞ as | z| → ∞, we conclude that | p(z)| → ∞ as

|z| → ∞. So, if M = infC |p(z)| and |p(wj )| → M, then there is an R < ∞ so that |wj | ≤ R,
for all j. Because {z : |z| ≤ R } is compact and because |p| is continuous, there exists z0 so
that | p(z0 )| = M. Moreover, M ± = 0 since p has no zeros. Now, by the “walking the dog”
argument in Section 2.1 (see Figure 2.2), for small ε > 0, as ζ traces a circle of radius ε, the
function p(z0 + ζ ) traces a path about p(z0) which must intersect the open disk centered at 0 of
radius M because | p(z0 )| = M. More explicitly, choose ε > 0 so that 0 < s < r < M, where
r, s are defined in Section 2.1. Because bk ζ k traces a circle of radius r, we can find ζ so that
|p(z0 ) + bk ζ k | = M − r. Then
|p(z0 + ζ )| ≤ |p(z0 + ζ ) − (p(z0 ) + bk ζ k )| + M − r ≤ s + M − r < M.
This contradiction proves that no such polynomial exists and Theorem 2.1 follows.
Corollary 2.2 If p is a polynomial of degree n ≥ 1, then there are complex numbers

z1 , . . . , zn and a complex constant c so that
n
¶
p(z) =c (z − zk ).
k=1
Corollary 2.2 does not tell us how to find the zeros, but it does say that there are exactly n
zeros, counting multiplicity.
Proof The proof is by induction. First note that
zk − bk= (z − b)(bk−1 + zbk−2 + . . . + zk −2b + zk −1).

∑
So if p(z) = nk=0 ak zk and p(b) = 0, then
⎛ ⎞
p(z) p(z) − p(b)
n
· −1
k·
q(z) ≡ = = ak ⎝ bk −1− j zj ⎠ . (2.3)
z −b z− b
k=1 j =0
The coefficient of zn− 1 in (2.3) is an so q is a polynomial of degree n − 1. Repeating this

argument n times proves the corollary.
For example, the polynomial zn − 1 has n zeros. If zn = 1 then 1 = |zn| = |z|n so that
|z| = 1. Write z = cos t + i sin t = eit (see Exercise 2.6). Then zn = eint = 1 so that nt = 2π k
for some integer k, and thus t = 2π k/n. The n distinct zeros of zn − 1 are then ei2π k/ n,
k = 0, 1, . . . , n − 1, which are equally spaced around the unit circle.
A rational function r is the ratio of two polynomials. By the fundamental theorem of
algebra, we can write r in the form
p(z)
r(z) = ¸N .
j=1 (z − zj )n j
The next corollary, also probably familiar, allows us to write a rational function in a form that
is easier to analyze. The form is also of practical importance because it allows us to solve
certain differential equations that arise in engineering problems using the Laplace transform
and its inverse.
Corollary 2.3 (partial fraction expansion)If p is a polynomial then there is a polynomial

q and constants ck,j so that
nj
N ·
p(z) · ck,j
¸N = q(z) + (z − zj )k
. (2.4)
j=1 (z − zj )n j
j =1 k =1
Proof There are two initial cases to consider: If p is a polynomial then

p(z)
z−a
= q(z) + zp(a)
− a, (2.5)
where q(z) = (p(z) − p(a))/(z − a) is a polynomial, as in (2.3). Secondly, if a ±= b, we can

write
1
(z − a)(z − b)
= z −A a + z −B b , (2.6)
for some constants A and B. For if this equation is true, then we can multiply each term on
the right by z − a and let z → a to obtain A on the right. The same process on the left
yields 1/(a − b), and hence A = 1/(a − b). Similarly B = 1/ (b − a). Now substitute these
values for A and B into (2.6) and check that equality holds. The full corollary now follows by
induction: suppose the corollary is true if the degree of the denominator is at most d. If we
have an equation of the form (2.4) of degree d then we can divide each term in the equation by
z − a. After division, the right-hand side consists of lower degree terms to which the induction
hypothesis applies, with one exception: when the denominator of the left-hand side of (2.4) is
(z − b)d . If a = b, then, after division by z − a, each term will be of the correct form. If a ± = b,
then we could have applied the inductive assumption to the decomposition of
2.3 Power Series 17
p(z)
(z − b)d−1(z − a)
and then divided this result by z − b instead of z − a.
The above proof also suggests an algorithm for computing the coefficients {ck,j }. First apply
(2.5) with a = z1. Multiply each term of the result by 1/ (z − b), where b is one of the zeros
of the denominator in (2.4), and apply either (2.5) or (2.6) to each of the resulting terms
on the right-hand side. Repeat this process, increasing the degree of the denominator by 1
until you have the desired expansion. At each stage, the terms to be expanded are of the form
1/[(z − a)k (z − b)]. These can be expanded by starting with (2.6), dividing the result by (z − a),
then using (2.6) again, repeating until you have reached the power k in the denominator.
The algorithm can be speeded up because we know the form of the solution. If powers
in the denominator nj are all equal to one, and if the numerator has smaller degree than the
denominator, then the form is
p(z) ·
N
cj
¸N = z − zj
. (2.7)
j=1 (z − zj ) j=1
If we multiply each term of the right-hand side by z − z1 then let z → z1 , we obtain c1. If we
multiply the left-hand side by the same factor, it cancels one of the terms in the denominator.
Letting z → z1 we obtain the value of the remaining part of the left-hand side at z1 . This
quickly gives c1 and can be repeated for c2 , . . . , cN . This method is sometimes called the
“cover-up method” because it can be done with less writing by observing that cj is the value
of the left-hand side at zj when you cover z − zj with your hand. If the denominator has terms
with degree bigger than one, first use a denominator with all terms of degree one as above.
Then, as in the proof of Corollary 2.3, multiply everything by 1/(z − zk ) and simplify all terms
on the right, repeating as often as needed. If the degree of the numerator of any term is not
less than the degree of its denominator, polynomial division can also be used to reduce the
degree instead of repeated application of (2.5).
Engineering problems typically have rational functions with real coefficients. See Exer-
cise 2.2 for a similar technique that decomposes rational functions with real coefficients into
terms whose denominators are either powers of linear terms with real zeros or powers of
irreducible quadratics with real coefficients.
2.3 Power Series
More complicated functions are found by taking limits of polynomials. Here is the primary
example:
∞
·
zn .
n=0
This series is important to understand because its behavior is typical of all power series
(defined shortly) and because it is one of the few series we can actually add up explicitly.
If |z| ≥ 1 then |zn| = | z|n ≥ 1, so the terms of the series do not tend to 0 as n → ∞. The
series then diverges. If |z| < 1, then the partial sums
m
·
Sm = zn = 1 + z + z2 + . . . + zm
n=0
satisfy
(1 − z)Sm = 1 − zm+1 ,
as can be seen by multiplying out the left-hand side and cancelling. If |z| < 1, then |zm +1| =
|z|m+1 → 0 as m → ∞ and so Sm(z) → 1/(1 − z). We conclude that if |z| < 1 then
∞
·
zn = 1 −1 z . (2.8)
n=0
It is important to note that the left- and right-hand sides of (2.8) are different objects. They
agree in |z| < 1, the right-hand side is defined for all z ± = 1, but the left-hand side is defined
only for | z| < 1.
The formal power series
∞
·
f (z) = an (z − z0)n = a0 + a1 (z − z0 ) + a2 (z − z0 )2 + . . .
n=0
is called a convergent power series centered (or based) zat 0 if there is an r > 0 so that
the series converges for all z such that |z − z0 | < r. Note: If we plug z = z0 into the formal
power series, then we always get a0 = f (z0). More formally, the definition of the summation
notation includes the convention that the n = 0 term equals a0, so that we are not raising 0 to
the power 0. The requirement for a power series to converge is stronger than convergence at
just the one point z0.
A variant of the primary example is:
1 1 1
z−a
= z − z0 − (a − z0)
= .
−(a − z0)(1 − ( az−−zz 0
0
))
Substituting
= az −− zz0
w
0
into (2.8) we obtain, when | w| = |(z − z0 )/(a − z0 )| < 1,
1 ·∞ −1 (z − z )n .
z−a
= (a − z0 )n+1
0 (2.9)
n= 0
By (2.8), this series converges if |z − z0 | < | a − z0 | and diverges if |z − z0 | ≥ |a − z0 |.

The domain of convergence is an open disk, and it is the largest disk centered at z0 which
is contained in the domain of definition of 1/(z − a). In particular, this function has a power
series expansion based at every z0 ± = a, but different series for different base points z0.
∑
Theorem 2.4 (Weierstrass M-test) If |an(z − z0 )n| ≤ Mn for |z − z0| ≤ r and if Mn < ∞
∑
then ∞n= 0 an (z − z0 ) converges uniformly and absolutely in { z : | z − z0 | ≤ r} .
n
2.3 Power Series 19
Proof If M > N then the partial sums Sn (z) satisfy

² ²
² · ² ·
² M ² M
|SM (z) − SN (z) | = ²
² an (z − z0 ) ² ≤
n²
M n.
²n=N+1 ² n=N +1
∑ ∑
Since Mn < ∞, we deduce M n=N +1 Mn → 0 as N, M → ∞, and so { Sn } is a Cauchy
sequence converging uniformly. The same proof also shows absolute convergence.
Note that the convergence depends only on the “tail” of the series so that we need only
satisfy the hypotheses in the Weierstrass M-test for n ≥ n0 to obtain the conclusion.
The primary example (2.8) converges on a disk and diverges outside the disk. The next
result says that disks are the only kind of domain in which a power series can converge.
∑
Theorem 2.5 (root test) Suppose an (z − z0 )n is a formal power series. Let
1
R = lim inf |an| − n = ∈ [0, +∞].
1
n→∞
lim sup | an | n
1
n→∞
∑∞
Then n=0 an (z − z0 )n
(a) converges absolutely in {z : | z − z0 | < R },
(b) converges uniformly in {z : |z − z0| ≤ r}, for all r < R, and
(c) diverges in {z : |z − z0 | > R}.
∑
See Figure 2.3; R is called the radius of convergenceof the series an (z − z0)n . It has an
interpretation in terms of the decay rate for the coefficients. If S < R then |an| ≤ S−n, for
large n, and this statement fails for any S > R.
∑
Proof The idea is to compare the given series with the example (2.8), zn . If | z − z0 | ≤
r < R, then choose r1 with r < r1 < R. Thus r1 < lim inf |an| − , and there is an n0 < ∞ so
1
n
that r1 < |an| − n for all n ≥ n0 . This implies that |an(z − z0)n | ≤ ( r )n. But, by (2.8),
1
r1
∞ ´ r µn
·
r1
= 1 −1r/r < ∞
n=0 1
since r/r1 < 1. Applying Weierstrass’s M-test to the tail of the series (n ≥ n0 ) proves (b). This
same proof also shows absolute convergence (a) for each z with |z − z0| < R. If |z − z0 | > R,
fix z and choose r so that R < r < | z − z0 |. Then |an|− n < r for infinitely many n, and hence
1
diverges
converges
R z0
Figure 2.3 Convergence of a power series.

´
|an (z − z0 ) | >n |z − z0 | µn
r
for infinitely many n. Since (| z − z0 |/r) n
→ ∞ as n → ∞, (c) holds.
The proof of the root test also shows that if the terms an(z − z0) n of the formal power series
are bounded when z = z1 then the series converges on {z : | z − z0 | < | z1 − z0|}.
The root test does not give any information about convergence on the circle of radius R.
The series can converge at none, some or all points of {z : |z − z0 | = R }, as the following
examples illustrate.
Examples
∞ zn
· ∞ zn
· ∞
· ∞
· ∞
·
2−n zn .
2 2
(i) , (ii) 2
, (iii) nz n, (iv) 2 n zn , (v)
n n
n=1 n=1 n=1 n=1 n=1
The reader should verify the following facts about these examples. The radius of conver-
gence of each of the first three series is R = 1. When z = 1, the first series is the harmonic
series which diverges, and when z = −1 the first series is an alternating series whose terms
decrease in absolute value and hence converges. The second series converges uniformly and
absolutely on {|z| = 1}. The third series diverges at all points of {|z| = 1}. The fourth series
has radius of convergence R = 0 and hence is not a convergent power series. The fifth example
has radius of convergence R = ∞ and hence converges for all z ∈ C.
∑
The radius of convergence of the series an zn , where
¹
3−n, if n is even
an = 4 n, if n is odd
is R = 1/4 by the root test. This is an example where ratios of successive terms in the series
do not provide sufficient information to determine convergence.
2.4 Analytic Functions
Definition 2.6 A function f is analytic at z0 if f has a power series expansion valid in a

∑
neighborhood of z0. This means that there is an r > 0 and a power series an(z − z0 )n which
converges in B = {z : | z − z0 | < r} and satisfies
∞
·
f (z) = an (z − z0 )n,
n=0
for all z ∈ B. A function f is analytic on an open set

± if f is analytic at each z0 ∈ ±.
Note that we do not require one series for f to converge in all of ±. The function (z − a)−1 is
analytic on C \ {a}, as shown in (2.9), and is not given by one series. Note that if f is analytic
on ± then f is continuous in ±. Indeed, continuity is a local property. To check continuity near
z0, use the series based at z0. Since the partial sums are continuous and converge uniformly
2.4 Analytic Functions 21
on a closed disk centered at z0, the limit function f is continuous on that disk. Occasionally it
is convenient to say that a function f is analytic on a set E which is not open. This means that
there is an open set ± ⊃ E and an analytic function g defined on ± with g = f on E.
A natural question at this point is: where is a power series analytic?
Theorem 2.7 If f (z) = ∑ an (z − z0 )n converges on {z : |z − z0 | < r } then f is analytic on

{z : |z − z0| < r}.
Proof Fix z1 with | z1 − z0 | < r. We need to prove that f has a power series expansion based
at z1. We saw in (2.2) how to rearrange a polynomial. The idea is the same for power series,
but we need to prove convergence. By the binomial theorem,
n ´ µ
· n
(z − z0 )
n
= (z − z1 + z1 − z0 ) = n
(z1 − z0 )n−k (z − z1 )k .
k
k =0
Hence
∞ º·
· n ´ µ
n
»
n−k
f (z) = an (z1 − z0 ) (z − z1 )k
. (2.10)
k
n=0 k =0
Suppose, for the moment, that we can interchange the order of summation, then
∞ º·
· ∞ ´ nµ »
an (z1 − z0)n−k (z − z1)k
k
k =0 n=k
will be the power series expansion for f based at z1. To justify this interchange of summation,
it suffices to prove absolute convergence of (2.10). By the root test,
∞
·
|an||w − z0|n
n=0
converges if |w − z0 | < r. Set
w = |z − z1| + | z1 − z0| + z0 .
Then |w − z0| = |z − z1| + | z1 − z0| < r provided |z − z1 | < r − |z1 − z0| . See Figure 2.4.
z
r − | z1− z0| z1 r
w
z0
Figure 2.4 Proof of Theorem 2.7.

Thus if | z − z1 | < r − |z1 − z0| then

∞
·
∞> |an ||w − z0 |n
n=0
∞
·
= |an | (|z − z1 | + |z1 − z0 |)n
n=0
∞ º·
· n ´ µ
n
»
= |an| k
|z1 − z0 |n−k |z − z1|k
n=0 k=0
as desired.
Another natural question is: can an analytic function have more than one power series
expansion based at z0 ?
Theorem 2.8 (uniqueness of series)Suppose

∞
· ∞
·
an(z − z0)n = bn(z − z0) n,
n=0 n=0
for all z such that |z − z0| < r, where r > 0. Then an = bn for all n.
∑
Proof Set cn = an − bn . The hypothesis implies that ∞ n=0 cn (z − z0 ) = 0, for all z with
n
|z − z0 | < r. We need to show that cn = 0 for all n. Suppose cm is the first non-zero coefficient.
If 0 < |z − z0 | < r then
∞
· ∞
·
(z − z0 )−m cn (z − z0 )n = cm +k (z − z0 )k ≡ F(z).
n=m k =0
The series for F converges in 0 < | z − z0| < r because we can multiply the terms of the series
on the left-hand side by the non-zero number (z − z0)−m and not affect convergence. By the
root test, the series for F converges in a disk and hence in {|z − z0 | < r}. Since F is continuous
and cm ± = 0, there is a δ > 0 so that, if |z − z0| < δ then
|F(z) − F(z0 )| = |F(z) − cm| < |cm|/2.

If F(z) = 0, then we obtain the contradiction |− cm | < |cm |/ 2. Thus F(z) ± = 0 when | z − z0 | <
δ . But (z − z0) m = 0 only when z = z0 , and thus
∞
·
cn(z − z0)n = (z − z0)m F(z) ±= 0
n=0
∑
when 0 < | z − z0 | < δ , contradicting our assumption on cn(z − z0 )n.
Note that the proof of Theorem 2.8 shows that if f is analytic at z0 then, for some δ > 0,
either f (z) ± = 0 when 0 < |z − z0| < δ or f (z) = 0 for all z such that |z − z0| < δ. If f (a) = 0,
then a is called a zero of f . A region is a connected open set.
Corollary 2.9 If f is analytic on a region ± then either f ≡ 0 or the zeros of f are isolated
in ±.
Corollary 2.9 is sometimes called the uniqueness theorem or identity theorem because
of the consequence that if two analytic functions on a region ± agree on a set with an
accumulation point in ± then they must be identical.
Proof Let E denote the set of non-isolated zeros of f . Since f is continuous, the set of zeros
of f is closed in ±. Each isolated zero is contained in an open disk containing no other zero
and hence E is closed in ±. In the proof of Theorem 2.8, we showed that if z0 is a non-isolated
zero of f then f is identically zero in a neighborhood of z0 . Thus E is open. By connectedness,
either E = ± or E = ∅.
There are plenty of continuous functions for which the corollary is false, for example
x sin(1/x). The corollary is true because, near z0 , the function f (z) behaves like the first
non-zero term in its power series expansion about z0. See Exercise 2.5.
2.5 Elementary Operations
Theorem 2.10 If f and g are analytic at z0 then so are
f + g, − g, cf (where c is a constant) and

f fg.
If h is analytic at f (z0) then (h ◦ f )(z) ≡ h( f (z)) is analytic at z0 .
Proof The first three follow from the associative, commutative and distributive laws applied
to the partial sums. To prove that the product of two analytic functions is analytic, multiply
∑ ∑
f (z) = an (z − z0 )n and g = bn (z − z0 )n as if they were polynomials to obtain
∞
· ∞
· ∞ ´·
· n µ
an(z − z0) n
bk (z − z0)
k
= ak bn−k (z − z0)n , (2.11)
n=0 k =0 n=0 k =0
which is called the Cauchy productof the two series. Why is this formal computation valid?
The series for f and the series for g converge absolutely for |z − z0 | sufficiently small. Because
we can rearrange non-negative convergent series,
∞
· ∞
· ∞ ´·
· n µ
∞> |an ||z − z0 |n |bk ||z − z0 |k = |ak ||bn−k | |z − z0 |n .
n=0 k=0 n=0 k =0
This says that the series on the right-hand side of (2.11) is absolutely convergent and therefore
can be arranged to give the left-hand side of (2.11). To put it another way, the doubly indexed
sequence an bk (z − z0 )n+k can be added up two ways: if we add along diagonals: n + k = m,
for m = 0, 1, 2, . . . , we obtain the partial sums of the right-hand side of (2.11). If we add
along partial rows and columns n = m, k = 0, . . . , m, and k = m, n = 0, . . . , m − 1, for
m = 1, 2, . . . , we obtain the product of the partial sums for the series on the left-hand side of
(2.11). Since the series is absolutely convergent (as can be seen by using the latter method of
summing the doubly indexed sequence of absolute values), the limits are the same.
we can compose analytic functions where it makes sense, suppose f (z) =
∑To prove that ∑
an(z − z0 )n is analytic at z0 and suppose h(z) = bn (z − a0 )n is analytic at a0 = f (z0).
The sum
∞
·
|am ||z − z0|m−1 (2.12)
m=1
converges in {z : 0 < | z − z0 | < r} for some r > 0 since the series for f is absolutely
convergent, and |z − z0 | is non-zero. By the root test (set k = m − 1), this implies that the
series (2.12) converges uniformly in {|z − z0 | ≤ r1 }, for r1 < r, and hence is bounded in
{|z − z0 | ≤ r1 }. Thus there is a constant M < ∞ so that
∞
·
|am||z − z0 |m ≤ M |z − z0|,
m=1
if |z − z0| < r1 . We conclude that

∞ ¼∞ ½m ∞
· · ·
|bm| |an ||z − z0| n
≤ |bm|(M|z − z0 |)m < ∞,
m=0 n=1 m=0
for |z − z0| sufficiently small, by the absolute convergence of the series for h. This proves
absolute convergence for the composed series, and thus we can rearrange the doubly indexed
series for the composition so that it is a (convergent) power series.
As a consequence, if f is analytic at z0 and f (z0) ± = 0 then 1/f is analytic at z0. Indeed, the
function 1/z is analytic on C \ { 0} by (2.9) with a = 0, and 1 /f is the composition of 1/z with
f . A rational functionr is the ratio
p(z)
r(z) = q(z) ,
where p and q are polynomials. The rational function r is then analytic on {z : q(z) ± = 0}
by Theorem 2.10. Rational functions and their limits are really what this whole book is
about.
Definition 2.11 If f is defined in a neighborhood of z then

f (w) − f (z)
f ² (z) = wlim
→z w − z
is called the (complex) derivative of f , provided the limit exists.
The function z does not have a (complex) derivative. If n is a non-negative integer,
(zn )² = nzn−1 .
The next theorem says that you can differentiate power series term-by-term.
Theorem 2.12 If f (z) = ∑∞n=0 an (z − z0 )n converges in B = {z : |z − z0 | < r} then f ² (z)

exists for all z ∈ B and
∞
· ∞
·
f ² (z) = nan(z − z0 )n−1 = (n + 1)an+1 (z − z0 )n,
n=1 n=0
for z ∈ B. Moreover, the series for f ² based at z0 has the same radius of convergence as the
series for f .
Proof If 0 < |h| < r then

∑∞ ∞ ∞
f (z0 + h) − f (z0) n=0 a nh
n − a0 − a = · ·
− a1 = 1 anhn−1 = an+1 hn .
h h
n=2 n=1
∑
By the root test, the region of convergence for the series an+1 hn is a disk centered at 0 and
∑
hence it converges uniformly in {h : | h| ≤ r1}, if r1 < r. In particular, an+1 hn is continuous
at 0 and hence
∞
·
lim
h→0
an+1 hn = 0.
n=1
This proves that f ² (z0 ) exists and equals a1 .

By Theorem 2.7, f has a power series expansion about each z1 , with | z1 − z0 | < r given by
∞ º·
· ∞ ´ µ
n
»
an (z1 − z0)n−k (z − z1)k .
k
k=0 n=k
Therefore f ² (z1) exists and equals the coefficient of z − z1:

∞ ńµ
· ∞
·
f ² (z1 ) = an (z1 − z0 )n−1 = an n(z1 − z0)n−1 .
1
n=1 n=1
→ 1, the series for f ² has exactly the same radius of

1
By the root test and the fact that n n
convergence as the series for f .
Since the series for f ² has the same radius of convergence as the series for f , we obtain the
following corollary.
Corollary 2.13 An analytic funtion f has derivatives of all orders. Moreover, if f is equal to
a convergent power series on B = {z : |z − z0| < r} then the power series is given by
∞ f (n)(z )
·
f (z) = 0
n!
(z − z0 )n,
n=0
for z ∈ B.
By definition of the symbols, the n = 0 term in the series is f (z0 ).

Proof If f (z) = ∑∞n=0 an (z − z0 )n , then we proved in Theorem 2.12 that a1 = f ² (z0) and
∞
·
f ² (z) = nan (z − z0)n−1 .
n=1
Applying Theorem 2.12 to f (z), we obtain 2a2 = ( f ²)² (z0 ) ≡ f ²² (z0), and, by induction,
²
n! an = f (n) (z0).
If f is analytic in a region ± with f ² (z) = 0 for all z in a neighborhood of z0 ∈ ±, then,
by Corollary 2.13 and Corollary 2.9 applied to f (z) − f (z0), f is constant in ±. A useful
consequence is that if f and g are analytic with f ² = g² , then f − g is constant.
A closer examination of the idea of the proof of Theorem 2.12 shows that power series
satisfy a stronger notion of differentiability at a point. Corollary 2.14 will be used in Chapter 3
for understanding the local behavior of power series.
∑
Corollary 2.14 If f (z) = an(z − z0)n converges in B = {z : | z − z0 | < r } then
f (z) − f (w)
f ² (z0) = z,wlim .
→z 0 z −w
Proof Set z = z0 + h and w = z0 + k. Then, for h − k ± = 0 and ε = max( |h|, | k|) < r,
f (z0 + h) − f (z0 + k)
∞
· h n − kn
∞
· n− 1
·
− a1 = an = an hj kn−j−1. (2.13)
h−k h−k
n=2 n=2 j=0
But
M
· n−1
· M
·
lim
N,M →∞
|an| |h|j |k|n−j−1 ≤ N,Mlim
→∞
|an |nεn−1 = 0,
n=N j=0 n=N
by the root test. Because of uniform convergence, the right-hand side of (2.13) is a continuous
function of (h, k) when ε < r, vanishing at (0, 0), and hence
f (z) − f (w) f (z0 + h) − f (z0 + k)
lim
z,w→z 0 z− w
= h,klim
→0 h−k
= a1 = f ² (z0).
Corollary 2.14 fails for the real-valued differentiable function x2 sin(1 /x) but holds for
continuously differentiable real-valued functions by the mean-value theorem of calculus.
Exercise 2.17 shows, however, that the mean-value theorem does not hold for all analytic
functions.
∑
Corollary 2.15 If f (z) = an(z − z0 )n converges in B = {z : |z − z0 | < r} then the power
series
∞ a
·
F(z) = (z − z0 )n+ 1
n
n+1
n=0
converges in B and satisfies
F ²(z) = f (z),
for z ∈ B.
2.6 Exercises 27
The series for F has the same radius of convergence as the series for f , by Theorem 2.12 or
by direct calculation.
2.6 Exercises
2.1 Check that Examples (i)–(v) in Section 2.3 are correct.

2.2 (a) If p is a polynomial with real coefficients, prove that p can be factored into a
product of linear and quadratic factors, each of which has real coefficients, such
that the quadratic factors are non-zero on R. Most engineering problems involving
polynomials only need polynomials with real coefficients.
(b) For rational functions with real coefficients, such as those that typically occur in
applications, it is sometimes preferable to use a partial fraction expansion without
complex numbers in the expression. The cover-up method can also be used in this
case. Here is an example to illustrate the idea. If a, b, c, d and e are real, show that
z2 + dz + e − b) + D ,
(z − a)((z − b)2 + c2)
= z −A a + B(z
(z − b)2 + c2
where A, B and D are real. Here we have completed the square for the irreducible
quadratic factor. Note also that we have written the numerator of the last term as
B(z − b) + D, not Bz + D. We can find A by the usual cover-up method. Then, to find
B and D, we multiply by (z − b)2 + c2 and let it tend to 0. Thus z → b ³ ic. Cover
up the quadratic factor in the denominator on the left and let z → b + ic. On the
right-hand side, when we multiply by the quadratic factor, the first term will tend
to 0, the denominator of the second term will be cancelled and B(z − b) + D tends to
Bic + D. Thus, the real part of the result on the left equals D and the imaginary part
equals Bc, and then we can immediately write down the coefficients B and D. Try
this process with two different irreducible quadratic factors in the denominator, and
you will see how much faster and accurate it is than solving many equations with
many unknowns. The choice of the form of the numerator as B(z − b) + D instead
of Bz + D made this computation a bit easier. It also turns out that it makes it a bit
easier to compute inverse Laplace transforms of these rational functions, because
the resulting term is a shift in the domain of a simpler function.
2.3 For what values of z is
∞ ´ z µn
·
1+z
n=0
convergent? Draw a picture of the region.

2.4 Prove the sum, product, quotient and chain rules for differentiation of analytic functions
and find the derivative of (z − a)−n, where n is a positive integer and a ∈ C.
2.5 (a) Prove that f has a power series expansion about z0 with radius of convergence r > 0
−f (z0) has a power series expansion about z0 , with the same
if and only if g(z) = f (z)z−z0
radius of convergence. (How must you define g(z0), in terms of the coefficients of
the series for f , to make this a true statement?)
(b) It follows from (a) that if f has a power series expansion at z0 with radius of conver-
gence R, and if r < R, then there is a constant C so that | f (z) − f (z0 )| ≤ C| z − z0 |,
∑
provided | z − z0 | ≤ r. Use the same idea to show that if f (z) = an (z − z0)n
then
² ²
² k ²
² f (z) − · a (z − z )n² ≤ Dk |z − z0 |k +1,
² n 0 ²
² ²
n=0
where Dk is a constant and |z − z0 | ≤ r < R.

(c) Use the proof of the root test to give an explicit estimate of Dk (for large k) and
therefore an estimate of the rate of convergence of the series for f if |z − z0 | <
r < R.
∑
2.6 Define ez = exp(z) = ∞ zn
n=0 n! .
(a) Show that this series converges for all z ∈ C.

(b) Show that ez ew = ez+w .
(c) Define cos θ = 21 (ei θ + e−iθ ) and sin θ = 2i1 (eiθ − e−iθ ), so that eiθ = cos θ + i sin θ .
Using the series for ez show that you obtain the same series expansions for sin and
cos that you learned in calculus. Check that cos2 θ + sin2 θ = 1, by multiplying out
the definitions, so that eiθ is a point on the unit circle corresponding to the cartesian
coordinate (cos θ , sin θ ).
(d) Show that |ez | = eRez and arg ez = Imz. If z is a non-zero complex number then
show that z = reit , where r = |z| and t = arg z. Moreover, show that zn = rn eint .
(e) Show that ez = 1 only when z = 2π ki, for some integer k.
2.7 (a) Using the definitions in Exercise 2.6, prove dzd ez = ez .

(b) Use (a) and the chain rule to compute the indefinite integral
¾
ent cos mt dt.
¿
Hint: Use Re e(n+im)t dt, which results in a lot less work than the standard calculus
trick of integrating by parts twice.
(c) Use (a), the chain rule and the fundamental theorem of calculus to prove
¿ 2π int
0 e dt = 0, if n is a non-zero integer.
(d) Suppose a = eb. If f ² (z) = 1/z and f (a) = b, find the series expansion for f about
a valid in | z − a| < | a|. Use (a) to prove f (ez ) = z. The function f is called the
complex logarithm with f (a) = b. See Corollary 5.8 and Definition 5.9.
2.8 Prove Theorems 2.7 and 2.10 without explicitly exhibiting the rearrangements, by using
the first sentence of Exercise 1.1. While this may yield slightly easier proofs, the
proofs in the text were chosen because the explicit rearrangements are useful, as in
Exercise 2.6(b).
2.9 (a) Suppose p and q are polynomials with no common zero, and suppose q(z0 ) ± = 0.
Let d denote the distance from z0 to the nearest zero of q. Then the rational function
r = p/q has a power series expansion which converges in {z : |z − z0| < d} and no
2.6 Exercises 29
larger disk. Hint: Use the partial fraction expansion, Exercise 2.4, Theorem 2.12,
and (2.9).
(b) Find the series expansion and radius of convergence of
z + 2i
(z − 6)2 (z2 + 6z + 10)
about the point 1. Hint: Set z = 1 + w, then expand in powers of w.
2.10 Let n be a positive integer. Prove that z n is analytic in B = {z : | z − 1| < 1} in
1
the following sense: there is a convergent power series f in B with the property that
∑
f (z) n = z and f (1) = 1. Hint: Write z = 1 + w, | w| < 1, and let g(w) = ak w k be
the (formal) Taylor series for (1 + w) n . Then prove | ak | ≤ 1/k so that g is analytic in
1
|w | < 1. Use Taylor’s theorem to show that g(x) = (1 + x)1/n for −1 < x < 1, and then
use the uniqueness theorem, Corollary 2.9, to show that g(w)n = 1 + w. Alternatively,
prove that z1/ n can be defined so that it has derivatives of all orders and prove that
Taylor’s theorem is true for complex differentiable functions using complex integration.
∑ ∑∞
2.11 Suppose ∞ j=0 | aj | < ∞. Show that f (z) =
2
j=0 a j z is analytic in {z : | z| < 1} .
j
Compute (and prove your answer):

¾ 2π
lim
r ´1 0
| f (rei )|2 2dπθ .
θ
¿ f has a power series expansion at 0 which converges in all of C. Suppose also

2.12 Suppose
that C | f (x+iy)| dxdy < ∞. Prove f ≡ 0. Hint: Use polar coordinates to prove f (0) = 0.
2.13 Suppose f is analytic in a connected open set U such that, for each z ∈ U, there exists
an n (depending upon z) such that f (n) (z) = 0. Prove f is a polynomial.
2.14 Let f be analytic in a region U containing the point z = 0. Suppose | f (1/ n)| < e−n for
n ≥ n0 . Prove f (z) ≡ 0.
2.15 (Newton’s methodfor solving f (z) = 0.) Suppose f is analytic in a neighborhood of a
and suppose f (z) = (z − a)g(z), where g(a) ± = 0. Set Nf (z) = z − f (z)/f ²(z).
(a) Prove that, for δ sufficiently small,
² ² ²
²
g (z) ² ²
sup ²² ≤ 21δ .
|z−a|≤δ g(z) ²
(b) Prove that if |z − a| <δ then
² ² ² ²
² Nf (z) − a ² ² z − a ²2
² ²≤² ²
² δ ² ² δ ² .
Thus if |z0 − a| < δ and zn = Nf (zn−1) then
² ² ² ²
² z n − a ² ² z 0 − a ²2
n
² ² ≤² ²
² δ ² ² δ ² .
The function Nf is called the Newton iterate of f . We say that zn converges to a
quadratically because the error is squared at each step. Roughly, if zn has the same
first k digits as a then zn+1 will have the same first 2k digits. Numerically, just a few
steps of Newton’s method gives virtually a formula for a, provided z0 is sufficiently
close to a. If z0 is not sufficiently close to a, then Newton’s method may not converge.
Studying the behavior of the iteration zn = r(zn−1) for analytic functions r is the basis
of the field called complex dynamics.
2.16 If f is analytic in a neighborhood of a then we can approximate f ² (a) by the differ-
ence quotient Dδ = f (a+δδ)−f (a) with error on the order of | δ |. However, on a computer,
where functions can only be evaluated approximately, there is a loss in precision called
round-off error. If functions can be evaluated with 16 decimal digits of accuracy and
if δ = 10− k then the evaluation of Dδ will only have roughly 16 − k digits of accuracy.
The approximation to f ² (a) by Dδ will have roughly min(16 − k, k) digits of accuracy.
Choosing k = 8 gives the best estimate of 8 digits of accuracy. Consider instead
·
3
À
Dδ = 41δ f (a + δ ij )i−j .
j =0
²
The numerical approximation of f (a) by À Dδ has min(16 − k, 4k) digits of accuracy,
roughly. When k = 3 this gives roughly 12 digits of accuracy. Explain this reasoning
using the power series expansion of f at a with estimates in terms of the derivatives of f .
An approximate Newton’s methodinvolves combining Exercises 2.15 and 2.16 when
an explicit formula for f ² is not available.
2.17 Suppose f is analytic in a convex open set U. Suppose that for each z, w ∈ U there exists
a point ζ on the line segment between z and w with
f (z) − f (w)
z −w
= f ² (ζ ).
Prove f is a polynomial of degree at most 2. (The point is that you have to be careful:
not all calculus theorems extend to similar complex versions.) Hint: First prove it for a
degree 3 polynomial on a very small disk.
∑∞
2.18 Let f (z) = n=0 an z have radius of convergence 1 and suppose a n ≥ 0 for all n.
n
Prove that z = 1 is a singular point of f . That is, there is no function g analytic in a

neighborhood U of z = 1 such that f = g on U ∩ D.
To illustrate that not everything is known about polynomials, here is an unsolved problem
posed by the Fields Medalist S. Smale [25] in 1981, which arose in connection with the com-
plex version of Newton’s method. Find the smallest constant K so that, for each polynomial p
and each z ∈ C, there exists c ∈ C so that p² (c) = 0 and
² ²
² p(z) − p(c) ² ²
² z − c ² ≤ K| p (z)| .
² ²
This is known to hold for all polynomials if K = 4 and fails for some polynomials if K <
Kd = 1 − 1/d, where d is the degree of p. The conjectured best constant is Kd , but even K = 1
would be an interesting result.
3 The Maximum Principle
3.1 The Maximum Principle
We can apply the same local (“walking the dog”) analysis to analytic functions that we applied
to polynomials in Section 2.1. If
∞
±
f (z) − f (z0 ) = an(z − z0)n , (3.1)
n=k
where ak ±= 0, then, setting z = z0 + h,
| f (z0 + h) − [f (z0 ) + ak hk ]| ≤ C|h|k+1 , (3.2)
for |h| sufficiently small by Exercise 2.5(b). This inequality says that the value f (z) lies inside
a disk centered at f (z0 ) + ak hk of radius at most C |h|k+1 , which is much smaller than |ak hk |.
See Figure 2.2. In other words,
f (z) ≈ f (z0 ) + ak (z − z0)k . (3.3)
The next result is perhaps the most important elementary result in complex analysis. It
follows from the “walking the dog” analysis much like the proof of the fundamental theorem
of algebra. See Exercise 3.1. Because of its importance, we give another proof. The proof is
less geometric, but we will apply it to a more general class of functions in Chapter 7.
Theorem 3.1 (maximum principle) Suppose f is analytic in a region ±. If there exists a

z0 ∈ ± such that
| f (z0)| = sup | f (z)|
z ∈±
then f is constant in ±.
In particular, a non-constant analytic function has no local maximum absolute value, by

Theorem 3.1 and the identity theorem, Corollary 2.9.
Proof If f has a series expansion given by (3.1) which converges in {z : |z − z0| < r0 }, then
for r < r0
² 2π
dt
f (z0) = f (z0 + reit ) , (3.4)
0 2π
because we can interchange the order of summation and integration by uniform convergence.
Equation (3.4) is called the mean-value propertyfor analytic functions. It says that f (z0) is
the average of f over a circle centered at z0 . Intuitively, if f (z0 + reit ) lies in a disk of radius M,
then the only way that an average of its values can have absolute value M is for the function
to be a constant of absolute value M. A more rigorous argument follows.
Suppose | f (z)| ≤ | f (z0)| for all z ∈ ±. If | f (z0 )| = 0, then f is the constant function 0. If
| f (z0 )| ± = 0, set λ = | f (z0)|/ f (z0 ), so that |λ| = 1 and | f (z0 )| = λf (z0 ). By the mean-value
property,
² 2π ( ) dt
0 = Re λf (z0 ) − λf (z0 + reit)
0 2π
² 2π ( ) dt
= | f (z0 )| − Reλf (z0 + reit) 2π
, (3.5)
0
for r < r0 . But the right-hand integrand in (3.5) is continuous and non-negative and hence
| f (z0 )| = Reλf (z),
for all | z − z0 | < r0 . Moreover, for |z − z0 | < r0,
| f (z0 )|2 = (Reλf (z))2 ≤ (Reλf (z))2 + (Imλf (z))2 = |λf (z)|2 ≤ | f (z0)|2
so that Imλf (z) = 0 when |z − z0 | < r0 . Thus λf (z) = | f (z0 )| in a neighborhood of z0. By the
identity theorem, Corollary 2.9, λ f (z) − | f (z0 )| = 0 for all z ∈ ±.
A colleague calls this the “nobody is above average” proof.

If f (z0 ) ± = 0 then the same argument shows that | f (z0 )| is not a local minimum if f is
non-constant. This fact can also be derived from the statement of the maximum principle by
considering the function 1/f , which is analytic off the zeros of f .
Another form of the maximum principle is Corollary 3.2.
Corollary 3.2 If f is a non-constant analytic function in a bounded region ±, and if f is

continuous on ±, then
max | f (z)|
z ∈±
occurs on ∂ ± but not in ±.
The reader should verify the alternative form: if f is analytic on ± then

lim sup | f (z)| = sup | f (z)|. (3.6)
z →∂ ± ±
We say that a sequence tends to ∂ ± if it is eventually outside each compact subset of ±.

The lim sup is then the largest subsequential limit of the values of | f | over all sequences that
are eventually outside of each compact subset of ±. If ± is unbounded, we view our sets as
lying on the Riemann sphere, so that the boundary includes the north pole (the point “at ∞”).
Equivalently, we can measure the distance to the boundary using the chordal metric given at
the end of Section 1.3.
The function f (z) = e−iz is analytic in the upper half-plane H = {z : Imz > 0}, continuous
on {z : Imz ≥ 0} and has absolute value 1 on the real line R, but is not bounded by 1 in H.
See Exercise 3.2.
3.2 Local Behavior 33
3.2 Local Behavior
The maximum principle allows us to give an improved description of the local mapping
property of analytic functions.
Corollary 3.3 A non-constant analytic function defined on a region is an open map.
In other words, if f is analytic and non-constant on a region ±, and if U ⊂ ± is open, then

f (U) is an open set.
Proof Suppose f has a power series expansion which converges on {z : | z − z0| < R}. Pick
r < R and set
δ = |z−inf
z |= r
| f (z) − f (z0 )|.
0
Since the zeros of f − f (z0 ) are isolated by Corollary 2.9, we may suppose that δ > 0 by
decreasing r if necessary. If |w − f (z0 )| < δ/2 and if f (z) ± = w for all z such that |z − z0| ≤ r,
then 1/( f − w) is analytic in |z − z0 | ≤ r and
³ ³
³ 1 ³ 1 1 2
³ f (z) − w ³ ≤ | f (z) − f (z0)| − | w − f (z0) | =
³ ³
δ − δ/2
<
δ
on |z − z0 | = r. By the maximum principle, the inequality |1/( f (z) − w)| < 2/δ persists
in |z − z0| < r. But evaluating this expression at z0 we obtain the contradiction 2 /δ < 2/δ.
Thus the image of the disk of radius r about z0 contains a disk of radius δ/2 about f (z0 ). This
implies that the image of an open set contains a neighborhood of each of its points.
The main ingredient in the proof of Corollary 3.3 is the maximum principle. An open
continuous function on a region always satisfies the maximum principle.
Definition 3.4 A function f is one-to-one if f (z) = f (w) only when z = w.
A more enlightened terminology proposed by one of my teachers, Richard Arens, is two-

to-two since two points go to two points. “One-to-one” should really be the definition of a
function, but the inertia of common usage is too large to overcome.
Proposition 3.5 If f is analytic at z0 with f ² (z0) ±= 0, then f is one-to-one in a sufficiently

small neighborhood of z0 .
Proof By Corollary 2.14, if zn, wn → z0 with f (zn ) = f (wn) then f ²(z0 ) = 0.

Corollary 3.3 and Proposition 3.5 show that if f is analytic at z0 with f ² (z0) ± = 0, then f
is a homeomorphism of a neighborhood of z0 onto a neighborhood of f (z0 ). That is, f has a
continuous inverse function in a neighborhood of f (z0 ). We will prove in Corollary 4.17 that
the inverse is analytic.
σ
f
α f(z0) f(σ)
γ α
z0
f(γ)
Figure 3.1 Conformality.
a(z − z0 )F (z) zn + f (z0 )
t tn
z0 0 f (z0 )
(a) (b) (c)
Figure 3.2 Local behavior of an analytic function.
Returning to our local analysis, if f ² (z0 ) ± = 0 and if |z − z0 | is small then by (3.3) the map
f approximately translates by −z0, dilates by the factor | f ² (z0)| and rotates by arg f ²(z0 ), then
translates by f (z0 ), and this approximation becomes more and more accurate as | z − z0| → 0.
To put it another way, if γ and σ are two curves passing through z0 with angle α from γ
to σ , then f (γ ) and f (σ ) will be curves passing through f (z0 ) and the angle from f (γ ) to f (σ )
will also be equal to α. See Figure 3.1.
Definition 3.6 We say that f is locally conformalif it preserves angles (including direction)
between curves.
An analytic function f with f ² ± = 0 on a region ± is locally conformal on ±, though not

necessarily one-to-one. Some books use the term “conformal” instead of locally conformal.
But modern usage of the phrase “conformal map” usually means a one-to-one and analytic
function. For that reason we add the qualifier “locally.”
We can now picture the local correspondence given by an analytic function. By repeated
division as in Exercise 2.5, if f is analytic at z0 we can write
f (z) − f (z0) = an (z − z0 )ng(z),
where an is the first non-zero power series coefficient after a0 and g is analytic at z0 with
g(z0 ) = 1. Choose a so that an = an . By Exercise 2.10, we can define z n to be analytic in a
1
neighborhood of 1. Set F(z) = g(z) n . Then

1
f (z) = f (z0 ) + [a(z − z0)F(z)]n .
By Proposition 3.5, a(z − z0 )F(z) is one-to-one in a neighborhood of z0 .

The case when n = 3 is illustrated in Figure 3.2. The composed function is equal to f (z)
near z0 . Each of the three regions in Figure 3.2(a) is mapped one-to-one onto the slit disk in
Figure 3.2(c). Asymptotically (as the radius tends to 0) the map “looks” like z3, translated,
rotated and dilated. Note that Figure 3.2 was constructed right-to-left by taking preimages.
By the analysis above, if γ and σ are two curves passing through z0 with angle α < 2π/n
from γ to σ , then f (γ ) and f (σ ) will be two curves passing through f (z0) and the angle from
f (γ ) to f (σ ) will be nα .
Corollary 3.7 Suppose f is analytic at z0. Then

∞
±
f (z) − f (z0) = am (z − z0 )m
m =n
with an ± = 0 if and only if, for ε sufficiently small, there exists δ > 0 so that f (z) − w has
exactly n distinct roots in {z : 0 < | z − z0 | < ε }, provided 0 < |w − f (z0)| < δ.
The condition in Corollary 3.7 states that f (Bε (z0)) covers Bδ (w 0) \ {w0 } exactly n times,
where Br (ζ ) is the ball centered at ζ with radius r, and w0 = f (z0). In particular, f is one-to-
one in a neighborhood of z0 if and only if f ² (z0) ± = 0.
3.3 Growth on C and D

In this section we will use the maximum principle to draw some conclusions about the growth
of analytic functions defined on the plane C or on the open unit disk D = {z : |z| < 1}.
Corollary 3.8 (Liouville’s theorem) If f is analytic in C and bounded, then f is constant.
Proof Suppose | f | ≤ M < ∞ . Set g(z) = ( f (z) − f (0))/z. Then g is analytic and |g| →0
as |z| → ∞. By the maximum principle, g ≡ 0 and hence f ≡ f (0).
Here is a typical use of Liouville’s theorem. If p is a polynomial of degree n, then | p(z)| ≤

C |z| n for sufficiently large | z|. If f is analytic in C, and if | f (z) | ≤ C| z| n for | z| > M, then let
∑
p(z) = nk=0 ak zk be the terms of the power series expansion of f at 0 up to degree n. Then
g(z) = ( f (z) − p(z))/zn is analytic in C and bounded. By Liouville’s theorem, g is constant
and hence f must be a polynomial. In fact, f = p since g(0) = 0.
Corollary 3.9 (Schwarz’s lemma) Suppose f is analytic in D and suppose | f (z)| ≤ 1 and
f (0) = 0. Then
| f (z)| ≤ |z|, (3.7)
for all z ∈ D, and

| f ²(0)| ≤ 1. (3.8)
Moreover, if equality holds in (3.7) for some z ± = 0 or if equality holds in (3.8), then f (z) = cz,
where c is a constant with | c| = 1.
In some sense, Schwarz’s lemma says that a bounded analytic function cannot grow too fast
in the disk.
Proof By Exercise 2.5, the function g given by

⎧
⎪ f (z)
⎪
⎨ , if z ∈ D \ { 0}
g(z) =
z
⎪
⎪
⎩ f ² (0), if z = 0
is analytic in D and, for 0 < r < 1,
1
sup |g(z)| ≤ .
| z|= r r
Fix z0 ∈ D, then for r > | z0| the maximum principle implies |g(z0)| ≤ r1 , so that, letting
r → 1, we obtain (3.7) and (3.8). If equality holds in (3.7) at z0 or holds in (3.8) then g(z) has
a maximum at z0 or 0 and hence is constant.
Corollary 3.10 (invariant form of Schwarz’s lemma)Suppose f is analytic in D = {z :

|z| < 1} and suppose | f (z)| < 1. If z, a ∈ D then
³ ³ ³ ³
³ f (z) − f (a) ³ ³ z − a ³
³ ³≤³ ³ (3.9)
³ 1 − f (a)f (z) ³ ³ 1 − az ³
and
| f ² (z)| ≤ 1 . (3.10)
1 − | f (z)|2 1 − |z|2
Proof If |c| < 1, the function

z−c
T c (z) =
1 − cz
is analytic except at z = 1/c. Moreover, for z = eit ∈ ∂ D,
³ it ³
|Tc (e )| = ³ 1 − ceit ³³ = ||ee−it−−cc||
³ e −c ³ it
it ³ = 1.
By the maximum principle (or direct computation), |T c | ≤ 1 on D. Setting c = f (a), the
composition Tc ◦ f is analytic on D and bounded by 1. Furthermore,
´ µ´ µ
T c ◦ f (z) f (z) − f (a) 1 − az
= z −a
T a(z) 1 − f (a)f (z)
is analytic on D, and
³ ³
³ T c ◦ f (z) ³
lim sup ³³ ³ = lim sup | Tc ◦ f (z)| ≤ 1.
| z|→1 Ta (z) ³ |z|→1
By the maximum principle, (3.9) holds. Inequality (3.10) follows by dividing both sides of
(3.9) by |z − a| and letting z → a.
An alternative proof is to apply Schwarz’s lemma to g(w) = Tc ◦ f ◦ T−a (w) then set
w = T a(z). The details are almost the same.
Corollary 3.11 If f is analytic on D, | f | ≤ 1 and f (zj ) = 0, for j = 1, . . . , n, then

n ´ µ
¶ z − zj
f (z) = g(z),
1 − zj z
j=1
where g is analytic in D and |g(z)| ≤ 1 on D.

Proof If f (a) = 0, then, by the proof of Corollary 3.10, g(z) = f (z)/T a(z) is analytic in the
disk and bounded by 1. Repeating this argument n times proves the corollary.
By the uniqueness theorem, if f is analytic on D, then the zeros of f do not cluster in D,

unless f is identically zero. If we have more restrictions on f , such as boundedness, then the
zeros cannot approach the unit circle too slowly by the next corollary.
Corollary 3.12 If f is non-constant, bounded and analytic in D, and if {zj } are the zeros of
f , then
±
(1 − |zj |) < ∞ .
j
The convention we adopt here is that if zj is a zero of order k, then (1 − |zj | ) occurs k times
in the sum in the statement of Corollary 3.12.
Proof We may suppose | f | ≤ 1, by dividing f by a constant if necessary. If f (0) ± = 0, then,

using the notation of the proof of Corollary 3.11,
⎛ ⎞
¶n ¶
n
| f (0)| = ⎝ |zj | |g(0)| ≤ |zj |,
⎠
j =1 j=1
so that, by taking logarithms (base e),
1 ±
n
1 ±
n
ln
| f (0)| ≥ ln
|zj | ≥ (1 − |zj |).
j=1 j=1
If f (0) = 0, then write f (z) = zk h(z), where h(0) ± = 0. Applying the preceding argument to h,
we obtain
±
n
1
(1 − |zj |) ≤ ln + k.
j =1
|h(0)|
The corollary follows by letting n → ∞ .
Much of what appears in this book takes place on D or on C, which look like rather special
domains, but we know that a power series converges on a disk, and, by translating and scaling
the domain, we can assume it is D or C. Also, in Chapter 15 we shall prove the uniformization
theorem which says that in some sense the only analytic functions we need to understand can
be defined on D or C.
3.4 Boundary Behavior
We conclude this chapter with some examples and a theorem about boundary behavior of
analytic functions on the unit disk.
The first example is
z +1
I(z) = e z−1 .
(See Exercise 2.6 for the definition of ez .) K. Hoffman called this the “world’s greatest func-
tion.” Since I is the composition of an analytic function on C \{1} and the exponential function,
which is analytic in C, I is analytic on C \ {1} by Theorem 2.10. Moreover, |ez | = eRez so
that by a computation
|z |2 −1
|I(z)| = e | − | . z 12
Thus |I(z)| ≤ 1 on D. On the unit circle I(eit ) = e−i cot(t 2), for 0 < t < 2π . In particular, if
/
ζ ∈ ∂ D \ {1 } then
lim I(z)
z→ζ
r +1
exists and has absolute value 1. However, for 0 < r < 1, I(r) = e r−1 → 0 as r → 1. On the
unit circle, I(eit ) is spinning rapidly as t → 0. Hence I(z) does not have a limit as z ∈ D → 1.
The proper way to take limits in the disk is through cones. For ζ ∈ ∂ D and α > 1, define
²α (ζ ) = {z ∈ D : |z − ζ | ≤ α(1 − |z|)}
to be a Stolz angleor non-tangential coneat ζ . See Figure 3.3.
The precise shape of ²α is not important except that it is symmetric about the line segment
[0, ζ ] and forms an angle less than π at ζ . For z ∈ ²α (1) we have
−(1−|z |) (1+|z|)
|I(z)| = e | − | | − | ≤ e− | − | → 0
1
z 1 z 1 α z 1
as z ∈ ² (1) → 1. Thus I(z) → 0 in every cone, and ∪ ² (1) = D, but there is still no
α α α
limit as z ∈ D → 1. The image of the disk by the function I can be better explained once we
understand linear fractional transformations. See Sections 6.1 and 6.6.
1 − |z |
ζ
z |z − ζ|
Γα (ζ )
0
2 sec−1 α
Figure 3.3 Stolz angle, ²α( ζ ).

3.4 Boundary Behavior 39
Another example is the function given by

∞
±
L(z) =
n
z2 .
n=0
This series converges uniformly and absolutely on compact subsets of D. It is sometimes

called a lacunary series because the spacing between non-zero coefficients increases as n
increases. The function L does not extend continuously to any point of ∂ D. For example, if
ζ
2k
= 1 then
k−1
± ∞
±
L(rζ ) = +
n n
(rζ )2 r2 .
n=0 n=k
The second sum is clearly positive and increasing to ∞ as r → 1. Since the 2k roots of 1 are
evenly spaced around ∂ D, if eit ∈ ∂ D, then we can find a ζ as close to eit as we like with
ζ 2 = 1, for some k. Thus, in any neighborhood in D of eit , f is unbounded.
k
The next theorem gives a connection between Fourier series and analytic functions in D.
∑
Theorem 3.13 (Abel’s limit theorem)If ∞ n=0 a ne
int converges for some ζ = eit ∈ ∂ D,
∑∞
then f (z) = n=0 anz converges for all z ∈ D. Moreover, if ² = ²α (ζ ) is any Stolz angle at
n
ζ then
∞
±
lim f (z) = aneint . (3.11)
z∈ ²→ζ
n=0
The convergence in Theorem 3.13 is called non-tangential convergence. Abel’s limit the-
orem says that you get what you would expect for a limit, but only non-tangentially. Note that
the limit does not depend on which Stolz angle we choose. When written as a power series,
the function
∞ 1·
± ¸
f (z) = − (z2) 3
n n
z3
n
n=1
converges at z = 1 and hence converges in | z| < 1 by the root test. By Abel’s limit theorem,
f has non-tangential limit 0 at 1. Set ζ k = eiπ/3 . Then | f (rζ k )| → ∞ as r ↑ 1, so we may
k
choose zk = rk ζ k → 1 so that | f (zk )| → ∞. This says that we cannot conclude unrestricted

convergence in Abel’s limit theorem.
Proof By the root test, the series for f converges in D since it converges at a point on the
unit circle. Replacing f (z) by f (ζ z), we may suppose ζ = 1, and by subtracting a constant
∑
from a0 we may suppose ∞ n=0 a n = 0. Suppose z ∈ ²α (1). Then
f (z)
∞
± ∞
± ∞ ¹±
± n
º
1−z
= z n
ak z k
= a k zn .
n=0 k=0 n=0 k=0
Set sn = ∑nk=0 ak . Then

³N−1 ³ ³∞ ³
³± ³ ³± ³
³ n³ ³ n³
| f (z)| ≤ |1 − z| ³³ sn z ³³ + |1 − z| ³³ snz ³³ .
n=0 n=N
Given ε > 0, there exists N < ∞ so that |sn| < ε for n ≥ N. Thus
−1 ∞ −1
|sn | + |1 1−−z|||zz||
N
± ± N
± Nε
| f (z)| ≤ |1 − z| |sn | + |1 − z|ε |z|n = |1 − z| .
n=0 n=N n=0
We can choose δ > 0 so that, if z ∈ ²α (1) and | z − 1| < δ ,
| f (z)| ≤ ε + αε.
Since ε > 0 was arbitrary,
lim
z∈ ²α →1
| f (z)| = 0.
∑
For example, the series ∞ n=1 z /n converges at z = −1 by the alternating series test. By
n
Taylor’s theorem (or by integrating the derivative on [0, x]), this series converges to − ln(1 − x)
∑
for −1 < x < 1. By Abel’s limit theorem, ∞ n=1 (−1) /n = − ln 2.
n
A Fourier series is a function of the form

∞
±
F(t) = an eint .
n=−∞
∑N
We say F converges at t if n=−M an eint converges as M, N → +∞ independently. If F(t)
converges, then |an | → 0 as |n| → ∞. Thus
∞
± ∞
±
f (z) = an zn and g(z) = a−n zn
n=0 n=1
converge and are analytic on D by the root test. By Abel’s limit theorem, f (reit ) + g(re it )
converges to F(t) at each t, where the series for F converges. Thus the function f + g “extends”
F to D, and the infinitely differentiable functions defined on [0, 2π ] by
fr (t) = f (reit ) and gr (t) = g(reit )

satisfy
fr + gr → F
as r → 1, provided the series for F converges at t.

Fourier series arose from attempting to solve certain differential equations. Each square
integrable function on [0, 2π ] has a Fourier series, and it was a famous problem for many
years to prove that the Fourier series converges almost everywhere on [0, 2π ]. The proof that
was eventually found remains perhaps one of the hardest proofs in analysis, but it used this
connection between Fourier series and analytic functions on the disk, and the relation between
convergence on the circle and nontangential convergence.
3.5 Exercises 41
3.5 Exercises
3.1 (a) Show geometrically why the maximum principle holds using a “walking the dog”
argument. Make it rigorous by imitating the last half of the proof of the fundamental
theorem of algebra.
(b) Use the maximum principle to prove the fundamental theorem of algebra by
applying it to 1/p.
3.2 (a) Prove Corollary 3.2.
(b) Prove the alternative form (3.6) of Corollary 3.2.
3.3 Suppose f is analytic in a connected open set U. If | f (z)| is constant on U, prove that f
is constant on U. Likewise, prove that f is constant if Ref is constant.
3.4 Suppose f and g are analytic in C and | f (z)| ≤ |g(z)| for all z. Prove that there exists a
constant c so that f (z) = cg(z) for all z.
3.5 Prove that if f is non-constant and analytic on all of C then f (C) is dense in C.
3.6 Let f be analytic in D and suppose | f (z)| < 1 on D. Let a = f (0). Show that f does not
vanish in {z : |z| < |a|}
3.7 Prove that if f is a one-to-one (two-to-two!) analytic map of an open set ± onto f (±),
and if zn ∈ ± → ∂ ±, then f (zn) → ∂ f (±), in the sense that f (zn ) eventually lies outside
each compact subset of f (±). A function with this property is called proper.
3.8 (a) Prove that ϕ is a one-to-one analytic map of D onto D if and only if
´ µ
z− a
ϕ (z) = c ,
1 − āz
for some constants c and a, with |c| = 1 and |a| < 1. What is the inverse map?
(b) Let f be analytic in D and satisfy | f (z)| → 1 as | z| → 1. Prove that f is rational.
3.9 The pseudohyperbolic metricon D is defined by
³ ³
³ z −w ³
ρ (z, w) = ³ ³
³ 1 − w̄z ³ .
The hyperbolic metricon D is given by
´ µ
1 + ρ (z, w)
δ (z, w) = 21 ln 1 − ρ (z, w)
.
(a) Prove that if τ is a map of the form in Exercise 3.8(a) then

ρ (z, w) = ρ(τ (z), τ (w)).
(b) Prove the identity
(1 − |z|2)(1 − |w| 2)
1 − ρ 2(z, w) =
|1 − wz |2 .
Parts (a) and (b) should be used as much as possible to reduce the work in
subsequent parts of this exercise.
(c) Prove ρ (|z| , |w| ) ≤ ρ (z, w) ≤ ρ (|z|, −|w|).

(d) Prove ρ and δ are metrics on D. Hint: To prove the triangle inequality for ρ , assume
one point is 0 and use (c). Then apply (a). The identity 1 − r = (1 − r2 )/(1 + r) is
useful in proving that δ is a metric.
(e) Suppose f is analytic on D and | f (z)| ≤ 1. Schwarz’s lemma says that f is a
contraction in the pseudohyperbolic metric:
ρ ( f (z), f (w)) ≤ ρ (z, w).
Deduce the same fact for the hyperbolic metric.
The shortest curves or geodesics in the pseudohyperbolic geometry will be identi-
fied in Exercise 6.8, where it will be shown that the hyperbolic geometry does not
satisfy Euclid’s parallel postulate.
3.10 (a) Suppose p is a non-constant polynomial with all its zeros in the upper half-plane
H = {z : Imz > 0}. Prove that all of the zeros of p² are contained in H. Hint: Look
at the partial fraction expansion of p² /p.
(b) Use (a) to prove that if p is a polynomial then the zeros of p² are contained in the
(closed) convex hull of the zeros of p. (The closed convex hull is the intersection of
all half-planes containing the zeros.)
3.11 Suppose f is analytic in D and | f (z)| ≤ 1 in D and f (0) = 1/2. Prove that | f (1/3)| ≥
1/5.
3.12 Suppose f is analytic and non-constant in D and | f (z)| ≤ M on D. Prove that the number
of zeros of f in the disk of radius 1 /4, centered at 0, does not exceed
³ ³
1 ³ M ³
ln ³³ ³.
ln 4 f (0) ³
3.13 Suppose f is bounded and analytic in the right half-plane {z : Rez > 0}, and
lim supz →iy | f (z)| ≤ M for all iy on the imaginary axis. Prove that | f (z)| ≤ M on
the right half-plane. Check that f (z) = ez satisfies all the hypotheses above, except for
boundedness, and fails to be bounded in the right half-plane.
4 Integration and Approximation
In this chapter we prove several important properties of analytic functions including equiv-
alent ways of determining analyticity in terms of complex derivatives, integrals around
rectangles and uniform approximation by rational functions.
4.1 Integration on Curves
In this section we give the basic definitions for integration along curves in C.
Definition 4.1 A curve is a continuous mapping of an interval I ⊂ R into C.

Different curves can have the same image. For example, if γ (t) : [0, 1] → C then γ (t2 ) :
[0, 1] → C and both curves have the same image. We will also use the symbol γ to denote
the image or range γ (I) of a curve γ : I → C when it is clear from the context that we mean
a set in the plane, not a function. Arrows, as in Figure 4.1, show how a parameterization γ (t)
traces the image as t ∈ I increases.
Definition 4.2 (i) A curve γ is called an arc if it is one-to-one.

(ii) A curve γ : [a, b] → C is called closed if γ (a) = γ (b).
(iii) A closed curve γ : [a, b] → C is called simple if γ restricted to [a, b) is one-to-one.
A simple closed curve γ : [0, 2π ] → C can also be viewed as a one-to-one (two-to-two!)

continuous mapping of the unit circle given by ψ (eit ) = γ (t).
Definition 4.3 A curve γ (t) = x(t) + iy(t) is called piecewise continuously differentiable
if γ ± (t) = x± (t) + iy± (t) exists and is continuous except for finitely many t, and x± and y± have
one-sided limits at the exceptional points.
Figure 4.1 A curve, γ .

If γ is piecewise continuously differentiable then

± t2 ± t2
γ (t2 ) − γ (t1) = (x(t2) − x(t1 )) + i(y(t2) − y(t1 )) = x± (t)dt + i y±(t)dt.
t1 t1
Note that γ (t2 ) − γ (t1) corresponds to the vector from γ (t1 ) to γ (t2 ), so that
± (t1 ) = lim γ (t 2) − γ (t1 )

t2 − t1
γ
t2 →t 1
is tangent to the curve γ at t 1, provided γ ±(t1) exists.
Definition 4.4 A curve ψ : [c, d] → C is called a reparameterization of a curve γ :

[a, b] → C if there exists a one-to-one, onto, increasing function α : [a, b] → [c, d] such that
ψ (α(t)) = γ (t).
For example, ψ (t) = t2 + it 4, for 0 ≤ t ≤ 1, is a reparameterization of γ (t) = t + it 2,

for 0 ≤ t ≤ 1, with α(t) = t 2 . Any curve can be reparameterized to be defined on [0, 1]. If
1
σ : [0, 1] → C is a curve then the curve β , defined by β (t) = σ (1 − t), is not a reparameteriza-
tion of σ because 1 − t is decreasing. If ψ is a reparameterization of a piecewise continuously
differentiable curve γ with ψ (α(t)) = γ (t), where α is also piecewise continuously differen-
tiable, then ψ ± (α (t))α ±(t) = γ ±(t), by the chain rule applied to the real and imaginary parts, or
by taking limits of difference quotients.
Definition 4.5 If γ : [a, b] → C is a piecewise continuously differentiable curve, and if f is

a continuous complex-valued function defined on (the image of) γ , then
± ± b
f (z)dz ≡ f (γ (t))γ ± (t)dt.
γ a
The reader can check that a piecewise continuously differentiable reparameterization of γ

will not change the integral by the chain rule, and so the integral really depends on
² the image
of γ , not the choice of parameterization. For that reason we use the notation γ fdz. Note,
however, that the direction of the image curve is important.
Definition 4.6 If γ : [a, b] → C is a curve, then −γ : [ −b, −a] → C is the curve defined by
−γ (t) = γ (−t).
The curve −γ has the same geometric image as γ , but it is traced in the opposite direction.
If f is continuous on a piecewise continuously differentiable curve γ , then
± ±
f (z)dz = − f (z)dz.
−γ γ
If a closed piecewise continuously differentiable curve γ is split into a curve γ1 followed

by another curve γ2 , as in Figure 4.2, then
4.1 Integration on Curves 45
γ1
γ2
Figure 4.2 A closed curve, γ = γ1 + γ2 .
∂ (S 1 ∪ S 2 )
S1 S2
(a) (b)
Figure 4.3 Integrals around squares.
± ± ±
f (z)dz = f (z)dz + f (z)dz.
γ γ1 γ2
The integral is the same if we use the curve formed by γ2 followed by γ1 . It follows that the
integral around a closed curve of a continuous function does not depend on the choice of the
“starting point.”
For the purposes of computing integrals, it is useful to extend the notion of a curve to
allow finite unions of curves. If γ1, . . . , γn are curves defined on [0, 1], then we can define
γ : [0, n) → C by γ (t) = γj (t − j + 1) for j − 1 ≤ t < j, j = 1, . . . , n. If f is continuous on
(the image of) each γj , and if each γj is piecewise continuously differentiable, then
± n ±
³
f (z)dz = f (z)dz.
j=1
γ γj
∑
For this reason we define j γj ≡ γ . The associative and commutative laws hold for sums
(unions) of curves
² in this sense. We² do not require
² the union to be connected.
In particular, γ +(−γ ) f (z)dz = γ f (z)dz − γ f (z)dz = 0, if γ is piecewise continuously
differentiable. This idea can be used to simplify some integrals. For example, the integral
around two adjacent squares, each in the counter-clockwise direction, is equal to the integral
around the boundary of the union of the squares.
In Figure 4.3(a), the boundaries of the squares, ∂ S1 and ∂ S2 , are parameterized in the
counter-clockwise direction and
± ± ±
f (z)dz + f (z)dz = f (z)dz,
∂ S1 ∂ S2 ∂ (S1 ∪S2 )
for every continuous function f defined on ∂ S1 ∪ ∂ S2 . This can be seen by writing the integrals
around each square as the sum of integrals on the bounding line segments. The common
boundary edge is traced in opposite directions, so the corresponding integrals will cancel. A
similar argument applies to a finite union of squares, so that, after cancellation, the sum of
the integrals around the boundaries of all the squares in Figure 4.3(b) is equal to the integral
around the boundary of the union of the squares.
For the most part, we will deal with finite unions of closed curves.
Definition 4.7 A cycle γ = ∑nj=1 γj is a finite union of closed curves γ1, . . . , γn .

²
Another reason for using the notation γ fdz is the following. Suppose γ : [a, b] → C is a
piecewise continuously differentiable curve, and suppose a = t 0 < t1 < t2 < · · · < tn = b.
Set γ (tj ) = zj . Then
³
n −1 n−1
³
f (zj)(zj+1 − zj ) = f (γ (tj ))[γ (tj+1 ) − γ (t j )]
j=0 j=0
n−1
³
≈ f (γ (tj ))γ ± (t j )[tj+1 − t j ]. (4.1)
j=0
²
The left-hand side looks like a Riemann sum for γ f (z)dz with independent variable z, and
the last sum is a Riemann sum, using the independent variable t, for
± b
f ( γ (t))γ ± (t)dt.
a
²
The left-hand side of (4.1) also converges to γ f (z)dz as the mesh µ({tj }) = maxj (t j+1 − tj )
of this partition tends to 0.
A closely related notion is integration with respect to arc-length.
Definition 4.8 If γ : [a, b] → C is a piecewise continuously differentiable curve, and if f is

a continuous complex-valued function defined on (the image of) γ , then we define
± ± b
f (z)|dz | = f (γ (t))|γ ± (t)|dt.
γ a
Thus
´± ´ ´´± b ´
´
´ ´ ´
´ f (z)dz´ = ´ f (γ (t)) γ (t)dt´´
±
´ γ ´ ´ a ´
± b ±
≤ | f (γ (t))||γ ± (t)|dt = | f (z) ||dz|.
a γ
Note that if zj = γ (t j) then

³
f (zj )| zj +1 − zj |
² ²b
is approximately a Riemann sum for γ
f (z)|dz | = a f (γ (t))|γ ± (t)|dt.
Definition 4.9 If γ : [a, b] → C is a piecewise continuously differentiable curve then the

length of γ is defined to be
± ± b
±(γ ) = |γ | = |dz| = |γ ± (t)|dt.
γ a
The following important estimate follows immediately from the definitions because all
piecewise continuously differentiable curves have finite length. If γ is piecewise continuously
differentiable, and if f is continuous on γ , then
´± ´ µ ¶
´ ´
´ f (z)dz´ ≤ sup | f (z)| ±(γ ). (4.2)
´ γ ´ γ
Consequently, if fn converges uniformly to f on γ , then

± ±
lim fn (z)dz = f (z)dz.
n γ γ
To prove this, just use the following estimate:

´± ´ µ ¶
´ ´
´ ( fn − f )dz´ ≤ sup | fn − f | ± (γ ).
´ γ ´ γ
² Finally, we note that

² integration
² is linear on piecewise continuously
² differentiable
² curves:
γ
( f (z) + g(z))dz = γ f (z)dz + γ g(z)dz, and if C is constant then γ
Cf (z)dz = C γ
f (z)dz,
for continuous functions f and g.
4.2 Equivalence of Analytic and Holomorphic
Definition 4.10 A complex-valued function f is said to be holomorphic on an open set U if

f (w) − f (z)
f ± (z) = lim
w→z w−z
exists for all z ∈ U and is continuous on U. A complex-valued function f is said to be
holomorphic on a set S if it is holomorphic on an open set U ⊃ S.
There are various a priori weaker conditions for analyticity. For example, many books do
not require continuity of the derivative in the definition of a holomorphic function. In almost
every situation encountered in practice, however, verifying that the derivative is continuous,
once you have proved it exists, is not hard. Exercise 4.12 removes the requirement that the
derivative is continuous. An important advance in partial differential equations was to consider
“weak” derivatives in the sense of distributions. Indeed, it led to the development of functional
analysis. See Exercise 7.13 for the corresponding definition of “weakly-analytic.”
As we saw in Section 2.5, analytic functions are holomorphic. In particular, polynomials
are holomorphic. A rational function is holomorphic except where the denominator is zero.
Linear combinations of holomorphic functions are holomorphic. The reader is invited to verify
that the chain rule for complex differentiation holds for the composition of two holomorphic
functions, and so the composition of two holomorphic functions is holomorphic, wherever the
composition is defined.
It also follows from the usual chain rule applied to real and imaginary parts that if γ :
[a, b] → C is a piecewise continuously differentiable curve, and if f is holomorphic on a
neighborhood of γ , then f ◦ γ is a piecewise continuously differentiable curve, and
d
f (γ (t)) = f ± (γ (t))γ ± (t),
dt
except at finitely many points t1 , . . . , tn . Then, by the fundamental theorem of calculus,

± ± b ± b d
f ± (z)dz = f ± (γ (t))γ ± (t)dt = f (γ (t))dt = f (γ (b)) − f (γ (a)).
γ a a dt
For example, the line segment from z to ζ can be parameterized by γ (t) = z + t(ζ − z), for
t ∈ [0, 1], so that, if f is holomorphic in a neighborhood of γ , then
± 1
f (ζ ) − f (z) = f ± (z + t(ζ − z))(ζ − z)dt. (4.3)
0
Corollary 4.11 If γ : [a, b] → C is a closed, piecewise continuously differentiable curve,

and if f is holomorphic in a neighborhood of γ , then
±
f ± (z)dz = 0.
γ
Proof Because γ is closed, f (γ (b)) − f ( γ (a)) = 0.

²
For example, if γ is a closed curve then γ
p(z)dz = 0 for every polynomial p.
∑
Corollary 4.12 If f (z) = ∞ n=0 a n(z − z0 ) converges in B = {z : | z − z0|
n < r}, and if γ ⊂B
is a closed, piecewise continuously differentiable curve, then
±
f (z)dz = 0.
γ
Proof Corollary 4.12 follows immediately from Corollaries 2.15 and 4.11.
Much of this chapter and the next center around extending Corollary 4.12 to larger sets than
disks B and more general curves.
If γ is a piecewise continuously differentiable closed curve and a ∈/ γ then, by
Corollary 4.11, for n ² = 1,
±
1
dz = 0.
γ (z − a)
n
By the partial fraction expansion, Corollary 2.3, in order to integrate a rational function along
² −1
γ we need only to be able to compute γ (z − a) dz for various values of a. The next example
will be key to understanding integrals of analytic functions. If r > 0 set
Cr = {z0 + reit : 0 ≤ t ≤ 2π }.
Then we have the following proposition.
Proposition 4.13
± ·
1 1 1, if |a − z0 | < r
dz =
2π i Cr z −a 0, if |a − z0 | > r.
Proof Suppose | a − z0 | < r. Then Cr± (t) = ireit and

± ± 2π
1 1 1 1
dz = it
2π i Cr z −a 2π i 0 reit − (a − z0 ) ire dt
± 2π
= 21π 1
−z0 ) dt
0 1 − ( areit
± 2π ∞ ¸ a − z ¹n
³
= 21π 0
reit
dt
n=0
0
∞ (a − z )n 1 ±
³ 2π
= e−int dt = 1.
0
rn 2π
n=0 0
Interchanging the order of summation and integration is justified because | (a − z0 )/(reit )| < 1
implies uniform convergence of the series.
If |a − z0 | > r, then write
¸ ¹ ∞ rn eint
³
reit reit 1
reit − (a − z0)
= z0 − a
=− (a − z0) n
,
1 − are−z0
it
n=1
so that
± ± 2π
1 1 1 reit
dz = dt
2π i Cr z−a 2π 0 reit − (a − z0 )
∞
³ rn 1
± 2π
=− (a − z0 )n 2π
eint dt = 0.
n=1 0
An immediate consequence of Corollary 4.11 and Proposition 4.13 is that there is no

function f defined in a neighborhood of ∂ D satisfying f ± (z) = 1/z.
Theorem 4.14 If f is holomorphic on {z : | z − z0 | ≤ r} then, for |z − z0| < r,

±
1 f (ζ )
f (z) =
2π i Cr ζ − z dζ ,
where Cr is the circle of radius r centered at z0 , parameterized in the counter-clockwise
direction.
Theorem 4.14 shows that it is possible to find the values of a holomorphic function inside
a disk from the values on the bounding circle.
Proof By (4.3) and Corollary 4.11, for |z − z0| < r,

± ± ±
f (ζ ) − f (z) 1
dζ = f ± (z + t(ζ − z))dtdζ
Cr ζ −z Cr 0
± 1±
= f ± (z + t(ζ − z))d ζ dt
0 Cr
± 1±
d
= lim
→0
ε dζ
f (z + t(ζ − z))dζ dtt = 0.
ε Cr
Thus
± ±
1 f (ζ )
2π i ζ −z
dζ = f (z) · 2π1 i dζ
− z = f (z),
Cr Cr ζ
by Proposition 4.13.
A consequence of Theorem 4.14 is the converse to Theorem 2.12.
Corollary 4.15 A complex-valued function f is holomorphic on a region ² if and only if f is

analytic on ². Moreover, the series expansion for f based at z0 ∈ ² converges on the largest
open disk centered at z0 and contained in ².
See Exercise 2.9 for a proof of the second statement in Corollary 4.15 for rational functions.
Proof If f is analytic in ² then f is holomorphic in ², by Theorem 2.12. To prove the

converse, suppose f is holomorphic on {z : |z − z0 | ≤ r }. If |z − z0| < r, then, by (2.9),
± ± µ³
∞ ¶
1 f (ζ ) 1 1
f (z) = dζ = (z − z0 )n f (ζ )dζ
2π i ζ −z 2π i (ζ − z0 )n+1
Cr Cr n=0
∞¸
³ 1
±
f (ζ )
¹
= 2π i (ζ − z0)n+1
dζ (z − z0 )n .
n=0 Cr
Interchanging the order of the summation and integral is justified by the uniform convergence
in ζ ∈ Cr of the series for z fixed. Thus f has a power series expansion convergent in {z :
|z − z0 | < r }, provided the closed disk is contained in ². By Theorem 2.7, f is analytic
in ².
In particular, if f is analytic in C then f has a power series expansion which converges

in all of C. Such functions are called entire. From now on, we will use the words analytic
and holomorphic interchangably. Note that, when interchanging the order of summation and
integration it is useful to think of Cr as a set rather than a function.
The Bernoulli numbers Bn are given by
z
∞ B
³
ez − 1
= n
n!
zn .
n=0
By Corollary 4.15 this series converges in {|z| < 2π } and no larger disk, so that, by the root
test,
¸
|Bn | ¹ = 1 .
1
n
lim sup
n→∞ n! 2π
The rate of convergence of the series on | z| = r < 2π can also be deduced from this estimate.
The proof of Corollary 4.15 yields a bit more information. Not only can we find the values
of an analytic function inside a disk from its values on the boundary, but also we have a
formula for each of its derivatives in the disk.
Corollary 4.16 If f is analytic in {z : |z − z0 | ≤ r} and Cr(z0 ) = {z0 + reit : 0 ≤ t ≤ 2π },

then
±
f (n) (z0) 1 f (ζ )
n!
= 2π i (ζ − z0 )n+1 dζ (4.4)
Cr (z 0 )
and
´ (n) ´
´ f (z0 ) ´ sup Cr (z0) | f |
´ n! ´ ≤
´ ´ . (4.5)
rn
Inequality (4.5) is called Cauchy’s estimate. We will show in Lemma 4.30 that (4.4) holds
for all z ∈ Cr (z0).
Proof Equation (4.4) follows from Corollary 2.13, the proof of Corollary 4.15, and the
uniqueness of series, Theorem 2.8. Inequality (4.5) follows from (4.4) by using inequality
(4.2).
The equivalence of analytic and holomorphic makes it easy to prove that the inverse of a
one-to-one analytic function is analytic.
Corollary 4.17 If f is analytic and one-to-one in a region ² then the inverse of f , defined
on f (²), is analytic.
Proof Since analytic functions are open by Corollary 3.3, f has a continuous inverse. Take
z0 ∈ ² and set w0 = f (z0 ). Then f (²) contains a disk centered at w0. If w ∈ f (²) tends to w0 ,
then z = f −1 (w) tends to z0 . By Corollary 3.7, f ± (z0 ) ² = 0, so that
f −1 (w) − f −1 (w0 )
w − w0
= f (z)z −− zf 0(z ) → f ± (z1 ) .
0 0
This proves, f −1 has a complex derivative at w0 equal to 1/f ± ( f −1(w0 )). This derivative is
continuous, so f −1 is holomorphic and hence analytic.
As an application of the second sentence of Corollary 4.15, we obtain a local version of a

theorem we will encounter in the next chapter, called Cauchy’s theorem.
Corollary 4.18 If f is analytic in an open disk B, and if γ ⊂ B is a closed, piecewise

continuously differentiable curve, then
±
f (z)dz = 0.
γ
Proof By Corollary 4.15, f has a power series expansion which converges on all of B. Now
apply Corollary 4.12.
Morera’s theoremis a useful converse to Corollary 4.18.

Theorem 4.19 (Morera) If f is continuous in an open disk B, and if

±
f (ζ )dζ =0
∂R
for all closed rectangles R ⊂ B with sides parallel to the axes, then f is analytic on B.
Proof We may suppose B = D. Define

±
F(z) = f ( ζ )dζ ,
γz
where γz is a curve from 0 to z consisting of a horizontal line segment followed by a vertical

line segment.
If |h| < 1 − |z| then γz +h = γz + σ + ∂ R, where σ is a curve from z to z + h consisting of a
horizontal line segment followed² by a vertical line segment and R ⊂ B is a closed rectangle.
See Figure 4.4. By assumption, ∂ R f (ζ )dζ = 0, so that
± ± ±
F(z + h) − F(z) = f (ζ )dζ − f (ζ )dζ = f (ζ )dζ .
γz h + γz σ
By
² the fundamental theorem of calculus, since the identity function has derivative equal to 1,
σ
dζ = z + h − z = h, and so
±
F(z + h) − F(z)
h
− f (z) = 1h ( f (ζ ) − f (z))dζ .
σ
By (4.2),
´ ± ´ √
´1 ´
´
´h ( f (ζ ) − f (z))d ζ ´ ≤ 2 sup | f (ζ ) − f (z)|,
´
σ ζ σ∈
√
because |σ | ≤ 2|h|. Since f is continuous, letting h → 0 proves that F is holomorphic on B
with F ± = f . By Corollary 4.15, F is analytic on B, and, by Theorem 2.12, f = F ± is analytic
on B. Finally, apply Corollary 4.11.
One consequence of Morera’s theorem is that the definition of holomorphic does not need
to include the continuity of the derivative. See Exercise 4.12.
z+h
σ
z
R
Figure 4.4 Proof of Morera’s theorem.

4.3 Approximation by Rational Functions
In this section we will show that Theorem 4.14 also holds if the circle Cr is replaced by the
boundary of a square, and then use it to prove Runge’s theorem that analytic functions can be
uniformly approximated by rational functions.
Proposition 4.20 If S is an open square with boundary ∂ S parameterized in the counter-

clockwise direction then
± º
1 1 1, if a ∈ S
dz =
2π i ∂S z−a 0, if a ∈ C \ S.
Proof If a ∈ C \ S, then we can find a disk B which contains S and does not contain a. See
Figure 4.5.
By (2.9) and Corollary 4.18,
±
1
dz = 0.
∂S z−a
If a ∈ S, then let C be the circumscribed circle to ∂ S parameterized in the clockwise direction.
See Figure 4.6. Then we can write
∂S = s1 + s2 + s3 + s4 ,
where sj , j = 1, . . . , 4, are the sides of ∂ S and
C = c1 + c4 + c3 + c2 ,
a
B S
Figure 4.5 If a is outside S.
c2
s2
c3 s3 s1 c1
a B1
s4
c4
Figure 4.6 The square S and its circumscribed circle C.
where cj , j = 1, . . . , 4, are the arcs of C subtended by the corresponding sides of ∂ S. Then

/ B j , for j = 1, . . . , 4. By Corollary 4.18,
sj + cj is a closed curve contained in a disk Bj with a ∈
±
1
dz = 0, (4.6)
s j +cj z − a
for j = 1, . . . , 4. Adding the four integrals (4.6) we obtain

± ± ±
1 1 1
dz + dz = dz = 0.
∂S z− a C z −a ∂S +C z − a
But we also have, by Proposition 4.13,
±
1
dz = −2π i,
C z −a
because −C is the circle parameterized in the counter-clockwise direction. This proves
Proposition 4.20.
Proposition 4.20 can also be proved by explicit computation, but we chose this proof
because the idea will be used later to compute the integral of 1/(z − a) for other curves.
Theorem 4.21 If f is analytic in a neighborhood of the closure S of an open square S, then,

for z ∈ S,
±
f (z) = 2π1 i f (ζ )
− z dζ ,
∂S ζ
where ∂ S is parameterized in the counter-clockwise direction.
Proof The proof of Theorem 4.21 is exactly like the proof of Theorem 4.14 except that
Proposition 4.20 is used instead of Proposition 4.13.
Corollary 4.22 If f is analytic in a neighborhood of the closure S of an open square S, then

±
1
2π i
f (ζ )dζ = 0.
∂S
Proof Fix z ∈ S and apply Theorem 4.21 to g(ζ ) = f (ζ )(ζ − z).

Theorem 4.23 (Runge) If f is analytic on a compact set K, and if ε > 0, then there is a
rational function r so that
sup | f (z) − r(z)| < ε.
z ∈K
Proof Suppose f is analytic on U open, with U ⊃ K. Let

d = dist(∂ U, K) = inf{|z − w| : z ∈ ∂ U, w ∈ K }.
Construct a grid of closed squares with side length d/ 2.

Figure 4.7 A union of closed squares covering K and contained in U.
√4.7. Note that each (closed)

Shade each square in the grid which intersects K. See Figure
shaded square is contained in U because each has diameter d/ 2. Let {Sk } be the collection
of shaded squares and let ³ denote the boundary of the union of the shaded squares,
( )
³ = ∂ ∪j Sj ,
formed from ∪∂ Sj , each parameterized in the counter-clockwise direction, by cancelling edges
which are traced in opposite directions. Then ³ ⊂ U is a cycle, a finite union of closed curves,
with ³ ∩ K = ∅. See Exercise 4.1. If z is in the interior of one of the squares, Sj0 , then
f (ζ )/(ζ − z) is analytic as a function of ζ on Sj , for j ² = j0 . Apply Theorem 4.21 to Sj0 and
Corollary 4.22 to all of the other squares, then cancel edges traced in opposite directions to
obtain ±
1 f (ζ )
f (z) = dζ . (4.7)
2π i ³ ζ − z
Both sides of (4.7) are continuous functions of z, for z ∈
/ ³, so that equality holds for all
z ∈ ∪Sk \ ³. Fix z0 ∈ K and write ³ = ∪ 1 ³j , where each ³j is a closed curve. Choose a
n
Riemann sum on each ³j for the integral on the right-hand side of (4.7), as in (4.1), to obtain
´ ´
´ ´
f (ζ j,k )(ζ j,k+1 − ζ j,k ) ´´
n ³ mj
´ ³
´ f (z0) − < ε.
´
´ 2π i(ζ j,k − z0) ´´
j=1 k=1
We can in fact choose the partition so that the inequality remains true for all refinements
of the partition. By uniform continuity, this inequality remains true for all z in a small disk
containing z0 and all refinements of the partition, if we replace ε with 4ε. See Exercise 4.5.
Cover K by finitely many such disks, and take a common refinement.
Definition 4.24 If r is a rational function, by the fundamental theorem of algebra we can

write r(z) = p(z)/ q(z), where p and q are polynomials with no common zeros. The zeros of q
are called the poles of the rational function r.
If b is a pole of r(z) then |r(z)| → ∞ as z → b. A rational function is analytic everywhere

in the plane, except at its poles.
It is possible to improve the statement of Runge’s theorem by restricting where the poles of
the rational function need to be placed using the next lemma.
Lemma 4.25 Suppose U is open and connected, and suppose b ∈ U. Then a rational func-
tion with poles only in U can be uniformly approximated on C \ U by a rational function with
poles only at b.
Proof Suppose a, c ∈ U and suppose | c − a| < dist(a, ∂ U). If z ∈ C \ U then |z − a| ≥

dist(a, ∂ U) > |c − a| , so that
1 1 1
∞ (c − a)n
³
z −c
= z − a − (c − a)
= −a )) = (z − a)n+1
, (4.8)
(z − a)(1 − ( zc−a n=0
where the sequence of partial sums approximates 1/(z − c) uniformly on C \ U. By taking

products we can also approximate (z − c)−n for n ≥ 1 on C \ U, and, by taking finite linear
combinations, we can uniformly approximate on C \ U any rational function with poles only
at c by rational functions with poles only at a.
Write c ∈ Rd if every rational function with poles only at c can be uniformly approximated
on C \ U by rational functions with poles only at d. This relation is transitive: if c ∈ Rd and
d ∈ Re then c ∈ R e. Set E = { a ∈ U : a ∈ R b}. By transitivity and the argument above, if
a ∈ E then E contains a disk centered at a with radius dist(a, ∂ U). Thus E is open. Moreover,
if an ∈ E converges to a∞ ∈ U, then we can choose n so large that, for all z ∈ C \ U,
|z − an| ≥ |z − a∞ | − |an − a∞ | ≥ dist(a∞, ∂ U) − |an − a∞ | > |an − a∞ |.

By (4.8), a∞ ∈ Ra and, by transitivity, a∞ ∈ E. This proves E is closed in U and, by
n
connectedness, E = U.
Now suppose that r is rational with poles only in U and fix b ∈ U. Each term 1 /(z − c)k in
the partial fraction expansion of r can be approximated by a rational function with poles only
at b. Adding the approximations gives an approximation of r by a rational function with poles
only at b.
Corollary 4.26 Suppose U is connected and open and suppose {z : |z| > R} ⊂ U for some
R < ∞. Then a rational function with poles only in U can be uniformly approximated on
C \ U by a polynomial.
Proof By Lemma 4.25, we need only prove that, if |b| > R, then a rational function with
poles at b can be uniformly approximated by a polynomial on C \ U. But
1 1 1
∞ » z ¼n
³
z−b
= z =−
−b(1 − ) b b
,
b n=0
where the sum converges uniformly on |z| ≤ R. As in the proof of Lemma 4.25, we can
approximate (z−b)−n for n ≥ 1, and, by taking finite linear combinations, we can approximate
any rational function with poles only at b by a polynomial, uniformly on {z : | z| ≤ R } ⊃
C \ U.
Theorem 4.23, Lemma 4.25 and Corollary 4.26 combine to give the following improvement
of Runge’s theorem (“one pole in each hole”).
Theorem 4.27 (Runge) Suppose K is a compact set. Choose one point an in each bounded
component Un of C \ K. If f is analytic on K and ε > 0, then we can find a rational function
r with poles only in the set {an} such that
sup | f (z) − r(z)| < ε .
z ∈K
If C \ K has no bounded components, then we may take r to be a polynomial.

See Exercise 4.2(a) for the definition of components of an open set. For example, if K1 and
K 2 are disjoint compact sets such that C \ (K 1 ∪ K2 ) is connected and ε > 0, then we can find
a polynomial p so that |p| < ε on K1 and |p − 1| < ε on K2 because the function which is
equal to 0 on K1 and equal to 1 on K 2 is analytic on K1 ∪ K2.
Corollary 4.28 (Runge) If f is analytic on an open set ² ² = C then there is a sequence of

rational functions r n with poles in ∂ ² so that rn converges to f uniformly on compact subsets
of ².
Proof Set º ½
1
Kn = z ∈ ² : dist(z, ∂ ²) ≥ and |z| ≤ n .
n
Then Kn is compact, ∪ Kn = ² and each bounded component U, of C \ Kn contains a point of
∂ ² . Indeed ∂ U ⊂ Kn ⊂ ² so that U ∩ ² ² = φ. If z ∈ U ∩ ², then | z| < n and | z − ζ | < 1 /n
for some ζ ∈ ∂ ². Let L be the line segment from z to ζ . If α ∈ L then | α − ζ | < 1/n, so that
α ∈/ Kn . Thus L is a connected subset of C \ K n, so L must be contained in one component of
C \ Kn. Because z ∈ L ∩ U, we must have L ⊂ U. But then ζ ∈ L ∩ ∂ ² ⊂ U.
By Theorem 4.27 we can choose the rational functions approximating f to have poles only
in ∂ ².
The improvement of Corollary 4.28 over Theorem 4.23 is that the poles of rn are outside of
², not just outside the compact subset of ² on which rn is close to f .
Corollary 4.28 says that every analytic function is a limit of rational functions, uniformly
on compact subsets. Weierstrass proved that the set of analytic functions on a region is closed
under uniform convergence on compact sets.
Theorem 4.29 (Weierstrass) Suppose { f n} is a collection of analytic functions on a region

² such that fn → f uniformly on compact subsets of ². Then f is analytic on ². Moreover,
f n± → f ± uniformly on compact subsets of ².
Lemma 4.30 If G is integrable on a piecewise continuously differentiable curve γ , then

±
G(ζ )
g(z) ≡ dζ
γ ζ −z
is analytic in C \ γ and
±
G(ζ )
g± (z) = dζ .
γ (ζ − z)2
Proof There are at least two ways to prove this lemma. One way is to write out a power series
expansion for 1/(ζ − z) based at z0, where z0 ∈ / γ , then interchange the order of summation
and integration to obtain a power series expansion for g based at z0 . The derivative of g at z0
is the coefficient of z − z0 by Corollary 2.13. The second proof is to write
± ±
g(z + h) − g(z) G(ζ ) h
h
− (ζ − z) 2
dζ = G(ζ )
(ζ − z)2 (ζ − (z + h)) dζ ,
γ γ
which → 0 as h → 0. Thus
±
G(ζ )
g± (z) = dζ
γ (ζ − z)2
exists and is continuous on C \ γ . By Corollary 4.15, g is analytic on C \ γ .
A similar proof shows that

±
g(n) (z) G(ζ )
n!
= (ζ − z)n+1 dζ .
γ
So, by Theorem 4.14, formula (4.4) in Corollary 4.16 holds with z0 replaced by any z such
that | z − z0 | < r.
Proof of Theorem4.29 Analyticity is a local property, so to prove the first statement we

may suppose B is a disk with B ⊂ ². Then, by Theorem 4.14, if z ∈ B,
±
1 fn (ζ )
fn (z) =
2π i ∂B ζ − z dζ .
The limit function f is continuous. Set
±
F(z) = 21π i f (ζ )
ζ −z
dζ .
∂B
Then
| fn (z) − F(z)| → 0
for each z ∈ B, because fn → f uniformly on ∂ B. Thus F = f on B and, by Lemma 4.30, F is
analytic on B.
By Theorem 4.14 and Lemma 4.30,
±
fn± (z) = 2π1 i fn (ζ )
(ζ − z)2
dζ
∂B
and
±
1 f (ζ )
f ± (z) = F± (z) = dζ.
2π i ∂B (ζ − z)2
Again, since f n → f uniformly on ∂ B, we have that fn± converges uniformly to f ± on compact
subsets of B. Thus fn± converges uniformly to f ± on closed disks contained in ². Given a
compact subset K of ², we can cover K by finitely many closed disks contained in ² and
hence fn± converges uniformly on K to f ± .
This proof of Theorem 4.29, which implicitly uses the equivalence of analytic and
holomorphic, is easier than the original proof of Weierstrass, which used doubly indexed
sums.
The reason for using piecewise continuously differentiable curves γ is that we want to
integrate continuous functions. If f is continuous, then the left-hand side of (4.1) converges
²
to γ f (z)dz for piecewise continuously differentiable γ , but the limit can fail to exist if γ
is assumed to be continuous only. However, Corollary 4.18 actually allows us to define the
integral of an analytic function over any continuous curve. See Section 12.1 for some badly
behaved curves.
We first need an elementary but useful covering lemma.
Lemma 4.31 Suppose ² is a region and suppose γ : [0, 1] → ² is continuous. Given ε,

0 < ε < dist(γ , ∂ ² ), we can find a finite partition 0 = t 0 < t 1 < · · · < tn = 1 so that
γ ([tj−1 , tj ]) ⊂ Bj ≡ {z : |z − γ (tj )| < ε } ⊂ ².
Proof If 0 < ε < dist(γ , ∂ ²), then by the uniform continuity of γ , we can choose δ > 0
so that |γ (t) − γ (s)| < ε whenever |t − s| < δ . Then any finite partition {tj } with mesh size
µ ({ tj }) = supj |tj−1 − tj | < δ will satisfy the conclusion of Lemma 4.31.
∑
Using the notation of Lemma 4.31, let σ be the polygonal curve σj , where σj is straight-
line segment from γ (tj−1 ) to γ (t j ). Let γj be the subarc γ ([t j−1, tj ]) of γ . Then βj = γj − σj is a
closed curve contained in Bj . If γ²is piecewise ²continuously differentiable, and if f is analytic
on ², then, by Corollary 4.18, γj f (z)dz = σj f (z)dz. If γ is not piecewise continuously
differentiable, we use this as the definition of the integral.
Theorem 4.32 Suppose ² is a region and γ : [0, 1] → C is continuous with γ ⊂ ². Let σ

be the polygonal curve defined above. If f is analytic on ², define
± ±
f (z)dz = f (z)dz.
γ σ
²
Then this definition of γ f (z)dz does not depend on the choice of the polygonal curve σ and
it agrees with our prior definition if γ is piecewise continuously differentiable.
²
Proof As seen above, this definition of γ f (z)dz agrees with our prior definition if γ is piece-
wise continuously differentiable. For an arbitrary continuous γ , suppose σ is the polygonal
curve associated with a partition {tj } chosen as above. Let α be the polygonal curve associated
∑
with a finite refinement {sk } ⊃ {t j } of the partition {tj }. We can write σ − α = βj , where
each β j : [0, 1] → C is a closed polygonal curve contained in B j . By Corollary 4.18 again,
² ²
α
f (z)dz = σ f (z)dz. Given any two partitions satisfying the conditions of the definition of
the integral, we can find a common refinement. Thus the definition does not depend on the
choice of the partition.
²
Theorem 4.32 says that if we want to prove something about γ f (z)dz, where f is analytic
on a region ² and γ ⊂ ², then it is enough to prove it for all polygonal curves γ . Note that
the choice of the polygonal curve σ depends on γ and the region ² but not on the analytic
function f . For example, if fn is a sequence of analytic functions converging uniformly on
compact subsets of ² to f , then f is analytic by Weierstrass’s Theorem 4.29 and
± ±
lim fn (z)dz = f (z)dz
n→∞ γ γ
for every continuous curve γ ⊂ ², even if γ has infinite length. We simply apply (4.2) to the
analytic function f n − f on the polygonal curve σ which has finite length.
4.4 Exercises
4.1 (a) A finite union of boundaries of squares, oriented in the usual counter-clockwise
direction, is a cycle, by definition. Prove that if an edge in the union is traced twice,
in opposite directions, then, after removal of the common edge, the union is still a
cycle.
(b) The boundary of a finite union of squares is a cycle, oriented so that the region lies
on the left. Hint: Use (a) and induction.
4.2 (a) Let U be an open set in C. A polygonal curveis a curve consisting of a finite union
of line segments. Define an equivalence relation on the points of U as follows:
a ∼ b if and only if there is a polygonal curve contained in U with edges parallel to
the axes and with endpoints a and b. Show that each equivalence class is open and
closed in U and connected. Show that there are at most countably many equivalence
classes. The equivalence classes are called the components of U. For open sets in
C∗, allow polygonal curves to contain a half-line and obtain a similar result.
(b) Let K be a compact set. Define an equivalence relation on the points of K as follows:
a ∼ b if and only if there is a connected subset of K containing both a and b. Prove
that the equivalence classes are connected and closed. The equivalence classes are
called the (closed) components of K. There can be uncountably many (closed)
components. In both parts (a) and (b) the components are the maximal connected
subsets.
4.3 Prove the following version of Weierstrass’s theorem. Suppose {Un} is an increasing
sequence of regions and suppose f n is defined and analytic on Un. If the sequence fn
converges uniformly on compact subsets of U ≡ ∪nUn to a function f , then f is analytic
on U and the sequence f n± converges to f ± uniformly on compact subsets of U (even
though perhaps none of the fn is defined on all of U).
4.4 (a) Use Cauchy’s estimate (4.5) to prove Liouville’s theorem.
(b) Use Cauchy’s estimate (4.5) to compute a lower bound on the radius of convergence
of the power series representation of a bounded holomorphic function.
4.5 At the end of the proof of Runge’s Theorem 4.23, we stated that an inequality remains
true for all z in a disk containing z0 and for all refinements of the partition. Supply the
details. Hint: If I = [a, b] ⊂ ³ is a line segment, show that we can choose a partition
4.4 Exercises 61
I = ∪Ij = ∪n1[ ζj−1 , ζj ] with ζ0 = a and ζn = b such that

´ ´
´ f (ζ ) ´
´
´ ζ − z0 − ζf −(ζj z) ´ < ε,
´ (4.9)
j 0
for all ζ ∈ [ζ j−1, ζj ], j = 1, . . . , n. Then (4.9) holds for any refinement of ∪Ij by the
triangle inequality, if ε is replaced by 2ε. Moreover, if δ is sufficiently small then (4.9)
holds for all z with |z − z0| < δ, replacing 2ε with 4ε , because z0 ∈/ ³. Now use
´ ´ ´ ´
´± ³ f (ζj) ´ ´³ ± ¸ ¹ ´
´ f (ζ ) ´ ´ f (ζ ) f (ζj ) ´
´
´ I ζ − z dζ − ζj − z
(ζ j − ζj−1 )´´ = ´ − dζ ´ .
´ j ´ ´ Ij ζ − z ζj − z ´
4.6 Show that Theorem 4.21 and Corollary 4.22 hold for bounded convex sets S with piece-
wise continuously differentiable boundary, ∂ S. A set S is convex if, for all z, w ∈ S and
0 < t < 1, we have tz + (1 − t)w ∈ S. Hint: To prove the analog of Proposition 4.20,
use Corollary 4.18 if there is an appropriate disk B or show that you can replace the
integral around ∂ S with an integral around the boundary of a small square S0 centered
at a by writing ∂ S − ∂ S0 as the sum of four curves chosen so that Corollary 4.18 applies
to each.
4.7 Suppose ² is a region which is symmetric about R. Set ²+ = ² ∩ H and ²− =
+ +
² ∩ (C \ H). If f is analytic on ² , continuous on ² ∪ (² ∩ R) and Imf = 0 on ² ∩ R,
then the function defined by
⎧
⎨ f (z), for z ∈ ² \ ²−
F(z) =
⎩
f (z), for z ∈ ²−
is analytic on ². Hint: Divide a small rectangle with sides parallel to the axes and
intersecting R into three rectangles. This is the analytic version of the Schwarz reflection
principle, which we will encounter in Section 8.3.
4.8 (a) For x > 0, define x−z = e−z ln(x) . Prove that the Riemann zeta function
∞
³
ζ (z) = n−z
n=1
converges and is analytic in {z : Rez > 1}.

(b) Show that, for Rez > 1,
± ∞ ∞ ± n+1 ±
³ x
ζ (z) − x−z dx = zt−z− 1dtdx.
1 n=1 n n
(c) Use (b) to prove that (z − 1)ζ (z) has a unique analytic extension to {z : Rez > 0}.
(d) Use the fundamental theorem of calculus to give a series for ζ (z) − 1/ (z − 1), valid
in Rez > 0, which does not involve an integral.
Probably the most famous problem in all of mathematics is to prove that if ζ (z) = 0
and Rez > 0, then Rez = 1/2.
4.9 Show that there is a constant C < ∞ so that if f is analytic on D then

±
| f ± (z)| ≤ C | f (x + iy)|dxdy
D
for all |z| ≤ 1/2.
4.10 Prove that there exists a sequence of polynomials pk such that
⎧
⎨ 1, if Re(z) > 0
lim pk (z)
k →∞
=⎩ 0, if Re(z) = 0
−1, if Re(z) < 0.
4.11 (Universal entire function) Prove that there is an entire function f with the property that,
if g is entire, and if K is compact and ε > 0, then there is a number r > 0, depending
on g, K and ε, so that supK |g(z) − f (r + z)| < ε .
4.12 Suppose f has a complex derivative at each point of a region ². Prove ² f is analytic in
². Outline (Goursat): Suppose R ⊂ ² is a closed rectangle. Prove ∂ R f (z)dz = 0 in
the following way. First show that the conclusion holds for linear functions Az + B.
²
Set C = | ∂ R f (z)dz|. Divide R into four equal rectangles. Then at least one of them,
²
call it R1 , satisfies | ∂ R1 f (z)dz| ≥ C /4. Repeat this idea to obtain a nested sequence of
²
rectangles Rn with | ∂ Rn f (z)dz| ≥ C4−n and perimeter 2−n |∂ R| . Set b = ∩Rn . Since f
has a derivative at b, we have for z ∈ R n, with n sufficiently large,
| f (z) − [f (b) + f ± (b)(z − b)]| < ε|z − b| ≤ ε2−n |∂ R|. (4.10)
Estimate the integral of the left-hand side of (4.10) on ∂ Rn to obtain C4−n ≤ 4−nε| ∂ R| 2.
Finally, deduce that f is analytic. Exercise 4.13 gives an application.
4.13 Suppose f is a function defined on D with the property that, given any three points
a, b, c ∈ D, there is an analytic function g (possibly depending on a, b, c) so that |g| ≤ 1
on D and g(a) = f (a), g(b) = f (b) and g(c) = f (c). Then f is analytic on D and bounded
by 1. Hint: Suppose f (0) = 0. Prove f (z)/z is Cauchy as z → 0. Use the maps T c of
the proof of Corollary 3.10 to deduce that f has a complex derivative at each point of D,
then apply Exercise 4.12.
5 Cauchy’s Theorem
5.1 Cauchy’s Theorem
Recall that if γ is a piecewise continuously differentiable closed curve then

±
q(z)dz = 0,
γ
for all polynomials q, and

±
1
dz = 0,
γ (z − a) k
when k ≥ 2 and a ∈/ γ , because the integrand in each case is the derivative of an analytic
function. See Corollary 4.11. These also hold for all continuous curves γ by Theorem 4.32.
If r is a rational function, then it has a partial fraction expansion
nk
N ²
² ck,j
r(z) = (z − pk ) j
+ q(z),
k=1 j=1
where q is a polynomial. So, if γ is a closed curve which does not intersect any of the poles
of r, then
± N
² ±
1
r(ζ )dζ = ck,1
(ζ − pk ) dζ . (5.1)
γ
k=1 γ
Theorem 5.1 (Cauchy) Suppose γ is a cycle contained in a region ±, and suppose

±
dζ
γ ζ −a = 0 (5.2)
for all a ∈/ ±. If f is analytic on ± then

±
f (ζ )dζ = 0.
γ
Corollary 4.18 is a local version of Cauchy’s theorem for piecewise continuously differen-
tiable curves.
Proof By Runge’s Corollary 4.28, we can find a sequence of rational functions rn with poles
in C \ ± so that rn converges to f uniformly on the compact set γ ⊂ ±. By Theorem 4.32,
z
γ
Figure 5.1 Integration on a cycle.
we may³ suppose γ is piecewise continuously differentiable and has finite length. By (5.1) and
(5.2), γ rn (z)dz = 0. But then
´± ´ ´± ´
´ ´ ´ ´
´ f (z)dz´ = ´ ( f (z) − rn (z))dz´ ≤ sup | f − rn |²(γ ) → 0.
´ ´ ´ ´ γ
γ γ
We can use Cauchy’s theorem to extend Theorems 4.14 and 4.21.
Theorem 5.2 (Cauchy’s integral formula)Suppose γ is a cycle contained in a region ±,

and suppose
±
dζ
γ ζ −a = 0
for all a ∈/ ±. If f is analytic on ± and z ∈ C \ γ then
± ±
1 f (ζ )
2π i ζ −z
dζ = f (z) · 2π1 i ζ
1
− z dζ .
γ γ
Proof For each z ∈ ±, the function g(ζ ) = ( f (ζ ) − f (z))/( ζ − z) extends to be analytic

on ±, by Exercise 2.5(a). By Cauchy’s theorem, it has integral over γ equal to 0. Theorem 5.2
follows by splitting the integral of g along γ into two pieces.
The cycle γ in Figure 5.1 consists of two circles, parameterized in opposite directions
as indicated. If f is analytic on the closed region bounded by the two circles, then, by
Proposition 4.13, the hypotheses in Cauchy’s integral formula are satisfied. Moreover,
±
1 f (ζ )
f (z) = dζ ,
2π i γ ζ −z
when z is between the two circles, again using Proposition 4.13 in addition to Theorem 5.2.
When z is outside the larger circle or inside the inner circle, the integral is equal to 0 by
Cauchy’s theorem, because f ( ζ )/(ζ − z) is then an analytic function of ζ on the closure of the
region between the two circles.
5.2 Winding Number
The important integrals (5.2) have a geometric interpretation which we will next explore.
Lemma 5.3 If γ is a cycle and a ∈/ γ , then

±
1 1
2π i γ ζ − a dζ
is an integer.
Proof Without loss of generality, we may suppose γ is a closed curve parameterized by a

continuous, piecewise differentiable function γ : [0, 1] → C. See the comments after the
proof of Theorem 4.32. Define
± x γ ±(t)
h(x) =
0 γ (t) − a dt.
Then h± (x) exists and equals γ ±(x)/ (γ (x) − a), except at finitely many points x. Then
d −h(x)
e (γ (x) − a) = −h± (x)e−h(x) (γ (x) − a) + e−h(x) γ ± (x)
dx
= −γ ± (x)e−h(x) + γ ± (x)e−h(x) = 0,
except at finitely many points. Since e− h(x) (γ (x) − a) is continuous, it must be constant. Thus
e−h(1) (γ (1) − a) = e−h(0) (γ (0) − a) = 1 · (γ (1) − a).

Since γ (1) − a ² = 0, e−h(1) = 1 and h(1) = 2π ki, where k is an integer. Thus
±
1 dζ h(1)
2π i ζ −a
= 2π i
= k,
γ
an integer.
Definition 5.4 If γ is a cycle, then the index or winding numberof γ about a (or with
respect to a) is
±
n(γ , a) = 2π1 i ζ
dζ
− a,
γ
for a ∈/ γ .
Note the following properties of n(γ , a):
(a) n(γ , a) is an analytic function of a, for a ∈ / γ , by Lemma 4.30 and Theorem 4.32. In
particular, it is continuous and integer valued by Lemma 5.3. Thus n(γ , a) is constant in
each component of C \ γ .
(b) n(γ , a) → 0 as a → ∞ . Thus n( γ , a) = 0 in the unbounded component of C \ γ .
(c) n(−γ , a) = −n(γ , a).
(d) n(γ1 + γ2, a) = n( γ1 , a) + n(γ2, a).
(e) If γ (t) = eikt, for 0 ≤ t ≤ 2π , where k is an integer, then
± ± 2π
1 dz 1 ike ikt
n(γ , 0) = = 2π i dt = k.
2π i γ z 0 eikt
We say that the curve γ in (e) winds k times around 0.

For example, in the proof of Runge’s Theorem 4.23 we constructed a finite union of shaded
squares {Sj }. Let γ = ∂ (∪Sj ). If a is in the interior of a square Sk then n(∂ Sk , a) = 1 and
/ ∪Sj then
n(∂ Sj , a) = 0 for j ² = k, by Proposition 4.20. By (d), n( γ , a) = 1. Similarly, if a ∈
n(γ , a) = 0. If a ∈ ∂ Sk , but a ∈/ γ then, by (a), n(γ , a) = 1.
There are a couple of ways to compute winding numbers that may provide more intuition.
Suppose γ (t) = r(t)eiθ (t) , 0 ≤ t ≤ 1, where r > 0 and θ are continuous and piecewise
continuously differentiable, with γ (0) = γ (1). Then
µ ± ¶ ± 1· ± ¸
1 dz
n(γ , 0) = Re 2π i z
= 21π Im r
r
+ iθ ± dt
γ 0
= 21π [θ (1) − θ (0)] . (5.3)
In this case the winding number of γ about 0 is equal to the total change in arg γ (t) divided
by 2π . More generally, if γ : [0, 1] → C is continuous, with γ (0) = γ (1) and 0 ∈/ γ , choose
δ > 0 so that if |t − s| < δ then | γ (t) − γ (s)| < dist(γ , 0). Take a partition 0 = t0 < t1 < · · · <
tn = 1 with | tj−1 − tj | < δ. Then γ ([tj−1 , tj ]) lies in a disk B j which does not contain 0. Thus
the line segment σj from γ (tj−1 ) to γ (tj ) also lies in B j . As in Theorem 4.32 with f (z) = 1/z,
n(γ , 0) = n(σ , 0),
where σ : [0, 1] → C is the polygonal curve σ = σ1 + . . . + σn. By (5.3)
¹ º
n(σ , 0) = 21π arg σ (1) − arg σ (0) ,
where arg σ (t) is defined so as to be continuous on [0, 1]. In fact, because γ ([tj−1 , t j]) is con-
tained in a disk Bj which omits 0, and hence omits a ray from 0 to ∞ , we can define arg γ (t)
to be continuous on [tj−1 , tj ], and hence arg γ (t) can be defined as a continuous function on
[0, 1]. Then
1 ¹ º
n(γ , 0) = arg γ (1) − arg γ (0) ,
2π
the total change in arg γ (t).
If z0 ∈ C then n(γ , z0) = n(γ − z0 , 0). If we imagine standing at z0 , watching a point trace
the closed curve γ then n(γ , z0 ) is the net number of revolutions we make, counting positive
if counter-clockwise and negative if clockwise.
There is an even easier way to compute winding numbers in most cases. Let Ra be any ray,
or half-line, from a to ∞ . Suppose that γ is a cycle such that a ∈/ γ , and suppose that γ ∩ Ra
consists of finitely many points zj . Each connected component of γ \ Ra contributes 1, 0 or
−1 1 2
Figure 5.2 Calculating the winding number.

Figure 5.3 Parity: which red point is not in the same component as the other two?
−1 to the winding number of γ about a. Because γ is oriented, we can count the winding of
γ about a by adding +1 if the crossing at zj is counter-clockwise and −1 if the crossing is
clockwise. See Exercise 5.1. If γ is the curve in Figure 5.2 then n( γ , −1) = 0, n(γ , 1) = 2,
n(γ , 2) = 0 and n(γ , i) = 1.
A two-meter by four-meter version of Figure 5.3 was painted by W. Thurston and D. Sulli-
van on the wall of the Berkeley Mathematics department, where it remained for decades. By
counting the number of crossings of three rays, it can be shown that one red point is not in the
same component of the complement as either of the remaining two. Are the remaining two
points in the same component? Be careful, how many components are there and why? See
Section 12.1 for some badly behaved curves.
Definition 5.5 Closed curves γ1 and γ2 are homologous in a region ± if n(γ1 − γ2, a) =0
for all a ∈
/ ±, and in this case we write
γ1 ∼ γ2 .
Homology is an equivalence relation on the curves in ± (Exercise 5.2). A closed curve
γ ⊂ ± is said to be homologous to0 in ± if n( γ , a) = 0 for all a ∈/ ±. In this case we write
γ ∼ 0. Cauchy’s theorem says that if γ ∼ 0 in ±, and if f is analytic in ±, then
±
f (z)dz = 0.
γ
Thus, if γ1 is homologous to γ2 in ±, then

± ±
f (z)dz = f (z)dz.
γ1 γ2
The most common application of Cauchy’s integral formula is when γ ⊂ ± with γ ∼ 0 and
n(γ , z) = 1. Then, for f analytic on ±,
±
f (z) = 2π1 i f (ζ )
ζ −z
dζ .
γ
Figure 5.4 shows a closed curve σ contained in ± = C \ {0, 1} with the property that
n(σ , a) = 0 for all a ∈ / ±. If you are familiar with homotopy (we will treat this subject in
Section 14.1), this curve shows that homotopy and homology are different because σ can-
not be shrunk to a point while remaining in ±. See Exercise 14.5. By Cauchy’s theorem,
³
σ
f (z)dz = 0 for every analytic function f on ±.
If ± is a bounded region in C bounded by finitely many piecewise differentiable curves,
then we can parameterize ∂ ± so that as we trace each boundary component, the region ±
0 1
Figure 5.4 n( σ , 0) = n(σ , 1) = 0.
lies on the left. In other words, iγ ± (t), where it exists, is an inner normal, rotated counter-
clockwise by π/2 from the tangential direction γ ± (t). In Exercise 5.4 you are asked to show
that n(∂ ±, a) = 1 for all a ∈ ± and n(∂ ±, a) = 0 for all a ∈/ ±. We call this the positive orien-
tation of ∂ ±. Thus, for all such regions, ∂ ± ∼ 0 in any region containing ±. So, by Cauchy’s
theorem, if ± is a bounded region, bounded by finitely many piecewise differentiable curves,
and if f is analytic on ±, then
±
f (ζ ) dζ
f (z) = ζ − z 2π i , (5.4)
∂±
for all z ∈ ±, where ∂ ± has positive orientation. A rigorous proof of the corresponding fact
for regions bounded by finitely many simple closed curves is given in Corollary 12.16 and
Exercise 12.10. If ± is the unbounded region C \ D, then ∂ ± is not homologous to zero in a
region containing ± and (5.4) does not hold.
Definition 5.6 A region ± ⊂ C∗ is called simply-connected if C∗ \ ± is connected in C∗ .
Equivalently, a region ± is simply-connected if S 2 \ π (±) is connected, where π is stereo-

graphic projection and π (∞) is defined to be the “north pole” (0, 0, 1) in S 2. Simply-connected
essentially means “no holes.” For example: the unit disk D is simply-connected; the ver-
tical strip {z : 0 < Rez < 1} is simply-connected; the punctured plane C \ {0} is not
simply-connected; the set C \ D together with ∞ is simply-connected, but C \ D is not
simply-connected; the open set inside a “figure 8” curve is not simply-connected because
it is not connected.
Theorem 5.7 A region ± ⊂ C is simply-connected if and only if every cycle in ± is homolo-

gous to 0 in ±. If ± is not simply-connected then we can find a simple closed polygonal curve
contained in ± which is not homologous to 0.
The point of Theorem 5.7 is that simply-connected is a geometric condition which is

sufficient for Cauchy’s theorem to apply.
Proof Suppose that ± is simply-connected, that γ is a cycle contained in ± and that a ∈ / ±.

Because ± = C \ ± is connected, it must be contained in one component of the complement
c ∗
of γ in C∗ . Because ∞ ∈ ±c , a must be in the unbounded component of C\ γ , and n(γ , a) = 0
by property (b).
Conversely, suppose that C∗ \ ± = A ∪ B, where A and B are non-empty closed sets in C∗

with A ∩ B = ∅. Without loss of generality, ∞ ∈ B. Since A is closed, a neighborhood of ∞
does not intersect A, and hence A is bounded. Pick a0 ∈ A. We will construct a curve γ0 ⊂ ±
such that n(γ0 , a0) ² = 0, proving Theorem 5.7.
The construction is the same construction used to prove Runge’s Theorem 4.23. Let
d = dist(A, B) = inf{|a − b| : a ∈ A, b ∈ B} > 0.

If B = {∞}, set d = 1. Pave the plane with squares of side d/2 such that a0 is the center of
one of the squares. Orient the boundary of each square in the positive, or counter-clockwise,
direction (like all storms in the northern hemisphere). Shade each square Sj with Sj ∩ A ² = ∅ .
Let γ0 denote the cycle obtained from ∪ ∂ Sj after performing all possible cancellations. Then
γ 0 ⊂ ± because γ0 does not intersect either A or B, and n(γ 0, a0 ) = 1. Because γ 0 is a
finite union of closed simple polygonal curves, at least one of these curves is not homologous
to 0.
Corollary 5.8 Suppose f is analytic on a simply-connected region ±. Then

³
(i) γ f (z)dz = 0 for all closed curves γ ⊂ ±;
(ii) there exists a function F analytic on ± such that F ± = f ;
(iii) if also f (z) ² = 0 for all z ∈ ± then there exists a function g analytic on ± such that f = eg .
Proof Corollary 5.8(i) follows from Cauchy’s theorem and Theorem 5.7. To prove ³ (ii), we
follow the proof of Morera’s Theorem 4.19. Fix z0 ∈ ±. We can then define F(z) = σ z f (ζ )dζ ,
where σz is any curve contained in ± connecting z0 to z. This definition does not depend on
the choice of σz because the integral along any closed curve is zero. If B ⊂ ± is a disk, then
for z ∈ B we can write F(z) as an integral from z0 to the center of B plus an integral along
a horizontal then vertical line segment from the center of B to z. By the proof of Morera’s
Theorem 4.19, F is analytic and F ± = f on B and hence on all of ±.
To prove (iii), note that f ± /f is analytic on ±, so that, by (ii), there is a function g analytic
on ± such that g± = f ± /f . To compare f and eg , set
f
h=
eg
= fe−g .
Then
h± = f ± e−g − fg± e−g = f ± e−g − f ± e−g = 0.
This implies h is a constant. Adding a constant to g, we may suppose eg(z0 ) = f (z0), so that
h ≡ 1 and f = eg . This proves Corollary 5.8.
Note that if ± is not simply-connected then, by Theorem 5.7, there is a point a ² ∈ ± and a
polygonal curve γ ⊂ ± such that
±
1
γ ζ − a dζ ²= 0. (5.5)
Then f (z) = 1/(z − a) is analytic in ± but cannot be the derivative of an analytic function, for
otherwise the integral in (5.5) would be 0 by the fundamental theorem of calculus. We also
cannot find a function g analytic on ± with eg(z) = 1/(z − a), for otherwise 1/(z − a) = −g± (z)
is the derivative of an analytic function.
The solutions to Corollary 5.8(ii) and (iii) are essentially unique, for if F ± = f = G± then
(F − G)± = 0 and hence F − G is constant. Likewise, if eg = f = eh then eg−h = 1 so that
g − h = 2π ki for some integer k, by continuity and the connectedness of ±.
Definition 5.9 If g is analytic in a region ± and if f = eg then g is called a logarithm of f

in ± and is written g(z) = log f (z). The function g is uniquely determined by its value at one
point z0 ∈ ±.
By Corollary 5.8(iii), if f is analytic and non-vanishing in a simply-connected region ±

then f has countably many logarithms, which differ by integer multiples of 2π i. So, to specify
log f uniquely as an analytic function, we must include in the definition its value at one point
z0 ∈ ±. For example, if ± is simply-connected and 0 ∈ / ± then log z can be defined as an
analytic function on ±. Given z0 ∈ ±, there are countably many choices for log z0 and hence
for the function log z.
A word of caution: the conclusion (iii) of Corollary 5.8 is that there is an analytic function
g such that
f (z) = eg(z).
We are not claiming that a logarithm can be defined on f (±), the range of f , and then composed
with f . The function g is locally the composition of a function log z and f , but there might not
be a function h defined on all of f (±) such that z = eh(z). For example, the rational function
(z − 1)/ (z + 1) maps the disk D onto the left half-plane {Rez < 0} and so the function
f (z) = e(z −1)/ (z+1) maps D onto D \ {0}. Thus f satisfies the hypotheses of Corollary 5.8(iii)
and indeed g(z) = log f (z) = (z − 1)/ (z + 1) works. However, we cannot define log z as an
analytic function on f (D) = D \ {0} since arg z = Im log z will increase by 2π as a circle is
traced counter-clockwise about 0.
Throughout this book, log denotes the natural log, not base 10. Some books call log z a
“multiple-valued function” in C \ { 0}, which is a bit of a contradiction in terms. We will only
consider log z on domains where it can be defined as a function, or we will define log f as an
analytic function, as in Corollary 5.8(iii).
5.3 Removable Singularities
The main result in this section is in some sense a generalization of Exercise 2.5.
Corollary 5.10 (Riemann’s removable singularity theorem)Suppose f is analytic in ± =

{z : 0 < |z − a| < δ} and suppose
lim (z − a)f (z) = 0.
z →a
Then f extends to be analytic in {z : |z − a| < δ}.
In other words, we can “remove” the “singularity” of f at a.

5.3 Removable Singularities 71
Proof Fix z ∈ ± and choose ε and r so that 0 < ε < |z − a| < r < δ. Let Cε and Cr denote
the circles of radius ε and r centered at a, oriented in the counter-clockwise direction. The
cycle Cr − Cε (see Figure 5.1) is homologous to 0 in ±, so that, by Cauchy’s integral formula,
± ±
1 f (ζ )
f (z) =
2π i ζ −z
dζ − 21π i f (ζ )
− z dζ .
Cr Cε ζ
But
´± ´
´ f (ζ) ´ 1
´ C ζ − z dζ ´ ≤ max | f (ζ )|
´ ´
ε
ζ ∈C ε |z − a| − ε 2π ε.
But if ζ ∈C ε then | f (ζ )| ε = | f (ζ )||ζ − a| → 0 as ε → 0 and hence
±
1 f (ζ )
f (z) = dζ . (5.6)
2π i Cr ζ −z
By Lemma 4.30, the right-hand side of (5.6) is analytic in {z : |z − a| < r}. Thus if we define
f (a) as the value of the right-hand side of (5.6) when z = a, then this extension is analytic at
a and we have extended f to be analytic in {z : |z − a| < δ}.
The most important special case of Riemann’s removable singularity theorem is as follows:
if f is bounded and analytic in a punctured neighborhood of a, then f extends to be analytic in
a neighborhood of a.
We say that a compact set E has one-dimensional Hausdorff measure equal to 0 if for
every ε > 0 there are finitely many disks Dj with radius rj so that
E ⊂ ∪j Dj
and
²
rj < ε.
j
Corollary 5.11 (Painlevé) Suppose E ⊂ C is a compact set with one-dimensional Hausdorff

measure 0. If f is bounded and analytic on U \ E, where U is open and E ⊂ U, then f extends
to be analytic on U.
Proof As in the proof of Runge’s Theorem 4.23 and Theorem 5.7, we can find a cycle
γ ⊂ U \E which is the boundary of a finite union of closed squares { Sj } so that n(γ , a) = 0 or 1
for all a ∈
/ γ , n(γ , b) = 1 for all b ∈ ∪ Sj \ γ ⊃ E, and n(γ , b) = 0 for b ∈ / ∪ Sj and hence
∑
for all b ∈ C \ U. Cover E by finitely many disks Dj of radius rj so that rj < ε . We
) for small ε, each Dj is contained in ∪Sj \ γ . Let
may assume each Dj intersects E( so that,
V = {z : n(γ , z) = 1}, let σ = ∂ ∪Dj , and let ± = V \ ∪Dj . Then γ + σ = ∂ ±, which we
parameterize so that ∂ ± has positive orientation. See Exercise 5.4. Then, as in (5.4), γ + σ ∼ 0
in U \ ∪Dj , so that, by Cauchy’s theorem,
± ±
f (z) = 2π1 i f (ζ )
ζ −z
dζ + 21π i f (ζ )
ζ −z
dζ ,
γ σ
for z ∈ V \ ∪Dj . Fix z ∈ V \ ∪Dj . Then the second integral tends to 0 as ε → 0 because
² (σ ) ≤ ²(∪ Dj ) < 2π ε and because f is bounded, exactly as in the proof of Riemann’s
theorem. Thus
±
1 f (ζ )
2π i γ ζ − z dζ
provides an analytic extension of f to E, by Lemma 4.30.
Painlevé asked for geometric conditions on a compact set E to be removable for bounded
analytic functions in 1888. Removable means that the second sentence of Corollary 5.11
holds. A major accomplishment in complex analysis within the last fifteen years was an
answer to this question.
We can use the Cauchy integral formula and Riemann’s theorem to give a formula for the
inverse of a one-to-one analytic function. Suppose f is analytic and one-to-one on a region ±.
Let γ be a cycle in ± such that n(γ , a) = 0 for all a ∈ ±c . Fix w and suppose f (z) − w has
exactly one zero in {z : n(γ , z) = 1}. Then ( f (z) − w)/(z − f −1 (w)) is a difference quotient for
f ± ( f −1 (w)), and so
z − f −1 (w) ±
lim f (z)z = f −1 (w).
z→f −1 (w) f (z) − w
By Riemann’s removable singularity theorem,

f (z)z ± f (w) −1
h(z) = f (z) − w − z − f −1 (w)
extends to be analytic in ±, and, by Cauchy’s theorem,
±
h(z)dz = 0.
γ
Since n(γ , f −1 (w)) = 1, we obtain

±
1 f ± (z)z
f −1 (w) = dz. (5.7)
2π i γ f (z) − w
We proved that f −1 is analytic in Corollary 4.17, but we can also use (5.7) to show that f −1 is
analytic, by imitating the proof of Lemma 4.30.
5.4 Laurent Series
An annulus is the region between two concentric circles. If f is analytic on the annulus A =
{z : r < |z − a| < R} then, by Runge’s Theorem 4.27, we can approximate f by a rational
function with poles only at a. The Laurent seriesis another version of this result, similar to
a power series expansion.
Theorem 5.12 (Laurent series) Suppose f is analytic on A = { z : r < |z − a| < R}. Then
there is a unique sequence {an} ⊂ C so that
∞
²
f (z) = an(z − a)n ,
n=−∞
where the series converges uniformly and absolutely on compact subsets of A. Moreover,
±
an = 2π1 i f (ζ )
(ζ − a)n+1
dζ , (5.8)
Cs
where C s is the circle centered at a with radius s, r < s < R, oriented counter-clockwise.
In the course of the proof of Theorem 5.12 we shall see that the integrals in (5.8) do not
depend on the choice of s, so long as r < s < R.
Proof Without loss of generality, a = 0. Set

±
1 f (ζ )
fs (z) = dζ ,
2π i Cs ζ −z
where C s = {seit : 0 ≤ t ≤ 2π }. Then, by Lemma 4.30, fs is analytic off Cs . If r < | z| <

s1 < s2 < R, then Cs2 − C s1 ∼ 0 with respect to A and n(Cs 2 − Cs1 , z) = 0. By Cauchy’s
integral formula, fs2 (z) − fs1 (z) = 0. This says that fs (z) does not depend on s, so long as
r < |z| < s < R.
1
Expanding ζ − z
in a power series expansion about 0, and interchanging the order of sum-
mation and integration, as we have done before, we conclude that fs has a power series
expansion
∞
²
f s (z) = an zn ,
n=0
valid in |z| < s, where an satisfies (5.8). Likewise, f s (z) does not depend on s so long as
r < s < |z| < R. Expanding ζ −
1
z
in a power series expansion about ∞, i.e., in powers of 1 /z,
and interchanging the order of summation and integration, we conclude that fs has a power
series expansion
∞
²
fs (z) = − a−n z−n,
n=1
valid in |z| > s, where a−n satisfies (5.8).

If r < s1 < |z| < s2 < R, then Cs2 − Cs 1 ∼ 0 in A, so that, by Cauchy’s integral formula,
± ∞
²
f (z) = 2π1 i f (ζ )
ζ −z
dζ = fs (z) − fs (z) =
2 1
a n zn .
Cs 2 −Cs 1 n=−∞
The uniqueness statement is Exercise 5.5.
One consequence of Theorem 5.12 is that an analytic function f on A can be written as

f = f1 + f2, where f1 is analytic in |z| < R and f2 is analytic in |z| > r.
For example, the function
z 2 1
r(z) = + −
z2 + 4 (z − 3) z − 4
2
is analytic in C except at ³2i, 3, 4 and hence is analytic on four regions centered at 0: A1 =

{z : |z| < 2}, A2 = { z : 2 < |z| < 3}, A3 = { z : 3 < |z| < 4} and A4 = { z : 4 < |z|}. Of
course, A1 is a disk, so that r has a power series expansion in A1 . We can use
1 1
∞
² − 1 zn ,
z−a
= −a(1 − z ) = an+1
a n=0
provided | z| < | a| , and

1 1
∞ an−1
²
z−a
= z(1 − a )
= zn
,
z n=1
provided | z| > |a|. These formulae give the Laurent expansion of 1/(z − 4) in each region.
Differentiating a series of this form with a = 3 will give the series for 1/(z − 3)2 . Setting
a = −4 and replacing z with z2 will give the series for 1/(z2 + 4), and multiplying by z will
give the series for z/(z2 + 4). If a rational function is given as a ratio of two polynomials,
p/ q, then use the techniques of Section 2.2 to find its partial fraction expansion, and then use
the series expansion for 1/(z − a) as above. Sometimes it might be easier to find the partial
fraction expansion for 1/q, then multiply the resulting series by the polynomial p and collect
together similar powers of z.
Laurent series are useful for analyzing the behavior of an analytic function near an isolated
singularity. We say that f has an isolated singularityat b if f is analytic in 0 < |z − b| < ε
for some ε > 0 and f (b) is not defined. Write
∞
²
f (z) = an(z − b)n .
n=−∞
Note the following:

(a) If an = 0 for n < 0 then f extends to be analytic at b with f (b) = a0. In this case we say
that f has a removable singularityat b.
(b) If an = 0 for n < n0 with n0 > 0 and an0 ² = 0, then we can write
∞
²
f (z) = (z − b)n0 an0+n (z − b)n = an (z − b)n + an +1(z − b)n +1 + · · · .
0
0
0
0
n=0
In this case b is called a zero of ordern0 .

(c) If an = 0 for n < −n0 with n0 > 0 and a−n0 ²= 0 then we can write
∞
²
f (z) = (z − b)−n 0
a−n0+n (z − b)n = (z a−−b)n
n
+ (z a−−b)
0n +1
n −1
0
+ ··· .
0
0
n=0
In this case b is called a pole of ordern0, and | f (z)| → ∞ as z → b.
In each of the above cases there is a unique integer k so that
lim (z − b)k f (z)
z→b
exists and is non-zero, and

(z − b)k f (z)
extends to be analytic and non-zero in a neighborhood of b.
(d) If an ² = 0 for infinitely many negative n, then b is called an essential singularity. For
example,
∞ (−1)n
²
f (z) = e
− z12 = z−2n
n!
n=0
has an essential singularity at 0, even though it is infinitely differentiable on the real line.
If f is analytic in {z : | z| > R }, then f (1/z) has an isolated singularity at 0, and we say that f
has an isolated singularity at∞ . We classify this singularity at ∞ as a zero, pole or essential
singularity if f (1/z) has a zero, pole or (respectively) essential singularity at 0. In terms of the
Laurent expansion of f in | z| > R,
∞
²
f (z) = bn zn,
n=−∞
f has an essential singularity at ∞ if bn ² = 0 for infinitely many positive n. The reader
can supply the corresponding statement for zeros and poles and their orders. For example,
a polynomial of degree n has a pole of order n at ∞.
Definition 5.13 A zero or pole is called simple if the order is 1.
Definition 5.14 If f is analytic in a region ± except for isolated poles in ± then we

say that f is meromorphic in ±. A meromorphic function in C is sometimes just called
meromorphic.
If f ² ≡ 0 is meromorphic in ±, then 1/f is meromorphic in ±, and a zero of order k becomes

a pole of order k, and a pole of order k becomes a zero of order k. If f and g are meromorphic
in ± then the order of a zero or pole, for the meromorphic function f /g at b, can be found
by factoring the appropriate power of z − b out of f and g then taking the difference of these
powers.
The next result gives an idea of the behavior near an essential singularity.
Theorem 5.15 If f is analytic in U = {z : 0 < |z− b| < δ}, and if b is an essential singularity
for f , then f (U) is dense in C.
In other words, every (punctured) neighborhood of an essential singularity has a dense

image.
Proof If f(U) is not dense in C, there exists A ∈ C and ε > 0 so that | f (z) − A| > ε for all
z ∈ U. Then
1
f (z) − A
is analytic and bounded by 1/ε on U. By Riemann’s theorem, 1/( f (z) − A) extends to be
analytic in U ∪ { b}. Thus f (z) − A is meromorphic in U ∪ {b} and hence f is meromorphic in
U ∪ {b}. The Laurent expansion for f then has at most finitely terms with a negative power of
z − b, contradicting the assumption that b is an essential singularity.
5.5 The Argument Principle
The next result is useful for locating zeros and poles of meromorphic functions.
Theorem 5.16 (argument principle) Suppose f is meromorphic in a region ± with zeros

{zj } and poles {pk }. Suppose γ is a cycle with γ ∼ 0 in ±, and suppose {zj } ∩ γ = ∅ and
{pk } ∩ γ = ∅. Then
±
1 f ± (z) ² ²
n( f (γ ), 0) = dz = n(γ , zj ) − n( γ , pk ). (5.9)
2π i γ f (z)
j k
In the statement of the argument principle, if f has a zero of order k at z, then z occurs k
times in the list {zj }, and a similar statement holds for the poles. If γ is a simple closed curve
with n(γ , z) = 0 or = 1 for all z ∈ / γ , then we say that z is enclosed by γ if n(γ , z) = 1. A
consequence of Theorem 5.16, is that if γ is a simple closed curve in ±, with n(γ , z) = 0 or
= 1 for all z ∈/ γ , and if γ is homologous to 0 in ±, then the number of zeros “enclosed” by
γ minus the number of poles “enclosed” by γ is equal to the winding number of the image
curve f (γ ) about zero.
Proof The first equality in (5.9) follows from the change of variables w = f (z). Note that
γ ∼ 0 and γ ⊂ ± imply that n(γ , a) = 0 if a is sufficiently close to ∂ ±. Thus n(γ , zj ) ² = 0
for only finitely many zj and for only finitely many pj because there are no cluster points of
{zj } or {pk } in ±. This implies that the sums in (5.9) are finite. Set ±1 = ± \ {zj : n(γ , zj ) =
0} ∪ { pk : n(γ , pk ) = 0}. Then γ ∼ 0 in ±1 .
If b is a zero or pole of f , we can write
f (z) = (z − b)k g(z),
where g is analytic in a neighborhood of b and g(b) ² = 0. Then
f ± (z) = k(z − b)k −1g(z) + (z − b)k g±(z)
and
f ± (z) ±
f (z)
= z −k b + gg(z)
(z)
.
Since g(b) ² = 0, g± /g is analytic in a neighborhood of b and hence f ± /f − k(z − b)−1 is analytic

near b. Thus
f ± (z) ² 1 ² 1
f (z)
− z − zj
+ z − pk
(5.10)
is analytic in ±1 . In the sums, we repeat zj and pk according to their multiplicity. By Cauchy’s

theorem, integrating (5.10) over γ gives (5.9).
Corollary 5.17 (Rouché) Suppose γ is a closed curve in a region ± with γ ∼ 0 in ± and

n(γ , z) = 0 or = 1 for all z ∈ ± \ γ . If f and g are analytic in ± and satisfy
| f (z) + g(z)| < | f (z)| + |g(z)| (5.11)

5.5 The Argument Principle 77
for all z ∈ γ , then f and g have the same number of zeros enclosed by γ .
Equation (5.11) says that strict inequality holds in the triangle inequality. The number of
zeros of f and g are counted according to their multiplicity.
f
Proof The function g
is meromorphic in ± and satisfies
´ ´ ´ ´
´f ´ ´ ´
´
´g + 1´´ < ´´ gf ´´ + 1 (5.12)
on γ . By (5.11), f ² = 0 and g ² = 0 on γ , so that the hypotheses of the argument principle are

satisfied.
The left-hand side of (5.12) is the distance from w = f (z)/g(z) to −1. But | w−(−1)| = |w|+
1 if and only if w ∈ [0, ∞ ). See Figure 5.5. Thus the assumption (5.11) implies that g (γ ) omits
f
the half-line [0, ∞ ) and so 0 is in the unbounded component of C\ gf (γ ). Hence n( gf (γ ), 0) = 0.

By the argument principle, the number of zeros of gf equals the number of poles, and so the
number of zeros of f equals the number of zeros of g, counting multiplicity.
Example 5.18 f (z) = z9 − 2z6 + z2 − 8z − 2.
How many zeros does f have in |z| < 1? The biggest term is −8z, so, comparing f and −8z,
we have
| f (z) + 8z| = |z9 − 2z6 + z2 − 2| ≤ 1 + 2 + 1 + 2 = 6 < |8z|
on |z| = 1. By Rouché’s theorem, f and 8z have the same number of zeros in | z| < 1, namely
one. How many zeros does f have in |z| < 2? In this case we compare f with z9 :
| f (z) − z9 | = |−2z6 + z2 − 8z − 2| ≤ 128 + 4 + 16 + 2 = 150 < |z9| = 512.

Therefore f and z9 have the same number of zeros in | z| < 2, namely 9. Thus 8 of the zeros
of f lie in 1 < |z| < 2, and the remaining zero lies in | z| < 1.
Example 5.19 g(z) = z4 − 4z + 5

How many zeros does g have in |z| < 1? Compare g with the constant function 5 on |z| = 1:
|z4 − 4z + 5 − 5| = |z4 − 4z| ≤ 5 ≤ |5| + |z4 − 4z + 5|.
w
| w+ 1|
| w|
−1 0
1
γ
Figure 5.5 Proof of Rouché’s theorem.
If equality holds in the first inequality, then 5 = | z4 − 4z| = |z3 − 4|. But |z| = 1, so z3 = −1,
and z4 − 4z + 5 = −z − 4z + 5 = 5(−z + 1). Since z ² = 1, |z4 − 4z + 5| > 0 and we have
|z4 − 4z + 5 − 5| < |5| + |z4 − 4z + 5|.

By Rouché’s theorem, z4 − 4z + 5 and 5 have the same number of zeros in | z| < 1, namely
none.
A more elaborate process for locating zeros of polynomials is given in Exercise 5.13.
5.6 Exercises
5.1 Suppose γ is a closed curve and a ∈/ γ . Suppose R a is a ray (half-line) from a to ∞

such that γ ∩ Ra consists of finitely many points. We say that the curve γ (t) crosses Ra
counter-clockwise (clockwise) at γ (t1 ) ∈ Ra if arg(γ (t) − a) is increasing (respectively,
decreasing) as a function of t in a neighborhood of t1 . Prove that n(γ , a) equals the
number of counter-clockwise crossings of R a minus the number of clockwise crossings.
The set γ \ Ra consists of finitely many pieces, but not all endpoints are necessarily
“crossings.”
5.2 Prove that homology is an equivalence relation on curves in a region ±.
5.3 The proof of Corollary 5.8(iii) should remind you of the proof that the winding num-
ber is an integer. Explain this using the increase in the imaginary part of a continuous
determination of log(z − a) along γ .
5.4 Suppose ± is a bounded region whose boundary consists of finitely many disjoint piece-
wise differentiable simple closed curves. Suppose that ∂ ± is oriented so that the region
lies on the left for each boundary component. This means that, except for finitely many
t, γ (t) + i ε(t)γ ± (t) ∈ ± and γ (t) − iε(t)γ ± (t) ∈/ ± for sufficiently small ε (t) > 0. Prove
n(∂ ±, a) = 1 if a ∈ ± and n(∂ ±, a) = 0 if a ∈ C \ ±. Hint: If L is a short line seg-
ment from a ∈ ± to b ∈/ ± that intersects γ in one point, use Exercise 5.1 to show
n(γ , a) = n(γ , b) + 1. Then use the fact that n( ∂ ±, a) is constant in each component of
C \ ∂ ±.
5.5 Prove the uniqueness of the Laurent series expansion; i.e., if
∞
² ∞
²
an zn = bn zn
n=−∞ n=−∞
for r < | z| < R then an = bn for all n. Convergence of the series on the region is part of
the assumption. Hint: Liouville or multiply by zk and integrate.
5.6 Note that in the proof of Laurent series expansions we proved that a function f which is
analytic on r < |z| < R can be written as f = f1 + f2, where f1 is analytic in | z| < R, f2 is
analytic in |z| > r, and f2 (z) → 0 as | z| → ∞. Suppose that ± is a bounded region in C
such that ∂ ± is a finite union of disjoint (piecewise continuously differentiable) closed
∑
curves ³j , j = 1, . . . , n. Suppose that f is analytic on ±. Prove that f = fj , where fj is
analytic on the component of C \ ³j which contains ±.
5.6 Exercises 79
5.7 Prove that if a sequence of analytic polynomials converges uniformly on a region ±

then the sequence converges uniformly on a simply-connected region containing ±.
Hint: See Exercise 4.2(b).
5.8 We define the singularity at ∞ of f (z) to be the singularity at 0 of g(z) = f (1/ z). Find
the singularity at ∞ of the following functions. If the singularity is removable, give the
value; if the singularity is a zero or pole, give the order.
(a) z e+z 1 ; (b) e1/1z−1 − z; (c) ez /(1−z); (e) z2 − z;
2 1
(d) ze1/z ; (f) 1 z
z3
e .
5.9 Find the expansion in powers of z for
z
(z2 + 4)(z − 3)2 (z − 4)
which converges in 3 < |z| < 4.
5.10 Prove that the number of roots of the equation
z2n + αz2n−1 + β2 = 0,
where α , β are real and non-zero and n is a natural number, which have positive real
part is equal to n if n is even. If n is odd, the number of roots is n − 1 for α > 0 and
n + 1 for α < 0. Hint: See what happens to z2n + αz2n−1 + β 2 as z traces the boundary
of a large half-disk.
5.11 (a) How many zeros does z4 + z3 + 5z2 + 2z + 4 have in the first quadrant? Hint: See
where the image of the boundary of a large circular sector crosses R.
(b) How many zeros does p(z) = 3z5 + 21z4 + 5z3 + 6z + 7 have in D? How many
zeros in {z : 1 < |z| < 2}?
5.12 Prove that all of the zeros of the polynomial
p(z) = zn + cn−1 zn− 1 + . . . + c1z + c0
lie in the disk centered at 0 with radius

»
R= 1 + |cn−1 |2 + . . . + |c1|2 + |c0| 2.
5.13 This exercise gives an algorithm for counting the number of zeros of a polynomial in a
disk. First translate and dilate the disk to the unit disk. Suppose p(z) = an zn+an−1zn−1 +
. . . + a 0 is a polynomial with an ² = 0. Set
· ¸
1
p∗ (z) = znp = a0 zn + a1zn−1 + . . . + an
z
and let q(z) = a0 p(z) − anp∗ (z).

Case (a): if |an| < |a0| , set p1 = q.
Case (b): if |an | > |a0| , set p1 = (a0 p∗ − an p)/ z.
Case (c): if |an| = |a0| and q ² = 0 on |z| = 1, set p1 = q.
Then the polynomial p1 has smaller degree than p. Let N(p) denote the number of
zeros of p in D = {z : | z| < 1}. Then prove
¼
N(p1 ), in cases (a) and (c)
N(p) = N(p1 ) + 1, in case (b).
Repeat the procedure until it stops. Try it for p(z) = 2z4 − 3z3 + z2 + z − 1. If none
of the cases applies, one possibility is to work with p((1 − ε )z).
The following problem has been open since the 1930s and is of current interest in ergodic
theory. Suppose p is a polynomial with integer coefficients
p(z) = anzn + . . . + a0
with |an| = |a0| = 1. Let z1, z2 , . . . , zk be the roots in | z| < 1. If p has at least one root in
|z| < 1, how big can ½ |zj | be? (It is conjectured that
k
¾
|zj | ≤ 0.8501371 . . .
p=1
This value is achieved for the polynomial

p(z) = z10 + z9 − z7 − z6 − z5 − z4 − z3 + z + 1.
Eight zeros lie on the unit circle, one zero z1 is outside the unit circle and the last is 1/z1,
which is inside the unit disk.)
6 Elementary Maps
In this chapter we will study the mapping properties of the elementary functions and their
compositions. The emphasis will be on the behavior of LFTs and the power, trigonometric
and exponential functions related to familiar elementary functions of a real variable. These
∑ n
functions are all built from linear functions a + bz, ez = ∞ 0 z / n! and its locally defined
inverse log z using algebraic operations and composition.
To facilitate our study, we will illustrate these functions using color pictures. Figure 6.1(b)
shows a polar grid on the plane, where rays are colored using a standard color wheel in
counter-clockwise order beginning along the negative reals: red, yellow, green, cyan, blue,
magenta, red. Circles of radius (1 + ε)n , n = − 6, . . . , 6, are also plotted using a gray scale,
increasing in darkness with the modulus except for the unit circle which is plotted in black
with a thicker line width for emphasis. We will call this picture the standard polar grid .A
“picture” of a complex-valued function f can be created by plotting points z using the same
color that f (z) has on the polar grid. For example, the Figure 6.1(a) shows the plot of a rational
function. The rational function is a map from Figure 6.1(a) to the polar grid in Figure 6.1(b).
Note that the colors near z = 3 cycle twice around in the same order as in the polar grid in
Figure 6.1(b). This means that there is a zero of order 2 at z = 3. The colors near 2i and near
−2i cycle once in the opposite, or clockwise, order. This means that the function has poles of
order one at ±2i. In fact, it is a picture of the function (z − 3)2/(z2 + 4). The preimage of the
unit circle is black.
2
4
1.5
3
2 1
1 0.5
0 0
−1 − 0.5
−2
−1
−3
− 1.5
−4
−2
− 3 − 2 −1 0 1 2 3 4 5 − 2 − 1.5 −1 −0.5 0 0.5 1 1.5 2
(a) (b)
Figure 6.1 A rational function.

82 Elementary Maps
As we shall see, we can better understand some functions by plotting the color of z from
the standard polar grid at the point f (z), where z lies in a region on which f is one-to-one. This
is of course a picture by the first method of the inverse function.
6.1 Linear Fractional Transformations
A linear fractional transformation(or LFT) is a non-constant rational function of the form

az + b
T (z) = , (6.1)
cz + d
where a, b, c, d are complex constants. An LFT is also called a Möbius transformation,
though in some texts a Möbius transformation is an LFT that maps the unit disk onto itself. To
avoid this confusion we will use the term LFT. The representation is not unique since the same
function is obtained if a, b, c, d are replaced by ta, tb, tc, td with t ² = 0. Various normalizations
are used, depending on the application. The four basic types of LFTs are:
translation: T (z) = z + b,
rotation: T (z) = eiθ z,
dilation: T (z) = az, for a > 0, and
inversion: T (z) = z1 .
The translation above shifts every point by the vector b. The rotation rotates the plane by
an angle θ . The dilation expands (if a > 1) or contracts (if a < 1). The inversion is best
understood by writing z = reit . Then the argument of 1 /z is −t and the length of 1 /z is the
reciprocal of the length of z.
An LFT can be built out of these examples using composition. If c = 0 in (6.1) then
a b
T (z) = z+ . (6.2)
d d
± ±
± ±
In this case, T is a dilation by ± da ±, a rotation by arg da , followed by a translation by db . If c ² = 0,
then we can rewrite (6.1) as
T(z) = bc c−2 ad 1
+ ac . (6.3)
(z + d
c)
± ±
± ±
In this case, T is a translation by cd , an inversion, a dilation by ± bcc−2ad ±, a rotation by arg bcc−2ad ,
followed by a translation by ac .
Note that, by (6.2) and (6.3), T is non-constant if and only if bc − ad ²= 0.
Proposition 6.1 The LFTs form a group under composition.
Proof If S is an LFT, then, for constants a, b, ei θ , it is not hard to check that S + b, eiθ S, aS
and 1/S are all LFTs, so, by (6.2) and (6.3), LFTs are closed under composition. The inverse
of (6.1) is easily found to be
dw − b
z=
−cw + a .
If T is given by (6.1) or (6.3) with c ² = 0, then T is meromorphic in C with a simple pole
at − dc . Moreover, T extends to be analytic at ∞ with T (∞ ) = ac . Thus T extends to be a
one-to-one map of the extended plane C∗ onto itself. The next theorem says that LFTs are the
only such maps.
Theorem 6.2 If f is analytic on C \ {z0 } and one-to-one then f is an LFT.

Proof Using a translation, we may assume z0 = 0. Then f has a Laurent series expansion
∞
²
f (z) = an zn.
n=−∞
If f has an essential singularity at 0, then, by Theorem 5.15, the image of every (punctured)
neighborhood of 0 is dense in C. In particular, if B = {z : |z − 1| < 21 } then there is a ζ ∈
/ B
with f (ζ ) in the open set f (B). But then there is z ∈ B with f (z) = f (ζ ), contradicting the
assumption that f is one-to-one. If f has a pole of order n at 0, then 1/f has a zero of order
n at 0. Since 1/ f is one-to-one, we must have n ≤ 1 by Corollary 3.7. Applying the same
argument to f (1/z), we conclude that
a
f (z) = + b + cz,
z
for some constants a, b, c. Set w = f (z) then multiply by z. If a ² = 0 and c ² = 0 then, for
each w,
cz2 + (b − w)z + a = 0
has two roots in C \ {0}, counting multiplicity, contradicting the assumption that f is one-to-
one. Thus either a = 0 or c = 0, but not both since f is non-constant. In either case, f is an
LFT.
In particular, Theorem 6.2 shows that if f is entire and one-to-one, then f (z) = cz + b, for
some constants c and b. Thus the linear functions are precisely the analytic automorphisms of
the plane. If f is an LFT, then f extends to be an automorphism of the extended plane C∗ , or,
via stereographic projection, an automorphism of the Riemann sphere S2 , and it is analytic in
C \ {z0 }, where z0 is the simple pole. By Theorem 6.2, the LFTs are the only analytic auto-
morphisms of the sphere in this sense. Exercise 3.8 characterized the analytic automorphisms
of the disk D. So we have now identified the analytic automorphisms of the disk, the plane
and the sphere. As we will see in Section 15.4, subgroups of these automorphisms will allow
us to transplant the study of analytic and meromorphic functions on any region, indeed on any
Riemann surface, to the study of functions on the disk, plane or sphere.
The LFTs are characterized geometrically by the following theorem. A “circle” or gen-
eralized circle is a circle or a straight line. As we saw in Section 1.3, “circles” correspond
precisely to the circles on the Riemann sphere. Lines lift to circles through the north pole.
Similarly, a “disk” is a region (in C∗ ) bounded by a “circle.”
84 Elementary Maps
Theorem 6.3 LFTs map “circles” onto “circles” and “disks” onto “disks.”
Proof We need only check this for the four basic types of LFTs. If |z −c| = r then |az −ac | =
|a|r and |(z − b) − (c − b)| = r so that rotations, dilations and translations map circles to circles.
The equation of a straight line is given by
Re(c(z − b)) = 0,
since we can translate the line so it passes through 0 then rotate it to correspond to the imag-
inary axis. Rotations, dilations and translations map lines to lines exactly as in the case of
circles. To check that inversions preserve “circles,” suppose | z − c| = r and set w = 1/z. Mul-
tiply out | w1 − c|2 = r2 . If r2 = | c| 2, then Re(2cw) = 1, the equation of a line. If r2 ² = |c| 2,
then by completing the square we obtain
± ±2 ³ ´2
± c ± r
±w + ± = ,
± r − |c| ±
2 2 r2 − |c|2
which is the equation of a circle. A similar reasoning for the image of a line is left to the reader
to verify.
The equation of a disk is found by replacing the equals sign in the equation for a circle with
< or >, so that the proof of the statement for “disks” follows in a similar way.
A useful restatement of Theorem 6.3 is that if T is an LFT and if D is a “disk” bounded by

a “circle” C then T(D) is a “disk” bounded by the “circle” T (C).
The most common example is the one-to-one analytic map from the upper half-plane H
onto the disk D, sometimes called the Cayley transform,
z− i
C(z) = .
z+ i
See Figure 6.2. Note the order of the coloring around i and −i, the zero and pole of C. The
absolute value |C(z)| can be interpreted as the ratio of the distance from z to i over the distance
from z to −i. Level curves (where C has constant modulus) and curves where C has constant
argument are circles, as Theorem 6.3 guarantees. The real line is colored black. Indeed, the
distance from x ∈ R to i is equal to the distance to −i, so that C maps R ∪ {∞} into ∂ D, and
therefore by Theorem 6.3 onto ∂ D. Since C(i) = 0, the image of H must be D by Theorem 6.3.
Another way to see that the image of H is D is to note that if z ∈ H then the distance from
z to i is less than the distance from z to −i, so that |C(z)| < 1, and similarly |C| > 1 on
the lower half-plane. Since C maps C∗ onto C∗ , the image of H must be D. The Cayley
transform can be used, for example, to transform an integral on R to an integral on ∂ D and
vice versa.
Conformality can also be used to determine the image of an LFT. For example, the LFT
given by T (z) = (z − 1)/(z + 1) is real valued on R so it maps the unit circle to a “circle”
which is orthogonal to R at the image of 1, namely 0, and passes through ∞, the image of
−1. Thus it maps the unit circle to the imaginary axis, and, since T(0) = −1, it maps D onto
the left half-plane {Rez < 0}.
The proof of the following theorem is useful for constructing LFTs.
2 2
1.5 1.5
1 1
0.5 0.5
0 0
−0.5 − 0.5
−1 −1
−1.5
1.5
−2 −2
−2 −1.5 −1 −0.5 0 0.5 1 1.5 2 −2 −1.5 − 1 − 0.5 0 0.5 1 1.5 2
Figure 6.2 The Cayley transform.
Theorem 6.4 Given z1 , z2, z3 distinct points in C∗ , and w1 , w2 , w3 distinct points in C∗ , there
is a unique LFT, T, such that
T(zi ) = wi , (6.4)
for i = 1, 2, 3.
In the statement of Theorem 6.4, C∗ is the extended plane, so that ∞ is included in the
possibilities.
Proof First suppose that w1 = 0, w2 = ∞ and w 3 = 1. Set

³ ´³ ´
z − z1 z3 − z2
T(z) = .
z − z2 z3 − z1
Then T (zi ) = wi , i = 1, 2, 3. For the general case, choose LFTs R and S so that R(z1 ) =
S(w1 ) = 0, R(z2) = S(w2) = ∞ and R(z3 ) = S(w3 ) = 1, then
T = S−1 ◦ R
satisfies (6.4) and is an LFT by Proposition 6.1. To prove that T is unique, suppose U is an LFT
satisfying (6.4) then V = S ◦ U ◦ R −1 is an LFT by Proposition 6.1 again, with V(∞ ) = ∞,
V (0) = 0 and V (1) = 1. It is easy then to check that V(z) ≡ z and so U = T .
One additional property of LFTs that is sometimes useful for determining their image is
the following: if T is an LFT, and if z1, z2, z3 are three points on the boundary of a “disk” D
such that D lies to the left of ∂ D as ∂ D is traced from z1 to z2 to z3 , then T (D) lies to the
left of T(∂ D) as it is traced from T(z1 ) to T (z2) to T (z3 ). Indeed, T is one-to-one and hence
locally conformal, so that it preserves the angle, including direction, between the tangent and
(inner) normal directions. For example, the unit disk D lies to the left of the unit circle as it is
traced from 1 to i to −1. If T(z) = (z − 1)/(z + 1) then T(1) = 0, T (i) = i and T(−1) = ∞.
86 Elementary Maps
Thus the image of ∂ D is the imaginary axis by Theorem 6.3, because the imaginary axis is the
unique “circle” through 0, i, ∞ . The image T (D) must lie to the left of the imaginary axis as
it is traced from 0 to i to ∞ , because angles are preserved, including direction. Thus T (D) is
the left half-plane {z : Rez < 0}.
See Penrose and Rindler [19] for the connection between Lorentz transformations and LFTs
in the theory of relativity.
6.2 Exp and Log
In Exercise 2.6, we encountered the function
ez = ex eiy
= ex (cos y + i sin y).
This function maps the horizontal line y = c onto the ray arg z = c from 0 to ∞ , and it maps
each segment of length 2π in the vertical line x = c onto the circle |z| = ec . Figure 6.3(a) is
mapped onto the standard polar grid by the map ez .
e = ez , which is non-zero, so ez has a (local) inverse in a neighbor-
d z
By Exercise 2.7(a), dz
hood of each point of C \{0}, called log z. As we saw in Corollary 5.8(iii), log z can be defined
as an analytic function on some regions which are not just small disks. For example, the func-
tion z is non-zero on the simply-connected region C \ (−∞ , 0]. Then, log z, with log 1 = 0, is
the function given by
log z = log |z| + i arg z, (6.5)
where −π < arg z < π . Figure 6.3(b) is mapped onto the standard polar grid by this function.
If instead we specified that log 1 = 2π i, then (6.5) holds with π < arg z < 3π . If ± =
C \ (S ∪ {0}), where S is the spiral given in polar coordinates by r = eθ , −∞ < θ < ∞, then
± is simply-connected and Im log z is unbounded on ±. In this case we can still specify, for
example, log(−1) = π i, and this uniquely determines the function log z on ±.
1.5
4
3 1
2
0.5
1
0 0
−1
− 0.5
−2
−3 −1
−4
− 1.5
−4 −3 −2 −1 0 1 2 3 4 −1 −0.5 0 0.5 1 1.5 2
(a) (b)
Figure 6.3 Maps ez and log z.

6.3 Power Maps 87
Figure 6.3(a) probably gives more insight into the log function than the inverse image of
a polar grid in Figure 6.3(b). If ± is any simply-connected region in the standard polar grid
which omits the point w = 0, then log z maps it to a corresponding region in Figure 6.3(a). A
horizontal strip of height 2 π in Figure 6.3(a) is the image of a subset C\ R of the standard polar
grid by the function log z, where R is a ray from 0 to ∞ . Any region contained in Figure 6.3(a)
with the property that each vertical line intersects the region in a segment of length at most 2 π
is the image of a simply-connected region in the polar grid by the map log z. The complement
of the spiral r = eθ in the polar grid is mapped by log z to the open strip lying between the
lines y = x and y = x + 2π in Figure 6.3(a). A vertical shift by an integer multiple of 2π is
also the image of the same region by log z, but with a different choice of log(−1).
Note that, with the definition (6.5), log(zw) ² = log(z) + log(w) if arg(z) + arg(w) is not in
the interval ( −π , π ); however, equality holds “modulo 2π i.”
6.3 Power Maps
If ± is a simply-connected region not containing 0, and if α ∈ C, we define
zα =e α log z
,
where log z can be specified by giving its value at one point z0 ∈ ±. Then zα is an analytic
function on ±.
For example, suppose ± = C \ (−∞, 0], and define log 1 = 0. If z = reit , where −π <
t < π , then z1/ 4 = r1/4 eit/ 4. Figure 6.4(a) is the preimage of the standard polar grid, slit along
( −∞, 0] by this map. The image of a sector of the form {z : | arg z| < β } by this map is the
sector {z : | arg z| < β/4}. Points z on the circle | z| = r are mapped to points on the circle
|z| = r 1/4. The map z1/4 is locally conformal in ±, but it is not conformal at 0. Indeed, angles
are multiplied by 1 /4 at 0.
As with the logarithm, it might be easier to understand this map using Figure 6.4(b), which
shows the image of the subset C \ (−∞ , 0] of the standard polar grid by this map. Note the
range of colors and the location of the level lines. There are four possible definitions of z1/4
on C \ (−∞ , 0], depending on the choice of log(1) = 2π ki, where k is an integer. Each of
the remaining three are rotations of Figure 6.4(b) by integer multiples of 2π/4. If we put
1.5 1.5 1.5
1 1 1
0.5 0.5 0.5
0 0 0
−0.5 − 0.5 −0.5
−1 −1 −1
−1.5 − 1.5 −1.5
−1.5 − 1 −0.5 0 0.5 1 1.5 − 1.5 − 1 −0.5 0 0.5 1 1.5 − 1.5 −1 −0.5 0 0.5 1 1.5
(a) (b) (c)
Figure 6.4 Maps z 4 and z1/4 .

88 Elementary Maps
1.5 4 4
1 3 3
2 2
0.5
1 1
0 0 0
−0.5 −1 −1
−2 −2
−1
−3 −3
−1.5 −4 −4
− 1.5 −1 − 0.5 0 0.5 1 1.5 − 4 −3 −2 − 1 0 1 2 3 4 −4 −3 −2 −1 0 1 2 3 4
(a) (b) (c)
Figure 6.5 The map z1+i and its inverse z (1−i)/2 .
the image of all four such functions together, we obtain Figure 6.4(c), which is of course
the picture in the sense of Section 6.1 of the inverse function, z4. Test your understanding by
showing that there are exactly four possible definitions of z1/ 4 in any simply-connected region
not containing 0.
What does the image of C \ (−∞ , 0] by the map z1/π , with log(1) = 0, look like? There
are in fact infinitely many definitions of z1/π on this region.
A more complicated function is
ϕ (z) = z1+i
defined on ±0 = C \ ( −∞, 0] with log(1) = 0. Figure 6.5(a) illustrates this function by the
method of Section 6.1, where we give z the color of ϕ (z) in the standard polar grid.
But perhaps a better way to understand ϕ is to view it as a composition of three functions:
log z, (1 + i)z and ez . The function log z is analytic on ±0 and has image equal to the horizontal
strip ±1 = {z : |Imz √| < π }. The map (1 + i)z rotates ±1 counter-clockwise by an angle of
π/4 and dilates by 2. The image of ±1 by this map is the strip
±2 = {x + iy : x − 2π < y < x + 2π }.
Each vertical line meets ±2 in a segment of length 4 π . The function ez maps a vertical segment
of length 4π onto a circle, covered twice. The line y = x is mapped onto a spiral S given in
polar coordinates by r = eθ . The image of the parallel line y = x + c is the rotation of S
by the angle c. Thus ez is an analytic two-to-one map of ±2 onto C \ {0}, except that S is
covered only once. So ϕ maps the rays arg z = 2c and arg z = 2c − π onto a rotation of S by the
angle c, 0 < c < 2π . Figures 6.5(b) and (c) illustrate this description of the map ϕ defined on
the lower and upper (respectively) half-planes in the standard polar grid. Indeed, they are the
1− i
illustrations by the method of Section 6.1 of the inverse functions z 2 on C \ (S ∪ {0}). Note
how each picture is a continuation of the other across the spiral S.
Test your understanding by describing the image of circles centered at 0 by the map ϕ .
A related function is ψ (z) = z1+i with log(1) = 0 defined on the region ±3 = C \ (S ∪{0}).
From the description above, we can see that the image of ±3 by log z is the strip parallel to the
line y = x which meets each vertical line in a segment of length 2 π , the middle half of ±2.
The image of this region by the map (1 + i)z is a vertical strip ±4 of width 2π . The image of
±4 by the map e is the annulus A = {e
z −π < |z| < eπ }, covered infinitely many times. Thus
ψ maps C \ (S ∪ {0 }) onto A, covering it infinitely many times.
6.4 The Joukovski Map 89
This example highlights the need to define both the domain and the choice of log z to make
sense of the function zα. The function zα has been used to explain a lithograph by M.C. Escher.
See Exercise 6.9.
6.4 The Joukovski Map
The next function we will consider is

1µ 1¶
w(z) = z+ ,
2 z
which is analytic in C \ { 0}. See Figure 6.6. By the quadratic formula,
·
z = w± w2 − 1, (6.6)
so that w is two-to-one unless w2 = 1. Since w(z) = w(1/z), the two roots in (6.6) are
reciprocals of each other, one inside D and one outside D, or else complex conjugates of each
other on ∂ D.
To understand the function w better, we view it as a one-to-one map on various subsets of
C. Fix r > 0 and write z = reit , then
µ ¶ µ ¶ iµ 1¶
w= 21 z + 1z = 12 r + r1 cos t +
2
r − sin t.
r
If we also write w = u + iv, then
¸ ¹2 ¸ ¹2
u v
+ = 1,
1
2
(r + r1 ) 1
2
(r − r1 )
which is the equation of an ellipse, unless r = 1. For each r ² = 1, the circles of radius r and
1/r are mapped onto the same ellipse. The circle of radius r = 1 is mapped onto the interval
[ −1, 1]. We leave as an exercise for the reader to show that the image of a ray from the origin
4 2
3 1.5
2 1
1 0.5
0 0
−1 − 0.5
−2 −1
−3 − 1.5
−4 −2
−4 −3 −2 −1 0 1 2 3 4 −2 −1.5 − 1 − 0.5 0 0.5 1 1.5 2
Figure 6.6 The map (z + 1/z)/ 2.

90 Elementary Maps
to ∞ which is not on a coordinate axis is a branch of a hyperbola which is perpendicular to

each ellipse given above, by conformality.
The function w is a one-to-one analytic map of C \ D onto C \ [−1, 1], and a one-to-one
analytic map of D \{ 0} onto C \ [−1, 1]. The function w is also analytic on H = {z : Imz > 0}
and one-to-one, because Im(1/z) < 0 if z ∈ H. The reader can check that the image of H by
the map w is the region ±2 = C \ {(−∞ , −1] ∪ [1, ∞)}. The inverse of w is given by (6.6)
in each case, but we must make the correct choice for the square root on the image region. If
U is any simply-connected region with ±1 ∈ / U then we can define log(w − 1) so as to be
2
analytic in ± by Corollary 5.8, and thus

·
w+ w2 − 1 = w + e 2 log(w −1)
1 2
is analytic and one-to-one on U with inverse function 21 (z + 1/z).

For example, if U = C∗ \ [−1, 1], write
· º
1
w2 −1 = w 1−
w2
.
·
The function 1 − 1/w2 is analytic and non-zero on U. Define ϕ(w) = 1 − 1/w2 so that
ϕ (∞ ) = 1. More concretely, the image of U by the map 1 − 1 /w is contained in C \ ( −∞, 0]
2
·
so if log ζ = log |ζ | + i arg ζ with −π < arg ζ < π then 1 − 1/w2 is the composition of
ζ = 1 − 1/ w2 and exp( log ζ ), and therefore analytic. Then
1
2
¸ º ¹
1
ψ (w) =w 1+ 1−
w2
is an inverse to (z + 1/z)/2 defined on U. By the preceding discussion, there are two inverses
to (z + 1/ z) /2 on U. Since ψ (w) → ∞ as w → ∞ , ψ must be the inverse to (z + 1/ z)/2 with
range C \ D. See Figure 6.7(a). The inverse with range D \ { 0} is
¸ º ¹
1
ψ1(w) =w 1− 1−
w2
,
which tends to 0 as w → ∞ . See Figure 6.7(b). Note that the pictures are not continuous
across (−1, 1), but are continuous across (−∞ , −1) ∪ (1, ∞ ). Readers are invited to test their
understanding by finding the inverses defined on C \{(−∞ , −1]∪[1, +∞ )}. See Figures 6.7(c)
and (d), both of which are continuous across (−1, 1) but not ( −∞, −1) ∪ (1, ∞). Note also the
ellipses and hyperbolas where the functions have constant modulus and constant argument.
In Section 6.5 we will use the function (z + 1/z)/2 and the closely related function
(z − 1/z)/2, which can be understood as a composition
³ ´
1µ 1¶ 1 1
z − = −i · (iz) + .
2 z 2 (iz)
This function is the composition of rotation by π/2, followed by the map 21 (z + 1/z), followed
by rotation by −π/2. Thus, circles and lines through 0 are mapped to ellipses and orthogonal
hyperbolas. The ellipses have semi-major axes along the imaginary axis. The unit circle is
mapped to the interval [−i, i].
6.5 Trigonometric Functions 91
(a) (b)
2 2
1.5 1.5
1 1
0.5 0.5
0 0
− 0.5 −0.5
−1 −1
− 1.5 −1.5
−2 −2
−2 −1.5 − 1 −0.5 0 0.5 1 1.5 2 − 2 − 1.5 −1 − 0.5 0 0.5 1 1.5 2
(c) (d)
2 2
1.5 1.5
1 1
0.5 0.5
0 0
− 0.5 −0.5
−1 −1
− 1.5 −1.5
−2 −2
−2 − 1.5 − 1 −0.5 0 0.5 1 1.5 2 − 2 − 1.5 −1 − 0.5 0 0.5 1 1.5 2
Figure 6.7 Some inverses of (z + 1/z) /2.
6.5 Trigonometric Functions
We define
eiz + e−iz eiz − e−iz
cos z = and sin z = .
2 2i
Then cos z and sin z are entire functions satisfying
cos z + i sin z = eiz ,
(cos z)2 + (sin z)2 = 1,
d d
cos z = − sin z and sin z = cos z.
dz dz
These functions agree with their usual calculus definitions when z is real. However, we know
by Liouville’s theorem that they cannot be bounded in C. The function cos z is best understood
by viewing it as the composition of the maps iz, ez and 21 (z +1/z). See Figure 6.8. For example,
92 Elementary Maps
6 2
1.5
4
1
2
0.5
0 0
− 0.5
−2
−1
−4
− 1.5
−6 −2
−6 −4 −2 0 2 4 6 −2 −1.5 −1 − 0.5 0 0.5 1 1.5 2
Figure 6.8 The map cos(z).
the vertical strip {z : |Rez| < π } is rotated to the horizontal strip {z : | Imz| < π } by the map
iz. This horizontal strip is mapped onto C \ (−∞, 0] by the map ez . The composition eiz maps
vertical lines to rays from 0 to ∞ and maps horizontal lines to circles. Rays and circles are
mapped by 21 (z + 1/ z) to branches of hyperbolas and ellipses, as we saw in Section 6.3.
Other trigonometric functions are defined using sin and cos, for example
sin z
tan z = .
cos z
Hyperbolic trigonometric functions are also defined using the exponential function:
ez + e−z ez − e−z
cosh z = and sinh z = .
2 2
The inverse trigonometric functions can be found by working backward. For example, to
find arccos z, set z = (eiw + e−iw) /2, multiply by eiw and obtain a quadratic equation in eiw ,
so that, by the quadratic formula,
·
eiw =z± z2 − 1.
Thus
·
w = −i log(z ± z2 − 1).
√
If ± is a simply-connected region such that √ 2±1 ∈/ ± then f (z)2 = 2z + z2 − 1 is an analytic
function, as defined in Section 6.4. If z + z − 1 = 0 then z = z − 1, which is impossible.
Thus f is a non-vanishing function on ±, and, by Corollary 5.8, we can define F(z) = log f (z)
as an analytic function on ±. It satisfies cos(F(z)) = z, for z ∈ ±. Thus cos(z) has an analytic
inverse function, arccos(z), on any simply-connected region which does not contain ±1. The
choice in the definitions of the square root and logarithmic functions are uniquely determined
√
by specifying their values at a point. The second solution to the quadratic equation above,
z − z2 − 1, is just a different choice for the analytic square root. To find the arccos with
arccos(0) = π2 , it may be better to write it in the form
·
arccos(z) = w = −i log(z + i 1 − z2 ),
√
with 1 = 1 and log(i) = iπ/2.
6.6 Constructing Conformal Maps
In this section we will use the functions we have studied in this chapter to construct conformal
maps. In modern usage, the phrase conformal mapmeans a one-to-one analytic map. The
entire function f (z) = ez is locally conformal everywhere, since its derivative is everywhere
non-zero, but it is not a conformal map on C because it is not one-to-one. It is a conformal
map on {z : |Imz| < π }, for example. The list below is not exhaustive, but rather is meant to
illustrate the techniques that can be used at this point in the book.
D onto D with f (z0 ) = 0 and f ³ (z0) > 0 As we saw in Exercise 3.8, the conformal maps of
D onto D are given by
³ ´
z −a
f (z) = c ,
1 − az
where a and c are constants with | a| < 1 and | c| = 1. Set a = z0 then f (z0) = 0. Divide f by
z − z0 and let z → z0 to obtain f ³ (z0 ) = c/(1 − |z0| 2). Setting c = 1 gives f .
Sector A conformal map of a sector ± = {z : a < arg z < b}, with 0 < b − a ≤ 2π , onto
D can be constructed in steps. The function
f (z) = zα =e α log z
,
where α = π/(b − a), will map ± onto a sector with angle at 0 equal to α (b − a) = π , a
half-plane. The choice of log z is already given in the description of ±. A rotation z → eit z
will map the half-plane onto H, and the Cayley transform (z − i)/(z + i) will map H onto D.
It is usually sufficient to describe a conformal map as a composition of a sequence of simpler
conformal maps.
Intersection or union of disks If ± is the intersection of two disks, then, in order to map
± onto D, find the two points c, d where the bounding circles meet. The map
z−c
z −d
1.5 1 1
1 0.5 0.5
0.5 0 0
0 −0.5 –0.5
–0.5 −1 −1
–0.5 0 0.5 1 1.5 −1 –0.5 0 0.5 1 − 1 –0.5 0 0.5 1
Figure 6.9 Four intersections of disks meeting at angle π/ 4.

94 Elementary Maps
will map each disk onto a “disk” with 0 and ∞ on its boundary, and hence the image of ± is
the intersection of two half-planes forming a sector at 0. Now apply the sector map constructed
above. The same construction works for the union of two disks, whose interiors have a point
in common. The region outside the union of two such disks is also the intersection of two
“disks” in the extended plane if we add the point at ∞ .
Figure 6.9 shows four regions, each of which has bounding circles meeting at 0 and 1 + i at
an angle of π/4. The point at ∞ is viewed as an interior point of one of the regions. The map
w = z/(z − (1 + i)) maps each region to a sector by Theorem 6.3. The map ζ = w4 will map
each sector onto the right half-plane, and T(z) = (z − 1)/(z + 1) maps the right half-plane onto
D. One way to see that the image of the right half-plane by the map T is D is to note that, for
each point z on the imaginary axis, the distance from z to 1 equals the distance from z to −1,
so that | T(z)| = 1. By Theorem 6.3, the right half-plane is mapped to a “disk” bounded by the
unit circle. Since 1 is mapped to 0, the image must be the unit disk.
Half-plane with a slit The region ± = H \ I, where I is the segment [0, i] on the imaginary
axis, can be mapped onto H by the map
·
z2 + 1,
√
where −1 = i. Indeed, z2 maps the half-line {reit : r > 0} onto the half-line {r2 e2it : r > 0},
doubling the angle at 0. Thus the image of ± is the slit plane C \ [−1, +∞ ) and z2 + 1
maps ± onto C \ [0, +∞ √ ).2 Similarly, the square root maps half-lines to half-lines, halving
the angle at 0, so that z + 1 maps √ ± onto H. The choice of the square root is uniquely
determined by the requirement that −1 = i. It can be given more explicitly as exp( 21 log z),
where 0 < arg z = Im log z < 2π . Note that the two “sides” of the slit [0, i] correspond to the
intervals [−1, 0] and [0, 1] in the closure of the upper half-plane. The interval [0, +∞ ) in the
√
boundary of the slit plane corresponds to the interval [1, +∞) in the image region, the upper
half-plane. In fact, the map z2 + 1 extends to be a one-to-one analytic map of C \ J onto
C \ [−1, 1], where J = i[−1, 1] is the√union of I and its reflection about R. See Exercise 6.4.
We will use the inverse map given by z2 − 1 in Section 8.1.
Half-plane minus a circular arc Suppose ± = H \ A, where A is an arc containing 0 lying
on a circle C which is orthogonal to R at 0. Then ± can be mapped to H by first applying the
LFT σ (z) = az /(1 − z/b), where b is the other point of intersection of C and R and a > 0. In
fact, b = |d| 2/Red, where d ∈ H is the tip of the arc A. The map σ is real valued on R and
has positive derivative at 0, so, by Theorem 6.3, σ maps H onto H. Also by Theorem 6.3, the
image of A must lie on a “circle” through σ (0) = 0 and ∞ which is perpendicular to R at 0,
since LFTs are conformal. Thus the image of A is an interval [0, ic] on the imaginary axis. We
can choose a > 0 so that c = 1, and then the slit half-plane example applies. The circular slit
is “opened up” to two intervals in R and the real line is mapped to the remaining portion of R.
See Figure 6.10.
Strip To map the strip {z : 0 < Rez < 1} onto D, first apply the map eπ iz . The image of
{Rez = c}, 0 < c < 1, is the ray rei π c , r > 0, so the image of the strip is H. Now apply the
Cayley transform, (z − i)/(z + i).
Half-strip To map the half-strip {z : 0 < Imz < π , Rez < 0} onto D, first apply the
map ez which has image D ∩ H. Indeed, if z = x + iy then ez = ex eiy, so that the half-line
1.5 1.8
2.5
1.6
1.4
2
1
1.2
1 1.5
0.8
1
0.5 0.6
0.4
0.5
0.2
0 0 0
-0.2
-0.2 0 0.2 0.4 0.6 0.8 1 1.2 −1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1 -1.5 −1 -0.5 0 0.5 1 1.5
Figure 6.10 Mapping a circularly slit half-plane.
3
2.5
2
2
1
1.5
0
1
−1
0.5
−2
0
−3
− 1.5 −1 − 0.5 0 0.5 1 −3 −2 −1 0 1 2 3
Figure 6.11 Below a parabola.
{x + iy : x < 0} is mapped onto the portion of the ray arg z = y which lies in |z| < 1. Now
apply the intersection of disks example, or use the map 21 (z + 1/z) and a map of a half-plane
to the disk.
Below a parabola To map H onto the region below the parabola y = ax 2 with a > 0, note
that the image of a vertical line by the map z2 is a parabola. To see this, fix b > 0. Write
u + iv = (b + iy)2 = b2 − y2 + i2by. Then u = b2 − (v/2b)2 , which is the equation of a
parabola that opens to the left. Thus, the image of the half-plane {z : Rez > b > 0} by the
one-to-one map z2 is the region P = {u + iv : u > b2 − (v/ 2b)2}. To find the desired map,
first apply z → −iz + b, which maps H onto the half-plane Rez > b. Then apply z2 to obtain
the region P. Finally, apply the map z → −iz + ib2, which rotates the region clockwise by an
angle of π/2 then shifts vertically by b2 to obtain the region {x + iy : y < (x/2b)2}. Choosing
√
b = 1/(2 a) will give the desired map. See Figure 6.11.
Exterior of an ellipse To map D onto the exterior of an ellipse in C∗ , apply the map
z → r/ z, r > 1, then 21 (z + 1/z). See Section 6.4. In Figure 6.12, the domain, instead of the
range, is given the standard (polar) coloring, so the inverse of the map is used to construct the
picture.
96 Elementary Maps
4 3
3
2
2
1
1
0 0
−1
−1
−2
−3 −2
−4 −3
−4 − 3 −2 − 1 0 1 2 3 4 − 3 −2 − 1 0 1 2 3
Figure 6.12 Exterior of an ellipse.
2.5 1.5
2.5
2 1
2
1.5 0.5
1.5
1 0
1
0.5 −0.5
0.5
0 −1
0
−1.5
−1.5 −1 −0.5 0 0.5 1 1.5 −1.5 − 1 − 0.5 0 0.5 1 1.5 − 1.5 −1 −0.5 0 0.5 1 1.5
Figure 6.13 Region between the branches of a hyperbola.
Region between the branches of a hyperbolaTo map H onto the region between two
branches of a hyperbola, first map H onto a sector symmetric about the y-axis using eit zα with
0 < α < 1 and t = π (1 − α)/ 2. Then apply 21 (z + 1/z). See Section 6.4. In Figure 6.13,
as in Figure 6.12, we have used the standard coloring scheme on the domain, instead of the
range.
See Exercise 6.11 for the construction of a conformal map to a region bounded by one
branch of a hyperbola and Exercise 6.12 for the conformal map to a region above a parabola
y = ax 2, a > 0. The only remaining regions bounded by a conic section are the interiors of
ellipses. We will learn how to map the unit disk onto the interior of an ellipse in Chapter 8.
See Exercise 8.8.
Many other examples can be constructed by using combinations of the above ideas. The
conformal map in each of the examples above is a composition of a sequence of simpler
conformal maps. The inverse map can be found by composing the inverses of the simpler
functions in the reverse order. To find the conformal map of D onto a region ±, it is usually
easier to discover the map from ± onto D, then compute its inverse.
A natural question at this point is: how unique are these maps? A conformal map of ± onto
D can be composed with an LFT of the form given in the first example of this section and still
map D onto D.
Proposition 6.5 If there exists a conformal map of a region ± onto D, then, given any
z0 ∈ ±, there exists a unique conformal map f of ± onto D such that
f (z0 ) = 0 and f ³ (z0 ) > 0.
Proof If g is a conformal map of ± onto D, set a = g(z0). Then

z−a
h(z) = c
1 − az
maps D onto D if |c| = 1, and f = h ◦ g maps ± onto D with f (z0) = 0. Then
f ³ (z0) = cg³ (z0) /(1 − |a|2 ),
so that the proper choice of the argument of c will give f ³(z0 ) > 0. If k also maps ± onto D,
with k(z0 ) = 0 and k³(z0 ) > 0, then H = k ◦ f −1 maps D onto D with H(0) = 0. By Schwarz’s
lemma, | H(z)| ≤ |z| and | H−1 (z)| ≤ |z| so that |H(z)| = |z| and H(z) = cz, with |c| = 1. Since
H ³ (0) = k³ (z0 )/f ³ (z0 ) > 0, we must have c = 1 and k = f .
For example, to find a conformal map ϕ of H onto D such that ϕ (z0) = 0 and ϕ ³ (z0) > 0,
first apply the Cayley transform, then apply an LFT of the form given in the first example of
this section. Here it is actually easier to first apply the map
f (z) = zz −− zz0
0
which maps H onto D because | f (z)| = dist(z, z0) /dist(z, z0) < 1, when z ∈ H, and is > 1
when z ∈ C \ H so that the image of H is D by Theorem 6.3. Then f ³ (z0) = limz→z 0 f (z)/(z −
z0 ) = 1/(2iImz0). So ϕ = if will work.
Another natural question is: what regions can be mapped conformally onto D? Proposi-
tion 6.6 gives a necessary condition.
Proposition 6.6 If ϕ is a conformal map of a region ± onto D, then ± must be simply-

connected.
Proof Suppose γ is a closed curve contained in ±, and suppose a ∈ C \ ±. Let f = ϕ−1 ,

which is analytic on D by Corollary 4.17. Then f ³ /( f − a) is analytic on D and
» »
1 dw f ³(z)
n(γ , a) =
2π i w−a
= 2π1 i f (z) − a
dz = 0,
γ ϕ (γ )
by Cauchy’s Theorem 5.1 or Corollary 4.18. By Theorem 5.7, ± is simply-connected.
In Chapter 8 we will prove that every simply-connected region ± ⊂ C∗, such that C∗ \ ±
contains at least two points, can be conformally mapped onto D. Equivalently, any simply-
connected region ± ⊂ C with ± ² = C can be conformally mapped onto D.
We conclude this chapter by giving a picture of the “world’s greatest function,”
z +1
I(z) = e z−1 ,
98 Elementary Maps
1.5 2
1.5
1
1
0.5
0.5
0 0
–0.5
−0.5
−1
−1
–1.5
−1.5 −2
− 0.5 0 0.5 1 1.5 2 2.5 − 2 −1.5 −1 − 0.5 0 0.5 1 1.5 2
Figure 6.14 “World’s greatest function,” exp ((z + 1) /(z − 1)).
which we have encountered a couple of times so far. This function is a composition of an LFT
which maps the unit disk onto the left half-plane followed by the exponential function. The
preimage of the unit circle is black in Figure 6.14. Each level line is a circle through z = 1,
where it has an essential singularity. The image of each level line covers the corresponding
circle in the standard polar grid infinitely many times.
A Matlab program is given in Section A.2 of the Appendix for creating these color pictures.
It also includes a 3D version, where the third coordinate is given by log | f |. For more color
pictures of analytic functions, see the beautiful text by E. Wegert [27].
6.7 Exercises
6.1 Provide details of the proof of the statement about “disks” in Theorem 6.3. Also provide
the details for the image of a line by an inversion. Then check that the restatement of
Theorem 6.3 given after the proof is correct.
6.2 The region ± = { z : Imz > 0 and |z| > 1} is the intersection of two disks. Map this
region to a sector, then to H. Find a conformal map of H onto H so that the composition
of these maps fixes 1, −1 and ∞ . You should get (z + 1/z)/2. Why?
√ 2 map f of ± = C \ (−∞, 0] onto D such that f (1) = 0 and f (1) > 0.
6.3 Find a conformal ³
6.4 Prove that z + 1 extends to be a conformal map of C \ [−i, i] onto C \ [−1, 1]. Hint:
Consider the map on the lower and upper half-planes separately, but make a choice
of the square roots so that the map extends continuously across R \ {0}. An alterna-
tive approach is to define ¼ J(z) = −iJ(iz) on | z| > 1, where J(z) = 21 (z + 1/z) is the
√
Joukovski map. Then show z2 + 1 = J ◦ ¼ J−1 (z) and use the properties of J developed
in Section 6.4.
6.7 Exercises 99
6.5 Prove there is no one-to-one analytic mapping of {z : 0 < |z| < 1} onto {z : 0 <
|z| < ∞}.
6.6 Find the image of H by the map ϕ given by
ϕ (z) = z0 + z11 +− zz20 ,

and show ϕ is one-to-one on H by writing it as a composition of the elementary maps
z2, z + 1, 1/ z, z0 + (z1 − z0 )z.
6.7 Construct a one-to-one analytic map f of the following regions ± onto D. If z0 is given,
find the map with f (z0 ) = 0 and f ³ (z0) > 0. You may leave your answer as an explicit
sequence of maps, but you must show that each map does what you claim it does.
(a) ± = D ∩ {z : Imz > 21 }, z0 = 34 i;
(b) ± = {z : |z| < 1} ∩ { z : | z − 1 | < 1 } ∩ {z : Imz > 0 };
(c) ± = C∗ \ [0, 1], z0 = ∞. Here f ³ (∞) means limz →∞ z( f (z) − f (∞)).
6.8 Recall Exercise 3.9.

(a) Prove that if σ is a conformal map of D onto D then σ (( −1, 1)) is a circular arc in
D which is orthogonal to ∂ D. Here, as elsewhere, we view a diameter as part of a
circle through ∞.
(b) Suppose that C is a circular arc in D which is orthogonal to ∂ D at a, b ∈ ∂ D. Prove
there is a conformal map of D onto D which maps the interval (−1, 1) onto C so
that the image of 0 is the closest point on C to 0. No computations are needed to
solve (b).
(c) Show that the balls {z : ρ (z, z0 ) < r} in the pseudohyperbolic metric are Euclidean
balls, though the center and radius in the two metrics are different.
(d) Prove that the hyperbolic metric on D is given by
»
δ (z, w) = inf | dζ | ,
γz,w 1 − |ζ |2
where the infimum is taken over all curves γz,w in D connecting z and w. Hint: First
prove it for w = 0, then use Exercise 3.9.
(e) Show that the geodesic, or the shortest curve, in the hyperbolic metric on D between
z and w is the arc between z and w on the unique circle orthogonal to ∂ D through z
and w.
(f) In this non-Euclidean hyperbolic geometry, the “points” are the points in D and the
“lines” are the circular arcs orthogonal to ∂ D. Two lines are parallel if they do not
intersect. Prove that, in this geometry, Euclid’s parallel axiom fails: through any
point p not on a line L there are infinitely many lines which do not intersect L.
6.9 Draw a picture in the square {z = x + iy : 0 < x < π , 0 < y < π }. Reflect the picture
about the line y = 0 and then reflect the union about x = 0. The picture can be extended
1 +i
to the plane by repeated reflections. Now apply the map e 2 z and describe the image.
Use the Matlab program in Section A.2 of the Appendix to visualize this map and its
100 Elementary Maps
inverse. A similar idea was used by M.C. Escher in one of his lithographs. See de Smit
et al. [7].
6.10 (Rain = complement of the Seattle Umbrella) Map C \ D onto
= C∗ \ ({z : |z| ≤ 1 and Imz ≥ 0} ∪ {it : −1 ≤ t ≤ 0})

±
so that ∞ is mapped to ∞ . Extra credit: put a handle of your choice on the umbrella.
6.11 To map a half-plane onto a region bounded by one branch of a hyperbola, note that, for
0 < α < 1, the map ϕ(z) = 21 (z + 1/z) maps a region of the form
½ ¾
Sα = z : | z| > 1, −
πα
2
< arg z <
πα
2
onto the region Hα to the right of a hyperbola, but slit from cos π2α to 1:
¿ ³ ´2 ³ ´2 À
x y Á Â
Hα = z = x + iy :
2 cos πα
− 2 sin π2α
> 1 \ cos
πα
2
,1 .
2
When α = 1, the image region is the right half-plane slit along (0, 1]; call it H1 . We can
map S1 onto Sα using the map zα . Thus f = ϕ ◦ zα ◦ ϕ −1 maps H1 onto Hα . This map
extends to be a one-to-one analytic map of the right half-plane onto the region to the
right of the hyperbola. Supply details for this outline.
6.12 To map H onto the region above the parabola y = ax 2, a > 0, note that z2 maps the
half-strip Sb = { z : −b < Rez < b, Imz > 0} onto the interior of a parabola, slit along
the segment [0, b2):
Ã µ y ¶2 Ä
Pb = z = x + iy : x < b − 2
2b
\ [0, b2 ).
As in Exercise 6.11, find a conformal map of a slit half-plane onto Pb which extends
continuously across the slit, using the inverse of 21 (z + 1/z), log z and z2, composing
√
with linear maps as needed. Supply details for this outline.
6.13 The function g(z) = log(2/ (1 − z)) can be defined so that it is analytic and one-to-
one on D. Draw a picture of the image by considering a sequence of simpler maps.
Show that Img is continuous on ∂ D but that Reg is not continuous. In fact, Reg is not
bounded. (For Fourier series enthusiasts, the imaginary part of the power series for g
gives a Fourier series of a continuous function whose conjugate Fourier series is not
even bounded.)
The following two exercises are not hard, if you get the right idea.
6.14 Suppose C and D are tangent circles, one inside the other. Find a circle C 1 which is
tangent to both C and D, then for n ≥ 2 find Cn tangent to C, D and C n−1 , with
{Cj , j = 1, . . . , n} bounding disjoint open disks. Prove that this process can be continued
indefinitely and that the points of tangency of the Cn all lie on a circle. See Figure 6.15.
6.15 Suppose C and D are non-intersecting circles, one inside the other. Choose C 1 tangent to
both C and D. Again choose Cj tangent to C, D and Cj−1 , with {Cj , j = 1, . . . , n} bound-
ing disjoint open disks. Under some circumstances, the chain of circles may “close up.”
In other words, Cn is also tangent to C1 for some n > 1. Show that, for each n ≥ 3, there
6.7 Exercises 101
C1 C2 C3 C
C3
1
C9
C C
D
D C4
Figure 6.15 Tangent circles.
is a choice of C and D so that the chain always closes up. Given circles C and D, show
that the property of “closing up” does not depend on the choice of C 1. See Figure 6.15.
It is known that, given a collection of disks in a simply-connected region whose
interiors are disjoint, there is collection of disks in the unit disk with the same pat-
tern of tangencies. If the disks are small enough, the map from one set of disks to the
other approximates the conformal map between the region and the unit disk. See K.
Stephenson [26] for an introduction to this subject.
6.16 (Mercator projection) Let π −1 denote the map from the unit sphere to the complex
plane given by the inverse of stereographic projection. Cut the plane along (−∞, 0] and
apply the map L(z) = i log z. By Exercise 1.11, the composition L ◦ π −1 is a map of
the (slit) sphere onto a vertical strip which preserves angles between curves. In other
words, if two curves on the unit sphere meet at an angle θ when viewed from inside the
sphere, then the images of these curves in the strip meet at the same angle. The map
M(w) = −i log π −1(w), w ∈ S2 , then preserves angles between curves when viewed
as if we are standing on the outside of the sphere or above the strip. The north pole
corresponds to the top of the strip, the south pole corresponds to the bottom of the strip,
and vertical lines correspond to great circles through the north and south poles, called
longitude lines. A horizontal line intersects the strip in a segment which corresponds
to a circle with constant angle of elevation, called a latitude line. The map M, called
the Mercator projection, was of great commercial value because it allowed ships to sail
beyond the sight of land. This is because a line between two points on the strip (the map)
corresponds to a path on the sphere with constant compass heading, by conformality.
The local stretch in distances, however, does depend on the direction.
Show that the ratio of the vertical stretch to the horizontal stretch on a latitude line
with elevation φ is given by sec φ , so that the height of the latitude line on the strip
(map) is given by
» φ
sec t dt. (6.7)
0
Mercator, perhaps unaware of the connection with stereographic projection, deduced
(6.7) geometrically via Riemann sums, which he used to estimate the value. It was an
unsolved problem for 40 years to find this integral in terms of elementary functions. Use
the map M to find the integral. Calculus students typically learn how to do the integral
with a trick, unaware of the great importance of the integral.
PART II
7 Harmonic Functions
Harmonic functions arise in applications as solutions of Laplace’s equation

2 2
±u ≡ ∂∂ xu2 + ∂∂ yu2 = 0.
They are useful for the study of analytic functions in two ways: if f is analytic then Ref is
harmonic and, more importantly, log | f | is harmonic where f ± = 0. The most useful property of
harmonic functions is the mean-value property, and its consequence, the maximum principle.
To avoid difficulties with proving differentiability properties, we take the mean-value property
as definition, then give the connection with Laplace’s equation in Theorem 7.12.
7.1 The Mean-Value Property and the Maximum Principle
In this section, we define harmonic and subharmonic functions, prove the maximum principle,
and then show how to find the values of a harmonic function on a disk from its values on the
bounding circle.
Definition 7.1 A continuous real-valued function u is harmonic on a region ² ⊂ C if for

each z ∈ ² there is an r z > 0 (depending possibly on z) such that
± 2π
1
u(z) = u(z + reit )dt (7.1)
2π 0
for all r < rz .
Equation (7.1) is called the mean-value property. The mean-value property says that the
value of u at the center of the disk is equal to the average of its values on the boundary of the
disk. This is interesting because it says that if you know u on the circle, then you can recover
its value inside the disk, at least at the center. We learn later how to recover values at other
points of the disk. Harmonicity is a local property because we only require the mean-value
equality on sufficiently small disks, not necessarily all disks contained in ². It is true though
that the mean-value property holds for all circles which bound a closed disk contained in ²,
if u is harmonic. See Corollary 7.6.
Definition 7.2 A continuous function u with values in [−∞, +∞) is subharmonic on a

region ² if for each z ∈ ² there is an r z > 0 (depending possibly on z) such that
± 2π
u(z) ≤ 21π u(z + reit )dt (7.2)
0
for all r < rz .
In some texts, the continuity assumption in Definition 7.2 is replaced by upper semi-
continuity, though we do not need this extra flexibility. See Exercise 7.14. Equation (7.2)
is called the mean-value inequality. Note that we allow a subharmonic function to take the
value −∞ but not +∞ . The mean-value inequality is trivially satisfied for those points z
where u(z) = −∞. If u is continuous and u(z) > −∞ then u is bounded below in some
neighborhood N of z, so that the integral in (7.2) is defined for sufficiently small r > 0. We
can also define the integral of u on any circle in ² as the decreasing limit of the integrals of
un (z) = max( −n, u(z)).
The reader can easily verify that if u and −u are subharmonic then u is harmonic. If u1
and u2 are harmonic then A 1u1 + A2 u2 is harmonic, for real numbers A 1, A 2, and |A 1u1| is
subharmonic. If u is subharmonic then Au is subharmonic provided A > 0. If u1 and u2 are
subharmonic then u(z) = max(u1 (z), u2(z)) is subharmonic.
For example, an analytic function f has the mean-value property, as can be seen by expand-
ing in a (local) power series and interchanging the order of integration and summation. Then,
by taking real and imaginary parts, u = Ref and v = Imf are harmonic. It also follows that | f |
is subharmonic. Harmonicity, like analyticity, is a local property, so that log | f | is harmonic
on ² \ { f = 0} since log | f | is the real part of log f , which can be defined to be analytic in
a neighborhood of any point where f ± = 0. This also shows that log | f | is subharmonic on
². Of these examples, perhaps the most important for the study of analytic functions is the
subharmonicity of log | f |. We will use subharmonic functions to study harmonic functions as
well as analytic functions.
The next result is perhaps the most important elementary result in complex analysis.
Theorem 7.3 (maximum principle) Suppose u is subharmonic on a region ². If there exists

z0 ∈ ² such that
u(z0) = sup u(z), (7.3)

z∈ ²
then u is constant.
A subharmonic function can have a local maximum, but the function will be constant in a
neighborhood of any point where the local maximum occurs, by Theorem 7.3. For example,
u(z) = max(0, log |z| ) is subharmonic in C and has a local maximum at each point of D.
Proof Suppose (7.3) holds. Set E = {z ∈ ² : u(z) = u(z0)}. Since u is continuous, E

is closed in ². By (7.3), the set E is non-empty. We need only show E is open, since ² is
connected. If z1 ∈ E, then by (7.2)
± 2π
1
[u(z1 ) − u(z1 + reit )]dt ≤ 0, (7.4)
2π 0
for r < rz1 . But the integrand in (7.4) is continuous and non-negative by (7.3) and hence
identically 0 for all t and all r < rz 1 . This proves E is open and hence equal to ².
Another form of the maximum principle is as follows:
Corollary 7.4 If u is a non-constant subharmonic function in a bounded region ² and if u

is continuous on ² then
max u(z)
z ∈²
occurs on ∂ ² but not in ².
The reader should verify the alternative form: if u is continuous and subharmonic on ² then
lim sup u(z) = sup u(z).

z→∂ ² ²
If ² is unbounded, then ∞ must also be viewed as part of ∂ ². The function u(z) = Rez is
harmonic on ² = {z : Rez > 0} and satisfies u = 0 on ∂ ² ∩ C but u is not bounded by 0.
Theorem 7.5 (Schwarz) If g is real valued and continuous on ∂ D, set

±
1 2π
1 − |z|2
u(z) = it
2π 0 |eit − z|2 g(e )dt,
for z ∈ D. Then u is harmonic in D and
lim u(z) = g(ζ ), (7.5)

z→ζ
for all ζ ∈ ∂ D.
Proof The function
±
G(z) = 1 2π
eit + z g(eit )dt
2π 0 eit −z
is analytic on D, as can be seen by expanding the kernel
eit
∞
+ z = 1 + e−it z = 1 + 2 ² e−int zn (7.6)
eit − z 1 − e−it z 1
and interchanging the order of summation and integration. The identity

³ ´
1 − |z|2
= Re eeit +− zz
it
(7.7)
|eit − z|2
shows that u = ReG and hence u is harmonic. Note also that if g ≡ 1, then G ≡ 1, since
µ −int
e dt = 0 if n ± = 0. Thus for all z ∈ D
±
1 2π 1 − |z|2
2π 0 |eit − z|2 dt = 1. (7.8)
To prove (7.5) fix t0 and ε > 0 then choose δ > 0 so that |g(eit ) − g(eit0 )| < ε if t ∈I =
δ
{t : |t − t0 | < δ }. Then, using (7.8),

¶ 1 ± 2π 1 − |z|2 ¶
|u(z) − g(e )| = ¶¶ 2π
it0
| eit − z| 2
(g(e it
) − g(e it 0
))dt
¶
¶
0
± ±
2
1 − |z|
≤ ε 21π |eit − z|2 dt + M(z) |g(eit) − g(eit )| 2dtπ ,
0
Iδ ∂ D\Iδ
where
1 − |z| 2
M(z) = sup .
{ t:|t −t0|≥ δ} |eit − z|2
The first term is at most ε by (7.8). Moreover, M(z) → 0 as z → eit 0 because |eit − z| is
bounded below for | t − t0| ≥ δ . Thus u(z) → g(eit0 ) as z → eit0 .
The proof of Schwarz’s theorem shows that we need assume only that g is integrable on ∂ D
and continuous at ζ for (7.5) to hold. The kernel
= 21π |1eit−−|zz||2
2
P z (t)
µ
is called the Poisson kerneland u = PI(g) ≡ Pz (t)g(eit )dt is called the Poisson integral
of g. If B is a disk and g is continuous on ∂ B, then PIB (g) denotes the harmonic function on
B which equals g on ∂ B, and is called the Poisson integral of g on B. The reader can give a
formula for PI B(g) by transplanting the problem to the unit disk using a linear map. Note that
by (7.8) if g is integrable and satisfies m ≤ g ≤ M then m ≤ PIB (g) ≤ M. The Poisson kernel
for the upper half-plane H is given by
³ ´
1 v 1 1
PH (s) = = π Im ,
w
π (u − s)2 + v2 s−w
where w = u + iv, v > 0 and s ∈ R. This can be proved using the change of variables
z = (w − i)/(w + i), eit = (s − i) /(s + i) in Schwarz’s Theorem 7.5.
The next result shows how to find the values of a harmonic function in the disk from its
values on the boundary.
Corollary 7.6 If u is harmonic on D and continuous on |z| ≤ 1, then for z ∈ D

±
1 2π 1 − |z|2
u(z) = 2π
it
|eit − z|2 u(e )dt. (7.9)
0
Proof Let U(z) denote the right-hand side of (7.9). Then by Schwarz’s theorem u − U is
harmonic on D, continuous on D and equal to 0 on ∂ D. By the maximum principle applied to
u − U and U − u, we conclude u = U.
If u is harmonic in a region ² and if D is a closed disk contained in ², then, by a linear

change of variables and Corollary 7.6, the mean-value property holds for ∂ D. See Exercise 7.1.
The next corollary shows how to recapture an analytic function from the boundary values
of its real part.
Corollary 7.7 If u is harmonic on | z| < 1 and continuous on |z| ≤ 1, then

±
1 2π
eit + z it
f (z) = u(e )dt
2π 0 eit − z
is the unique analytic function on D with Ref = u and Imf (0) = 0.
The function (eit + z)/ (eit − z) is called the Herglotz kerneland the integral is called the
Herglotz integral.
Proof By the first part of the proof of Schwarz’s Theorem 7.5, f is analytic. The real part of
f is then equal to u by (7.7) and Corollary 7.6. If g is another analytic function with Reg = u,
then f − g is purely imaginary and hence not an open mapping. Thus f − g is constant. Finally
µ
note that f (0) = u(eit )dt is real, so that if g(0) is real then g = f .
It follows from Corollary 7.7 that a harmonic function on a region ² is the real part of
an analytic function on each disk contained in ². One consequence is that it is not hard to
prove that if u is harmonic on a disk and if f is analytic then u ◦ f is harmonic. Indeed,
harmonicity is local and on a disk u = Reg for some analytic function g, by Corollary 7.7.
Hence u ◦ f = Re(g ◦ f ) is harmonic. If u = Rez and f = z2 then f ◦ u is not harmonic. The
function u = log |z| is harmonic on ² = {z : 0 < |z| < ∞} and is the real part of an analytic
function on each disk that does not contain 0, but it is not the real part of an analytic function
on all of ² because arg z is not continuous on ².
An application to analytic functions is given in Corollary 7.8.
Corollary 7.8 (jump theorem) Suppose f is an integrable function such that f : ∂ D → C.

Let ±
f (ζ ) d ζ
F(z) = .
|ζ |=1 ζ − z 2π i
Then F is analytic on C \ ∂ D and, for | ζ | = 1,
· ³ ´¸
1
lim F(z) − F = f (ζ )
z →ζ z
at all points of continuity ζ of f .
The function F is called the Cauchy integralor Cauchy transformof f . The jump theorem
says that the analytic function F jumps by f (ζ ) as z crosses the unit circle at ζ . Note that if
f is analytic on D then F = f in D and F = 0 in | z| > 1 by Cauchy’s integral formula,
Theorem 5.2.
Proof We already proved that F is analytic off ∂ D in Lemma 4.30. To prove the corollary,
just manipulate the integrals:
³ ´ ± ³ ´
1 1 dζ 1
F(z) − F = f (ζ ) −
z | |=1ζ ζ −z ζ 2π i − 1/z
±
z − 1/z dζ
= f (ζ )
( − z)( − 1 z) 2 πi
| |=1ζ ζ ζ /
± 2 π
1 − |z| 2 dt
= f (eit ) it
|e − z|2 2π .
0
Applying Schwarz’s Theorem 7.5 to the real and imaginary parts of f completes the proof.
µ 2π
If g is integrable on ∂ D, set an = 21 π 0 g(eit )e−int dt. Then
∞
²
aneint
n=−∞
µ 2π
is called the Fourier series ofg. Note that | an | ≤ 1
2π 0
|g(eit )|dt. By (7.6) and (7.7)
1 − |z| 2
∞
² ∞
²
=1+ e−int zn + eint zn.
|eit − z|2 n=1 n=1
Interchanging the order of summation and integration, the harmonic “extension” of g to D is
given by
∞
² ∞
²
G(z) ≡ PI(g)(z) = a0 + a nz n
+ a−n zn.
n=1 n=1
In other words, G is found from the Fourier series of g by replacing eit with z and e−it with
z. The interplay between convergence of the Fourier series of g and the harmonic functions
ReG and ImG is delicate. One connection is given by Schwarz’s Theorem 7.5, and another by
Abel’s limit theorem, Theorem 3.13, which implies that if the Fourier series for g converges
at eit 0 then G has that series sum as its non-tangential limit at eit 0 .
Fourier series were first developed by Fourier because he wanted to solve the following
problem. Find a harmonic function u which satisfies 0 < u < 1 in the half-strip S = {x + iy :
x > 0, 0 < y < π } with u(0, y) = 1, 0 < y < π , and u(x, 0) = u(x, π ) = 0 for x > 0. If a
thin plate of homogeneous material in the shape of a long strip is heated to one temperature
on the top and bottom, and another temperature on the end, and allowed to reach a state of
equilibrium, then the resulting temperature is harmonic on the strip. Fourier’s idea was that
e−nx sin ny = Im(−e−nz ) is harmonic and equal to 0 on the top and bottom edges of S. So, set
∑
u(z) = ∞ n=1 cn e
−nx sin ny and choose cn so that
∞
²
1 = u(0, y) = cn sin ny. (7.10)
n=1
The coefficients ck can be determined by integrating both sides of (7.10) against sin ky by
orthogonality. The right-hand side of (7.10) is called a Fourier sine series. Alternatively,
because the right-hand side of (7.10) is odd, define F(eiy ) = 1 for y ∈ (0, π ) and F(eiy ) = −1
for y ∈ (−π , 0). Then the coefficients cn can be found by adding the n and −n terms of the
Fourier series for F. There are convergence issues though. We will not prove these facts, but
instead give another solution to Fourier’s problem in Section 7.3.
7.2 Cauchy–Riemann and Laplace Equations
If f (z) = u(z) + iv(z) we will sometimes use the notation f (x, y) = u(x, y) + iv(x, y), where
z = x + iy with x, y real and u(x, y) and v(x, y) are real valued. If f is analytic then, by the
definition of the complex derivative,
f (x + h, y) − f (x, y)
f ²(z) = hlim = f x (x, y) = ux (x, y) + ivx (x, y)
→0 h
= lim f (x, y + k) − f (x, y) = 1 f (x, y) = v (x, y) − iu (x, y).
k→0 ik i
y y y
Thus
ux = vy and uy = −vx . (7.11)
Equations (7.11) are called the Cauchy–Riemann equations.
There is another useful notation for first-order derivatives. Write z = x + iy and define
∂f ( )
fz ≡
∂z
≡ 12 f x − if y
and
∂f 1 ( )
fz ≡ ≡ fx + ify .
∂z 2
Then it is an easy exercise to verify the chain rule:
(f ◦ g)z = f z ◦ g gz + fz ◦ g gz
and
(f ◦ g)z = fz ◦ g gz + f z ◦ g gz .
As a mnemonic device, think of z and z as independent variables (even though one is the
complex conjugate of the other). View f as f (z, z) and view the composition of f and g as
f (g(z, z), g(z, z)). Then formally differentiate using the two-variable chain rule. It may also
help to write 1i instead of −i and − 1i instead of i since the “denominators” contain z = x + iy
and z = x − iy, respectively.
The Cauchy–Riemann equations can be restated in this terminology as
fz = 0.
The Cauchy–Riemann equations characterize analytic functions by the following theorem.
Theorem 7.9 Suppose u and v are real valued and continuously differentiable on a region ².
Then u and v satisfy the Cauchy–Riemann equations (7.11) if and only if f = u + iv is analytic
on ².
Proof We have already proved that if f = u + iv is analytic then u and v satisfy the Cauchy–
Riemann equations. Conversely, if u and v satisfy the Cauchy–Riemann equations, then, by
Taylor’s theorem applied to the real-valued functions u and v,
f (x + h, y + k) − f (x, y) = hux (x, y) + ku y (x, y) + i(hvx (x, y) + kvy (x, y))
+ ε(h, k),
¹
where ε (h, k)/ |h|2 + |k|2 → 0 as |h|, | k| → 0. Dividing by h + ik and applying the Cauchy–
Riemann equations, we obtain
f (x + h, y + k) − f (x, y)
√ lim
h + ik
= ux (x, y) − iuy (x, y).
|h| 2+|k |2→0
So f ² (z) exists and is continuous, and therefore f is analytic.
A harmonic function on a region ² is the real part of an analytic function on each disk
contained in ² by Corollary 7.7. So harmonic functions have continuous partial derivatives
of all orders. A harmonic function is not always the real part of an analytic function on all of
². However, the Cauchy–Riemann equations allow us to associate another analytic function
with a harmonic function as in the following theorem.
Theorem 7.10 A function u is harmonic on a region ² if and only if ux − iuy exists and is
analytic on ². If ² is simply-connected then u is harmonic on ² if and only if u = Ref for
some f analytic on ².
Proof If u is harmonic on ² and if B is a disk contained in ² then, by Corollary 7.7, f =

u + iv for some analytic function f on B. Moreover, f ² is analytic and f ² = ux + ivx = ux − iuy
by the Cauchy–Riemann equations. This proves that ux − iuy exists and is analytic on each B
and hence on ². Conversely if g = ux − iuy exists and is analytic on ², then g has a power
series expansion on any disk B ⊂ B ⊂ ². Integrating the series term by term gives an analytic
function f with f ² = g. If w = Ref then, by the Cauchy–Riemann equations, wx = ux and
wy = uy , so that u = w + c on B, where c is a constant. Since w = Ref is harmonic, u must
also be harmonic on B and hence on all of ².
If u is harmonic on a simply-connected region ², then, by Corollary 5.8(ii) and the third
sentence in this proof, there is an analytic function f on all of ² such that f ² = ux − iuy . By
the Cauchy–Riemann equations, w = Ref and u have the same partial derivatives on ² and
so u = Re( f + c) for some constant c.
Finally, if f is analytic on ², then the mean-value property holds for f and hence for u =
Ref . So u is harmonic.
If f = u + iv is analytic in a region ² then v is called a harmonic conjugateof u in ².

Because −if is analytic, −u is a harmonic conjugate of v. If v1 and v2 are harmonic conjugates
of u on a region ² then v1 − v2 is constant because the difference of two analytic functions is
an open map, if non-constant, and hence cannot be purely imaginary on an open set.
If f is analytic then f has continuous partial derivatives of all orders. By the Cauchy–
Riemann equations applied to u = Ref and v = Imf ,
= (vy )x = (vx )y = (−uy )y = −uyy ,
uxx (7.12)
and hence uxx + uyy = 0. Similarly, vxx + vyy = 0.
Definition 7.11 The Laplacian of u is the second-order derivative given by ± u = uxx + uyy .
We say that u satisfies Laplace’s equationon a region ² if u has continuous second-order
partial derivatives (including the mixed partials) and ± u = 0 on ².
The next result relates Laplace’s equation to harmonicity, and shows that both are
essentially the same as the maximum principle.
Theorem 7.12 Suppose u is real valued and continuous on a region ². Then the following
are equivalent:
(i) u is harmonic on ²,
(ii) u satisfies Laplace’s equation on ²,
(iii) if B is an open disk with B ⊂ B ⊂ ² and if v is harmonic on B, then u − v and v − u
satisfy the maximum principle on B.
Proof We have already seen that (i) implies (ii). By the maximum principle (i) implies (iii).
If (iii) holds and if B ⊂ B ⊂ ², then let v be the Poisson integral of u|∂ B on B. Then v is
harmonic on B and u − v and v − u are equal to 0 on ∂ B by Schwarz’s Theorem 7.5. By (iii),
u = v on B. This proves u is harmonic on each B ⊂ ², and hence (i) holds.
Finally we show that (ii) implies (i). Set g = ux − iuy . Now if R is a rectangle with sides
µ
parallel to the axes contained in ², we claim that ∂ R g(ζ )dζ = 0. To see this, note that
± ± ±
(ux − iuy )(dx + idy) = ux dx + uy dy + i ux dy − uy dx.
∂R ∂R ∂R
By the fundamental theorem of calculus applied to each segment in ∂ R, the first integral is
zero. Also by the fundamental theorem of calculus, integrating along horizontal lines in R and
vertical lines in R, the second integral can be rewritten as
±
(uxx + uyy )dxdy,
R
which is also equal to 0 by (ii). (We have just followed a proof of the simplest form of Green’s
theorem.) By Morera’s theorem, g = ux − iuy is analytic on ², and by Theorem 7.10, u is
harmonic.
See Exercise 7.13 for weaker conditions sufficient for analyticity and harmonicity.
For an application of the Cauchy–Riemann equations, recall that if f is analytic and
f ² (z0) ± = 0 then f is locally conformal. That is, f preserves angles between curves passing
through z0, in both magnitude and direction. See Section 3.2. Corollary 7.13 gives a converse
to this statement.
Corollary 7.13 If f is continuously differentiable (with respect to x and y) on a region ²,

and if f preserves angles between curves at each point of ², then f is analytic in ² and f ² ± = 0
on ².
Proof Suppose z0 ∈ ² and θ ∈ [0, 2π ]. Set γ (t) = z0 + teiθ and w(t) = f (γ (t)).
Because f preserves angles between curves at z0 , the angle between w(t) and γ (t) at t = 0,
arg(w² (0)/γ ² (0)), does not depend on θ . By the chain rule,
w² (t) = fzγ ² (t) + f z γ ² (t)
= fzei + fz e−i ,
θ θ
so that
w² (0)
γ ² (0)
= f z + f z e−2i ,θ
which traces a circle with radius | fz (z0)| as θ varies from 0 to π . Because arg(w² (0)/γ ² (0)) is
constant, fz (z0) = 0. By Theorem 7.9, f is analytic. An analytic function does not preserve
angles where f ² = 0, as described in Section 3.2, and the corollary follows.
As a consequence of the proof of Corollary 7.13, note that the local stretching of γ (t) at
t = 0 is given by |w ²(0)/γ ² (0)|. So if instead of preserving angles we assume that the local
stretch is the same in all directions and non-zero at each point z0 ∈ ², then either fz = 0 or
fz = 0 at z0 . Because f is continuously differentiable, either f or f is analytic in a neighborhood
of each point of ². The closed sets {w ∈ ² : fz (w) = 0} and {w ∈ ² : fz (w) = 0} are disjoint
because the local stretch is non-zero at each point. Because ² is connected, either f or f is
analytic in ². Moreover, the non-zero local stretch at z0 is given by | f ² (z0)| or |( f ) ²(z0 )|.
7.3 Hadamard, Lindelöf and Harnack
The maximum principle can be used to give more information than is immediately apparent
from its statement.
Theorem 7.14 (Hadamard’s three-circles theorem)Suppose f is analytic in the annulus

A = {z : r < |z| < R}. Let m = lim sup|z |→r | f (z)| and M = lim sup| z|→R | f (z)|, and suppose
m, M < ∞. If z ∈ A, then
| f (z)| ≤ M ω (z)
m1−ω(z),
where ω(z) = log(|z|/r) / log(R/r).
Theorem 7.14 is called the three-circles theorem because ω (z) depends only on t = |z|, so
the theorem relates the max on the circle of radius t to the max on the two bounding circles.
Another way to state the conclusion is that if M(s) = sup |z|= s | f (z)| then log M(s) is a convex
function of log s.
Proof The function ω log M + (1 − ω) log m is harmonic on A, equal to log M on |z| = R

and equal to log m on |z| = r. Thus u = log | f | − (ω log M + (1 − ω) log m) is subharmonic
on A and lim supz→∂ A u(z) ≤ 0. By the maximum principle, u ≤ 0 in A.
Another way to solve Fourier’s problem in Section 7.1 is to note that
ϕ (z) = (z + 1/z) ◦ ez = ez + e−z

is a conformal map of the half-strip S = {x + iy : x > 0, 0 < y < π } onto the upper half-
plane H, which maps the vertical end of the strip to the interval (−2, 2). If −∞ < a < b < ∞
−z is harmonic in H and equal to the angle θ formed at z ∈ H
then the function θ (z) = Im log ab−z
between line segments from z to b and from z to a. See Figure 7.1.
a b
Figure 7.1 The angle θ (z) as a harmonic function.
The angle θ approaches π as z → (a, b) and approaches 0 as z → R \ [a, b]. Thus

³ ´
1 2−z
u(z) =
π
arg
−2 − z ◦ (ez + e−z)
solves Fourier’s problem.
It is natural to ask if this function is the only solution. The answer is provided by an
extension to the maximum principle, called Lindelöf’s maximum principle .
Theorem 7.15 (Lindelöf) Suppose ² is a region and suppose {ζ1 , . . . , ζ n } is a finite subset
of ∂ ², not equal to all of ∂ ². If u is subharmonic on ² with u ≤ M < ∞ on ² and if
lim sup u(z) ≤ m,
z ∈²→ζ
for all ζ ∈ ∂ ² \ {ζ1, . . . , ζn}, then u ≤ m on ².

In the statement of Lindelöf’s theorem, if ² is unbounded, then we view ∞ as a boundary
point, which may or may not be one of the exceptional points {ζ j }.
Proof First suppose that ² is bounded and let d = diam(²). For ε > 0 set
¶ ¶
n
² ¶ z − ζj ¶
uε (z) = u(z) + ε ¶
log ¶ ¶. (7.13)
d ¶
j=1
Then uε is subharmonic in ², uε ≤ u and uε → −∞ as z → ζ j , for j = 1, . . . , n. Thus

lim supz →∂ ² uε (z) ≤ m, and so by the maximum principle uε ≤ m on ². Fix z and let ε → 0
in (7.13) to obtain u(z) ≤ m.
If ² is not bounded, we may suppose that ζ j ± = ∞ for j = 1, . . . , n by composing with an
LFT if necessary. Given ε > 0, we can choose R so that R > maxj |ζ j| and u(z) ≤ m + ε for
z ∈ ² ∩ {|z| > R }. Now apply the bounded case to u − ε on ² ∩ {|z| < R } to conclude that
u − ε ≤ m on ² ∩ {|z| < R} and hence on ². Let ε → 0 to conclude u ≤ m on ².
Lindelöf’s maximum principle shows that there is a unique bounded harmonic function
solving Fourier’s problem since the difference of two such solutions is bounded and harmonic
with boundary values equal to 0 except at three boundary points, 0, π i and ∞ . Similarly, if
I j = (aj , bj ), j = 1, . . . , n, are disjoint intervals on R then
³
u(z) =
n
² cj
arg
bj − z´
j=1
π aj −z
is the unique bounded harmonic function on H with boundary values cj on I j and 0 on R \∪I j .
Another consequence of Lindelöf’s maximum principle is the three-lines version of

Hadamard’s theorem, which plays an important role in complex interpolation theory of opera-
tors. It follows from the proof of Hadamard’s three-circles theorem and Lindelöf’s maximum
principle.
Corollary 7.16 If u is subharmonic on the strip S0 = { z = x + iy : 0 < x < 1} set

m0 = lim supRez →0 u(z) and m1 = lim supRez →1 u(z). If u ≤ M < ∞ on S0 then u(z) ≤
m1x + m0(1 − x).
For example, suppose g and h are continuous functions on a compact set X and suppose dµ
is a positive measure on X. Then the function
±
F(z) = |g|pz |h|q(1−z) dµ
X
is analytic on S0 for p, q > 0. This follows from Morera’s theorem after interchanging the
order of integration. Thus
¶± ¶
¶ pz q(1 −z)
¶
log ¶ | g| |h|
¶ dµ¶¶
X
is subharmonic on S0, and bounded above. By Corollary 7.16, for 0 < x < 1,
¶± ¶ ± ±
¶ pz q(1 −z)
¶
log ¶ |g| |h|
¶ dµ¶¶ ≤ x log | g| pdµ + (1 − x) log |h|q dµ.
X X X
If 1/p + 1/ q = 1 with p > 1 then set z = x = 1/p and exponentiate to obtain
± ³± ´ p1 ³± ´ q1
|gh |dµ ≤ |g| dµ
p
|h| dµ
q
,
X X X
which is called Hölder’s inequality.
Positive harmonic functions cannot grow too quickly in the unit disk, as the following result,
known as Harnack’s inequality, shows.
Theorem 7.17 (Harnack) Suppose u is a positive harmonic function on D. Then

³ ´ ³ ´
1−r 1+r
u(0) ≤ u(z) ≤ u(0),
1+r 1−r
for | z| = r < 1.
Proof We may assume u is harmonic on D by replacing u with u(sz), s < 1, and then
letting s → 1. Because u then satisfies the Poisson integral formula (7.9), we first estimate the
Poisson kernel:
1−r
= (11 −+ rr)2 ≤ |1eit−−|zz||2 ≤ (11 −− r)r 2 = 11 −+ rr .
2 2 2
1+r
Then, because u is positive and the mean-value property holds,
³ ´ ± ³ ´
1−r 2π 1 − |z|2 dt 1+r
u(0) ≤ u(eit ) ≤ u(0).
1+r 0 | e − z|
it 2 2π 1−r
A similar estimate can be used to show that two positive harmonic functions on D vanishing
on a interval of ∂ D must approach 0 at the same rate.
Theorem 7.18 (boundary Harnack inequality)Suppose u and v are positive harmonic

functions on D which extend to be continuous and equal to 0 on a closed arc I ⊂ ∂ D. Let
Uδ = {z ∈ D : dist(z, ∂ D \ I) > δ > 0}. Then for z ∈ Uδ
2 ³ ´ ³ ´
u(0) u(z) 4 u(0)
δ
4 v(0)
≤ v(z)
≤ δ2 v(0)
.
Proof Fix z ∈U δ with |z| < r < 1 and set δ r = dist(z/r, ∂ D \ I). By the Poisson integral
formula,
±
u(z) 1 u(reit )
1 − |z/r| 2
= dt
∂ D |e − z/r|
2π it 2
± ±
1 u(reit ) 1 u(reit )
≤ 2π ∂ D\I δ r 2
dt + 2π I | e − z/r |2
it
dt
±
u(0) 1 u(re it )
≤ δr2
+ 2π |eit − z/r|2 dt.
I
Similarly,
±
v(z) v(reit )
1 − |z/r| 2
= 21π |
eit − z/r|2 dt
∂ D
± ±
v(reit ) v(reit )
≥ 21π 4
dt +
1
2π |eit − z/ r|2 dt
∂ D\ I I
± ±
v(reit ) v(reit )
= v(0)
4
− 21π 4
dt +
1
2π | eit− z/ r|2 dt.
I I
Taking the ratio and letting r → 1, we obtain the right-hand inequality, since δr → dist(z, ∂ D\
I) > δ > 0, and u(reit ) and v(reit ) converge uniformly to 0 on I. The left-hand inequality is
proved by reversing the roles of u and v.
Corollary 7.19 Let K be a compact subset of a region ². Then there exists a constant C
depending only on ² and K such that if u is positive and harmonic on ² then for all z, w ∈ K
1
u(w) ≤ u(z) ≤ Cu(w). (7.14)
C
The point of Corollary 7.19 is that the constant C can be taken to be independent of the
function u.
Proof If B is a disk, let 2B be the disk with the same center as B and twice the radius.
Suppose B is a disk such that 2B ⊂ ². Let ϕ be a linear map of D onto 2B, then by Harnack’s
inequality applied to u ◦ ϕ we have that (7.14) holds for z ∈ B and w equal to the center of
B, with C = 3. Thus (7.14) holds for all z, w ∈ B with C = 9. Cover K by a finite collection
of disks B = {B j } with 2Bj ⊂ ². Add more disks if necessary so that ∪Bj is connected. If
Bj , B k ∈ B with Bj ∩ Bk ± = ∅ then (7.14) holds on B j ∪ Bk with C = 81. Because there are

only finitely many disks and because their union is connected, (7.14) holds on ∪{Bj : Bj ∈ B },
and therefore on K, with a constant C depending only on the number of disks in B , and not
on u.
We end this section with a simple but powerful consequence. We will use Theorem 7.20 in
Sections 8.2 and 13.1.
Theorem 7.20 (Harnack’s principle) Suppose {un} are harmonic on a region ² such that
un (z) ≤ un+1 (z) for all z ∈ ². Then either
(i) limn→∞ un(z) ≡ u(z) exists and is harmonic on ², or
(ii) limn→∞ un(z) = +∞ ,
where convergence is uniform on compact subsets of ². In case (ii), this means that given
K ⊂ ² compact and M < ∞ there is an n 0 < ∞ such that un (z) ≥ M for all n ≥ n0 and
z ∈ K.
Proof By assumption, if n > m then un − um ≥ 0, and by the maximum principle un − um

is strictly positive or identically 0. Fix z0 ∈ K, where K is compact. By Corollary 7.19 there
is a constant C such that
1
(un(z0 ) − um (z0 )) ≤ un(z) − um (z) ≤ C(un(z0 ) − um (z0 )),
C
for all z ∈ K. Thus {un(z0 )} is Cauchy if and only if {un } is uniformly Cauchy on K. Thus if the
increasing sequence {un (z0 )} converges, then {un} converges uniformly on compact subsets of
². Similarly if u n(z0 ) → ∞ then un (z) → ∞ uniformly on compact subsets of ². The limit
function u is harmonic by the mean-value property (see Exercise 7.1).
The assumption that the sequence un is increasing is essential. Note that one needs only
to prove the sequence is bounded at one point to prove uniform convergence on compact
subsets.
7.4 Exercises
7.1 (a) If u is harmonic on a region ² and if B is a closed disk contained in ² then the
average of u on the boundary of B is equal to the value of u at the center of B. Hint:
Use Corollary 7.6 and a linear change of variables. It is also possible to prove this
using Exercise 7.5 and Theorem 7.12.
(b) Use (a) to show that if {un } are harmonic on a region ² and converge uniformly on
compact subsets of ² to a function u then u is harmonic. This approach using the
mean-value property is easier than working with Laplace’s equation.
7.4 Exercises 119
7.2 Suppose v is subharmonic on D and continuous on D. Suppose u is harmonic on D and

continuous on D such that v ≤ u on ∂ D. Then v ≤ u in D. Use this to prove that
± 2π ± 2π
v(reit )dt ≤ v(seit )dt,
0 0
for r < s ≤ 1. Conclude that if v is subharmonic on a region ² then v satisfies the mean-
value inequality for each circle that bounds a closed disk contained in ². In particular,
if f is analytic on D then
± 2π ± 2π
| f (re )| dt ≤
it p
| f (seit)|pdt
0 0
for r < s < 1 and p > 0. A similar inequality holds for v(z) = log | f (z)|. Hint: Apply
the Poisson integral formula to max( −n, v).
7.3 Show that the one-variable analogs of harmonic and subharmonic are linear and convex
in two ways: using the second derivative and using the maximum principle on intervals.
7.4 (a) Prove that in polar coordinates (r, θ )
³ ´³ ´ ³ ´³ ´
r 2
±u = r
∂
∂r
r
∂
∂r
u+
∂
∂θ
∂
∂θ
u.
(b) Prove that if u is harmonic in C \ {0} and if u depends only on r (not θ ) then
u = a log r + b, where a and b are constants.
7.5 (a) Prove the following version of Green’s theorem: suppose that ² is a bounded region,
bounded by finitely many C 2 simple closed curves. If u and v are twice continuously
differentiable on ² then
± ³ ´ ±
∂v ∂u
u
∂η
− v ∂ η |dz| = (u±v − v±u)dxdy,
∂² ²
where η(ζ ) is the unit normal at ζ ∈ ∂ ² pointing out of the region ², |dz| is arc-
length measure on ∂ ² and dxdy is area measure on ². If you already know a proof
of some version of Green’s theorem, you may use it to derive this version.
(b) Prove that (ii) implies (i) in Theorem 7.12 by directly verifying that the mean-value
property holds. Hint: In D \ B(0, ε ) set v = log | z|, apply (a) and let ε → 0.
(c) Similarly, show that if v is twice continuously differentiable and ±v ≥ 0 then v is
subharmonic because the mean-value inequality holds.
(d) If g is differentiable on R and g²² > 0 then prove that v(reit ) = g(log r) is
subharmonic.
7.6 Prove that the following are equivalent for a real-valued twice continuously differen-
tiable function v on a region ²:
(a) v is subharmonic,
(b) ±v ≥ 0,
(c) if B is a disk with B ⊂ B ⊂ ² and if u is harmonic on B then v − u satisfies the
maximum principle.
Hint: See Exercise 7.5 and the proof of Theorem 7.12.
7.7 (a) Prove a “Schwarz’s lemma” for harmonic functions. If u is real valued and harmonic
in D with |u| ≤ 1 and u(0) = 0, then
|u(z)| ≤ π4 arctan |z|.

When can equality hold?
(b) Formulate and prove an upper bound for u(z), where u is harmonic on D with u(0) =
0 and u ≤ 1.
Hint: Use a conformal map.
7.8 If u(x, y) is harmonic in the unit disk D = {z = x + iy : |z| < 1}, define
ºz z »
f (z) = 2u , − u(0, 0).
2 2i
In other words, formally replace the real variables x and y in the two-variable Taylor
series for u by the complex variables α and β then set α = z/2 and β = z/2i. Where
will the resulting series converge? Prove that f is well defined (i.e., that this substitution
makes sense) and analytic in D and Ref = u. Note that this procedure does not work
for the functions u = log(x2 + y2) and Re z1 which are harmonic in D \ {0}. However, if
u is harmonic in a neighborhood of (x0, y0 ) then, for z0 = x0 + iy0 ,
f (z) = 2u((z − z0)/ 2 + x0, (z − z0 )/(2i) + y0 ) − u(x0, y0)
satisfies Ref = u. Try this procedure for a few familiar functions such as ex cos y and
Re(zn ).
7.9 Suppose u is harmonic in C and satisfies
|u(z)| ≤ M |z|k ,
when |z| > R. Show that u is the real part of a polynomial of degree at most k.
7.10 (a) Suppose u is harmonic in a punctured ball B(z0, r) \ { z0 }. Prove
u(z)
lim
z→z0 log |z − z0|
=0 (7.15)
if and only if u extends to be harmonic in B(z0, r). This is the “harmonic” ver-
sion of Riemann’s removable singularity theorem. In particular, if u is bounded and
harmonic on the punctured ball, then u extends to be harmonic on B(z0 , r).
(b) Suppose u is non-constant and harmonic in C. Prove
M(r)
lim
r →∞ log r
= ∞,
where M(r) = sup|z |=r u(z).
7.11 Suppose v is twice continuously differentiable and subharmonic in C. Prove that
M1(r)
lim
r→∞ log r
exists (possibly infinite), where M1(r) is the mean value of v on |z| = r.
7.4 Exercises 121
7.12 Suppose u is harmonic in C and satisfies
u(z) ≤ A|z| + B,
where A and B are constants. Prove u = ax + by + c, where a, b and c are constant. Note
that the assumption is about u not | u| .
7.13 (Weyl’s lemmafor analytic functions.) A continuous function f is weakly-analytic on
a region ² provided
±
f
∂ϕ
∂z
dA =0
²
for all compactly supported continuously differentiable functions ϕ on ². Here dA

denotes area measure. Prove that a continuous function is weakly-analytic if and only if
it is analytic. Fill in the details of the following outline.
(a) If R is a rectangle and ψ is continuously differentiable on R then

± ±
ψz dA = −i ψ (z)dz.
R ∂R
Apply this to ψ = f ϕ on a union of squares to deduce that an analytic function is

weakly-analytic.
( )2
(b) Set ϕ (z) = π3 1 − |z|2 on µ|z| < 1 and 0 elsewhere. Then show that ϕ is
continuously differentiable and C ϕ dA = 1.
(c) Suppose f is weakly-analytic on ². Set ϕδ (z) = δ12 ϕ( zδ ) and
±
f δ (z) = f (w)ϕδ (z − w)dA(w).
C
If R is a rectangle with R ⊂ R ⊂ ² prove that for δ sufficiently small
±
fδ (z)dz = 0.
∂R
Hint: Use (a) and ∂∂w ϕδ (zµ − w) = − ∂∂z ϕδ (z − w).

(d) Prove | fδ (z) − f (z)| ≤ ε ² ϕδ (z − w)dA(w), for δ sufficiently small.
(e) Conclude that fδ is analytic and converges uniformly to f on compact subsets of ²,
and so f is analytic on ².
(f) Formulate and prove a similar result for harmonic functions using the Laplacian
instead of ∂/∂ z̄.
7.14 In the definition of subharmonic, the continuity assumption can be replaced by upper
semi-continuity. Let ϕδ be as in Exercise 7.13. Suppose v is upper semi-continuous on
a region ² and suppose v satisfies the mean-value inequality on sufficiently small disks
contained in ². Define
±
vδ (z) = v(w)ϕδ (z − w)dA(w),
C
for z ∈ ² and δ < dist(z, ∂ ²).
(a) Prove that vδ is subharmonic and decreases to v (eventually) on compact subsets

of ².
(d) Use part (a) or check the proofs directly using the definition of upper semi-
continuity to show that the results of this chapter remain valid for this weaker
definition of subharmonic.
8 Conformal Maps and Harmonic
Functions
Conformal maps can be used to transfer problems on a complicated region to related problems
on a simpler region. Using the inverse of the conformal map, the solutions on the simpler
region can be transferred back to the original region. For example, suppose ± is a simply-
connected region such that a conformal map ϕ of D onto ± extends to be a one-to-one and
continuous map of D onto ±. If f is a continuous function on ∂ ±, then we can find a harmonic
function u on ± which extends to be continuous on ± and equal to f on ∂ ±. To accomplish
this, let g = f ◦ ϕ on ∂ D and let v be the Poisson integral of g. Then u = v ◦ ϕ −1 is the desired
solution.
In this chapter we give the geodesic zipper algorithmfor computing a conformal map
of a simply-connected region whose boundary contains a prescribed (finite) collection of
points, onto the upper half-plane H. Lindelöf’s maximum principle for harmonic functions
is used to bound the constructed curve between these points. We then use the geodesic zipper
algorithm and Harnack’s principle to prove the Riemann mapping theorem, which says that
every simply-connected region, other than C, can be conformally mapped onto the unit disk.
In Section 8.3, we explore symmetric regions and how symmetry of the region is reflected
in symmetry of conformal maps. Finally, we give the Schwarz–Christoffel formula for the
conformal map of the upper half-plane onto a region bounded by a polygonal curve.
8.1 The Geodesic Zipper Algorithm
Given distinct points z0 , z1, . . . , zn ∈ C, we can construct a simple closed curve γ whose
image, also denoted by the symbol γ , contains these points and conformal maps of the half-
plane H onto the two regions in C∗ bounded by γ as follows. Let γ1 be the line segment from
z0 to z1 and let ±1 = C∗ \ γ1. Then
ϕ1 (z) = z0 + z11 +− zz20 (8.1)
defines a conformal map of H onto ±1 such that ϕ1 (∞) = z0 and ϕ1(0) = z1 . See Exercise 6.6.
If z2 ∈/ γ1, let σ2 be the unique circular arc (image) in H from 0 to ϕ1−1(z2 ) which is orthogonal
to R. As described in Section √ 6.6 (see Figure 6.10), we can find a conformal map ϕ2 of H
onto H \ σ 2 using the map z2 − 1 of H onto H \ [0, i], followed by an LFT which maps
H onto H and maps the vertical segment [0, i] onto σ2 . Note that ϕ2 (0) = ϕ1−1(z2 ). Then
γ 2 ≡ ϕ1 (σ2) is a simple arc (image) from z1 to z2 and ϕ1 ◦ ϕ2 is a conformal map of H onto
∗
±2 ≡ C \ (γ 1 + γ2 ). Similarly if z3 ∈ / γ1 ∪ γ 2 , let σ3 be the unique circular arc (image) in H
from 0 to ϕ− 2
1
◦ ϕ
−1(z3 ) which is orthogonal to R and let ϕ3 be the conformal map of H onto
1
124 Conformal Maps and Harmonic Functions
ϕ1 ◦ ϕ2 ◦ ϕ3
γ4
z4
z3
z2
0 z1
ϕ4 z0
−1 0 1
Figure 8.1 Construction of γ4 and ϕ4.
H \ σ3 with ϕ3(0) = ϕ2−1 ◦ ϕ1−1 (z3). Then γ3 = ϕ1 ◦ ϕ2(σ3 ) is a simple arc (image) from z2 to
z3 and ϕ1 ◦ ϕ2 ◦ ϕ3 is a conformal map of H onto ±3 ≡ C∗ \ (γ1 + γ2 + γ3). Repeating this idea,
we obtain a conformal map ϕ = ϕ1 ◦ ϕ2 ◦ · · · ◦ ϕn of H onto C∗ \ (γ1 + γ2 + · · · + γn), where
γ1 + γ2 + · · · + γ 2 is (the image of) a simple curve from z0 to zn passing through z1, . . . , zn−1.
See Figure 8.1.
Next, let σn+1 be the semi-circle (or half-line) in H from 0 to ϕ−1(z0 ) ∈ R ∪ {∞} which is
orthogonal to R. Let γn+1 = ϕ (σn+1). Then γ ≡ γ1 + γ2 +· · ·+ γn+1 is (the image of) a simple
closed curve containing z0, z1, . . . , zn. The semicircle (or half line) σn+1 divides H into two
regions, D+ and D− , and so ϕ(H \ σn+1) is the union of two disjoint simply-connected regions
+ − ∗ + −
± and ± with C = ± ∪ ± ∪ γ . Finally we can find an LFT, τ , of H onto H so that
√ √ √
τ (σn+1 ) is the positive imaginary axis. Using the maps z and − −z, with 1 = 1, followed
by τ −1 we obtain conformal maps of H onto D+ and D−. Composing with ϕ yields conformal
maps ϕ + and ϕ− of H onto ±+ and ±− , respectively. Each map in the full composition has an
explicit inverse map, and composing the inverse maps in the reverse order produces conformal
maps of ±+ and ±− onto H.
This process can be applied to any finite sequence of distinct points z0, . . . , zn unless the
points are out of order, in the sense that some zj already belongs to γ1 + · · · + γj−1 . When this
happens, zj can√ 2be skipped.
The map z − 1 extends to be analytic and one-to-one in a neighborhood of each of the
intervals in R\{− 1, 0, 1}. LFTs are conformal everywhere. It follows that ϕ+ and ϕ− extend to
be one-to-one continuous maps of H ∪ R ∪{∞} onto the closures of ±+ and ±− , respectively.
This is perhaps clearer if the map ϕ is examined on D+ and D− separately.
The curve γ also has a continuously turning (unit) tangent. This follows because if σ is
a curve whose image is orthogonal to R at 0, then z2 − 1 applied to σ has an image which
is tangential to R√
√
at z = − 1. Because z is analytic in a neighborhood of −1, the image α
of σ by the map z2 − 1 is tangential to the interval [0, i] at z = i, and thus α meets [0, i]
at an angle of π . Because conformal maps preserve angles between curves, any subsequent
composition with conformal maps √ 2 on H will preserve this angle of intersection. Because γ
is built from compositions of z − 1 and LFTs, it must have a continuously turning (unit)
tangent. The tangent is also continuous at z0 because σn+1 is orthogonal to R not only at 0
but also at its other endpoint on R ∪ {∞}, and that angle is preserved by ϕ2 ◦ ϕ3 ◦ · · · ◦ ϕn.
But ϕ1 then doubles the angle to π because of the presence of z2 in (8.1). It might be easier
to understand orthogonality and the behavior of z2 when the point in question is at ∞ by
using the LFT −1/z of H onto H, which interchanges ∞ and 0. Because the (unit) tangent
direction is continuous, the curve γ is continuously differentiable when it is parameterized by

arc-length.
The constructed curve γ can be described in the language of hyperbolic geometry.
Definition 8.1 If ± ⊂ C∗ is simply-connected, then γ is called a hyperbolic geodesicin ±

if there is a conformal map ψ of the upper half-plane H onto ± such that γ = ψ (I), where I
is the positive imaginary axis. The image of an arc contained in a hyperbolic geodesic will be
called a geodesic arc.
For example, a subarc in H of a circle orthogonal to R is a geodesic arc using an LFT

and Theorem 6.3. See Exercises 3.8, 3.9 and 6.8 for more information about the hyperbolic
geometry on D.
Definition 8.2 An (open) simple arc γ is called analytic if there exists a function g which
is one-to-one and analytic in a neighborhood N of the open interval (0, 1) with g(t) = γ (t),
0 < t < 1. A simple closed curve σ is called analytic if there exists a one-to-one analytic
function g defined in a neighborhood of ∂ D with g(eit ) = σ (eit ), 0 ≤ t ≤ 2π . When the
parameterization of an arc or closed curve is not specified, we say that it is analytic if it is
analytic for some parameterization.
See Exercise 8.2(f) for an equivalent local definition of an analytic arc. By definition, then,
a geodesic arc is the image of an analytic arc. In particular, it has a parameterization that is
infinitely differentiable.
The image of the curve γ = γ1 + · · · + γn+1 consists of a line segment from z0 to z1
followed by the images of arcs γj , j = 2, . . . , n + 1. The image of the arc γj is, by construction,
a hyperbolic geodesic in the region ±j = C∗ \ (γ1 + · · · + γj −1) and therefore γj is analytic
except at its endpoints. We have proved the following:
Theorem 8.3 Given distinct points z0 , z1 , . . . , zn ∈ C, the geodesic algorithm constructs a

continuously differentiable and piecewise analytic simple closed curve γ whose image con-
tains these points, together with conformal maps of H onto the two regions in C∗ \ γ , as well
as the inverses of these maps.
Given a large number of data points z0, . . . , zn on a simple closed curve we can use this
algorithm on a computer to find conformal maps from H onto the two regions bounded by the
piecewise geodesic curve. A natural question, then, is: what does the curve look like between
the data points? Is it close to the closed curve from which we took the data points? The proof
of Theorem 8.4 is a typical use of harmonic functions and the maximum principle to estimate
analytic functions.
Theorem 8.4 (Jørgensen) Suppose ² is a closed disk contained in a simply-connected

region ±. If J is a hyperbolic geodesic in ±, then J ∩ ² is connected and J ∩ (∂ ²) consists of
at most two points.
a b 0
ψ
d 0
c
Δ τ
Figure 8.2 Proof of Jørgensen’s theorem.
Jørgensen’s theorem says that closed disks contained in ± are strictly convex in the
hyperbolic geometry on ±.
Proof Because J is a hyperbolic geodesic, there is a conformal map ψ of C \ [0, ∞) onto

± such that ψ ((−∞ , 0)) = J. Here we composed the conformal map in Definition 8.2 with
√z. We may suppose that J ∩ ² consists of more than one point. Let c be the first intersection
of J with ² and let d be the last intersection. So there are a < b < 0 with ψ (a) = c ∈ ∂ ²,
ψ (b) = d ∈ ∂ ², ψ (( −∞, a)) ∩ ² = ∅ and ψ ((b, 0)) ∩ ² = ∅. By Theorem 6.3, we can
choose a constant λ ∈ C so that the LFT τ (z) = λ (z − c)/(z − d) maps the interior of ² onto
the right half-plane, {z : Rez > 0}. Then τ (∂ ± ∪ J \ ² ) ⊂ {z : Rez < 0}. See Figure 8.2.
Set
± ²± ²
ψ (z) − c z−b
f (z) = λ .
ψ (z) − d z−a
Then f is bounded and analytic in U = C \ ((−∞, a] ∪ [b, ∞)). Moreover, if x ∈ (−∞, a) ∪
(b, ∞) then
± ²
x −b
lim sup Ref (z) = lim sup Re τ (ψ (z)) < 0.
z∈ U→x x − a z∈ U→x
By Theorem 7.15, Lindelöf’s maximum principle, Ref (z) < 0 in U. But if a < x < b, then
(x − b)/(x − a) < 0 so that
± ²
Re
ψ (x) − c
= Re τ (ψ (x)) > 0.
ψ (x) − d
λ
Thus ψ (x) ∈ interior(²), proving Theorem 8.4.
It follows from Jørgensen’s theorem that the geodesic arc between c, d ∈ ± lies in the
intersection of all closed disks ² ⊂ ± with c, d ∈ ∂ ². This is a “lens” shaped region given
by the intersection of two disks.
Corollary 8.5 Suppose D is an open disk contained in a simply-connected region ± ⊂ C∗

and suppose J is a geodesic in ± which is tangent to ∂ D at z 0 ∈ ±; then J ∩ D = ∅ .
Proof If z1 ∈ J ∩ D then we can find a slightly smaller closed disk ² ⊂ D ∪{z0 } ⊂ ± tangent
to J at z0 and with z1 in the interior of ² . By Jørgensen’s theorem, the portion of J between
z0 and z1 is contained in ². Because ² is a compact subset of ±, we can rotate ² slightly
about the point z0 and remain inside ±, yet still contain both z0 and z1 . Then the portion of
J between z0 and z1 must be contained in the intersection of these rotations by Jørgensen’s
theorem. This is impossible if J is tangent to ∂ D at z0.
Corollary 8.5 can be helpful in locating a geodesic. If z0 ∈ ± and if J is a geodesic in ±

containing z0 , then let D1 and D2 be the largest possible disjoint open disks contained in ±
and tangent to J at z0. Then J does not intersect D1 ∪ D2.
Using Jørgensen’s theorem we can build approximations to conformal maps on a simply-
connected region.
Definition 8.6 A disk-chain D0, D1 , . . . , Dn is a sequence of pairwise disjoint open disks

such that ∂ Dj is tangent to ∂ Dj+1 , for j = 0, . . . , n − 1. A closed disk-chainis a disk-chain
such that ∂ Dn is also tangent to ∂ D0 .
Any simple closed polygon P, for example, can be covered by a closed disk-chain of disks
with arbitrarily small radii and centers on P, for example by placing disks centered at each
vertex, then covering each of the remaining open line segments with disks centered on the
segments. Another method for constructing a disk-chain is to cover the plane with a hexagonal
grid of disks with radius ε > 0. Then select a sequence of these disks to form a disk-chain.
Yet another method is to use the curves γ constructed in the proof of Runge’s Theorem 4.23,
for a bounded simply-connected region ±. Pave the plane with squares of side length d and
shade the squares whose closure is contained in ±. Let σ be the boundary of a connected
component of the interior of the union of the closed shaded squares. Because σ is the union of
edges of squares with sides of length d, disks of diameter d centered at the endpoints of each
square edge in σ then form a closed disk-chain. See Figure 8.3.
If D0, D1 , . . . , Dn is a closed disk-chain, set
zj = ∂ Dj ∩ ∂ Dj+1 ,
for j = 0, . . . , n, where Dn+1 ≡ D0 . The next theorem gives one geometric method for
describing the contours in the constructed curve γ from the geodesic zipper algorithm.
Figure 8.3 A disk-chain.

Theorem 8.7 (Marshall-Rohde) If D0, D1 , . . . , Dn is a closed disk-chain, then the geodesic

zipper algorithm applied to the data z0 , z1 , . . . , zn produces explicit formulae for the confor-
mal maps ϕ ± of the upper half-plane H onto the two regions bounded by a continuously
differentiable and piecewise analytic curve γ with
n
³
γ ⊂ (Dj ∪ zj ).
0
The algorithm also produces explicit formulae for the inverses of ϕ± .
We emphasize, though, that the geodesic zipper algorithm does not require the construction
of a disk-chain. Disk-chains give a simple geometric method for locating the constructed
curve. See also Exercise 8.10.
Proof Set
j
³
Uj = { z0 } ∪ (Dk ∪ {zk }),
k =1
for j = 1, 2, . . . , n.
Since γ1 is a straight-line segment, γ1 ⊂ U1 and γ1 is not tangent to ∂ D1 at z1 . We proceed
by induction. Suppose now that γ1 + · · · + γj −1 ⊂ Uj −1. By Corollary 8.5, γj−1 is not tangent
to ∂ Dj−1 at zj−1 . Because γ1 + · · · + γj has a continuous tangent and Dj is tangent to Dj −1 at
zj−1, the initial portion of the curve γj near zj−1 is contained in Dj . If j < n + 1, find a slightly
smaller closed disk ² ⊂ Dj ∪{zj } so that zj ∈ ∂ ² and ² ∩ γj contains at least two points. Then
² is a closed disk in ±j−1 . See Figure 8.4. By Jørgensen’s theorem, γ j ⊂ Uj . If j = n + 1, note
that both γn and γ0 are not tangent to ∂ Dn+1. Since γ1 + · · · + γn+1 has a continuous tangent,
we can apply Jørgensen’s theorem on slightly smaller closed disks ² ⊂ Dn+1 to conclude
that γn+1 ⊂ Dn+1 ∪ { zn, z0 }. Theorem 8.7 follows.
Figure 8.5 shows conformal maps of grids on the disk and its complement to the interior
and exterior of a simple closed curve using the geodesic zipper algorithm. Given a simply-
connected region ± in the plane, the geodesic zipper algorithm can be used to approximate
a conformal map of D onto ±, by using lots of points on the boundary. Exercise 8.10 shows
that, for a given number of points, a more judicious choice of these points can lead to more
accurate approximations than using, for instance, approximately equally spaced points on the
boundary or the points of tangency of disks constructed from the approximation of ± by a
union of squares.
z −1
j
γ j
z
j
γ −1
j Δ
D
D −1
j
j
Figure 8.4 A smaller disk ² ⊂ Dj .

Figure 8.5 Conformal mapping using the geodesic zipper algorithm.
8.2 The Riemann Mapping Theorem
Our next major goal is to prove the Riemann mapping theorem, that every simply-connected
region contained in the plane (except the plane itself) can be mapped, one-to-one and ana-
lytically, onto the unit disk. The original approach to this problem, due to Riemann, was to
minimize a certain integral (with side conditions) over a collection of functions. This method
had the advantage of a physical interpretation, but it assumed that there was a minimizer.
Weierstrass then showed that there are similar minimization problems with no minimizer.
Existence proofs for this and other related problems dominated mathematical thinking in
analysis for the next 50 or so years. Around 1900, Hilbert finally patched up the difficul-
ties in Riemann’s approach, though the proof does not apply to all simply-connected regions.
See J. Gray [11] for a discussion of some of the history.
We give a somewhat constructive proof of the Riemann mapping theorem in this section.
Other proofs will be given in Chapters 10 and 13. The idea of the proof here is to construct,
using the geodesic zipper algorithm, conformal maps from an increasing sequence of regions
onto the disk, then prove that the sequence converges to a conformal map of the union of these
regions onto the disk.
We first prove some preliminary results. The next two results, due to Hurwitz, are typical
applications of the argument principle and will be key to finding solutions to many extremal
problems.
Theorem 8.8 (Hurwitz) Suppose {gn}∞ n=1 is a sequence of analytic functions on a region ±
and suppose g n(z) ² = 0 for all z ∈ ± and all n. If gn converges uniformly to g on compact
subsets of ±, then either g is identically zero in ± or g is never equal to 0 in ±.
Proof The limit function g is analytic on ± by Weierstrass’s theorem. In particular, if g is

not identically zero, then the zeros of g are isolated in ±. Moreover, again by Weierstrass’s
theorem, g³n converges to g³ uniformly on compact subsets of ±. If ² ⊂ ± is a closed disk
with g ² = 0 on ∂ ², then g³n /gn converges to g³ /g uniformly on ∂ ² . By the argument principle,
for n sufficiently large, the number of zeros of gn in ² is the same as the number of zeros of
g in ². Because gn is never zero, the theorem follows.
Corollary 8.9 (Hurwitz) If {gn }∞ 1 is a sequence of one-to-one and analytic functions on

a region ±, and if gn converges to g uniformly on compact subsets of ±, then either g is
one-to-one and analytic on ± or g is constant.
Proof Fix w ∈ ± and apply Hurwitz’s Theorem 8.8 to g − g(w) on ± \ {w}.

The following lemma says that, to prove convergence of a sequence of analytic functions,
it is enough to prove convergence of their real parts, together with the full sequence at one
point.
Lemma 8.10 Suppose f n is analytic on a region ±n with ±n ⊂ ±n+1 ⊂ ± = ∪ n±n . Sup-

pose: z0 ∈ ±1 ; {Ref n} converges uniformly on compact subsets of ±; and { fn (z0)} converges.
Then { fn } converges uniformly on compact subset of ±.
In the statement of Lemma 8.10, uniform convergence of f n on a compact set K ⊂ ± means

that f n is defined on K for n ≥ n0 = n0(K) and { fn }∞
n0 converges uniformly on K.
Proof If f is analytic on D, then, by Corollary 7.7 for | z| ≤ r < 1,

´µ ´
´ 2π eit + z ´
| f (z) − iImf (0)| = ´´ eit − z Ref (eit)dt ´´´ ≤ 11 +− rr sup |Ref (z)|.
´ (8.2)
0 |z|=1
If gn is analytic on D and {Regn } is a Cauchy sequence in the uniform norm on D then, by
(8.2) applied to gn − gm , we conclude {gn − iImgn (0)} is Cauchy in the uniform norm on
|z| ≤ r.
Now suppose {Refn } converges uniformly on compact subsets of ±. If z1 ∈ ±, and if
B(z1 ) = {z : | z − z1 | ≤ 21 dist(z1 , ∂ ±)}, then, by a linear change of variables,
{ fn − iImf n(z1 )} is a Cauchy sequence on B(z1). (8.3)
Let E = {z1 ∈ ± : { fn(z1 )} converges}. Then E is open by (8.3). If z1 ∈ ± \ E then, by (8.3)
and the convergence of {Refn (z1) }, we have B(z1) ⊂ ± \ E. By hypothesis, z0 ∈ E and ± is
connected so E = ±. Because any compact subset of ± can be covered by finitely many balls
of the form B(z1 ), { fn } converges uniformly on compact subsets of ±.
Next we combine these results with Harnack’s principle and Schwarz’s lemma. See [11]
for the history of contributions to this result by Harnack, Schwarz, Carathéodory, and others.
Theorem 8.11 (Carathéodory) Suppose ± is a simply-connected region with z0 ∈ ± ² = C.

Suppose also that ±n are simply-connected regions with
∞
³
z0 ∈ ± n ⊂ ±n+1 ⊂ ±n = ±,
n=1
and suppose fn is a conformal map of ±n onto D, with fn (z0) = 0 and fn³ (z0) > 0, n = 1, 2, . . . .
Then { fn } converges uniformly on compact subsets of ± to a conformal map f of ± onto D,
with f (z0 ) = 0 and f ³ (z0) > 0.
Proof We first show that we may suppose ± ⊂ D and z0 = 0, without loss of generality,
by constructing a conformal map g0 of ± into D, with g0 (z0) = 0 and g³0(z0 ) > 0. Indeed, if
gn is a conformal map of g0(±n ) onto D with gn (0) = 0 and g³n (0) > 0 then fn = gn ◦ g0.
By assumption, there is a point z1 ∈ C \ ±. Because ± is simply-connected we can define

log(z − z1 ) so as to be analytic on ±, by Corollary 5.8. Set
√
g1 (z) = ≡ e log(z−z ) .
z − z1
1
2 1
If g1 (z) = g1 (w) then, by squaring, z − z1 = w − z1 and hence z = w. Thus g1 is one-to-one.

The same argument shows that if z ² = w then g1 (z) ² = −g1(w). If g1 (z2) ² = 0 then g1 (±)
covers a neighborhood of g1(z2 ) because g1 is open, and hence g1(±) omits a neighborhood
of −g1 (z2 ). Thus
C
g2 (z) =
g1(z) + g1 (z2 )
is analytic and one-to-one on ± and bounded by 1 for C sufficiently small. The function
g2(z) − g2 (z0 )
g3 (z) =
1 − g2 (z0)g2(z)
is the composition of g2 with an LFT, and hence one-to-one with g3(z0 ) = 0. Now set g0 (z) =
| g³3 (z0)| g (z). Henceforth we will assume ± ⊂ D and z = 0.
3g³ (z0 ) 3 0
Because ±n is simply-connected and z/ fn(z) is a non-vanishing function on ±n, we
can define log(z/f n(z)) as an analytic function on ±n with value log(1/ fn³ (0)) ∈ R at 0.
Then un (z) ≡ Re log(z/ fn(z)) = log | z/fn(z) | is harmonic on ±n . By Schwarz’s lemma
applied to fn+1 ◦ f n−1, we have that un(z) ≤ un+1(z) for all z ∈ ±n . Moreover, un (0) =
log | 1/ fn³ (0)| ≤ 0 by Schwarz’s lemma applied to fn−1 , because ±n ⊂ D. Applying Har-
nack’s principle, we conclude that un converges uniformly on compact subsets of ± to a
harmonic function u. The value of log(z/f n(z)) at 0 is equal to un (0), which converges to
u(0). By Lemma 8.10, log(z/fn (z)) converges uniformly on compact subsets of ±, and hence
f n(z) = z exp(− log(z/fn (z))) converges uniformly on compact subsets of ± to an analytic
function f . The limit function f is bounded by 1 on ± with f (0) = 0 and f ³ (0) = e−u(0) > 0.
By Hurwitz’s theorem (Corollary 8.9), f is a one-to-one conformal map.
It remains to prove that f maps ± onto D. Set gn = f ◦ f n−1. Then gn is analytic on D with
|gn | ≤ 1 and g³n (0) → 1. If ζ0 ∈ D \ f ( ±) then gn ²= ζ0 so that
z − ζ0
T (z) =
1 − ζ 0z
is a non-vanishing analytic function on gn(D), which is simply-connected
√ by Proposition 6.6.
Then, as in the first part of this proof, T has an analytic square root T = exp( 21 log T ) on
gn (D) which is one-to-one. Let
√
z− T(0)
S(z) = √ .
1− T(0)z
√
Thus hn =S◦ T ◦ gn is an analytic function mapping D into D, vanishing at 0 with
¶ ·
1 1 1 + |ζ0 | ³
|h³ (0)| = 1 − |ζ0 | 2| ζ0 | 12
(1 − |ζ0 | 2
)g³n (0) = gn(0) > 1,
2|ζ 0| 2
n 1
for n sufficiently large. This contradicts Schwarz’s lemma and hence f maps ± onto D.
Theorem 8.12 If ± ⊂ C is a simply-connected region with ± ² = C and z0 ∈ ±, then we

can find conformal maps fn constructed using the geodesic zipper algorithm, with regions
−1
±n = f n (D) such that ±n ⊂ ±n+1 and ∪n±n = ±. Moreover, we can choose fn so that
fn (z0) = 0 and fn³ (z0 ) > 0.
Proof By the first part of the proof of Theorem 8.11, we may suppose that ± ⊂ D and
z0 = 0. If δ > 0 then, as in the proof of Runge’s Theorem 4.23, pave the plane with squares
of side length δ . If S is a square, 3S will denote the square with the same center as S but edge
length three times as long. Shade each square S with the property that 3S ⊂ ±. Let U be the
component of the interior of the union of the closed shaded squares with 0 ∈ U. The set U is
non-empty if δ is sufficiently small. Then ∂ U is the union of edges of squares of side length
δ . Let Cδ = D0 ∪ D1 ∪ · · · ∪ Dm be the closed disk-chain of disks of diameter δ centered at
the endpoints of each square edge in ∂ U. See Figure 8.3. Then
δ/2 ≤ dist(ζ , ∂ ±) ≤ 4δ , (8.4)
for each ζ ∈ Cδ . Let ±δ be the simply-connected region constructed via the geodesic zipper
algorithm, as in Theorem 8.7, with ∂ ±δ ⊂ C δ , and let fδ be the corresponding conformal map
of ±δ onto D, which we can normalize so that fδ (0) = 0 and f δ³ (0) > 0.
Choose a sequence δ n > 0 with δ n+1 < 8δn . Then, by (8.4), ±δn ⊂ ±δn+1 and z0 ∈ ±δ1
for δ 1 sufficiently small. If z1 ∈ ± then there is a polygonal curve γ ⊂ ± from z0 to z1. If
4δ n < dist(γ , ∂ ±) then γ ⊂ ±n and hence ± = ∪n ±n.
The reason for reducing to the case where ± ⊂ D is to guarantee Cδ is a finite union of
disks. Finally we deduce the Riemann mapping theorem.
Theorem 8.13 (Riemann mapping theorem)Suppose ± is a simply-connected region with

z0 ∈ ± ² = C. Then there is a unique one-to-one analytic map f of ± onto D with f (z0) = 0
and f ³ (z0) > 0.
Proof Theorem 8.13 follows from Theorem 8.12, Theorem 8.11 and Proposition 6.5.
8.3 Symmetry and Conformal Maps
In this section we will explore the connection between symmetry in a simply-connected region
and symmetry of a conformal map from the region to the disk D.
We say that a region ± is symmetric aboutR provided z ∈ ± if and only if z ∈ ±. If ±
is symmetric about R and if f is analytic on ±, then f (z) is analytic on ±. Indeed, the power
series expansion for f (z) based at b has coefficients which are the complex conjugates of the
coefficients in the power series for f based at b. If f is also real valued on an interval in ± ∩ R,
then
f (z) ≡ f (z) (8.5)
by the uniqueness theorem, Corollary 2.9. Geometrically, (8.5) says that f maps symmetric
points to symmetric points. For one-to-one functions we have the following related result:
8.3 Symmetry and Conformal Maps 133
Proposition 8.14 If f is a conformal map of a simply-connected symmetric region ± onto D

such that f (x0) = 0 and f ³ (x0) > 0 for some x 0 ∈ R ∩ ±, then f (z) = f (z).
Proof Apply the uniqueness conclusion in the Riemann mapping theorem to the functions f
and f (z).
Proposition 8.14 says that if the region is symmetric then we can take the mapping function
to be symmetric. We would like to use this idea to help construct conformal maps.
Theorem 8.15 (Schwarz reflection principle)Suppose ± is a region which is symmetric

about R. Set ±+ = ± ∩ H and ±− = ± ∩ (C \ H). If v is harmonic on ±+ , continuous on
±+ ∪ (± ∩ R) and equal to 0 on ± ∩ R, then the function defined by
¸
v(z), for z ∈ ± \ ±−
V(z) =
−v(z), for z ∈ ±−
is harmonic on ±. If also v(z) = Imf (z), where f is analytic on H ∩ ±, then the function
¸
f (z), for z ∈ ±+
g(z) =
f (z), for z ∈ ±−
extends to be analytic in ±.
Note that there is no assumption in Theorem 8.15 about boundary values of the real part of
f on ± ∩ R. The region ± is not necessarily simply-connected either.
Proof The extended function V is continuous on ±. To prove the first claim, we need only
prove that V has the mean-value property for small circles centered on ± ∩ R. But, for x0 ∈
R ∩ ±, V (x0 + reit) = −V(x0 + re−it ) so that the mean value over a circle centered at x0
contained in ± is zero, the value of V at x0. Thus V is harmonic in ±.
Suppose now that v = Imf , where f is analytic on H ∩ ±. If D is a disk contained in ± and
centered on R, then V = Im h for some analytic function h on D. It is uniquely defined by
requiring that h = f on ±+ ∩ D. By (8.5) h(z) = h(z) on D, and so h = g on D ∩ ± \ R. Thus h
provides the unique analytic extension of g to all of D, and g extends to be analytic on ±.
Corollary 8.16 If the function f in the Schwarz reflection principle is also one-to-one in ±+
with Imf > 0 on ±+ , then its extension g is also one-to-one on ±.
Proof By definition, g is one-to-one on ±− with Im g < 0 on ±− . So if g(z1 ) = g(z2) then

z1 , z2 ∈ R. Since g is open, it maps a small disk centered at zj , j = 1, 2, onto a neighborhood
of g(z1 ) = g(z2 ). If z1 ² = z2 then there are two points ζ1 , ζ 2 ∈ ±+ near z1, z2 (respectively)
with g(ζ 1) = g(ζ 2), contradicting the assumption that f is one-to-one on ±+ .
The function f (z) = z2 is one-to-one on D+ , and real on D ∩ (−1, 1), but does not have a
one-to-one extension to D, so the assumption Imf > 0 on ±+ cannot be removed.
As an application of Corollary 8.16 we outline how to find a conformal map of H to the

region ± = {x + iy : y > x2 }, which is somewhat different from the approach in Exercise 6.12.
First find a map f1 of H onto the half-strip S = {z = x + iy : 0 < x < b, y > 0} so that
f1 (∞) = ∞ (i.e., lim| z|→∞ | f1 (z)| = ∞), where b is chosen later. The function f2 (z) = z2 is
one-to-one on S with image equal to half of the interior region of a parabola. The composed
function f2 ◦ f1 maps an interval Ia = (−∞ , a] onto Ib2 = (−∞ , b2] and [a, +∞) onto the half-
parabola. By Corollary 8.16 we can extend this function by reflecting the domain across the
interval Ia and the range across I b2 to obtain a conformal map of C \ [a, ∞) onto the interior
√
of a parabola. The slit plane can be mapped onto H using z − a and the parabola can be
dilated, translated and rotated using a linear map. With the proper choice of b we obtain the
desired region.
This same idea can be used to construct a conformal map from H to the region outside one
branch of a hyperbola, which was also done in Exercise 6.11.
It is sometimes useful to be able to reflect across curves more general than line segments.
Definition 8.17 Let I = (0, 1). An open analytic arc γ contained in the boundary of a region
± is called a one-sided arc if there exists a function g which is one-to-one and analytic in a
neighborhood N of I with g(I) = γ and g(N ∩ H) ⊂ ± and g(N \ H) ⊂ ±c . If g(N \I) ⊂ ±
then γ is called a two-sided arc.
Corollary 8.18 Suppose ± is a simply-connected region and suppose γ ⊂ ∂ ± is a one-

sided analytic arc. Let f be a conformal map of ± onto D. Then f extends to be analytic and
one-to-one in a neighborhood of ± ∪ γ .
Proof By Exercise 3.7 the map f is proper so that | f (z)| → 1 as z → ∂ ±. If γ ² = ∂ ±, choose

g and N as in Definition 8.17. Then f ◦ g is analytic in N ∩ H. By shrinking N if necessary, we
may suppose N is symmetric about R, N ∩ H is simply-connected and f ◦ g ² = 0 on N. We
can then define h = i log f ◦ g to be analytic in N ∩ H and
lim Imh(z) = lim − log | f ◦ g(z)| = 0.

z∈ N→(0,1) z ∈N →(0,1)
By Corollary 8.16, h extends to be one-to-one and analytic in N. Therefore f extends to be

one-to-one and analytic in ± ∪ g(N). See Exercise 8.2(c) if γ = ∂ ±.
Corollary 8.18 remains valid if we replace D by H, although ∞ presents a small complica-

tion. It is not true that a conformal map f of ± onto H satisfies Imf → 0 as z → γ because
∞ is a boundary point of H in C∗. Thus we cannot directly apply the Schwarz reflection prin-
ciple. However, composing f with an LFT τ of H onto D, then applying Corollary 8.18, we
conclude that τ ◦ f extends to be analytic and one-to-one in a neighborhood of ± ∪ γ , so that f
extends to be analytic and one-to-one in a neighborhood of ± ∪ γ with the possible exception
of a simple pole at a single point of γ .
If h is analytic, but not necessarily one-to-one, in a region ± whose boundary contains a
one-sided analytic arc γ , and if h(z) tends to an analytic arc σ as z → γ , then h extends to be
analytic in a neighborhood of ± ∪ γ . See Exercise 8.2(d).
8.4 Conformal Maps to Polygonal Regions 135
Care must be taken if γ is a two-sided arc in ∂ ±. For example, if ± = C \ I, where

√
I = (−∞, 0], then the function z defined in ± has an analytic extension across I o = (−∞ , 0)
from the upper half-plane, and across I o from the lower half-plane, but these extensions do
not agree with the original function.
8.4 Conformal Maps to Polygonal Regions
In this section we will use the Schwarz reflection principle to find a conformal map, called a
Schwarz–Christoffel formula or integral, from the upper half-plane onto a region bounded by
a polygon.
Suppose ± is a bounded simply-connected region bounded by a simple closed curve con-
sisting of finitely many straight-line segments with vertices {vj }n1 . If we give ∂ ± a positive
(counter-clockwise) orientation, then at vj the tangent vector turns by an angle π αj , where
−1 < αj < 1. See Figure 8.6.
By the Schwarz reflection principle, Theorem 8.15, if ϕ is a conformal map of ± onto D,
then ϕ extends analytically and one-to-one across (the interior of) each boundary segment.
If Bj is a small ball centered at vj , then the map ϕj (z) = (z − vj )1/ (1− αj ) is one-to-one and
analytic in ± ∩ Bj , and maps ∂ ± ∩ Bj onto a straight-line segment. The inverse of this map
composed with ϕ then extends to be analytic and one-to-one in a neighborhood of 0, again
by the Schwarz reflection principle. Thus ϕ extends to be a one-to-one and continuous map
of ± onto D. Similarly, if f (z) is a conformal map of H onto ± then f (z) extends to be one-
to-one and continuous on H and analytic on R except at the prevertices xj = f −1(vj ), j =
1, . . . , n.
We may assume that the vertices are ordered so that −∞ < x1 < x2 < · · · < xn < ∞ . If
f (x) and f (x + h) belong to the same edge of the polygon ∂ ± then f (x + h) − f (x) is parallel
to ∂ ±. Writing
f (x + h) − f (x)
f ³ (x) = lim ,
h>0→0 h
we deduce that f ³ (x) points in the tangential direction to ∂ ±. In other words, for xj < x < xj+1 ,
arg f ³ (x) is given by the direction of the line segment from vj to vj+1 . Since f ³ ² = 0 on a simply-
connected region containing H \ {xj }n1 , we can define log f ³ (z) so as to be analytic on H \ {xj }n1 ,
and hence arg f ³ (z) is a bounded harmonic function on H which is continuous on H \ {xj }n1 ,
with a jump discontinuity on R of π αj at xj , where π αj is the turning angle at vj = f (xj ). See
Figure 8.6.
Ω
παj
vk παk
vj
Figure 8.6 Turn angles at the vertices of a polygon.

We can build a bounded harmonic function on H with exactly these jump discontinuities on
R as follows. The function arg(z − a), for a ∈ R, with 0 ≤ arg(z − a) ≤ π on H ∪ R \ {a}, is
harmonic on H and has a jump discontinuity of −π at a. Let c0 be the constant c0 = arg f ³ (x)
for x > xn . Then
¹
n
u(z) = c0 − αj arg(z − xj )
j=1
is a bounded harmonic function on H which equals arg f ³ on R \ {x1 , . . . , xn }. By Lindelöf’s

maximum principle applied to u − arg f ³ , we conclude that u(z) = arg f ³ (z) = Im log f ³ (z) on
H. Thus
n
º
f ³ (z) = c (z − xj ) −αj ,
j =1
for some constant c. The function f can now be found by integration.

We have proved the following theorem:
Theorem 8.19 (Schwarz–Christoffel) Suppose ± is a bounded simply-connected region

whose positively oriented boundary ∂ ± is a polygon with vertices v1, . . . , vn . Suppose the
tangent direction on ∂ ± increases by π α j at vj , −1 < αj < 1. Then there exists x 1 < x2 <
· · · < xn and constants c1, c2 so that
µ º
n
f (z) = c1 (ζ − xj ) −αj dζ + c2
γz
j=1
is a conformal map of H onto ±, where the integral is along any curve γz in H from i to z.
The definition of f in Theorem 8.19 does not depend on the choice of γz ∈ H by

Cauchy’s theorem, because H is simply-connected. We remark that this proof of the Schwarz–
Christoffel theorem uses the existence of the desired conformal map, so that it cannot be used
to replace the geodesic zipper algorithm in the proof of the Riemann mapping theorem in
Section 8.2.
One way to understand the powers in the product in Theorem 8.19 is to note that, near xj ,
f (z) − vj behaves like a power cj (z − xj )βj , where πβj is the interior angle at vj . So f ³ behaves
like cj βj (z − xj )β j −1. If the turning angle at vj is π αj , then πβj + π αj = π , so that βj − 1 = −αj .
The remarkable part of Theorem 8.19 is that this local behavior gives a formula for f ³ .
The difficulty in using a Schwarz–Christoffel formula is that it does not tell us how to find
the prevertices xj . As in the proof of Morera’s Theorem 4.19, the integral in Theorem 8.19
is always analytic, in fact locally conformal, but in general it does not represent a conformal
map. In the formulation above, −1 < αj < 1 and the tangent direction has a total increase of
∑
2π around ∂ ±, so that nj=1 αj = 2. Thus the integral is absolutely convergent, even if the
integration takes place on R instead of H. The length of the segment from vj to vj +1 can be
expressed as
µ xk +1 º
n
Lk = |c1| |x − xj |− αj
dx, (8.6)
j=1
xk
8.5 Exercises 137
because arg f ³ is constant on this segment. So the prevertices can be written as a solution of
the system (8.6) of (highly non-linear) singular integral equations. For numerical applications,
because the derivative is singular at some of the prevertices, Riemann sums are too inaccurate
to compute the integrals, so that Gauss–Jacobi quadrature is needed.
The Schwarz–Christoffel theorem on the unit disk has a similar form. See Exercise 8.4.
The Schwarz–Christoffel theorem can also be extended to cover unbounded regions whose
boundary consists of line segments and half-lines or full lines. Allowing an angle π αj to
equal π corresponds to an infinite channel between two parallel half-lines. The case when
the angles π αj are allowed to equal −π corresponds to a “cut” along a line segment into a
polygonal region. See Exercises 8.5 and 8.13.
We end this chapter with one of the more important open problems in complex analysis
called the Brennan conjecture. Suppose that T is a polygonal tree, a connected finite union
of line segments and the half-line [0, ∞], whose complement is a simply-connected region
±. The conformal map ψ of the upper half-plane H onto ± can be normalized so that it is
asymptotic to z2 for | z| large. Thus ψ ³ (z) is asymptotic to 2z near ∞. Indeed, the total turning
∑
of the tangent direction following a positive orientation of ∂ ± is −π , so that αj = −1,
where π α j is the change in the tangent direction at the vertex vj = ψ (xj ). By the Schwarz–
Christoffel theorem,
³(z) = » 2 .
(z − xj )αj
ψ
The Brennan conjecture is about the second derivative of ψ at the “tips” of the tree: prove that
for every polygonal tree
¹ º
|xk − xj |2 ≤ 1.
αj
k: αk =−1 j²= k
This
¼ ³ is equivalent to proving that, for any conformal map f of the disk into the plane,
−
| f | dxdy < ∞ for 0 < p < 2. The conformal map f (z) = (1 − z) 2 defined on D has
p
a derivative whose reciprocal is not square integrable.
8.5 Exercises
8.1 Let f be an analytic function on | z| < 2 such that f is real valued on a subarc of | z| = 1.
Prove that f is a constant.
8.2 (a) Using the Cayley transform, show that reflection z → z across R corresponds to
“reflection” z → 1/z across ∂ D.
(b) Suppose g is analytic in D and | g(z)| → 1 as z → I, where I ⊂ ∂ D is an open arc.
Show that g extends to be meromorphic in C \ (∂ D \ I) using 1/g(1/z). The extension
will have poles at the reflection of the zeros of g.
(c) Suppose ± is a simply-connected region and suppose ∂ ± is a closed analytic curve.
Prove that if f is a conformal map of ± onto D then f and f −1 extend to be one-to-one
and analytic in a neighborhood of ±, respectively D.
(d) Prove that if g is analytic in a region ±1 whose boundary contains a one-sided

analytic arc γ , and if g(z) tends to an analytic arc σ as z → γ , then g extends to be
analytic in a neighborhood of ± ∪ γ .
(e) Suppose ± is a region whose boundary contains a simple arc γ which is locally
analytic, in the sense that each point of γ is contained in a one-sided analytic subarc
of γ . Prove γ is a one-sided analytic arc. Hint: Find a simply-connected region
±1 ⊂ ± so that γ ⊂ ∂ ±1.
(f) If γ : [0, 1] → C is a simple arc whose image can be covered by a finite union of
(images of) analytic arcs, then γ is (the image of) an analytic arc.
8.3 Find a conformal map of H onto {x + iy : y > x2 } so that z0 = i is mapped to the focus.
See the outline in Section 8.3.
8.4 Derive a formula for a conformal map of the unit disk onto a region bounded by a
polygon by a change of variables in the Schwarz–Christoffel formula in Theorem 8.19.
Use the fact that the sum of the turning angles is 2π to simplify your expression. This
formula is also called a Schwarz–Christoffel formula.
8.5 Show that the Schwarz–Christoffel formula still holds if angles π αj are allowed to equal
−π (slits) as described in Section 8.4.
8.6 If f is a conformal map of a rectangle of width R and height 1 onto the upper half-plane,
then prove it can be reflected across a vertical edge to obtain a map from a rectangle
twice as long onto the slit plane. Prove that, by repeated reflections, the map f can be
extended to be a meromorphic function on C such that f (z + 2R) = f (z) and f (z + 2i) =
f (z). In general, an elliptic function is a meromorphic function invariant under two
linearly independent shifts. See Example 11.5. The inverse of an elliptic function, such
as the Schwarz–Christoffel integral for a rectangle map, is an example of an elliptic
integral. We will investigate elliptic functions more thoroughly in Section 16.3.
8.7 Suppose g is a non-constant analytic function on a region ± and suppose {gn} is a

sequence of analytic functions converging uniformly on compact subsets of ± to g. If
K is a compact subset of g( ±) then prove there is an n0 < ∞ so that K is a subset of
gn( ±) for all n ≥ n0 . Hint: Apply the proof of Hurwitz’s Theorem 8.8 to gn − w and
g − w for all w in a small ball containing g(z0 ).
8.8 Find a conformal map of D onto the interior of the ellipse given by x2/4 + y2 /9 = 1.
Hint: Consider the map 21 (z + 1z ) on the top half of an annulus. Your answer will involve
a Schwarz–Christoffel integral. The prevertices can be put in a symmetric position (at
the vertices of a rectangle on ∂ D) by choosing the value at 0 and the argument of the
derivative at 0, so that it involves only one unknown parameter. Give an equation whose
solution is this parameter. In general, it is not possible to find a formula for the parameter
in terms of elementary functions.
8.9 (a) Given a continuous function f (eit ), 0 ≤ t ≤ π , find a harmonic function u on
D+ = D ∩ {Imz > 0} such that u is continuous on D+ , u = f on ∂ D+ ∩ {Imz > 0}
and ∂∂uy = 0 on ( −1, 1). Hint: Find u so that u(z) = u(z).
¼
(b) Suppose g is continuous on ∂ D and 02π gd θ = 0. Find a harmonic function v on D
with the property that ∂∂ ηv = g on ∂ D, where η is the unit inward normal to ∂ D. This
8.5 Exercises 139
¼t
is called a Neumann problem. Hint: Consider f (eit ) = 0 g(eiθ )dθ , u = PI( f ), and
use the Cauchy–Riemann equations on |z| = r, then let r → 1.
(c) Given bounded continuous functions f (x), −∞ < x < 0, and g(x), 0 < x < ∞ ,
find a harmonic function u on H such that u is continuous on H ∪ R \ {0}, u = f on
(−∞ , 0) and ∂∂uy = g on (0, ∞).
(d) Suppose ϕ is analytic and one-to-one on a neighborhood of H and ± = ϕ(H).
Suppose E is a connected subset of ∂ ±. Given continuous functions f on E and g
on ∂ ± \ E, describe how to find a harmonic function u on ± so that u = f on E ◦
and ∂∂ ηu = g on (∂ ± \ E)◦ , where η = ηζ is the unit inner normal to ∂ ± at ζ ∈ ∂ ±.
This is called a mixed Dirichlet–Neumann problem .
8.10 The intersection of two disks is called a lens. Suppose zi , i = 1, . . . , n, are distinct points
in C. Suppose D+ −
i and Di are (open) disks whose intersection is a lens Li = Di ∩ Di
+ −
with endpoints zi−1 and zi . If (D+ − +
i ∪ Di ) ∩ L j = ∅ for j = 1, . . . , i − 1, and if ∂ Di is
− − +
tangent to ∂ Di−1 and ∂ Di is tangent to ∂ Di−1 , then we say that {L i } is a tangential lens-
chain. Prove that if {L i } is a tangential lens-chain then the geodesic zipper algorithm
constructs a conformal map of the half-plane onto a simply-connected region whose
boundary is contained in ∪Li . This idea can be used to reduce the number of data points
zi in the geodesic zipper algorithm by using long thin lenses √where the boundary is flat.
8.11 Prove that the basic map in the geodesic zipper algorithm z2 + 1 of H \ [0, i] onto H
can be½ extended as a conformal map of C \ [−i, i] onto C \ [−1, 1], by writing it in the
form z 1 + 1/z2 , where the square
√ 2 root is defined on C \ (−∞, 0]. Show that it can also
be computed by setting w = z + 1 (any choice of the square root), then replacing w
with −w when Rez · Rew < 0 or when Rez = √ 0 and Imz · Imw < 0. The latter method
is faster numerically. Find similar methods for z2 − 1.
8.12 Find a conformal map of D onto a regular n-gon. Use symmetry to find the prevertices
explicitly.
8.13 Formulate and prove the Schwarz–Christoffel theorem for an unbounded simply-
connected region whose boundary consists of line segments and half-line(s). Hint: The
turning angle can be understood by intersecting the region with a large disk.
8.14 Suppose ϕ is a conformal map of A1 = {z : 1 < |z| < R1 } onto A2 = {z : 1 < | z| < R 2}.
Prove R1 = R2 and ϕ (z) = λ z or ϕ (z) = λ R1 /z for some constant λ with | λ| = 1. Hint:
Reflect.
8.15 Find a conformal map f of the upper half-plane H onto H \ I, where I is the line segment
from 0 to ei π a, 0 < a < 1, as follows. First show that such a map can be chosen so that
| f | → ∞ as z ∈ H → ∞, then prove f extends to be continuous on R. Show that f
can be chosen so that f (a) = 0 and f (a − 1) = 0. Note that arg f is bounded and is
constant on each of the three intervals comprising R \ {a, a − 1}. Construct a bounded
harmonic function in H which has the same values on these intervals using arg(z − a)
and arg(z + 1 − a). Apply Lindelöf’s maximum principle then exponentiate to find Cf ,
for some positive constant C. Using logarithmic differentiation, find the preimage of the
tip of the slit I. Use this information to find C.
These maps can be used in place of the basic maps ϕj in the geodesic zipper algo-
rithm; however, the inverses are not elementary functions. The maps are very important
for a recent application of conformal mapping and probability to statistical mechanics
called SLE.
8.16 The geodesic zipper algorithm finds conformal maps of the upper half-plane H onto
two complementary regions ±± ⊂ C∗. Suppose ±+ is a bounded region and ±− is
unbounded.
(a) Describe how to find conformal maps f+ of D onto ±+ and f− of C \ D onto

±− \ {∞}.
(b) Check that the map f+ extends to be continuous and one-to-one on D and that f−
extends to be continuous and one-to-one on C \ D. See Figure 8.5.
(c) The extended maps f± do not agree on ∂ D. However, ϕ = f+−1 ◦ f − is an increasing
homeomorphism of ∂ D onto ∂ D (called a welding homeomorphism). Prove that
any increasing homeomorphism of ∂ D onto ∂ D can be uniformly approximated by a
welding homeomorphism constructed via the geodesic zipper algorithm. Hint: Two
increasing maps of [0, 2π ] to itself that agree on a fine mesh are uniformly close.
8.17 For h ≥ 1, let ϕh be a conformal map of the upper half-plane H onto the rectangle
R h = {z : |Rez| < π/ 2, 0 < Imz < h}, with ϕh(0) = 0 and ϕ(∞ ) = ih. Prove:
(a) there is a unique vh > 1 so that ϕ can be chosen to map vh, 1 /vh, −1/vh , −vh onto
the vertices of R h;
(b) ϕh (z) = ϕ2h(J −1(1/(vhz))), where J is the Joukovski map J(z) = 21 (z + 1/z) and
J−1 is chosen to map the lower half-plane onto H ∩ {|z| < 1};
(c) v2h = J(v2h ).
(d) Derive a formula for ϕh in terms of ϕ2n h.
(e) Let ψ (z) = i log(J −1 (z)) + π/2, where J −1 is chosen to map H onto H ∩ {|z| > 1}.
Prove that ψ is a conformal map of H onto the half-strip {z : |Rez| < π/2, Imz > 0}.
(f) Replace ϕ2nh(z) with ψ (v2n hz) in the formula in (d) to obtain an explicit approxima-
tion ψn for ϕh .
(g) Estimate the error ϕh − ψn when n = 4 and h > 1. The approximation is accurate
enough that it is virtually a formula for ϕh .
9 Calculus of Residues
9.1 Contour Integration and Residues
In this chapter we will learn one of the main applications of complex analysis to engineering
problems, but we will concentrate on the math, not the engineering. Cauchy’s theorem says
that if± f is analytic in a region ± and if γ is a closed curve in ± which is homologous to 0,
then γ f (z)dz = 0. What happens if f has an isolated singularity at a ∈ ±, but otherwise is
analytic? Expanding f in its Laurent series about a, we have
∞
²
f (z) = bn (z − a)n
n=−∞
∞
²
= zb−−1a + dzd (n + 1)
bn
(z − a) n+1 .
n=−∞
n±=−1
If ² is a disk centered at a with ² ⊂ ± then, by the fundamental theorem of calculus and

Proposition 4.13,
³ ³
dz
f (z)dz = b−1 = 2π ib−1 , (9.1)
∂² ∂² z−a
provided ∂ ² is oriented in the positive or counter-clockwise direction.
Definition 9.1 If f is analytic in {0 < |z − a| < δ} for some δ > 0, then the residue off at
a, written Resa f , is the coefficient of (z − a)−1 in the Laurent expansion of f about z = a.
Theorem 9.2 (Residue theorem) Suppose f is analytic in ± except for isolated singularities
at a 1, . . . , an. If γ is a cycle in ± with γ ∼ 0 and a j ∈/ γ , j = 1, . . . , n, then
³ ²
f (z)dz = 2π i n(γ , ak )Resak f .
γ
k
Usually the residue theorem is applied to curves γ such that n(γ , ak ) = 0 or 1, so that the
sum on the right is 2π i times the sum of the residues of f at points enclosed by γ . If f has
infinitely many singularities clustering only on ∂ ± then we can shrink ± slightly so that it
contains only finitely many aj and still have γ ∼ 0.
Proof Let ²k be a disk centered at ak , k = 1, 2, . . . , n, such that ²m ∩ ² k = ∅ if m ±= k.

Orient ∂ ² k in the positive or counter-clockwise direction. Then
²
γ − n(γ , ak )∂ ²k ∼0
k
in the region ± \ {a1, . . . , an}. By Cauchy’s theorem,

³ n
² ³
f (z)dz − n(γ , ak ) f (z)dz = 0.
γ
k =1 ∂ ²k
Then Theorem 9.2 follows from (9.1).
9.2 Some Examples
In this section we first illustrate several useful techniques for computing residues, then use
the residue theorem to compute some integrals. The emphasis is on the techniques instead of
proving general results that can be quoted.
e3z
Example 9.3 f (z) = has a simple pole at z = 2 and hence
(z − 2)(z − 4)
e 6
Res2 f = zlim
→2
(z − 2)f (z) =
−2 .
The residue at z = 4 can be calculated similarly.
e3z
Example 9.4 g(z) = .
(z − 2)2
Expand e3z in a series expansion about z = 2:
e6e3(z−2) 6 ∞ 3n
² 6 6
g(z) =
(z − 2)2
= (z −e 2)2 n!
(z − 2)n = (z −e 2)2 + z3e− 2 + . . . ,
n=0
so that
Res2g = 3e6.
In this case limz→2(z − 2)2 g(z) is not the coefficient of (z − 2)−1 and lim z→2(z − 2)g(z) is
infinite. Of course the full series for e3z was not necessary to compute the residue. We can
find the appropriate coefficient in the series expansion for e3z by computing the derivative of
e3z . More generally, if G(z) is analytic at z = a then
(n−1)
G(z)
Resa
(z − a)n
= G(n − 1)!
(a)
.
Example 9.5 Another technique that can be used with simple poles, when the pole is not
already written as a factor of the denominator, is illustrated by the example h(z) = eaz /(z4 + 1).
Then h has simple poles at the fourth roots of −1. If ω4 = −1, then
9.2 Some Examples 143
(z − ω)eaz eaω
Resω h = lim = .
z→ω z4 + 1 limz →ω zz −+ω1
4
Note that the denominator is the limit of difference quotients for the derivative of z4 + 1 at
z = ω and hence
eaz eaω ωe
aω
Resω 4
z +1
= 4ω 3
= − 4
.
Example 9.6 Another method using series is illustrated by the example

cot π z
k(z) =
π
.
z2
To compute the residue of k at z = 0, note that cot π z has a simple pole at z = 0 and hence k
has a pole of order 3, so that
cot π z
π
z2
= bz−33 + bz−22 + b−z 1 + b0 + . . . .
Then
´ µ
sin π z
π cos π z = z
(b−3 + b−2z + b−1 z2 + b0z3 + . . . ).
Inserting the series expansions for cos and sin we obtain

´ 2 µ ´ 3 µ
1− +... = − +... (b−3 + b−2z + b−1 z2 + . . . ).
π 2 π 2
π z π z
2 6
Equating coefficients,
3 3
π = π b−3, − π2 = − π6 b−3 + π b−1 ,
and Res0 k = b−1 = − π3 .
2
Now let’s apply these ideas and the residue theorem to compute some integrals.
Example 9.7 If γ is the circle centered at 0 with radius 3, then

³
e3z e6
dz = −2π i ,
γ (z − 2)(z − 4) 2
by the residue theorem and Example 9.3.
³ ∞
dx
Example 9.8
−∞ x4 + 1.
Construct a contour γ consisting of the interval [ −R, R] followed by the semi-circle CR in H
of radius R, with R > 1. See Figure 9.1.
By the residue theorem with f (z) = 1/ (z4 + 1),
³ R ³
f (z)dz + f (z)dz = 2π i(Resz 1 f + Resz 2
f ), (9.2)
−R CR
CR
R
z1 z2
Figure 9.1 Half-disk contour.
where z1 and z2 are the roots of z4 + 1 = 0 in the upper half-plane H. Note that
¶³ ¶ ³ 2π
¶ ¶ Rd θ
¶ C f (z)dz¶ ≤ 0 R4 − 1 → 0
¶ ¶
R
± 4 ±
as R → ∞. Since the integral R(x + 1)−1dx is convergent, it equals limR→∞ −RR (x4 +
1)−1 dx, so that, by (9.2) and example 9.5,
³ ∞
1
4 +1
dx = − 2π4 i (z1 + z2 ) = √π .
−∞ x 2
The technique in Example 9.8 can be used to compute the integral of any rational function with
no poles on R if the degree of the denominator is at least 2 plus the degree of the numerator.
This latter condition is needed for the absolute convergence of the integral.
³ 2π
1
Example 9.9 dθ .
0 3 + sin θ
Set z = eiθ . Then
³ 2π ³ ³
1 1 dz 2dz
3 + sin θ
dθ = = + 6iz − 1 .
0 |z|=1 (3 + 2i (z − 1/ z))
1 iz |z|=1 z2
√ √
The roots of z2 + 6iz − 1 occur at z1 , z2 = i(−3 ² 8). Only i(−3 + 8) lies inside |z| = 1. By
the residue theorem and the method for computing residues in Example 9.3 or Example 9.5,
³ 2π
1 2 2π
3 + sin θ
dθ = 2π iResi(−3+√8) z2 + 6iz −1 = √ .
0 8
The technique in Example 9.9 can be used to compute
³ 2π
R(cos θ , sin θ )dθ ,
0
where R(cos θ , sin θ ) is a rational function of sin θ and cos θ , with no poles on the unit circle.
An integral on the circle, as in Example 9.9, can be converted to an integral on the line using
the Cayley transform z = (i − w)/(i + w) of the upper half-plane onto the disk. It is interesting
to note that you obtain the substitution x = tan 2θ which you might have learned in calculus.
³ ∞
cos x
Example 9.10 dx.
−∞ x2 + 1
A first guess might be to write cos z = (eiz + e−iz) /2, but if y = Imz then | cos z| ∼ e|y | /2 for
large | z|. This won’t allow us to find a closed contour where the part off the real line makes
only a small contribution to the integral. Instead, we use eiz /(z2 + 1) then take real parts of
the resulting integral. Using the same contour as in Example 9.8, we have the estimate
¶³ ¶ ³
¶ eiz ¶ e−y πR
¶ ¶
¶ C z2 + 1 dz¶ ≤ R2 −1 | dz| ≤ 2
R −1
→0
R CR
as R → ∞ , where y = Imz > 0. By the residue theorem and the method in Example 9.5,
³ ∞ ² i ·i
eix eiz
dx = 2π i Resa = 2π i e2i = πe .
−∞ x2 + 1 Ima>0
z2 + 1
In this particular case, we did not have to take real parts. The integral itself is real because
sin x/(x2 + 1) is odd.
9.3 Fourier and Mellin Transforms
The technique in Example 9.10 can be used to compute

³ ∞
f (x)eiλ x dx (9.3)
−∞
for λ > 0, provided f is meromorphic in the closed upper half-plane H ∪ R with no poles on
R and | f (z)| ≤ K /|z|1+ε for some ε > 0 and all large |z| with Imz > 0. If the integral (9.3) is
desired for all real λ , then for negative λ use a contour in the lower half-plane, provided f is
meromorphic and | f (z)| ≤ K /| z|1+ε in Imz < 0. As the reader can deduce, if f is meromorphic
in C and satisfies this inequality for all |z| large, then f is rational and the degree of the
denominator is at least 2 plus the degree of the numerator. The integral in (9.3), usually with
λ replaced by −2π λ, is called the Fourier transform off , as a function of λ.
³ ∞
x sin λ x
Example 9.11 dx.
−∞ x2 + 1
This example cannot be done ± by the method in Example 9.10 because the integrand does not
decay fast enough to prove CR | f (z)||dz | → 0, where f (z) = ze /(z + 1). Indeed, it is not
iλ z 2
even clear a priori that the integral in Example 9.11 exists.

We may suppose λ > 0, because sine is odd. Let γ = γ1 + γ2 + γ3 +γ4, where γ1 = [−A, B],
A, B > 0, γ2 = {B + iy : 0 ≤ y ≤ A + B}, γ3 = {x + i(A + B) : B ≥ x ≥ −A } and
γ 4 = {−A + iy : A + B ≥ y ≥ 0 }, orienting γ as indicated in Figure 9.2.
To prove convergence of the integral in Example 9.11, we will let A and B tend to ∞
independently, and use the estimate
γ3
B + i(A + B)
γ4 γ2
−A γ1 B
Figure 9.2 Rectangle contour.

|z/(z2 + 1)| ≤ |z|/ (|z|2 − 1) ≤ 2/|z|

when |z|2 > 2. For A and B large,
¶³ ¶ ³ B −λ(A+B)
¶ zei λz ¶¶ 2 −λ(A+B) dx = 2e
¶
¶ z2 + 1 ¶ dz ≤ e (A + B) → 0
γ3 −A A + B A+ B
as A + B → ∞. Also
¶³ ¶ ³ A+B
¶ zeiλ z ¶¶ 2 −λy 2 (1 − e−λ (A+B))
¶ dz
¶ γ z2 + 1 ¶ ≤ B
e dy ≤ B λ
→0
2 0
as B → ∞. A similar estimate holds on γ4 as A → ∞. By the residue theorem,
³ B
xe i x λ
zei z 2π i · ie− λ λ
lim
A,B→∞ −A x2 + 1
dx = 2 π i Resi
z2 + 1
= 2i
= i π e− . λ
(9.4)
Indeed, by our estimates, the integrals over γ2 , γ3 and γ4 tend to 0 as A and B tend to ∞ so
that the limit on the left-hand side of (9.4) exists and (9.4) holds. Example 9.11 follows from
(9.4) by taking the imaginary parts.
The technique in Example 9.11 can be used to compute (9.3) with the (weaker) assumption
| f (z)| ≤ K / |z| for large |z|.
³ ∞
sin x
Example 9.12 dx.
−∞ x
The main difference between Examples 9.11 and 9.12 is that the function f (z) = eiz/z has a
simple pole on R. The function sin x/x is integrable near 0 since sin x/x → 1 as x → 0, but
f (x) is not integrable. However, the real part of f (x) is odd so that
³ −δ ³ 1 ix
e
δ
lim
→0 −1
+ x
dx
δ
exists.
Definition 9.13 Suppose f is continuous on γ \{a}, and suppose γ ³ is continuous at γ −1(a).

Then the Cauchy principal valueof the integral of f along γ is defined by
³ ³
PV f (z)dz = lim f (z)dz,
γ δ →0 γ ∩{|z−a|≥δ}
provided the limit exists.
Note that we have deleted points in a ball centered at a. If the limit exists for all balls
containing a, then the usual integral of f exists. For example,
³ 1
cos x
PV dx = 0,
−1 x
because the integrand is odd, but the integral itself does not exist.
Proposition 9.14 Suppose f is meromorphic in {Imz ≥ 0}, such that
| f (z)| ≤ |Kz|
γ3
B + i(A+B)
γ2
Cj
−A aj γ1 B
Figure 9.3 Contour avoiding poles on R.
when Imz ≥ 0 and | z| > R. Suppose also that all poles of f on R are simple. If λ > 0, then
³ ∞ ² ² 1
PV f (x)ei λx dx = 2π i Resaeiλz f (z) + 2π i
2
Resaeiλ z f (z). (9.5)
−∞ Ima>0 Ima=0
Part of the conclusion of Proposition 9.14 is that the integral exists even though the rate
of decay at ∞ is possibly slower than our assumptions in (9.3). If λ < 0, then a similar
result holds using the lower half-plane. The integral may not exist if λ = 0, as the example
f (z) = 1/z shows.
Proof Note that f has at most finitely many poles in {Imz ≥ 0} because | f (z)| ≤ K / |z|, so
that both sums in the statement of Proposition 9.14 are finite. Construct a contour similar to
the rectangle in Figure 9.2, but avoiding the poles on R using small semi-circles Cj of radius
δ > 0 centered at each pole a j ∈ R. See Figure 9.3.
The integral of f (z)ei λz along the top and sides of the contour tend to 0 as A, B → ∞, as
in Example 9.11. The semi-circle Cj centered at aj can be parameterized by z = aj + δ eiθ ,
π > θ > 0, so that if
bj
f (z)eiλz = + gj (z),
z − aj
where gj is analytic in a neighborhood of aj , then
³ ³ 0 ³
bj
f (z)eiλz dz = δ ie d θ +
iθ
g(z)dz.
Cj π δ ei θ Cj
Because gj is continuous at aj and the length of Cj is π δ, we have

³
lim f (z)eiλ z dz = −iπ bj .
δ →0 Cj
By the residue theorem,

³ ² ²
PV f (z)eiλz dz − iπ bj = 2π i Resa f (z)eiλz ,
R j Ima>0
and (9.5) holds.
One way to remember the conclusion of Proposition 9.14 is to think of the real line as
cutting the pole at each aj in half. The integral contributes half of the residue of f at aj .
For the present example,

³ ∞ ix
e
PV dx = π i, (9.6)
−∞ x
so that, by taking imaginary parts,
³ ∞
sin x
dx = π .
−∞ x
±
Note that sin x/x → 1 as x → 0 so that sin x/xdx exists as an ordinary (improper) Riemann
integral, if we extend the function sin x/ x to equal 1 at x = 0. For this reason we can drop the
PV in front of the integral.
³ ∞
xα
Example 9.15 dx,
0 x2 + 1 ±∞
where 0 < α < 1. An integral of the form 0 f (x)xβ −1dx is called a Mellin transform. By
the change of variables x = ey , the Mellin transform of f is the Fourier transform of f (ey )
when β is purely imaginary. Here we work with the original form of the integral and define
zα = eα log z in C \ [0, +∞ ), where 0 < arg z < 2π and set f (z) = 1/(z2 + 1). Part of the
difficulty here is constructing a closed contour that will give the desired integral. Consider the
“keyhole” contour γ consisting of a portion of a large circle CR of radius R and a portion of a
small circle Cδ of radius δ , both circles centered at 0, along with two line segments between
Cδ and CR at heights ²ε, oriented as indicated in Figure 9.4.
By the residue theorem, for R large and δ small,
³
zα f (z)dz = 2π i(Resi zα f (z) + Res−i zα f (z))
γ
´ µ (9.7)
α log i
eα log ( −i)
= 2π i e 2i + −2i = π (e − e
i π2α i 3 π2 α
).
We will first let ε → 0, then R → ∞ and δ → 0. Even though the integrals along the
horizontal lines are in opposite directions, they do not cancel as ε → 0. For ε > 0,
lim (x + iε)α f (x + iε ) = eα log |x| f (x)

ε →0
and
lim (x − iε) αf (x − i ε) = eα(log |x|+ 2π i)f (x),
ε →0
CR
R
Cδ
2ε
Figure 9.4 Keyhole contour.

9.4 Series via Residues 149
because of our definition of log z. Thus the integral over the horizontal line segments
tends to
³ R
(1 − e2π iα )xα f (x)dx. (9.8)
δ
For R large,
¶³ ¶ ³
¶ ¶ 2π Rα
¶
¶ zα f (z)dz¶¶ ≤ R2 − 1 Rdθ → 0 (9.9)
CR 0
as R → ∞ . Similarly,
¶³ ¶ ³
¶ ¶ 2π δα
¶ zα f (z)dz¶ ≤ δdθ →0 (9.10)
¶
Cδ
¶
0 1 − δ2
as δ → 0. By (9.7)–(9.10),
³ ∞
= π e 1 −−e2e i = 2 cosπαπ/2 .
πα 3π α
i i
xα 2 2
dx
0 x2 + 1 π α
This line of reasoning works for meromorphic f satisfying | f (z)| ≤ C| z|−2 for large | z| and
with at worst a simple pole at 0. The function zα can be replaced by other functions which
are not continuous across R, such as log z. In this case real parts of the integrals along [0, ∞ )
will cancel, but the imaginary parts will not. See Exercise 9.11. Mellin transforms are used in
applications to signal processing, image filtering, stress analysis and other areas.
In summary, computing integrals using the residue theorem usually involves choosing a
meromorphic function and a contour, computing residues, estimating integrals then passing to
a limit. This section is the first encounter in this text with estimating integrals, a useful skill.
For that reason we have not given general results which could be simply quoted to find special
integrals.
9.4 Series via Residues
We can turn this game around. Instead of using sums to compute integrals, we can use integrals
to compute sums.
Example 9.16
∞
² 1
n=0
n2 + 1.
Set f (z) = 1
z 2 +1
and consider the meromorphic function f (z)π cot π z. Write
´ µ ´ µ
eiπ z + e−iπ z
= π i ee2π iz −+ 11 .
2π iz
π cot π z = π i (9.11)
eiπ z − e−iπ z
Multiplying (9.11) by z − n and letting z → n shows that π cot π z has a simple pole with
residue 1 at each integer n. Because the poles are simple, f (z)π cot π z has a simple pole with
residue f (n) at z = n.
Consider the contour integral of f (z)π cot π z around the square SN with vertices
(N + 21 )(²1 ² i), where N is a large positive integer. The function π cot π z is uniformly
bounded on SN , independent of N. Indeed, the LFT (ζ + 1)/(ζ − 1) maps the region
|ζ − 1| < δ onto a neighborhood of ∞ and is one-to-one, so it is bounded on |ζ − 1| > δ .
The estimate |e2π iz − 1| > 1 − e−π holds on SN , as can be seen by considering the horizontal
and vertical segments separately, so that π cot π z is bounded on SN . Because | f (z)| ≤ C|z| −2,
we have ³
f (z)π cot π zdz → 0.
SN
By the residue theorem,

∞
²
0 = Resi f (z)π cot π z + Res−i f (z)π cot π z + f (n),
−∞
and hence
·
∞
² 1
= π eπ + e− ¸ + 1 .
π
n2 + 1 2 eπ − e− 2
π
n=0
This technique can be used to compute

∞
²
f (n),
n=−∞
provided f is meromorphic with | f (z)| ≤ C| z|−2 for |z| large. If some of the poles of f occur
at integers, then the residue calculation at those poles is slightly more complicated because
the poles of f (z)π cot π z will not have order 1 at these integers. See Example 9.6. If only
the weaker estimate | f (z)| ≤ C| z|−1 holds, then f has a removable singularity at ∞ and so
g(z) = f (z) + f (−z) satisfies | g(z)| ≤ C |z|−2 for large |z| . Applying the techique to g, we can
find the symmetric limit
N
²
lim f (n).
N→∞
n=−N
9.5 Laplace and Inverse Laplace Transforms
The Laplace transformof a function f defined on (0, +∞ ) is given by

³ ∞
L( f )(z) = f (t)e−zt dt
0
provided ±the integral exists. For example, if | f (t)| ≤ Mect on (0, +∞ ) then L ( f )(z) ≡
limT →∞ 0T f (t)e−ztdt exists for Rez > c. The convergence is uniform for Rez ≥ a > c
because |e−zt| ≤ e−at . By Weierstrass’s Theorem 4.29, L ( f ) is analytic in {Rez > c}.
Analyticity also follows from Morera’s theorem. The Laplace transform is similar to the
Fourier transform, except that the domain is rotated and it applies only to functions defined
on (0, ∞).
The Laplace transform can be used to convert an ordinary linear differential equation
with constant coefficients for a function f into an algebra problem whose solution is the
Laplace transform of f . It roughly converts differentiation into multiplication because of the
elementary identity:
L( f ³ )(z) = zL( f )(z) − f (0).
The algebra problem for L ( f ) is generally easy to solve, but, to find the solution to the orig-
inal differential equation, we need to find a function whose Laplace transform is the solution
to the algebra problem. In other words, given a function F, we would like to find f so that
L( f ) = F. The function f is called the inverse Laplace transformof F. See Exercise 9.4.
One method for computing the inverse Laplace transform f for a rational function F is to
use the partial fraction expansion from Section 2.2 and the fact that the Laplace transform of
f (t) = eat t n/ n! is (z − a)−(n+1). We can also compute the inverse Laplace transform using
residues.
Theorem 9.17 Suppose F is a rational function and |F(z)| → 0 as |z| → ∞. Set

² ( )
f (t) = Resp F(z)ezt ,
where the sum is taken over all poles p of F. Then L f (z) = F(z) for Rez > maxp Rep, where
the maximum is taken over all poles of F.
Proof Choose a ∈ R so that F(ζ ) is analytic in Reζ ≥ a and suppose Rez > a. Let γR
denote the positively oriented curve consisting of the vertical segment from a − iR to a + iR
followed by the semi-circular arc of the circle CR (a) of radius R centered at a in Reζ < a. By
the residue theorem, for R sufficiently large,
³
1
f (t) = F(ζ )eζ t dζ .
2π i γR
Thus, for M > 0,

³ ³ · ¸
M
1 e(ζ − z)M − 1
f (t)e−zt dt = F(ζ ) dζ . (9.12)
0 2 π i γR ζ −z
Because Reζ ≤ a < Rez on γR , the integral in (9.12) converges uniformly on compact subsets
±
of {Rez > a} as M → ∞ to − γR F(ζ )/(ζ − z)dζ . This proves the existence of the Laplace
transform of f for Rez > a.
Let σR denote the positively oriented curve consisting of the vertical segment from a + iR to
a − iR followed by the semi-circular arc of the circle CR (a) in Reζ > a. Then γR + σR = CR (a)
and, by Cauchy’s integral formula for R sufficiently large,
³ ³
1 F(ζ ) 1 F( ζ )
F(z) − L( f )(z) = dζ + 2π i dζ
2π i σR ζ −z γR ζ −z
³
= 2π1 i F(ζ )
− z dζ . (9.13)
CR (a) ζ
Because F tends to 0 at ∞, the integral around CR (a) in (9.13) tends to 0 as R → ∞, proving

Theorem 9.17.
The ideas behind the proof of Theorem 9.17 can be used to give a formula for the inverse
Laplace transform for functions F which are not necessarily rational.
±∞
Theorem 9.18 Suppose F is analytic in Rez ≥ c, −∞ | F(c + iy)| dy = M < ∞ and
|F(z)| → 0 as |z| → ∞ with Rez ≥ c. Then
³∞
1
f (t) = F(c + iy)e(c+iy)t dy
2π −∞
exists, | f (t)| ≤ Me ct , and L ( f )(z) = F(z) for Rez > c.
Proof We may suppose c = 0 because L (g)(z) = L( f )(z + c) if g(t) = e−ct f (t). The function
f (t) exists because |eiyt | = 1 and M < ∞ . Moreover,
³ ¶³ ¶
¶ K ¹ º ¶
| f (t1 ) − f (t2 )| ≤ 2 |F(iy)|dy + ¶¶ F(iy) e iyt 1
−e
iyt 2
dy¶¶ ,
| y|>K −K
so that, by uniform continuity and M < ∞ , f (t) is continuous. Then, for Rez > 0,
³ K ∞³ ³ K
1
f (t)e−zt dt = F(iy) e(iy−z)t dtdy
0 2π −∞ 0
³ ∞
e(iy −z)K − 1
= 21π F(iy)
iy − z
dy (9.14)
−∞
∞ F(iy) ³
→ − 21π dy,
−∞ iy − z
as K → ∞ . Thus L( f )(z) exists for Rez > 0.
If |z| < r and Rez > 0 then, by Cauchy’s integral formula,
³ π ³
F(reit ) it r
= 2π1 i − 1 F(iy)
2
F(z) ire dt idy.
−2 π reit − z 2π i −r iy − z
Because | F( ζ )| → 0 as |ζ | → ∞ with Reζ ≥ 0, we conclude from (9.14) that
F(z) = L ( f )(z)
for Rez > 0.
The Laplace transform and the residue theorem are used in the next theorem to give a suffi-
cient condition for the existence of an improper integral. The proof is a nice illustration of how
to deform a contour to take advantage of analyticity. Theorem 9.19 is used in Exercise 11.15
to prove the prime number theorem.
Theorem 9.19 Suppose f is piecewise continuous on [0, T ] for each T < ∞ and | f | ≤
M < ∞ on [0, +∞ ). If there exists a function g which is analytic in a neighborhood of
{Rez ≥ 0} such that± g(z) = Lf (z) for all Rez > 0, then, for Rez = 0, Lf (z) exists and equals
∞
g(z). In particular, 0 f (t)dt exists and equals g(0).
For students familiar with measure theory, the proof below is valid if the assumptions on f
are replaced by f ∈ L ∞.
L U
–ε 0 R
Figure 9.5 Enlarged half-disk.
±T
Proof Set gT (z) = 0 f (t)e−zt dt. Then gT is entire by Morera’s theorem. By a translation, it
is enough to show that limT →∞ gT (0) = g(0).
Fix R < ∞ and let U = {z : |z| < R and Rez > −ε}, where ε = ε (R) > 0 is so small that
g is analytic in a neighborhood of U. See Figure 9.5. Then
³ ´ µ
1 z
g(0) − gT (0) =
2π i
(g(z) − gT (z))ezT
R
+ Rz dz
R
(9.15)
∂U
by the residue theorem. On the semi-circle ∂ U ∩ {Rez > 0},

¶³ ∞ ¶ ³ ∞ −RezT
¶ ¶
|g(z) − gT (z)| = ¶¶ f (t)e− ztdt¶¶ ≤ M |e−zt |dt = M e Rez , (9.16)
T T
and ¶ ´ µ¶¶
¶ zT z
¶e
¶ R
+ Rz ¶¶ = eRezT 2Rez
R
(9.17)
because |z| = R. By (9.16) and (9.17), the contribution to (9.15) from the integral over
∂ U ∩ {Rez > 0 } is bounded by M /R.
For the integral over the rest of ∂ U, we treat the integrals involving g and gT separately.
Because gT is entire, by Cauchy’s theorem we can replace the path of integration with the
semi-circle C− = {z : |z| = R and Rez ≤ 0}. Then, as before,
¶³ ¶
¶ T ¶ e−RezT
|gT (z)| = ¶¶ f (t)e−ztdt ¶¶ ≤ M (9.18)
0 −Rez
and ¶ ´ µ¶¶ ´ µ
+ Rz ¶¶ = eRezT −2Rez
¶ zT z
¶e . (9.19)
¶ R R
By (9.18) and (9.19) we conclude that the integral over C− is bounded by M /R. To estimate the
integral involving g, write the rest of ∂ U as the sum of a vertical line segment L ⊂ {Rez = −ε}
and two small arcs of length ≤ K ε . Because R is fixed and g is independent of T , we can bound
the integrand on L,
¶ ´ µ ¶¶
¶
¶g(z)ezT z + R 1 ¶ ≤ K1 e−εT ,
¶ R z R¶ ε
where K1 is a constant depending on R and g. We can bound the integrand on the two small
arcs by a constant K2 , again depending on R and g. Choose ε so small that the sum of the
integrals over the small arcs is at most M/R, then choose T so large that the integral over
the line segment L is at most M /R. Thus there exists T (R) < ∞ such that | g(0) − gT (0)| ≤
4M/ R for T > T(R). Letting R → ∞, we conclude limT →∞ |g(0) − gT (0)| = 0, proving
Theorem 9.19.
9.6 Exercises
9.1 Find all non-zero residues of the following functions:

z− 1
(a) ,
(z + 1)2(z − 2)
z2 − 2z
(b) ,
(z + 1)2(z2 + 4)
(c) ez csc 2 z,
(d) ze− 1/ z ,
2
cot π z
(e) .
z6
9.2 Let C be the circle of radius 3 centered at 0, oriented in the positive sense. Find
³
eλz
dz.
C (z + 4)(z − 1)2(z2 + 4z + 5)
9.3 Suppose that f and g are analytic in a neighborhood of the closed unit disk. Let {an}
denote the zeros of f , including multiplicity, and suppose that |an| < 1 for all n. Prove
³ 2π
ei θ g(eiθ )f ³ (eiθ ) dθ ²
f (eiθ ) 2π
= g(an ).
0 n
9.4 Solve y³³ − 3y³ + 2y = 12e4t , with y(0) = 1 and y³ (0) = 0, using Laplace transforms
and Theorem 9.17.
B
In each of the exercises below, be sure you have defined your functions carefully. Prove all
claims about integrals. Draw a picture of any contour you use.
9.5 Find the Fourier transform of x3/(x2 + 1)2 . (Note that the integral is not absolutely
convergent, so part of the exercise is to prove that the integral converges, i.e., let the
limits of integration tend to ²∞ independently.)
9.6 For 0 < α < 1, find
³ ∞
xα
dx.
0 x(x + 1)
9.7 Find the inverse Laplace transform of
= (z +3z2)2+(z 12z +8
2
F(z)
+ 4)(z − 1)
in two ways: using partial fractions and using Theorem 9.17.
9.8 Verify
∞ (−1)n
² 3
(2n + 1)3
= π32 .
n=0
9.6 Exercises 155
9.9 Compute
∞ 1
²
ζ (6) = n6
n=1
by calculus of residues. You can use your answer to Exercise 9.1(e). (Remark: This
process can be used to compute ζ (2n). The value of ζ at an odd integer is another
story.)
9.10 By means of the calculus of residues, evaluate
³ ∞ √x log x
dx.
0 (1 + x2 )
9.11 Find
³ ∞ 3
x +8
dx
0 x5 + 1
using a contour integral of (log z)(z3 + 8)/(z5 + 1).
9.12 Find
³ 1
√ 1 dx.
0 x(1 − x)
Put a “dog bone” around the interval [0, 1] and add a large circle. Carefully define
the integrand so that it is analytic at ∞, then the integral over the large circle can be
found from the series expansion at ∞ . Redo the exercise by first making the substitution
x = 1/w.
10 Normal Families
One of the main goals in this chapter is to give another proof of the Riemann mapping theorem
(Theorem 8.13) which is rather elegant and useful for solving other problems. It is based on a
notion of convergence of analytic functions, called normal convergence. We also introduce a
powerful tool due to Zalcman that characterizes non-normal families in terms of an associated
convergent sequence. This result is used to prove classical results of Montel and Picard about
sets omitted by analytic and meromorphic functions.
10.1 Normality and Equicontinuity
Definition 10.1 A collection, or family, F of continuous functions on a region ± ⊂ C is

said to be normal on ± provided every sequence { fn } ⊂ F contains a subsequence which
converges uniformly on compact subsets of ±.
The family F1 = { fc (z) = z + c : |c| < 1} is normal in C but not countable. The family
F2 = {zn : n = 0, 1, . . . } is normal in D but the only limit function, the zero function, is not in
F2. The sequence zn converges uniformly on each compact subset of D, but does not converge
uniformly on D. The family F3 = {gn }, where gn ≡ 1 if n is even and gn ≡ 0 if n is odd, is
normal but the sequence {gn} does not converge.
The first lemma says that normality is local.
Lemma 10.2 Suppose ± is a region and suppose ± = ∪∞ j=1 ²j , where ²j ⊂ ± are closed
disks. A family of continuous functions F is normal on ± if and only if, for each j, every
sequence in F contains a subsequence which converges uniformly on ²j .
Proof The only if part follows by definition. Suppose that, for each j, every sequence in F
contains a subsequence which converges uniformly on ²j . Suppose { f n} ⊂ F . Then there is
a subsequence { fn } ⊂ { fn } such that { fn } converges uniformly on ²1 . Likewise, there is a
(1) (1)
sequence { f n } ⊂ { f n } which converges uniformly on ²2 , and indeed there is a sequence

(2) (1)
(k−1)
{ fn } ⊂ { f n } which converges uniformly on ²k . Then the “diagonal” sequence
(k)
{ fk(k)}
converges uniformly on each ²j , since it is a subsequence of { fn } for each k. A compact set
(k)
K ⊂ ± can be covered by finitely many ² j , so the diagonal sequence converges uniformly

on K.
The proof of the above lemma shows that we could have used an ostensibly weaker, but
equivalent, definition of normality: for each sequence and each compact subset there is a
subsequence which converges uniformly on the compact set.
We can define a metric on the space C(±) of continuous functions on a region ± as follows.
Write
= ∪∞j =1²j ,
±
where each ²j is a closed disk with ² j ⊂ ±. If f , g ∈ C(±), set
ρ j ( f , g) = sup
| f (z) − g(z) |
z∈ 1 +
²j | f (z) − g(z)|
and
∞
±
ρ ( f , g) = 2−j ρj ( f , g).
j=1
Then ρ is a metric:
(i) if ρ ( f , g) = 0, then f = g on each ²j and hence f = g on ±,
(ii) ρ ( f , g) = ρ (g, f ) for f , g ∈ C(±),
(iii) ρ ( f , g) ≤ ρ ( f , h) + ρ (h, g) for f , g, h ∈ C(±).
The triangle inequality follows from the observations that 1+x x = 1 − 1+1 x is increasing for
x ≥ 0 and, if a and b are non-negative numbers, then
a+b
1+a+b
≤ 1 +a a + 1 +b b .
Note that ρj ≤ 1 and ρ ≤ 1.
Proposition 10.3 A sequence { fn} ⊂ C(±) converges uniformly on compact subsets of ± to
f ∈ C( ±) if and only if
lim ρ ( fn , f ) = 0.
n
In other words, the space C(±) with the topology of uniform convergence on compact
subsets is a metric space. Compactness and sequential compactness are the same for metric
spaces. A family is normal if and only if its closure is compact in this topology. A sequence
which converges uniformly on compact subsets of a region ± is sometimes said to converge
normally on ± or converge locally uniformlyon ±. The term “normally convergent” has
a different meaning in functional analysis, related to convergence of norms, so it should be
avoided. Yet another term that is used is “c.c.,” which stands for compact convergence.
Proof If ρ ( f n, f ) → 0, then ρ j ( fn , f ) → 0 for each j. This implies fn converges uniformly

to f on ²j for each j, because a/(1 + a) < b is the same as a < b/(1 − b), for a > 0 and
0 < b < 1. Since each compact set K ⊂ ± can be covered by finitely many ² j , the sequence
f n converges uniformly on K. Conversely, if fn converges to f uniformly on each ² j , then,
given ε > 0, choose nj so that for z ∈ ² j
| f n(z) − f (z)| < 2ε , whenever n ≥ nj ,

158 Normal Families
for j = 1, 2, 3, . . . . Choose m so that

∞
±
2−j < .
ε
2
j =m
Then for n ≥ max{n1 , . . . , nm −1}

m
± −1 ∞
±
≤ 2−j + 2−j
ε
ρ ( fn , f ) < ε.
2
j=1 j=m
Normality is related to a strong concept of continuity, called uniform equicontinuity.
Definition 10.4 A family of functions F defined on a set E ⊂ C is

(a) equicontinuous at w ∈ E if for each ε > 0 there exist a δ > 0 so that if z ∈ E and
|z − w| < δ, then | f (z) − f (w)| < ε for all f ∈ F ;
(b) equicontinuous onE if it is equicontinuous at each w ∈ E;
(c) uniformly equicontinuous onE if for each ε > 0 there exists a δ > 0 so that if z, w ∈ E
with | z − w| < δ then | f (z) − f (w)| < ε for all f ∈ F .
For example, if FM is the family of analytic functions on D such that | f ±| ≤ M < ∞ for all
z ∈ D, then FM is uniformly equicontinuous on D. Writing f as the integral of its derivative
along a line segment from z to w, we obtain | f (z) − f (w)| ≤ M| z − w|. Thus | f (z) − f (w)| < ε
whenever | z − w| < δ = ε/ M.
Continuity, equicontinuity, uniform continuity and uniform equicontinuity are all related to
the following statement:
if |z − w| < δ then | f (z) − f (w)| < ε.
For continuity, δ is allowed to depend on ε, the function f and the point w. For equicontinuity,
δ does not depend on f . For uniform continuity, δ depends on ε and the function f , but works
for all w. For uniform equicontinuity, δ depends on ε but works for all w and for all f ∈ F .
A family of functions which is equicontinuous on a compact set K is uniformly equicontin-
uous on K. When the family has just one function, this is a familiar result about continuous
functions; moreover, the proof for equicontinuous families is the same. The next result relates
normality and equicontinuity.
Theorem 10.5 (Arzela–Ascoli) A family F of continuous functions is normal on a region

± ⊂ C if and only if
(i) F is equicontinuous on ± and

(ii) there is a z0 ∈ ± so that the collection { f (z0) : f ∈ F } is a bounded subset of C.
Proof Suppose F is normal. If F is not equicontinuous at z0 ∈ ± then there is an ε > 0,
zn ∈ ± and fn ∈ F such that |zn − z0 | < 1/n and
| fn (zn) − fn (z0)| ≥ ε (10.1)

for n ≥ n0. By normality, { fn } contains a subsequence which converges to a continuous

function f , uniformly on |z − z0 | ≤ 1/n0. Discarding the functions f n which are not in this
subsequence, and then relabeling the sequence and the points zn , we may suppose (10.1) holds
and fn converges to f uniformly on |z − z0| ≤ 1/n0. Thus
ε ≤ | fn (zn ) − f n(z0 )|
≤ | fn (zn ) − f (zn)| + | f (zn) − f (z0 )| + | f (z0 ) − fn(z0 )|. (10.2)
For sufficiently large n, the first and third terms on the right-hand side of (10.2) are less than
ε/ 3 by uniform convergence, and the second term is less than ε/3 because the limit function
f is continuous on |z − z0| ≤ 1/n0 . This contradiction shows that F is equicontinuous on ±.
If z0 ∈ ±, then S = { f (z0) : f ∈ F } is a bounded set, for if fn ∈ F and | fn(z0 )| → ∞ then, by
normality, { f n} contains a subsequence converging at z0 , which is a contradiction.
Conversely, suppose Theorem 10.5 (i) and (ii) hold. We first show that (ii) holds at each
z ∈ ±. By equicontinuity, each w ∈ ± is contained in the interior of a closed disk ² w ⊂ ±
such that
| f (z) − f (w)| < 1 for all z ∈ ²w and all f ∈ F . (10.3)
Let U = {z0 ∈ ± : (ii) holds }. If w ∈ U, then by (10.3) ²w ⊂ U, so that U is open. If

w ∈ ± \ U then, by (10.3) again, ²w ⊂ ± \ U, so that ± \ U is open. But ± is connected and
U ² = ∅ by (ii), so that U = ±.
Now suppose { fn } ⊂ F . Let D = {zk } be a countable dense subset of ±. Since { f (z1 ) :
f ∈ F } is a bounded set, we can find a subsequence { fn } such that { fn (z1)} converges.
(1) (1)
Likewise we can find a sequence { fn } ⊂ { fn } such that { f n (z2 )} converges, and indeed
(2) (1) (2)
(k−1)
there is a sequence { fn } ⊂ { f n } such that { fn(k) (zk )} converges. Then, as in the proof of
(k)
Lemma 10.2, the sequence

{ f n(n)}
converges at zk , for each k. We may relabel this sequence as { fn }.
Fix a closed disk ² with ² ⊂ ± and fix ε > 0. By (i) the family F is uniformly equicon-
tinuous on ², so we can find δ > 0 such that if f ∈ F and if z, w ∈ ² with |z − w| < δ
then
| f (z) − f (w)| < ε/3. (10.4)
Choose a finite set zk(1) , . . . , zk(N ) ∈ D ∩ ² so that for each z ∈ ² we have minj {|z − zk(j)|} < δ.
Then find M < ∞ so that for m, p ≥ M and 1 ≤ j ≤ N
| f m(zk(j) ) − f p(zk(j))| < ε/3. (10.5)
Given z ∈ ² and m, p ≥ M, choose zk(j) so that | z − zk(j) | < δ. Then
| fm(z) − fp (z)|
≤ | f m(z) − f m(zk(j) )| + | fm(zk(j) ) − fp(zk(j)) | + | f p(zk(j)) − fp (z)|. (10.6)
The first and third terms on the right-hand side of (10.6) are at most ε/ 3 by (10.4), and the
middle term is at most ε/ 3 by (10.5). Thus, for m, p > M,
| fm(z) − fp(z)| < ε

160 Normal Families
on ² . Since ε > 0 is arbitrary, this proves that { fn } is a Cauchy sequence in the uniform
topology on ². Because C is complete, the relabeled subsequence { fn} converges on ². By
Lemma 10.2, F is normal on ±.
In the discussion above, we used the Euclidean distance |α − β| to measure the distance
between α = f (z) and β = g(z). We could similarly consider families of continuous functions
with values in a complete metric space. The Arzela–Ascoli theorem holds in this context with
the same proof, replacing | f (z) − f (w)| with the metric distance between f (z) and f (w). For
example, we could consider continuous functions with values in the Riemann sphere using
the chordal distance between any two points on the sphere. In this case, we must allow the
north pole (0, 0, 1) as a possible value of a function. Equivalently, we can consider functions
with values in the extended plane C∗ = C ∪ {∞} with the chordal metric
⎧
⎨ √ 2| −√ |
α β
, if α, β ∈C
1+| | 1+| |
χ (α, β ) =⎩ α2
√1+|2 | ,
β 2
if β = ∞.
α2
Then, to say that a function f is “continuous” at z0 with f (z0) = ∞ means that for all R < ∞
there is a δ > 0 so that | f (z)| > R for all z with |z − z0| < δ . A family F of continuous
functions on ± is normal in the chordal metric if and only if F is equicontinuous in the
chordal metric. The analog of Theorem 10.5(ii) is not needed because C∗ is compact in this
topology. See Exercise 10.2. We remind the reader that the topology induced on C by χ is the
same as the usual Euclidean topology, by Corollary 1.5.
For families of analytic functions, the Arzela–Ascoli theorem, together with Cauchy’s
estimate, Corollary 4.16, gives simple criteria for normality, Theorem 10.7.
Definition 10.6 A family F of continuous functions is said to be locally boundedon ± if

for each w ∈ ± there is a δ > 0 and M < ∞ so that if |z − w | < δ then | f (z)| ≤ M for all
f ∈ F.
An equivalent definition is that F is locally bounded on ± if and only if, for each compact
K ⊂ ±, the set { f (z) : f ∈ F , z ∈ K } is bounded.
Theorem 10.7 The following are equivalent for a family F of analytic functions on a
region ±:
(i) F is normal on ±;
(ii) F is locally bounded on ±;
(iii) F ± = { f ± : f ∈ F } is locally bounded on ± and there is a z0 ∈ ± so that { f (z0) : f ∈ F }
is a bounded subset of C.
Proof Suppose F is normal. By the proof of Theorem 10.5, for each w ∈ ±, { f (w) : f ∈ F }
is bounded. By (10.3), F is locally bounded.
Now suppose F is locally bounded on ±. If | f | ≤ M on a closed disk B(z1, r) ⊂ ± centered
at z1 with radius r > 0, then, by Cauchy’s estimate, | f ±(z)| ≤ 4M/r 2 on B(z1, r/2). It follows
that F ± is locally bounded, and (iii) holds.
Finally, if (iii) holds and z1 ∈ ±, then | f ±(z) | ≤ L < ∞ for z in a disk ² centered at z1 .
Integrating f ± along a line segment from z1 to z ∈ ² , we have | f (z) − f (z1 )| ≤ L| z − z1| for all
f ∈ F . So F is equicontinuous at z1. By the Arzela–Ascoli theorem, F is normal.
We remark that the space of analytic functions on a region ± ⊂ C is closed in the topology
of uniform convergence on compact subsets by Weierstrass’s Theorem 4.29.
For families of meromorphic functions, we use the chordal metric χ (see Equation (1.4))
instead of the Euclidean metric. Lemma 10.8 is the analog of Weierstrass’s theorem for
meromorphic functions.
Lemma 10.8 If { fn} is a sequence of meromorphic functions which converges uniformly on

compact subsets of a region ± ⊂ C in the chordal metric, then the limit function is either
meromorphic on ± or identically equal to ∞.
Proof Note that

χ ( α, β ) = χ ( α1 , β1 ). (10.7)
If { fn} is a sequence of meromorphic functions which converges uniformly on compact subsets

of ± in the χ metric, then the limit function f is continuous as a map into the extended plane
C∗ . If | f (z0 )| < ∞ then f is bounded in a neighborhood of z0 , and hence fn converges to
f uniformly in the Euclidean metric in a neighborhood of z0 . By Weierstrass’s theorem, f is
analytic in a neighborhood of z0 . If f (z0) = ∞, then 1 /fn is bounded in a neighborhood of z0 ,
for n sufficiently large, and hence extends to be analytic by Riemann’s removable singularity
theorem. By (10.7) and Weierstrass’s theorem, 1/f is analytic in a neighborhood of z0. By the
identity theorem, if 1/f is not identically 0 in a neighborhood of z0, then the zero of 1/f at z0
is isolated, and hence f has an isolated pole at z0 . The set of non-isolated zeros of 1/f is then
both open and closed in ±. Because ± is connected, either f is identically equal to ∞ or f is
analytic except for isolated poles. This proves Lemma 10.8.
For example, by (10.7)

² ³
1
z − 1/n
is a normal family in C in the χ metric with limit function 1 /z. Be careful because
1/(z − 1/ n) − 1/z does not converge to 0 in the χ metric.
The equivalence of (i) and (ii) in Theorem 10.7 does not have an analog for meromor-
phic functions in the chordal metric because all continuous functions are bounded by 2 in
the chordal metric. However, there is an analog of the equivalence of (i) and (iii) using the
spherical derivative.
Definition 10.9 If f is meromorphic on a region ± ⊂ C then

χ ( f (z), f (w))
f # (z) = lim
w→z |z − w|
is called the spherical derivativeof f .
162 Normal Families
If z is not a pole of f then

2| f ± (z)|
f #(z) = .
1 + | f (z)|2
By (10.7), (1/ f )# = f # so that f # (z) is finite and continuous at each z ∈ ±. It is positive at z if
and only if f is one-to-one in a neighborhood of z.
The spherical distanced(p∗ , q∗) between p∗ , q∗ ∈ S2 is the arc-length of the shortest curve
on the sphere containing p∗ and q∗ . The quantity f # is called the spherical derivative because
d( f (z)∗, f (w)∗ )
f # (z) = lim
w→z |z − w| ,
where f (z)∗ denotes the stereographic projection of f (z).
Theorem 10.10 (Marty) A family F of meromorphic functions on a region ± ⊂ C is normal

in the chordal metric if and only if F # = { f # : f ∈ F } is locally bounded.
Proof Suppose F is normal in the chordal metric on ± and suppose the spherical derivatives
are not bounded in any neighborhood of z0 . Then there is a sequence fn ∈ F and zn → z0
so that fn# (zn) → ∞ . Taking a subsequence, we may suppose that fn converges uniformly on
compact subsets of ± in the chordal metric to a meromorphic function f , or to ∞ , by normality
and Lemma 10.8. If f (z0) ² = ∞ then f is bounded in the Euclidean metric in a neighborhood
Nz0 of z0. Because fn converges to f in the chordal metric, fn must also be bounded in Nz0 and
thus fn converges to f in the Euclidean metric on Nz 0 . By Weierstrass’s theorem, fn± converges
to f ± uniformly on compact subsets of Nz 0 and thus f n# converges uniformly on compact subsets
of Nz0 to f #. This contradicts the unboundedness of f n#(zn ). If f (z0 ) = ∞, then we can apply
this same argument to 1/f and 1 /fn using (10.7).
Conversely, suppose the spherical derivatives are bounded by M in a disk ² centered at z0.
If z, w ∈ ², let zj = z + (j/n)(w − z), 0 ≤ j ≤ n. Then, for n large,
±
n ±
n
χ ( f (z), f (w)) ≤ χ ( f (zj ), f (zj−1 )) ≈ f #(zj )| zj − zj−1 | ≤ M |z − w|.
j =1 j=1
Thus F is equicontinuous, with the chordal metric, on ² . By the version of the Arzela–Ascoli
theorem with the chordal metric (see Exercise 10.2) F is normal on ², and, by the chordal
version of Lemma 10.2, normal on ±.
A closer examination of the proof of Marty’s theorem shows that if meromorphic functions
fn converge to f in the chordal metric then fn# converges to f # . The converse fails as a sequence
of constants, or a sequence of rotations of a single function, show.
10.2 Riemann Mapping Theorem Revisited
In this section we give another proof of the Riemann mapping theorem, Theorem 8.13. Much
of the proof is the same. We avoid the use of the geodesic algorithm by using normal families
instead of Harnack’s principle.
10.2 Riemann Mapping Theorem Revisited 163
Theorem 10.11 (Riemann mapping theorem)Suppose ± ⊂ C is simply-connected and

± ² = C. Then there exists a one-to-one analytic map f of ± onto D = {z : |z| < 1 } . If z0 ∈ ±
then there is a unique such map with f (z0) = 0 and f ± (z0) > 0.
Proof Fix z0 ∈ ±. The idea of the proof is to show that there is a conformal map of ± into
D, vanishing at z0, with the largest possible derivative at z0. This forces the image region to
be as large as possible and hence to be equal to all of D.
Set
F = {f : f is one-to-one, analytic, | f | < 1 on ±, f (z0) = 0, f ± (z0) > 0}.
By the first paragraph of the proof of Theorem 8.11, F is non-empty. By Theorem 10.7, F is
normal. Let { f n} ⊂ F such that
±
lim f (z0) = M = sup{ f ± (z0) : f ∈ F }. (10.8)
n→∞ n
Replacing fn with a subsequence, we may suppose that fn converges uniformly on compact

subsets of ±. By Weierstrass’s theorem, the limit function f is analytic and { fn± } converges
to f ± . Thus f ± (z0 ) = M. In particular, f is not constant, and M < ∞. By Hurwitz’s theorem
(Corollary 8.9), f is one-to-one. Also f (z0 ) = lim fn (z0) = 0, so that f ∈ F .
Next we show that f must map ± onto D. The proof is similar to the proof in Theorem 8.11,
though we can avoid an explicit computation. Suppose there is a point ζ0 ∈ D such that f ² = ζ0
on ±. Then
f (z) − ζ0
g1 (z) = ≡ T1 ◦ f (z)
1 − ζ 0f (z)
is a non-vanishing function on the simply-connected region ± and hence it has an analytic
√
square root which is, again, one-to-one. Set g2 (z) = g1 (z) and
g2(z) − g2 (z0)
g(z) = ≡ T2 ◦ g2 (z).
1 − g2 (z0)g2(z)
Then λg(z) ∈ F , where λ = |g± (z0 )|/g± (z0). Because T1 and T2 are LFTs of D onto D,
−1
ϕ = T 1 ◦ S ◦ T2 ,
−1
where S(z) = z2 , is a two-to-one analytic map (counting multiplicity) of D onto D with

ϕ (0) = 0, and
f (z) = ϕ ◦ g(z).
By Schwarz’s lemma (or direct computation), | ϕ± (0)| < 1, so that
f ± (z0) = | f ± (z0 )| = |ϕ± (0)||g± (z0) | < | g± (z0 )| = λ g± (z0 ),
contradicting the maximality of f ± (z0 ), and hence f maps D onto D.

Uniqueness follows from Proposition 6.5.
We remark that, because the limit function in Theorem 10.11 is unique, every sequence fn
of conformal maps of ± onto D with fn (z0) = 0 and fn± (z0) → M must converge uniformly on
compact subsets of ± to the map f . See Exercise 10.1.
164 Normal Families
10.3 Zalcman, Montel and Picard
In this section we will prove Zalcman’s clever lemma characterizing non-normal families,
and use it to prove far-reaching extensions of Liouville’s theorem and Riemann’s theorem
on removable singularities. See Exercise 10.6 for a version for families of analytic functions
using the Euclidean metric.
Theorem 10.12 (Zalcman’s lemma) A family F of meromorphic functions on a region ±

is not normal in the chordal metric if and only if there exists a sequence {zn } converging to
z∞ ∈ ±, a sequence of positive numbers ρn converging to 0, and a sequence { f n} ⊂ F such
that
gn (ζ ) = fn (zn + ρn ζ ) (10.9)
converges uniformly in the chordal metric on compact subsets of C to a nonconstant function

g which is meromorphic in all of C. Moreover, if F is not normal then {zn} and {ρn } can be
chosen so that
g# (ζ ) ≤ g# (0) = 1, (10.10)
for all ζ ∈ C.
It is remarkable that non-normality can be described in terms of a convergent sequence. If
{ fn } were convergent then the functions gn given by (10.9) would converge to a constant on
compact subsets of C, since the radii ρn tend to 0. Zalcman’s lemma says that arbitrarily small
disks centered at zn can be found where fn is close to a non-trivial meromorphic function, after
rescaling.
For example, by Marty’s theorem, the family F = {zn } is normal on D and on C \ D but
not on {z : |z| < 2}. Uniform convergence fails in every neighborhood of any point on the unit
circle. So, set zn = 1 and ρn = 1/ n. Then for ζ ∈ C
´ µn
fn(zn + ρnζ ) = 1 + →e
ζ ζ
n
as n → ∞ , because
n log(1 + ζ /n)
ζ
= log(1 +ζ /ζn/n) − 0
is the difference quotient for the derivative of log(1 + z) at z = 0. It is also easy to check that
the spherical derivative of eζ satisfies (10.10).
Proof One direction of Zalcman’s lemma is easy. If F is normal then any sequence { fn}
contains a convergent subsequence, which we can relabel as { fn }, and call the limit function
f . If zn → z∞ ∈ ± and ρ n → 0, then
= fn (zn + ρn ζ ) → f (z∞),
gn (ζ )
for each ζ ∈ C as n → ∞ since the family { fn } is uniformly equicontinuous in a neighborhood

of z∞ . Thus the limit of gn is constant.
Conversely, suppose F is not normal. By Marty’s theorem, there exists wn → w∞ ∈ ± and

f n ∈ F such that the spherical derivatives satisfy
f n#(w n) → ∞ .
Without loss of generality, we may suppose w ∞ = 0 and {|z| ≤ r} ⊂ ±. Then
Mn = max
|z|≤r
(r − |z| )fn# (z) = (r − |zn |)f n#(zn ),
for some |zn | < r, since f n# is continuous. Note that Mn → ∞, since wn → 0. Then
gn (ζ ) = fn(zn + ζ /fn#(zn ))
is defined on {|ζ | ≤ Mn} because
|zn + ζ /fn# (zn)| ≤ |zn | + Mn /fn# (zn ) = |zn | + r − |zn| = r.

Fix K < ∞. If |ζ | ≤ K < Mn , then
f # (zn + ζ /fn#(zn ))
g# (ζ ) = n
n
f n#(zn )
≤ r − |z +Mnζ /f #(z )| r −M|zn|

n n n n
(10.11)
≤ r − |z r| −− ||zζn||/f #(z )
n n n
= 1 − |ζ1|/M → 1,
n
as n → ∞. By Marty’s Theorem 10.10, and the chordal version of Lemma 10.2, the family
{gn} contains a convergent subsequence in the chordal metric. Relabeling the subsequence,
(10.9) holds with ρ n = 1/ fn# (zn). By (10.11), the limit function g satisfies the inequality
and equality in (10.10) and is meromorphic by Lemma 10.8. Because g# (0) = 1, g is non-
constant. Because {zn } is contained in a compact subset of ±, we can arrange that zn → z∞ ∈
± by taking subsequences.
The first consequence of Zalcman’s lemma is an improvement of the locally bounded

condition for normality.
Theorem 10.13 (Montel) A family F of meromorphic functions on a region ± that omits

three distinct fixed values a, b, c ∈ C∗ is normal in the chordal metric.
Proof Normality is local by the chordal version of Lemma 10.2, so we may assume ± = D.
An LFT and its inverse are uniformly continuous in the chordal metric. So we may suppose
a = 0, b = 1 and c = ∞ by composing with an appropriate LFT. Without loss of generality,
we may assume that F is the family of all analytic functions on D which omit the values 0
and 1. Set
−
Fm = { f analytic on D : f ²= 0 and f ²= e2 π ik2 m
,k = 1, . . . , 2m}.
166 Normal Families
Then
F = F0 ⊃ F1 ⊃ F2 ⊃ . . . .
∈ Fm then f is analytic and f ²= 0, so that we can define f so as to be analytic. Moreover,
If f
1
2
f2
1
∈ Fm+1 . If F is not normal then there exists a sequence { fn} ⊂ F with no convergent
1
subsequence. Moreover, { f n } is then a sequence in F1 with no convergent subsequence. By
2
induction, each Fm is not normal. Thus for each m we can construct a limit function hm as
in Zalcman’s lemma. The functions hm are entire by Exercise 10.3 and non-constant since
h#m (0) = 1. By (10.10) and Marty’s theorem {hm } is a normal family. If h is a limit of a
subsequence, uniformly on compact subsets of C in the chordal metric, then h is entire by
Exercise 10.3 and non-constant since h# (0) = 1. By Hurwitz’s theorem, h omits the 2m roots
of 1 for each m. These points are dense in the unit circle. Since h( C) is connected and open,
either h(C) ⊂ D or h(C) ⊂ C \ D. Thus, either h or 1/h is bounded. By Liouville’s theorem,
h must be constant, which contradicts h# (0) = 1, and hence F is normal.
We remark that the families Fm in the proof of Montel’s theorem ¶ are·n not closed. Indeed,
1+z
the constant functions 0, 1 and ∞ are in the closure. For example, 2 omits 0 and 1 in D
and tends to 0 in D as n → ∞ . However, in the proof of Montel’s theorem, Zalcman’s lemma
yielded limit functions which are not constant, and therefore in the family Fm by Hurwitz’s
theorem.
The next consequence of Zalcman’s lemma can be viewed as an improved version of
Riemann’s theorem on removable singularities.
Theorem 10.14 (Picard’s great theorem)If f is meromorphic in ± = {z : 0 < |z −z0 | < δ },

and if f omits three (distinct) values in C∗ , then f extends to be meromorphic in ± ∪ {z0 }.
The reader can verify that an equivalent formulation of Picard’s great theorem is that an
analytic function omits at most one complex number in every neighborhood of an essential
singularity. The function f (z) = e1/ z does omit the value 0 in every neighborhood of the
essential singularity 0, so that Picard’s theorem is the strongest possible statement in terms
of the range of an analytic function. The weaker statement that a non-constant entire function
can omit at most one complex number is usually called Picard’s little theorem, and can be
viewed as an extension of Liouville’s theorem.
Proof As before we may assume a = 0, b = 1, c = ∞ and z0 = 0. Let εn → 0. If f omits

0 and 1 in a punctured neighborhood of 0, then, by Montel’s theorem, the family { f (εnz)}
is normal in the chordal metric on compact subsets of C \ {0}. Relabeling a subsequence,
we may suppose f (εn z) converges uniformly on compact subsets of C \ {0} to a function g
analytic on C \ {0} or to g ≡ ∞ by Lemma 10.8 and Hurwitz’s theorem. If g is analytic, then
|g(z)| ≤ M < ∞ on |z| = 1 and hence | f | ≤ M + 1 on |z| = εn for n ≥ n0 . But then, by
the maximum principle, | f | ≤ M + 1 on εn+1 < | z| < εn for n ≥ n0. Thus | f | ≤ M + 1
on 0 < |z| < εn0 , so that, by Riemann’s theorem on removable singularities, f extends to be
analytic in a neighborhood of 0.
If g ≡ ∞, we can apply a similar argument to 1/f ( εn z) to conclude that 1/f extends to be

analytic at 0 and hence f is meromorphic in a neighborhood of 0.
Picard’s little theorem follows from Picard’s great theorem by considering f (1/z). Picard’s
little theorem also follows from exactly the same proof by letting εn → ∞ and using
Liouville’s theorem instead of Riemann’s theorem to conclude that f is constant. Test your
understanding of the proof by writing out the details. Zalcman’s lemma has been used in
many situations to render precise the heuristic principle that if P is a property of meromorphic
functions then
{ f meromorphic on ± : f has P }
is normal on ± if and only if no non-constant meromorphic function on C has P . For example,
P could be “omits three values.”
We end this chapter with a comment about normal families and some open problems.
Normal families can be used to prove results such as Theorems 10.15–10.17.
Theorem 10.15 (Koebe) There exist K, K > 0, such that, if f is analytic and one-to-one on
D with f (0) = 0 and f ± (0) = 1, then f (D) ⊃ {z : |z| < K }.
Theorem 10.16 (Landau) There exists L, 0 < L < ∞, such that if f is analytic on D with
f ± (0) = 1 then f (D) contains a disk of radius L.
Theorem 10.17 (Bloch) There exists B, 0 < B < ∞, such that if f is analytic on D with
f ± (0) = 1 then there is a region ± ⊂ D such that f is one-to-one on ±, and f (±) is a disk of
radius B.
In each theorem, we already know that there is a constant for each f . Normal families
can be used to prove that the constants can be taken to be independent of the functions f ,
but this method of proof does not give any information about what the constants are. See
Exercises 10.12, 10.13 and 10.15. The largest K is known to be 1/4. The conformal map of
D onto C \ [1/4, ∞) is an example where 1 /4 is achieved. The largest L is called Landau’s
constant, and it is known that
¶ · ¶ ·
1 5
³ 3 ³ 6
0.5 < L ≤ L0 = ¶ · = 0.544 . . . .
1
³
6
See Section 11.4 for the definition of ³ (z). It is conjectured that Landau’s constant is L 0. The
largest B is called Bloch’s constant, and it is known that
¶ · ¶ ·
√ ³
1
³
11
3 1
= 0.433 . . . < B ≤ B0 = ¸ √
3
¶ ·
12
= 0.472 . . . ,
4 1+ 3 1
³ 4
and it is conjectured that B = B0.

168 Normal Families
10.4 Exercises
10.1 Prove that if a normal family has only one normal limit, then it converges. Hint: View
the elements of the family as points in a precompact metric space.
This simple observation is sometimes a useful bootstrap technique to prove conver-
gence of a sequence: find enough properties of a subsequential limit function to prove it
is unique. For example, if F is a normal family of analytic functions on a region ±, and
if B is a disk with B ⊂ B ⊂ ±, then any sequence { f n} ⊂ F which converges uniformly
on B must also converge uniformly on every compact subset of ±.
10.2 (a) Prove that a family of continuous functions (in the extended sense) on ± ⊂ C∗ is
normal in the chordal metric if and only if it is equicontinuous in the chordal metric.
(b) State and prove the Arzela–Ascoli theorem for continuous functions with values in
a complete metric space.
10.3 Prove that if { f n} is a sequence of analytic functions which converges uniformly in
the chordal metric on compact subsets of a region ± ⊂ C, then the limit function
is either analytic or identically ∞ . Moreover, if the limit function is analytic then the
convergence is uniform in the Euclidean metric on compact subsets of ±. Hint: Hurwitz.
10.4 Give details for the second proof of Picard’s little theorem mentioned in the text.
10.5 Suppose ± is a simply-connected region, ± ² = C, and z0 ∈ ±. Let

F = { f : f is analytic and | f (z)| < 1 on ±}.
Prove that there exists f 0 ∈ F such that
| f0± (z0 )| = sup{| f ± (z0)| : f ∈ F }.
Prove also that f0 is one-to-one on ± and f0 (z0) = 0.
10.6 Prove the following analytic version of Zalcman’s lemma. Suppose F is a family of
analytic functions on a region ± whose members vanish at z0 ∈ ±. Then F is not
normal (in the Euclidean metric) if and only if there is a sequence {zn} converging to
z∞ ∈ ±, a sequence of positive numbers ρn converging to 0, and a sequence { fn } ⊂ F
such that gn (ζ ) = fn (zn + ρnζ ) − f n(zn ) converges uniformly on compact subsets of C
to the function g(ζ ) = cζ , where c is a constant with |c| = 1. Hint: Imitate the proof of
Zalcman’s lemma using fn± instead of fn# .
10.7 Let { f n} be a sequence of analytic functions on a region ± with | fn| ≤ 1 on ±. Let K
be compact and contained in ±. Suppose { fn } converges at infinitely many points in K.
Then is it true or false that { fn } necessarily converges at every point of ±?
10.8 Let F M be the set of functions analytic on the (open) unit disk D and continuous on the
closed unit disk which satisfy
¹ 2π
| f (ei )|dθ ≤ M < ∞.
θ
0
Show that FM is a normal family on D with respect to the Euclidean metric.
10.4 Exercises 169
10.9 Let B be the set of functions f which are analytic on the unit disk D and satisfy both
f (0) = 0 and f (D) ∩ I = ∅ , where I is the interval [1, 2]. Prove that B is a normal family
(as maps from D into the complex plane with the Euclidean metric) which contains all
of its limit functions.
10.10 Suppose f and g are entire functions with the property that
f n + gn = 1.
If n = 2, prove that there is an entire function h so that f = cos h and g = sin h. Prove
that if n > 2 then f and g are constant.
10.11 Prove that the set of one-to-one analytic functions on D with the property that f (0) = 0
and f ±(0) = 1 is normal in the Euclidean metric. Show also that this set contains all its
limit points. Hint: The ingredients are Schwarz’s lemma applied to f −1 to show f ² = c
for some c = c( f ) ∈ ∂ D, Montel on D \ { 0}, Hurwitz and the maximum principle.
10.12 Prove Koebe’s theorem. Hint: See Exercises 8.7 and 10.11.
10.13 Prove Landau’s theorem. Hint: See Exercises 8.7 and 10.6.
10.14 This exercise gives an alternative proof of Picard’s little theorem, so don’t use Picard
to answer the questions below.
(a) Prove that a non-constant entire function covers disks of arbitrarily large radii. Hint:
See Exercise 10.13.
(b) If g is analytic on a simply-connected region ± such ¶ that g omits the two·values 1
¸
and −1, then prove that we can define h(z) = − π log g(z) + g(z)2 − 1 so as to
i
be analytic on ±. Furthermore, show that h omits the integers.

(c) Repeat ¶ the idea in part (b)· by constructing an analytic function k with k(z) =
¸
− π log h(z) + h(z)2 − 1 . Prove that k does not cover arbitrarily large disks.
i
(d) Deduce Picard’s little theorem from (c).
10.15 Prove Bloch’s theorem. Hint: See Exercise 10.6.

10.16 Give a proof of the Koebe, Landau or Bloch theorem with an explicit constant.
10.17 (A term project.) As a review of many of the concepts covered so far in this text, give
all details (from scratch) of the proof in this chapter of the Riemann mapping theorem:
trace backward through all of the required results and their proofs.
11 Series and Products
11.1 Mittag-Leffler’s Theorem
So far we have learned a little bit about polynomials, rational functions, exponentials and
logarithms. Now we will look at functions that arise naturally as limits of these.
Suppose f is meromorphic in a region ± with a pole at b ∈ ±. Then, recalling the Laurent
expansion,
f (z) =
cn −1 + · · · + c1 + a + a (z − b) + a (z − b)2 + . . . ,
+ (z −cnb)
(z − b)n n−1 (z − b)
0 1 2
for z near b. The sum of the first n terms
Sb(z) =
cn
+ cn−1
(z − b) n (z − b)n− 1
+ · · · + (z −c1 b)
is called the singular part off at b. If f is rational, then, by a partial fraction expansion,
m
±
f (z) = Sbk (z) + p(z),
k =1
where p is a polynomial and {bk } are the poles of f . If f is meromorphic in a region ± with
only finitely many poles {bk } and singular parts Sbk , k = 1, . . . , m, then
m
±
f (z) = Sbk (z) + g(z),
k =1
∑
where g is analytic in ±. This follows because f (z) − Sbk (z) is analytic at each bk and there-
fore in all of ±. In this section, we will find a similar expansion for meromorphic functions
in ± with infinitely many poles. As before, we say that an infinite sequence bk ∈ ± → ∂ ± as
k → ∞ if each compact K ⊂ ± contains only finitely many bk .
Theorem 11.1 (Mittag-Leffler) Suppose bk ∈ ± → ∂ ±, with bk ±= bj if k ±= j. Set

nk
± cj,k
Sk (z) = ,
(z − bk ) j
j=1
where each nk is a positive integer and cj,k ∈ C. Then there is a function meromorphic in ±
with singular parts Sk at bk , k = 1, 2, . . . , and no other singular parts in ±.
If ± is unbounded, we allow |bk | → ∞ ∈ ∂ ± ⊂ C∗ .

Proof Let
² ³
1
Kn = z ∈ ± : dist(z, ∂ ±) ≥ n
and | z| ≤ n .
Then, as in the proof of Corollary 4.28, Kn is a compact subset of ± such that each bounded
component of C \ Kn contains a point of ∂ ± and K n ⊂ K n+1 ⊂ ∪j K j = ±. Because bk → ∂ ±,
each Kn contains only finitely many bk . By Runge’s Theorem 4.27, we can find a rational
function fn with poles in C \ ± so that
´⎧ ⎫ ´
´⎨ ± ⎬ ´
´ ´
´
´⎩ Sk (z) − fn (z)´´ < 2−n
´ bk ∈Kn+1\Kn ⎭ ´
for all z ∈ Kn . Then for each m = 1, 2, . . .
⎛ ⎧ ⎫ ⎞
± ⎨ ± ⎬
⎝
⎩
Sk (z)
⎭
− fn (z)⎠
n≥m bk ∈Kn+1 \Kn
converges uniformly on K m to an analytic function on Km by the Weierstrass M-test and

Weierstrass’s Theorem 4.29. Set
⎛⎧ ⎫ ⎞
± ∞
± ⎨ ± ⎬
f (z) = Sk (z) + ⎝
⎩
Sk (z)
⎭
− fn (z)⎠ . (11.1)
bk ∈K1 n=1 bk ∈Kn +1 \ Kn
Then f is a well-defined analytic function on ± \ { bk } and f − Sk has a removable singularity

at bk for each k = 1, 2, . . . .
A more constructive proof of Mittag-Leffler’s theorem is useful in some circumstances.

We will illustrate the technique in some special cases, and leave the more general case to the
reader.
Suppose we would like to find a function meromorphic in C with singular part ak /(z − bk )
at {bk }, where |bk | → ∞. Then
∞ a
± k
z − bk
k =1
will work, provided the sum converges. When does it converge? If | z| ≤ R < ∞, write
∞ a
± ±
ak ± ak
k
z − bk
= + . (11.2)
k=1 {k:|b |≤2R} z − bk { k:|b |
k k >2R } z − bk
The first sum has only finitely many terms. For the second sum, |z| ≤ R < |bk |/2, so that
´ ´
´ 1 ´ 2
´ z − bk ´ ≤ | bk | .
´ ´
Thus, if
∞ |a |
±
|bk | < ∞,
k
k =1
then the second sum in (11.2) converges uniformly on {|z| ≤ R } to an analytic function, by
Weierstrass again. The right-hand side of (11.2) then is meromorphic in | z| < R with singular
part ak /(z− bk ) at bk , provided |bk | < R. Since R is arbitrary, the sum in (11.2) is meromorphic
in C.
What if |bk | tends to ∞ more slowly? If we examine the proof, we just need the tail of
the series on the left-hand side of (11.2) to converge for each R. Mittag-Leffler’s idea was to
subtract a polynomial from each term so that the result converges. As we have seen before, if
bk ± = 0 and | z| < | bk | , then
1 1
∞µ ¶
1 ± z j
z − bk
= −bk (1 − z
)
= −bk bk
.
bk j=0
So it is natural to subtract a few terms of the expansion to make it smaller:

´ ´ ´ ´
´ µ ¶±
nk µ ¶j ´ ´ ± ∞ µ ¶j ´
´ 1 ´ ´ z ´´
´
´ z − bk − −1b z ´ ´ 1
=
bk ´ ´ −bk
´ ´ bk ´´
´ k
j=0 j=nk +1
´ ´ ´ ńk +1
´ ´´ ´
≤ ´´ b1 ´´ ´´ bz ´´ 1 −1| z | , (11.3)
k k bk
provided | z| < | bk | .
For example, the following proposition holds:
Proposition 11.2 If 0 < | bk | → ∞ and if, for some integer n ≥ 0,

± |ak | < ∞,
k
|bk |n+2
then
⎛ ⎞
∞
± µ ¶±
n µ ¶j
f (z) = ⎝ ak − ak z ⎠
z − bk −bk bk
k =1 j=0
is meromorphic in C with singular part Sbk (z) = z−a b k

k
at bk , k = 1, 2, . . . , and no other poles.
To prove Proposition 11.2, if | z| < R, split the sum into two pieces: a finite sum of the terms
with | bk | ≤ 2R and a convergent sum of the terms with | bk | > 2R. Then use the estimate in
(11.3) with nk = n for all k.
Example 11.3
2 ∞
± 1
π
2
= (z − n)2
. (11.4)
sin πz n=−∞
Proof The right-hand side of (11.4) converges uniformly on compact subsets of C. The limit
is meromorphic with singular parts Sn (z) = 1/(z − n)2 at z = n and with no other poles. To
see this, follow the proof of Proposition 11.2. Fix R < ∞ and split the sum into two pieces:
a finite sum of terms with |n| < 2R and the remaining infinite sum. In the second sum when
|z| ≤ R, the nth term is uniformly bounded by 1/(|n| − R)2 ≤ 4/n2 . Thus the second (infinite)
sum converges uniformly and absolutely on |z| ≤ R to an analytic function by Weierstrass’s
theorem.
d
Note that sin π z/z is the difference quotient for dz sin π z at z = 0, and so sin π z/π z
approaches 1 as z → 0. The function π z/ sin π z has a removable singularity at z = 0 and
is an even function so that
πz
sin π z
= 1 + O(z2 )
near 0. By squaring and dividing by z2 , we conclude that the singular part of π 2 / sin2 π z at
z = 0 is 1/z2. If n is an integer then sin2 π (z − n) = sin2 π z. Thus the singular part of the
left-hand side of (11.4) at z = n is the same as the singular part of the right-hand side of (11.4)
at z = n for each integer n. Set
2 ∞
± 1
F(z) = −
π
.
sin 2
πz n=−∞
(z − n)2
Then F is entire and F(z + 1) = F(z).

Write z = x + iy and suppose 0 ≤ x ≤ 1. We claim that
|F(z)| → 0, (11.5)
as | y| → ∞. If so, F is bounded in the strip 0 ≤ x ≤ 1, and, since F(z + 1) = F(z), the

function F is bounded in C. By Liouville’s theorem, F is constant and, by (11.5), F ≡ 0,
proving Example 11.3. To see (11.5), first observe that
| sin π z| = 21 |e− πy+iπ x − eπ y−iπ x | → ∞
as |y| → ∞. Thus the left-hand side of (11.4) tends to 0 as |y| → ∞, with 0 ≤ x ≤ 1.

Likewise, in the strip 0 ≤ x ≤ 1, the right-hand side of (11.4) is dominated by
1 ± 1
y2
+ (|n| − 1)2 + y2
,
n±=0
which
· ∞ also2 tends to 0 as | y| → ∞ , as can be seen by comparing the sum with the integral
dx /(x + y2). This proves (11.5) and Example 11.3.
1
Example 11.4
1 ±µ 1 1
¶
cot π z = + + . (11.6)
z− n
π
z n
n±=0
Proof First observe that

d 2 ± 1
π cot π z = − =−
π
.
dz sin2 π z (z − n)2
But
± 1
cot π z ± = ,
z −n
π
because the latter sum does not converge. The difficulty is that, for | z| ≤ R, the nth term
behaves like −1/n for large n. However,
1
z−n
+ n1 = (z −z n)n ∼ n12 ,
∑
for large n and n−2 < ∞. To prove Example 11.4, suppose |z| ≤ R and split the sum on
the right-hand side of (11.6) into the sum of the terms with |n| ≤ 2R and the sum of the terms
with |n| > 2R. The first sum is finite, and the second sum has terms satisfying
´ ´ ´ ´
´ 1 ´ ´ ´
´
´z − n + n1 ´´ = ´´ (z −z n)n ´´ ≤ n R· n = 2R
n2
.
2
∑
Because 2R/n2 < ∞, the right-hand side of (11.6) is meromorphic in | z| ≤ R with poles
only at the integers and with prescribed singular parts 1/(z − n), for each R < ∞. Furthermore,
by Weierstrass’s Theorem 4.29 and Example 11.3, the right-hand side of (11.6) has derivative
± 2
− z12 − (z −
1
n)2
=− π
2
= dzd π cot π z.
n± = 0
sin πz
Thus, the two sides of (11.6) differ by a constant, C. Convergence of the right-hand side of
(11.6) is absolute, so that we can add the terms involving n and −n to obtain
1
∞
± 2z
z
+ z2 − n2 , (11.7)
n=1
which is clearly odd. Since π cot π z is also odd, we must have C = 0, proving (11.6) as well
as the equality with (11.7).
A slightly subtle point is that the convergence of (11.7) is not the same as the convergence
of (11.6). Convergence of (11.7) is the same as convergence of symmetric partial sums in
(11.6) from −N to N, which is a priori weaker than allowing the upper and lower limits of the
partial sums to tend to ∞ independently.
We used the function π cot π z in Section 9.4 to compute sums using residues.
P
Example 11.5 (Weierstrass function) Suppose w1 , w2 ∈ C \ { 0} with w1 /w2 not real. In
other words, w1 and w 2 are not on the same line through the origin. There is no non-constant
entire function f satisfying f (z + w1) = f (z + w2 ) = f (z) for all z, by Liouville’s theorem.
See Exercise 11.6. But there are meromorphic functions with this property. The Weierstrass
P function is defined by
± µ ¶
1 1 1
P (z) = + − , (11.8)
z2 (z − mw1 − nw2)2 (mw1 + nw2 )2
(m,n)±=(0,0)
where the sum is taken over all pairs of integers except (0, 0).
To prove convergence of this sum, we first observe that there is a δ > 0 such that | mw1 +
nw2| ≥ δ unless m = n = 0, for, if |mj w1 + nj w2| → 0, then
´ ´
´ w1 nj ´´
´
´ w2 + mj ´
→ 0,
contradicting the assumption that w1 /w2 is not real. Thus {ζ m,n = mw1 + nw2 : m, n ∈ Z},
where Z denotes the integers, forms a lattice of points in C with no two points closer than δ .
If we place a disk of radius δ/2 centered at each point of the lattice, then the disks are disjoint.
The area of the annulus k ≤ |ζ | ≤ k + 1 is (2k + 1)π so there are at most Ck lattice points in
this annulus, for some constant C depending on δ .
For |z| < R, we split the sum in (11.8) into a finite sum of terms with | ζm,n | ≤ 2R and the
sum of terms with |ζ m,n| > 2R. Note that if |z| < R and | ζ | > 2R, then
´ ´ ´ ´
´ 1 1 ´´ ´ 2zζ − z2 ´ |ζ | + R) ≤ 10R .
´
´ (z − ζ )2 − ζ2 ´
= ´ ζ 2 (z − ζ )2 ´´
´ ≤ R(2
|ζ | |ζ /2|2 |ζ |3
2
We conclude that for K ≥ R and |z| < R

´ ´ ∞ ´ ´
± ´´ 1 1 ´´ ± ± ´
´ 1 1 ´´
´
´ (z − ζm,n )2
− ζ 2 ´´ = ´
´ (z − ζm,n )2
− ζ2 ´
|ζm,n|>2K m,n k =2K k <|ζm,n |≤k+1 m,n ´
∞
± 10K
≤ Ck
k3
< ∞.
k=2K
By Weierstrass’s theorem, the Weierstrass P function is meromorphic in C with singular part
S(z) = 1/(z − mw1 − nw2 )2 at mw1 + nw2 and no other poles.
Next we show that P(z + w1 ) = P (z). By Weierstrass’s theorem,
±
P ² (z) = − 23 − 2
.
z (z − ζm,n)3
±=
ζm,n 0
By the same estimate, this series converges absolutely so that we can rearrange the terms,
obtaining P ² (z + w1) = P ² (z), and hence P (z + w1 ) − P(z) is a constant. The series for P is
even, so P (z + w1 ) = P (z) when z = −w1 /2, and thus P (z + w1 ) = P (z) for all z. A similar
argument shows that P(z + w2 ) = P (z).
It is possible to replace the proof in the preceding paragraph with an estimate of the differ-
ence of two partial sums. However, when z is replaced by z + w1 , the term involving z will
be paired with the wrong value at 0. An estimate of the error made by replacing it with the
correct value must be made, as well as an estimate of the terms in the difference of the partial
sums which do not cancel, since the summation is shifted by w1.
The Weierstrass P function is a basic example of an elliptic function. Elliptic functions are
important in the application of complex analysis to number theory. See Exercises 11.6 and
8.6, and Section 16.3.
11.2 Weierstrass Products
Just as infinite sums are used to create meromorphic functions with prescribed poles, infinite
products are used to create analytic functions with prescribed zeros. If { fj } are analytic and
¸
f j (zj ) = 0, then nj=1 f j vanishes at z1 , z2 , . . . , zn . If we want a function to vanish at infinitely
many zj , it is natural to try
n
¹
lim fj .
n→∞
j=1
However, this limit may not exist. Moreover, it is possible for the product of infinitely many
non-zero numbers to converge to zero, which might create additional zeros. To resolve this
difficulty, we will first treat the convergence of infinite products of complex numbers.
¸n
Proposition 11.6 Suppose pj ∈ C \ {0}. Then j=1 p j converges to a non-zero complex
number P as n → ∞ if and only if
∞
±
log pj
j=1
converges to a complex number S, where log pj is defined so that −π < arg pj ≤ π . Moreover,
if convergence holds then P = eS and lim pj = 1.
∑ ¸
Proof Let Sn = nj=1 log pj and P n = nj=1 pj . If Sn → S then P n = eSn → eS ∈ C \ {0}.
Conversely, if Pn → P ∈ C \ {0}, then, by altering p1 if necessary, we may suppose P > 0.
Then log Pn → log P, where the logarithm is defined so that −π < arg Pn ≤ π . Note that
|Pn−1||pn − 1| = |Pn − Pn−1 | → 0,
so that pn → 1 and hence log pn → 0. But
⎛ ⎞
±n
log Pn = ⎝ log pj ⎠ + 2π ikn
j=1
→ 0 and
for some integers kn. Because log pn
log pn + 2π i(kn − kn−1 ) = log Pn − log P n−1 → 0,
∑
we must have kn = kn for n0 sufficiently large and n ≥ n0 . It follows that limn nj=1 log pj
0
exists, proving Proposition 11.6.
¸
One possibility for a definition of absolute convergence of an infinite product pj would
¸n ¸n n(n +1)
be the convergence of the partial product j =1 |pj |. However, j =1 = eij ei
does not con- 2
verge even though the product of the absolute values is 1. The notion of absolute convergence
was useful for infinite sums because it implies that the terms in the sum can be rearranged
without affecting convergence.
¸∞
Definition 11.7 If pj are non-zero complex numbers then we say j=1 p j converges
∑
absolutely if | log pj | converges.
¸
It follows that if pj are non-zero complex numbers such that ∞ j=1 pj converges
¸ absolutely
then any rearrangement of the terms in the infinite product will not affect limn nj=1 pj . The
next lemma is useful for determining absolute convergence.
¸∞
Lemma 11.8 If pj are non-zero complex numbers then j=1 pj converges absolutely if and
only if
∞
±
|pj − 1|
j =1
converges.
Proof
´
log z d ´
lim
z→1 z − 1
= dz
log z´´ = 1,
z=1
so that if pj → 1 then, for j sufficiently large,

´ ´
´ log pj ´
´
´ pj − 1 − 1´´ < 12
and thus
1
2
|pj − 1| ≤ | log pj | ≤ 23 |pj − 1|,
and Lemma 11.8 follows.
¸
11.8 says that the product ∞
∑Lemma j =1 pj converges absolutely if and only if the sum
∞ (pj − 1) converges absolutely. The analogous
j =1 statement for (non-absolute) convergence
√j ∑ ¸
is false. If pj =∑e(−1) j/
then log pj converges by the alternating series test and hence
∑
pj
converges, but j (pj − 1) diverges. Similarly, if pj = 1 + i (−√1)j
then (pj − 1) converges but
j
∑ ¸
log pj diverges, and hence pj diverges. See Exercise 11.1.
As mentioned at the start of this section, we would like to consider infinite products of
¸
analytic functions fj (z), but we would like to allow the f j to have zeros. For that reason, we
make the following definition.
¸∞
Definition 11.9 Suppose { f j } are analytic on a region ±. We say that j =1 fj (z) converges
on ± if
n
¹
lim fj (z)
n→∞
j=0
converges uniformly on compact subsets of ± to a function f which is not identically equal

to 0.
= ¸mj=0 fj (z) converges to f ±≡ 0 on ± then f is analytic on ± by Weierstrass’s

If Fm (z)
Theorem 4.29. Moreover, Fm converges to F (k) uniformly on compact subsets of ±, for k =
(k)
1, 2, . . . . By Hurwitz’s Theorem 8.8, the zero set of f is the union of the zero sets, counting
multiplicity, of { fn }. Also note that f m converges uniformly on compact subsets of ± to the
∑
constant function 1. For example, by the proof of Lemma 11.8, if
¸∞
| fj (z) − 1| converges
uniformly on compact subsets of ± then j=1 fj (z) converges on ±.
The next result, called the Weierstrass product theoremis the analog for products of
Mittag-Leffler’s theorem.
Theorem 11.10 (Weierstrass) Suppose ± is a bounded region. If {bj } ⊂ ± with bj → ∂ ±,

and if n j are positive integers, then there exists an analytic function f on ± such that f has a
zero of order exactly n j at bj , j = 1, 2, . . . , and no other zeros in ±.
Theorem 11.10 is also true for unbounded regions ±. See Exercise 11.7.
Proof Let
² ³
Kn = z ∈ ± : |z − w| ≥ n1 for all w ∈ ∂ ± .
Then, as in the proof of Corollary 4.28, Kn is a compact subset of C such that each component
of C \ K n contains a point of ∂ ± and Kn ⊂ Kn+1 . Choose aj ∈ ∂ ± so that
dist(bj , ∂ ±) = |bj − aj |.
As in the proof of Corollary 4.28, if bj ∈
/ Kn then the line segment from bj to aj does not
intersect Kn . Thus we can define log((z − bj )/(z − aj )) so as to be analytic in C \ Kn.
Each Kn contains at most finitely many bk because bk → ∂ ±. By Runge’s Theorem 4.27,
we can find a rational function rn with poles in C \ ± so that
´⎧ ⎫ ´
´⎨ µ ¶ ´
´ ± z − bk ⎬ ´
´ nk log − rn(z)´´ 2−n , (11.9)
z − ak
´⎩ <
´ ⎭ ´
bk ∈ Kn +1 \ Kn
for all z ∈ Kn . Then

⎛ ⎧ ⎫ ⎞
µ ¶
± ⎨ ± z − bk ⎬
⎝
⎩
nk log
z − ak ⎭
− rn (z)⎠
n≥m bk ∈Kn+1 \Kn
converges uniformly on Km to an analytic function on Km by Weierstrass’s Theorem 4.29. Set

⎛ ⎞
¹ µ z − bk ¶nk ¹
∞ ¹ µ
z − bk
¶nk
f (z) = ⎝ ⎠ e−rn (z) .
z − ak z − ak
bk ∈K1 n=1 bk ∈ Kn +1 \ Kn
Then f is a well-defined analytic function on ± with a zero of order nk at bk , k = 1, 2, . . . ,

and no other zeros.
Here is a more constructive version when ± = C.
Theorem 11.11 Suppose {an} ⊂ C \ {0} and suppose g is a non-negative integer such that
∞
± 1
n=1
|an|g+1 < ∞.
Then
⎛ ⎞
∞º µ ¶j
z»
g
¹ ±1 z
1− exp ⎝ ⎠ (11.10)
an j an
n=1 j=1
converges and represents an entire function with zeros at {an } and no other zeros.
The number g in Theorem 11.11 is called the genus of the infinite product. In the case
when g = 0, we interpret the sum in (11.10) as equal to 0. The an in Theorem 11.11 do not
need to be distinct, so that the theorem can be used to construct functions with zeros of order
more than one.
Proof By integrating the series for 1/(1 − w) we have

∞ 1
±
log(1 − w) =− wj ,
j
j=1
where the sum converges and therefore is analytic for | w| < 1. If |z| < R and |a| > 2R, then
´ ´
´ º » ± g º z »j ´´ ´ ´g+1
´ z 1 ´R´
´log 1 − + ´ ≤ C ´´ ´´ ,
´ a j a ´ ´ a
´ j=1
º ∑g »
where C is a constant. This follows from the fact that log(1 − w) + j=1 1j wj /w g+1 is
analytic on the unit disk and therefore bounded on the disk of radius 1 /2. (The interested
reader can prove that C = 2 works by estimating the tail of the series.)
If |z| ≤ R then
´ ´
± ´´ µ ¶ ± g µ ¶j ´´ ±
´log 1 − z + 1 z ´
≤ CRg+1 1
∞.
|an |>2R n |
| <
´ a j a ´ a g+1
| an |>2R ´ n
j =1
n ´
Thus, by Weierstrass’s theorem,

¹ º z » ∑gj=1 1j ( anz ) j
1− e
an
|an| >2R
is analytic and non-zero in {|z| < R}. The finite product with terms | an | ≤ 2R then gives the
zeros in |z| ≤ R.
Example 11.12 The function

∞ º
¹ z» z
f (z) = z 1− en
n=−∞
n
n±=0
∑ −2
converges since n < ∞ (g = 1) and vanishes precisely at the integers. Thus we can
write
∞ º
¹ z» z
sin π z = eG(z) z 1−
n
en,
n=−∞
n±= 0
where G is entire. Indeed, sin π z has a simple zero at each integer and no other zeros, so, if
we divide sin π z by z times the product, the resulting function is entire and never equal to 0
in C, and hence has an analytic logarithm. To find G, we take the logarithmic derivative f ² /f :
µ ¶
1 ± 1
π cot π z = G² (z) +
z
+ z−n
+ n1 .
n±=0
But then, by (11.6), we must have G² (z) ≡ 0, and hence G is constant. Since
sin π z
lim
z →0 z
= π,
we must have eG(0) = π , and we obtain

¹º z» z
sin π z = π z 1− en.
n
n±=0
We now give a few corollaries of Theorem 11.10, the Weierstrass product theorem.
Corollary 11.13 If ± is a region then there is a function f analytic on ± such that f does
not extend to be analytic in any larger region.
Proof Take a sequence {an} ⊂ ± → ∂ ± such that ∂ ± ⊂ {an }. By the Weierstrass product
theorem, we can find f analytic on ±, with f (an ) = 0 but f not identically zero. If f extends
to be analytic in a neighborhood of b ∈ ∂ ± then the zeros of the extended function would not
be isolated.
In several complex variables, a similar result is not true. If B r = {(z, w) : | z|2 + |w |2 < r2 },
then any function which is analytic on B2 \ B1 extends to be analytic on B 2.
Corollary 11.14 Suppose ± is a region and an → ∂ ±, with a n ± = am when n ±= m, and

suppose {cn } are complex numbers. Then there exists f analytic on ± such that
f (an ) = cn, n = 1, 2, . . . .
Results like Corollary 11.14 are usually called interpolation theorems. See Corollary 11.16
for a version of this result when the zeros are not distinct.
Proof By the Weierstrass product theorem, we can find G analytic on ± with a simple zero
at each an. Let
G(z)
dn = lim
z→an z − a n
= G² (an ).
Since the zero of G at an is simple, dn ±= 0. By Mittag-Leffler’s theorem, we can find F
meromorphic on ± with singular part
cn/dn
Sn (z) =
z − an
at an and no other poles in C. Then f (z) = F(z)G(z) is analytic on ± \ { an } and
G(z)
lim F(z)G(z) = lim (z − an)F(z)
z →an z →an z − an
= cdn dn = cn .
n
Thus the singularity of f at each an is removable and f extends to be analytic on ± with
f (an) = cn, n = 1, 2, . . . .
Corollary 11.15 If f is meromorphic in ± then there are functions g and h, analytic on ±,

such that
g
f = .
h
Proof Let {an} be the poles of f , where the list is written such that a pole of order k occurs k
times in this list. By the Weierstrass product theorem, there is a function h analytic on ± with
zeros {an} and no other zeros. Then g = fh is analytic on ± \ {an } with removable singularity
at each an. Thus g extends to be analytic in ± and f = g/h.
The idea behind the proof of Corollary 11.14 can be used to prove the analog of Mittag-
Leffler’s theorem for Taylor polynomials.
Corollary 11.16 Suppose pn is a polynomial of degree d n for n = 1, 2, . . . , and suppose

an → ∂ ±. Then there is a function f analytic on ± such that, for each n = 1, 2, . . . ,
f (z) − pn (z − an )
has a zero of order at least d n + 1 at an. In other words, the Taylor polynomial for f at an of
degree dn is the prescribed function pn (z − an ).
Proof By the Weierstrass product theorem, we can find h analytic on ± with zeros
an of order dn + 1. Let Sn (z) be the singular part of pn (z − an)/h(z). This means that
Sn (z) − pn(z − an )/h(z) is analytic in a neighborhood of an. By Mittag-Leffler’s theorem,
we can find g meromorphic on ± with singular part Sn (z) at an . Then g(z) − pn(z − an )/h(z)
extends to be analytic near an and so g(z)h(z) − pn (z − an) is analytic near an with a zero of
order at least dn + 1 at an . This proves the corollary with f = gh.
11.3 Blaschke Products
Weierstrass’s product theorem shows that the zero set of an analytic function on the unit disk
can be any sequence tending to the unit circle. More information on the growth of the analytic
function will give us more information on how quickly the zeros approach the circle. This is
the content of Jensen’s formula.
Theorem 11.17 (Jensen) Suppose f is meromorphic on | z| ≤ R with zeros {ak } and poles
{bj }. Suppose also that 0 is not a zero or a pole of f . Then
¼ ± ±
1 π
R R
log | f (Reit )| dt = log | f (0)| + log − log (11.11)
2π −π
| ak | <R |ak | |bj| <R |bj |
Proof Replacing f (z) by f (Rz), we may assume R = 1. First suppose that f has no poles or
zeros on | z| = 1. Write
¸ z−ak
k 1− ak z
f (z) = ¸ z −bj
g(z),
j 1−bj z
where g is analytic on |z| ≤ 1 and has no zeros on |z| ≤ 1. Then

¸
|aj |
| f (0)| = ¸j |g(0)|
|bk |
k
and log | f (eit )| = log | g(eit )| = Re log g(eit ), where log g(z) is analytic on |z| ≤ 1. Note that
if z = eit then dz /(iz) = dt. So, by Cauchy’s theorem,
¼ ¼
1 π
1 π
log | f (eit ) |dt = Re log g(eit )dt
2π − π 2π −π
± ±
= Re log g(0) = log | f (0)| − log |aj | + log |bk |.
j k
Thus (11.11) holds if f has no zeros or poles on |z| = R.

To complete the proof of Jensen’s theorem, by Corollary 11.15 we may suppose f is analytic
on |z| < R1 , for some R 1 > R. The right side of (11.11) is continuous as a function of R,
0 < R < R 1, and equality holds in (11.11) except for isolated R. So it suffices to prove that the
left side of (11.11) is non-decreasing when f is analytic. The function vM = max( −M, log | f |)
is subharmonic, for M < ∞. As · in the paragraph following Definition 7.2, to· avoid a foray
into measure theory, we define log | f (Reit )|dt to be the decreasing limit of vM (Reit )dt. If
r < R1 then using the Poisson integral formula, find uM harmonic on | z| < r with uM = vM
on |z| = r. Then vM − uM is subharmonic and ≤ 0 in |z| < r by the maximum principle. Thus
for 0 < s < r
¼ 2π ¼ 2π ¼ 2π
vM (se )dt ≤
it
uM (se )dt = 2π uM (0) =
it
vM (reit )dt.
0 0 0
Letting M → ∞ shows that the left-hand side of (11.11) is non-decreasing.
See Exercise 11.12 for a more direct proof of the equality in (11.11) when f has a zero or
pole on |z| = R.
Corollary 11.18 Suppose f is analytic in D, f ±≡ 0, with zeros {an }, and suppose

¼ π
sup log | f (reit )| dt < ∞ .
r <1 −π
Then
∞
±
(1 − |an |) < ∞.
n=0
Proof Since f has only finitely many zeros at 0, we may divide them out and suppose
f (0) ± = 0. Then, by Jensen’s formula,
∞
± 1
log
|an| < ∞.
n=0
∑ ∑
But limx→1 x−1
log x
= 1, so that log |a1n | < ∞ if and only if (1 − |an |) < ∞.
We remark that ¼ ¼
log | f (re ) |dt ≤ Cp
it
| f (reit )|p dt
·
for some constant Cp depending on p > 0. So if |z |=r | f | pdt is uniformly bounded, then the
hypothesis of Corollary 11.18 holds.
As a partial converse to Corollary 11.18, we have the following theorem.

∑
Theorem 11.19 If {an } ⊂ D such that (1 − |an |) < ∞ then
∞ |a | z − a
¹
B(z) =
n n
n=0
−an 1 − anz
converges uniformly and absolutely on compact subsets of D, where we define the convergence
factor |an| /(−an ) to be equal to 1 if an = 0. The function B is analytic on D, is bounded by 1
and has zero set exactly equal to {an }.
The function in Theorem 11.19 is called a Blaschke product.
Proof Without loss of generality, an ±= 0 for all n. Then for |z| ≤ r < 1
´ ´
´ |an | µ z − an ¶ ´ (1 − |an |)||an| z + an|
´ −an 1 − an z − 1´ =
´ ´
|an (1 − anz)| ≤ (inf2(1|a −|)(1
|an |) .
− r)
n
By Weierstrass’s M-test and Lemma 11.8, B converges uniformly and absolutely on |z| ≤ r.
Note that the partial products for B are all bounded by 1 and analytic on D.
Theorem 11.19 gives the following factorization result, which allows us to divide out the
zeros of a bounded analytic function without affecting the supremum norm.
Theorem 11.20 If f is bounded and analytic on D with zero set {an} (counting multiplicity),
and if B is the Blaschke product with zero set {an }, then
f (z) = B(z)eg(z)
for some analytic function g with
sup |eg(z)| = sup | f (z)| .

z ∈D z∈D
Proof Without loss of generality, supD | f | = 1. Write
BN (z) =
N
¹ |an | z − an
n=1
−an 1 − an z .
Then, by Corollary 3.11, | f (z)/BN (z)| ≤ 1. Choose rn so that B ± = 0 on | z| = rn. Then f /B N
converges uniformly to f /B on |z| = rn and, by the maximum principle, the convergence
is uniform on | z| ≤ rn. Now let rn → 1. Thus h = f / B is bounded by 1 on D and non-
vanishing. This implies g = log h can be defined as an analytic function on D. By construction,
sup D | h| ≤ supD | f | = 1, but also |B| ≤ 1, so supD |h| = supD | f |.
It is difficult to overestimate the importance of Blaschke products in function theory on D.

11.4 The Gamma and Zeta Functions
In this section we introduce two important functions, Euler’s gamma function and the
Riemann zeta function.
As we saw in Example 11.12, the simplest entire function with zeros at the integers is
sin π z. By the same convergence proof,
∞º
¹ z » −z /n
G(z) ≡ 1+ e (11.12)
n
n=1
is the simplest entire function with zeros only at the negative integers. Moreover,
1
zG(z)G(−z) = sin π z. (11.13)
π
Note that G(z − 1)/(zG(z)) has no zeros and hence
eγ (z)zG(z) = G(z − 1), (11.14)
where γ (z) is entire. By Weierstrass’s theorem and the uniform convergence of the product
(11.12), we can take logarithmic derivatives and
∞µ 1 ¶ ± ∞µ ¶
² (z) + 1 + ± − 1
= 1
− 1
z +n n z− 1+n n
γ
z
n=1 n=1
1 ±∞µ 1 1
¶
= z −1+ −
z +n n+1
.
n=1
Thus
±∞ µ1 1
¶
²
γ (z) = −1 + − =0
n n+1
n=1
and γ is a constant. By (11.12) and (11.14),
m µ
¹ ¶
1 −1/n
1 = G(0) = eγ G(1) = lim eγ 1+ e
m →∞ n
n=1
= mlim (m + 1)e−(1+ 2 + 3 +···+ m ) .

γ 1 1 1
→∞ e
Thus µ ¶
1 1 1
γ = mlim
→∞
1 + + + · · · + − log(m + 1) ≈ 0.57722 . . . .
2 3 m
The proof above shows that this limit, called Euler’s constant, exists.
The gamma functionis defined by
1
∞
¹
²(z) = zG(z)e zγ
= e− z 1z γ 1
1 + z/n
ez/ n.
n=1
Then ² is a non-vanishing meromorphic function with poles at the non-positive integers and
no other poles. Moreover, by (11.14),
² (z + 1) = z² (z) (11.15)
and, by (11.13),
²(z)² (1 − z) = sinππ z . (11.16)
It follows from (11.12) and (11.14) that ²(1) = 1, and, by (11.15) and induction,
²(n) = (n − 1)! .
So ²(z + 1) is a meromorphic extension of the factorial function n!. See Exercise 11.14, where
² is used to estimate the growth of n!.
The gamma function also has an integral representation, given by the next theorem.
Theorem 11.21 If Rez > 0 then

¼ ∞
²(z) = tz −1e−t dt. (11.17)
0
Proof The integral converges uniformly and absolutely on compact subsets of Rez > 0
because |tz −1 | = |t |Rez−1 is integrable near t = 0 and because |t |Rez−1 e−t ≤ e−t /2 for t large.
By Morera’s theorem, the right-hand side of (11.17) is analytic for Rez > 0.
Set
¼ n º t »n
²n(z) = tz −1 1 − dt.
0 n
Then we claim that for Rez = x > 0
¼ ∞
lim ²n(x) = t x−1 e−t dt (11.18)
n→∞ 0
and
1
lim
n→∞ ²n(z)
= ²1(z) (11.19)
uniformly on compact subsets of C. If (11.18) and (11.19) hold then Theorem 11.21 follows
from the uniqueness theorem.
Lemma 11.22 If M < ∞ then, as n → ∞,

º t »n t
1− e increases to 1
n
uniformly for 0 ≤ t ≤ M.
Proof The slope of the secant line to the graph of log u with one end at u = 1 − t /n and the
other end at u = 1 decreases to 1 as n → ∞ since the graph of log u is concave. Thus
½ ¾
º t» log(1 − nt ) − log 1
n log 1 − +t = − 1 (−t) → 0
n (1 − nt ) − 1
uniformly for 0 ≤ t ≤ M. Lemma 11.22 now follows since −t ≤ 0 and es is an increasing

function of s.
By Lemma 11.22,
¼ ∞
²n (x) ≤ ²n+1 (x) ≤ tx −1e−t dt,
0
and (11.18) holds for x > 0. Substitute t = ns in the definition of ²n(z) and obtain
¼ 1
²n(z)n
−z = sz −1 (1 − s)nds ≡ f n(z). (11.20)
0
Note that f0 (z) = 1/z and fn+1 (z) = fn (z) − fn(z + 1), so that, by induction,
n!
fn (z) = .
z(z + 1) · · · (z + n)
Thus
n º
½ n ¾
1 ¹ z » ¹ º z » ∑n
²n (z)
= n−z z 1 + = z 1 + e−z/k ez( k =1 k1 −log n)
k k
→ ²1(z) ,
k=1 k=1
as n → ∞ , uniformly on compact subsets of C, proving (11.19) and Theorem 11.21.
The Riemann zeta functionis defined by

∞
±
ζ (z) = n−z ,
n=1
where n−z = e−z log n. The zeta function is important because of its connection with the
prime numbers, as illustrated in Theorem 11.23 below. By the integral test, the series for
ζ (z) converges uniformly and absolutely on compact subsets of Rez > 1. By Weierstrass’s
theorem, it is analytic in Rez > 1. The zeta function can be extended to Rez > 0 by comparing
the sum to the appropriate integral. For Rez > 1,
∞ ½ ¼ n+1 ¾
1 ±
ζ (z) − z −1 = n−z − t−z dt
n=1 n
∞ ¼ n+1 ¼ (11.21)
± t
= zs−z −1ds dt.
n=1 n n
The estimate
¼ n+1 ¼
|zs−z−1 |ds dt ≤ |2z| n−Rez−1
t
n n
shows as above that (11.21) converges uniformly and absolutely on compact subsets of
Rez > 0 to an analytic function. Thus ζ (z) extends to be meromorphic in Rez > 0 with a
simple pole at z = 1 and no other poles. Figure 11.1 shows the image of {z = 21 + iy : −40 ≤
y ≤ 40} by the function ζ (z). As y increases, the “loops” are traced counter-clockwise, pass-
ing through 0, except for one loop that crosses (0, 1) before going clockwise around 0. The
Riemann hypothesis, perhaps the most famous problem in mathematics, is that ζ (z) ± = 0 for
Rez > 21 . The next theorem proves the absence of zeros in Rez > 1 and gives the connection
between ζ (z) and the prime numbers.
Figure 11.1
1
ζ (2 + iy), −40 ≤ y ≤ 40.
Theorem 11.23 Let {pn } be the primes, with p1 = 2, p2 = 3, . . . . If Rez > 1 then
1
∞
¹
= (1 − p−z
n ). (11.22)
ζ (z)
n=1
In particular, ζ (z) ± = 0 for Rez > 1.
Proof The infinite product in (11.22) converges uniformly and absolutely for Rez ≥ x0 > 1
because
|p−n z| = p−n Rez ≤ n−x . 0
By Lemma 11.8 and the Weierstrass M-test, the right-hand side of (11.22) is analytic in
Rez > 1. Because the series defining ζ (z) is absolutely convergent, we can rearrange the
terms to obtain
± ± ±
ζ (z)(1 − 2−z) = n−z − (2n)−z = m−z .
m odd
Similarly,
± ± ±
ζ (z)(1 − 2−z )(1 − 3−z) = m−z − (3m)−z = m−z .
m odd m odd m not divisible
by 2 or 3
By induction,
¹
k ±
ζ (z) (1 − p−
n )=
z
m−z = 1 + p−k+z1 + . . . .
n=1 m not div. by p1 ,...,pk
Thus
´ ´
´ k ´
´ζ (z) ¹ (1 − p−z ) − 1´ ≤ ± | n−z | .
´ n ´
´ ´ n≥p
n=1 k+1
∑ −z
Since pk+1 → ∞, and since |n | converges, the product converges and
∞
¹
ζ (z) (1 − p−
n ) = 1,
z
n=1
for Rez > 1.
As one might expect from (11.20), there is a connection between ²(z) and ζ (z). By the
same substitution used to establish (11.20), it follows from (11.17) that
¼ ∞
²(z) =n z
sz−1 e−ns ds
0
and
∞
± ∞¼∞
±
n−z ² (z) = sz −1e−ns ds
n=1 n=1 0
¼ ∞
e−s
= sz −1 ds.
0 1 − e−s
Thus, if Rez > 1 then
¼ ∞ z−1
ζ (z) = ²1(z) s
es − 1
ds.
0
The Riemann zeta function is used to prove the celebrated prime number theorem, which
describes the rate of growth of the number of primes which are at most x, as x → ∞ . The
complex analysis ingredients are Theorem 9.19 and Theorem 11.24 See Exercise 11.15.
Theorem 11.24 The Riemann zeta function ζ (z) is non-zero for Rez ≥ 1.
Proof We proved in Theorem 11.24 that ζ (z) ± = 0 in Rez > 1. By (11.22) for Rez > 1
²(z) ∞ log p
± ∞
± log pn
− ζ
ζ (z)
= n
pzn − 1
= ³(z) + pzn (pzn − 1)
, (11.23)
n=1 n=1
where
∞ log p
± ± ∞
³(z) = n
pzn
= − dzd p−n z.
n=1 n=1
The last sum in (11.23) converges and is analytic for Rez > 21 , so that ³(z) extends to be
meromorphic in Rez > 21 , with poles only at the zeros of ζ (z) in Rez > 21 and at z = 1, the
pole of ζ . If ζ (z) = (z − z0)k eg(z) , where g is analytic near z0, then
= z −−kz − g² (z).
²(z)
− ζζ (z)
0
Thus ³ has a simple pole with integer residue −k at z0. The residue equals 1 at z0 = 1 by
(11.21).
Note that ζ (z) = ζ (z), so that if α ± = 0 then
lim ε³ (1 + ε ) = 1, lim ε³(1 + ε ³ i α) = −M, and lim ε³ (1 + ε ³ i2α) = −N,

ε↓0 ε ↓0 ε ↓0
where M and N are non-negative integers. Multiplying the magical inequality
2 µ
± ¶ ∞ log p
±
4
+ ε + ijα ) = (pinα/2 + p− ≥0
n iα/2 4
³(1 )
2+j p1+ε
n
j=−2 n=1 n
by ε > 0, and letting ε → 0, shows that 6 − 8M − 2N ≥ 0. Since M and N are non-negative

integers, we must have that M = 0, i.e., ζ (1 + iα) ± = 0.
11.5 Exercises 189
11.5 Exercises
¸∞ ¸
11.1 (a) Prove that n=1 (1 + ni ) diverges but ∞n=1 |1 + n | converges.
i
√ ∑ ∑
(b) If pj = e(−1) / j prove that log pj converges but (pj − 1) diverges.
j
√ ∑
(c) If pj = 1 + i(−1) j/ j prove that (pj − 1) converges but
∑
log pj diverges.
11.2 Prove directly from the definition that
∞ n2 + π
¹
n2 − 1
n=3
converges to a non-zero real number. (Convergence can be proved by quoting the

right result in the text. Test your understanding of the material by trying to make the
estimates yourself.)
11.3 Find an explicit bounded analytic function f ± ≡ 0 defined on the unit disk with zeros
{∑zn} such that every point of the unit circle is a cluster point of {zn } and such that
(1 − |zn| ) < ∞.
11.4 If B is the Blaschke product in Theorem 11.19 with zero set {an} and if E = {1/an},
show that B converges uniformly on compact subsets of C \ E. In particular, if the zeros
do not accumulate on an open interval I ⊂ ∂ D then B is analytic on a neighborhood
of I and |B | = 1 on I. Hint: The LFT (z + an/| an |)/(1/an − z) is bounded on these
compact sets.
11.5 Find a meromorphic function with singular parts
n 4
Sn (z) = + ,
z − n (z − n)2
for n = 1, 2, 3, . . . . Prove convergence of your sum.
11.6 Suppose w1, w2 are non-zero complex numbers such that w1/ w2 is not real. Prove
there are no non-constant entire functions satisfying f (z + w1 ) = f (z + w 2) = f (z) for
all z. What happens if w1 /w2 is real?
11.7 (a) Prove Weierstrass’s Theorem 11.10 for ± = C by using z − bk instead of
(z − bk )/(z − ak ). Define log(z − bk ) by cutting the plane along a line segment
from bk to ∞ .
(b) Prove Weierstrass’s Theorem 11.10 for unbounded regions ±. In this case, view
∗
∂ ± as a subset of C so that we allow subsequences b j with |b j | → ∞. Hint: Use
the chordal distance instead of the Euclidean distance, and use z − bk instead of
(z − bk )/(z − ak ) if ∞ is the closest point in ∂ ± to bk .
11.8 Suppose {bk } → ∞ and |ak /bk | ≤ M < ∞ for all k ≥ k0 . Prove that
⎛ ⎞
± µ ¶±
k µ ¶j
⎝ ak − −abk z ⎠
z − bk bk
k k j=0
is meromorphic in C with singular part ak /(z − bk ) at bk and no other poles in C. The

point of this exercise is that the upper limit of the inside sum is equal to k, no matter
how slowly |bk | → ∞. In practice, it is usually best to take as few terms as possible,
so in many examples fewer terms are taken.
11.9 Find an explicit entire function g with g(n log n) = nπ , n = 1, 2, 3, . . . . Prove that your
function works. Hint: Make use of an entire function that vanishes at the integers.
11.10 Find an entire function of least possible genus, with simple zeros at the Gaussian
integers:
{m + in : m, n integers },
and no other zeros. In other words, find the smallest integer g such that (11.10) con-
verges uniformly and absolutely on compact sets to an analytic function with simple
zeros at the Gaussian integers.
11.11 If ± is a region, then H(±) is the algebra of analytic functions on ±. One of the first
questions you might ask about an algebra is: what are its ideals? Suppose g1 , g2 ∈
H(±) with no common zeros. Prove that there are functions f1 , f 2 ∈ H(±) with
f 1(z)g1(z) + f2 (z)g2(z) = 1 (11.24)
for all z ∈ ±. If g1 and g2 had a common zero then we could not find f1 and f2 satis-
fying (11.24). This says that the ideal generated by g1 and g2 consists of all analytic
functions on ± if and only if g1 and g2 have no common zeros. A proof can be based
on Corollary 11.16, but there is also a more direct proof using only the Mittag-Leffler
Theorem 11.1.
11.12 (a) Show that
¼ π
log |eit − 1|dt = 0.
−π
Hint: Compare the integral with the integral of log |reit − 1| when r < 1, using
different estimates for |t | < δ and |t | ≥ δ .
(b) If f is meromorphic in |z| ≤ R, show that we can assume f has no zero or pole on
|z| = R when
² (z)
proving Jensen’s theorem because of part (a).
11.13 Let ´(z) = ²² (z) .
(a) Show ´ ² (z) + ´ ² (z + 21 ) = 4´ ² (2z).

(b) Integrate (a) twice to obtain
µ ¶
az+b 1
²(2z) =e ²(z)² z+
2
for some constants a and b. (Note: log ²(z) is only defined locally.)
(c) Find a and b. The resulting formula is called Legendre’s duplication formula.
11.14 Stirling’s formula for the gamma function (used in combinatorics, physics and
elsewhere).
√
(a) Suppose x > 21 . Substitute t = ( x + v)2 in the integral formula (11.17) for ²(x)
to show that
√
² (x)ex x
¼ ∞
= 2 ϕ x (v)e
−v2 dv,
x x
−∞
11.5 Exercises 191
where
⎧ √
if v ≤ − x
⎨ 0, µ ¶2x−1
ϕx (v) =⎩ e−2vx 2
1
1+
v √
if v ≥ − x.
√x ,
(b) Prove that

² (x)ex
√x ¼ ∞ √
=2 e−2v dv =
2
lim 2π .
x→∞ xx −∞
Hint: Show log ϕx (v) ≤ 1 for each x > 1/2. Then show log ϕx (v) → −v2 as
x → ∞, for each v. Use (11.16) to find ² ( 12 ) and then the last equality.
(c) Use part (b) to estimate n! (Stirling’s formula).
11.15 (Prime number theorem) In this exercise p will always denote a prime number. Let
π (x) denote the number of primes ≤ x. Set
± ± log p
θ (x) = log p and ³(z) = pz
.
p≤x p
( )
(a) Prove θ (x) ≤ Cx. Hint: The binomial coefficient 22k− 1 is divisible by the product
k
of the primes p satisfying 2k−1 < p ≤ 2k , so that, for 2 k−1 < x ≤ 2k ,

k
± k
±
≤ ) − θ (2j−1 ) ≤ log(1 + 1)2 ≤ C2k ≤ 2Cx.
j j
θ (x) θ (2
j =1 j=1
·∞ ·∞
(b) Prove 1 ( θ (x)
x
− 1) dxx = 0 (θ (et )e−t − 1)dt converges. Hint:
L
(
θ (e
t
)e−t − 1 (z) =
) ³(z + 1) − 1 = − ζ ² (z + 1) − 1 + g(z)
z+ 1 z ζ (z + 1) z
is analytic in {Rez > 0}. Apply Theorems 11.24 and 9.19.
(c) Prove limx→∞ θ (x)x = 1. Hint: If θ (x) ≥ λ x, for λ > 1 then, because θ is non-
decreasing,
¼ λx
µ ¶
θ (t) dt
t
−1 t
≥ λ − 1 − log λ > 0.
x
If θ (x) ≤ λx for λ < 1, find a similar bound for the integral from λ x to x.
(d) Prove that, for ε > 0,
θ (x) ≤ π (x) log x ≤ 1θ−(x)ε + x1− ε

log x.
Hint: θ (x) ≥ θ (x) − θ (x1− ε ) and π (x1−ε ) ≤ x1−ε .

π (x) log x
(e) Prove lim
x →∞ x
= 1.
C
11.16 (a) Prove that, for b ∈ ±,

Ib = { f ∈ H(±) : f (b) = 0}
is a closed maximal ideal in H(±), the holomorphic functions on ±, and all closed
maximal ideals are of this form. The topology is given by uniform convergence on
compact subsets of ±. Hint: First prove the analog of Exercise 11.11 for n functions
instead of 2. Then use that H(±) is closed under uniform convergence on compact
sets.
(b) If zn ∈ ± and zn → ∂ ±, then { f ∈ H(±) : ∃m = m( f ) such that f (zk ) =
0 for all k ≥ m} is an ideal in H(±) which is dense and not generated by a single
function.
(c) Find the principal ideals.
Here is an open problem. Suppose ak > 0, |zk | → ∞, as k → ∞, and ∑ |az | < ∞. Then
k
k
∞ a
±
f (z) =
k
z − zk
k=1
is meromorphic in C. Prove f (w) = 0 for some w. The three-dimensional analog (also open)
∑
is that if ak > 0, |xk | → ∞ as k → ∞ and ∞ k=1 a k /|xk | < ∞ , then
2
∞ a (x − x )
±
F(x) =
k k
k =1
|x − xk |3
has at least one zero x ∈ R3 . The R3 -valued function F is the gradient of a real-valued function
and can be thought of as the gravity force due to the positive masses ak at xk ∈ R3 . The
problem asks you to prove that there is always some place in space where you will experience
no force. See [5] and [8].
PART III
12 Conformal Maps to Jordan Regions
We define a Jordan curve in C∗ to be the homeomorphic image of the unit circle. It may
seem obvious, but by no means simple to prove, that a Jordan curve divides C∗ into exactly
two regions. We begin the chapter with some examples of strange curves and regions. Then,
following Pommerenke [20], we use complex analysis to prove a topological lemma due
to Janiszewski, which is quite useful for tackling subtle topological problems in the plane.
Janiszewski’s lemma is used many times in Section 12.3 to prove the Jordan curve theorem.
In Section 12.4, following Garnett and Marshall [10], we prove the useful and important the-
orem, due to Carathéodory, that a conformal map of the unit disk onto a region bounded by a
Jordan curve extends to be a homeomorphism of the closed disk onto the closure of the region.
12.1 Some Badly Behaved Regions
In this section we give some examples of rather badly behaved regions which are useful to
keep in mind as we prove the Jordan curve theorem.
Example 12.1 (Jordan curve with positive area)Our first example is a Jordan curve with
positive area. First we construct a totally disconnected compact set E of positive area, then we
describe how to pass a curve through this set.
The compact set E will be the intersection of compact sets Ej , j = 0, 1, 2, . . . . The initial
set E0 is the closed unit square E 0 = [0, 1] × [0, 1]. Choose σ1 < 1 and let E1 ⊂ E0 be the
union of four squares of side length σ1 /2 located in each of the corners of E0 . For example,
the lower left square is [0, σ1 /2] × [0, σ1 /2]. The area of E 1 is σ12 . Next choose σ2 < 1 and let
E 2 ⊂ E1 be the union of 16 squares, one in each of the corners of E1 with side length σ1σ2 /4
and total area ( σ1 σ2)2 . Repeating this process we obtain compact sets En ⊂ En−1 consisting
± ± ±
of 4 n squares with side length nj=1 (σj /2) and area nj=1 σj2. Choose σj so that ∞ j=1 σj > 0.
For instance, we can take σj = 1 − 1/ (j + 1)2 . Set
∞
²
E = E n.
n=0
Then E is a compact set of positive area. See Figure 12.1.

The Jordan curve J is constructed in stages also. A Jordan arc is the homeomorphic
image of the unit interval [0, 1]. First we construct a Jordan arc containing the set E, then we
close it up to form a Jordan curve. Connect the four squares in E1 using three line segments
Figure 12.1 E2 and connecting segments.
which leave the lower right corner of each square and connect to the upper left corner of an
adjacent square, as shown in Figure 12.1. There is a natural left-to-right ordering of these three
intervals. Inside each square in E1 we insert three line segments connecting the four squares
in E 2 using the same pattern. Repeat this process for each square in each En .
Next construct a mapping of a dense subset of the unit interval onto the union of these line
segments. Let I 1 consist of three disjoint closed intervals contained in (0, 1) with total length
1/ 2. Let I2 consist of three disjoint closed intervals in each of the four intervals of [0, 1] \ I1.
Repeat this process so that In consists of 3 · 4n−1 disjoint intervals contained in [0, 1] \ In−1 of
∑∞ −n
total length 2−n. Then ±(∪∞ 1 In) = 1 2 = 1. Construct a map ϕ mapping each interval in
In linearly onto one of the segments connecting a square in E n, preserving the natural ordering
left to right within each square of En−1, n = 1, 2, . . . . Because the side lengths of the squares
in E n tend to 0 and the lengths of the intervals in [0, 1] \ I n also tend to 0, as n → ∞, the map
ϕ extends to be continuous and one-to-one. It maps the unit interval to a curve which contains
the set E. If we then connect the upper left corner of E0 to the lower right corner of E 0 (in
C \ E0 ) we obtain a (closed) Jordan curve J with positive area.
Example 12.2 (Dense set of spirals)Another example of a Jordan arc J0 is found by spi-
raling in to 0 in a clockwise direction then spiraling back out again counter-clockwise with a
Jordan arc. For example γ (x) = e−(1+i)x , 0 ≤ x < ∞ , followed by −e(1+i)x , −∞ < x ≤ 0.
See Figure 12.2.
Note that arg z is neither bounded above nor bounded below on J0 . Starting with the unit
circle ∂ D, replace a small arc of of ∂ D with a rotated, scaled and translated version of J0
having the same endpoints. Next choose another subarc of the resulting curve which does not
contain the limit point of the spiral and replace it by a rotated and scaled version of J0 with
the same endpoints. If the replaced arc is small enough, the resulting curve J1 will still be
Figure 12.2 Constructing a dense set of spirals.

12.2 Janiszewski’s Lemma 197
ζr ζb
ζg
Figure 12.3 Constructing the Lakes of Wada.
a Jordan curve. We can repeat this process countably many times to obtain a closed Jordan
curve J with a dense set of spirals. So, there is a dense subset of points {an} ⊂ J such that
arg(z − an ) is neither bounded above nor bounded below for z ∈ J. This curve can even be
constructed so that it has finite length. In this example it may be rather difficult to tell whether
a given point near the Jordan boundary lies in the “inside” or “outside” region.
Example 12.3 (Lakes of Wada) The third example is three disjoint simply-connected
regions with the same boundary. Suppose lakes L r , Lg , Lb are three simply-connected sub-
sets of D with disjoint closures and suppose each closure intersects the unit circle in a single
point, ζ r , ζg , ζ b ∈ ∂ D. For example, we can take three disjoint disks in D which are tangent
to the unit circle. Then the dry land D0 = D \ (L r ∪ Lg ∪ L b) is compact, connected and has
a simply-connected interior. Increase the size of one of the lakes by allowing water to flow
out along an interval on its boundary so that the three lakes remain simply-connected with
disjoint closures, each intersecting the unit circle in a single point and such that the remaining
dry land D1 = D \ (Lr ∪ Lg ∪ L b) ⊂ D0 is compact, connected and has a simply-connected
interior. See Figure 12.3. Repeat this process, cycling through the lakes infinitely many times.
We can also arrange that, after expanding each lake n times, the distance from each point
of dry land z ∈ D3n to each lake is at most 1/n. The final dry land D = ∩Dn is compact,
connected and equal to the boundary of each lake.
The set D in this third example is not a Jordan curve. As we shall see in the proof of the
Jordan curve theorem, at most two complementary components of a Jordan curve can have
a boundary point in common. A similar construction works for n lakes, giving n disjoint
simply-connected regions, all possessing the same boundary.
12.2 Janiszewski’s Lemma
The reader may wish to review the definition and elementary properties of the components of
an open set U ⊂ C∗ as given in Exercise 4.2(a). It is not hard to verify also that
(a) if E ⊂ U is connected, then E is contained in a component of U;
(b) each component Ua of U has boundary ∂ Ua contained in C∗ \ U;
(c) if F is a homeomorphism of C∗ then Ua is a component of U if and only if F(Ua ) is a
component of F(U).
See Exercise 12.1.
Definition 12.4 A closed set E ⊂ C∗ separates points a, b ∈/ E if a and b belong to distinct

components of C∗ \ E.
Lemma 12.5 (Janiszewski) Suppose K1 and K2 are compact subsets of C such that K 1 ∩ K2
is connected and 0 ∈
/ K1 ∪ K2. If K 1 does not separate 0 and ∞ , and if K2 does not separate
0 and ∞, then K1 ∪ K2 does not separate 0 and ∞ .
The hypothesis that K1 ∩ K2 is connected is essential, for, if K1 is the closure of the top half
of the unit circle and if K2 is the closure of the bottom half of the unit circle, then K 1 ∪ K2
separates 0 and ∞.
Janiszewski’s lemma follows from Lemmas 12.6 and 12.7 below.
Lemma 12.6 If a compact set E separates 0 and ∞ then we cannot define log z to be analytic
in a neighborhood of E.
In other words, there is no function g which is analytic in a neighborhood of E satisfying

eg(z) = z, if E separates 0 and ∞ .
Proof Suppose g is analytic in an open set W ⊃ E and eg(z) = z, with 0, ∞ ∈ C∗ \ W ⊂

C∗ \ E. Since 0 and ∞ belong to distinct components of C∗ \ W, by Theorem 5.7 we can find a
±
closed polygonal
³ ³ σ± ⊂ W so that n(σ , 0) = 1. But, by the chain rule, g (z) = 1/z, so that
curve
2π i = σ 1/z dz = σ g (z)dz = 0 by the fundamental theorem of calculus. This contradiction
proves Lemma 12.6.
Lemma 12.7 Suppose K 1 and K2 are compact sets such that K1 ∩ K2 is connected and
0, ∞ ∈/ K1 ∪ K2 . If K1 does not separate 0 and ∞ , and if K2 does not separate 0 and ∞, then
we can define log z to be analytic in a neighborhood of K 1 ∪ K 2.
Proof By hypothesis, we can find simple polygonal curves σ1 and σ2 connecting 0 to ∞

with σj ∩ Kj = ∅ , j = 1, 2. Then
(σ1 ∪ σ2 ) ∩ (K1 ∩ K2) = ∅.

By item (a) above, the connected set K1 ∩ K2 is contained in one component U of C∗ \ (σ1 ∪σ 2).
See Figure 12.4.
Since C∗ \ σj is simply-connected, j = 1, 2, we can find functions fj analytic on C∗ \ σj
such that ef j (z) = z on C∗ \ σj . Then ef1 −f2 = 1 on C∗ \ (σ1 ∪ σ2). Thus, on each component of
C∗ \ (σ1 ∪ σ 2), f1 − f2 is a constant of the form 2π ki for some integer k. We may then add a
constant to f 2 so that f1 = f2 on U. Because K1 \ U and K2 \ U are disjoint compact sets, we
can find open sets V j so that
Kj \ U ⊂ Vj ⊂ Vj ⊂ C∗ \ σj ,
12.3 Jordan Curve Theorem 199
K2
V2
K1 V1
σ2
0
U
σ1
Figure 12.4 Janiszewksi’s lemma.
for j = 1, 2 and V1 ∩ V2 = ∅. Set

´
f (z) =
f1 (z), for z ∈ V1 ∪ U
f2 (z), for z ∈ V2 ∪ U.
Then f is analytic on V1 ∪ V2 ∪ U, an open neighborhood of K1 ∪ K2, and ef (z) = z.
Janiszewski’s lemma follows immediately from Lemmas 12.6 and 12.7. Moreover, by item
(c), using LFTs, we can replace 0 and ∞ by any two points not in K1 ∪ K2 in the statement of
Janiszewski’s lemma.
Corollary 12.8 If J is a Jordan arc in C∗ then C∗ \ J is open and connected.
Proof Since LFTs are homeomorphisms of C∗ , it suffices to show that if 0, ∞ ∈/ J then J

does not separate 0 and ∞. Write J = ∪n1 Jk , where Jk are subarcs of J such that Jk ∩ Jk−1 is
a single point. We may choose each Jk so small that, for each k, there is a half-line from 0 to
∞ contained in C∗ \ Jk and hence no Jk separate 0 and ∞. But J1 ∩ J2 is a single point, and
hence connected, so that, by Janiszewski’s lemma, the arc J1 ∪ J2 does not separate 0 and ∞ .
The intersection of this arc with J3 again is a single point, and hence J1 ∪ J2 ∪ J3 does not
separate 0 and ∞ by Janiszewski’s lemma. By induction, J does not separate 0 and ∞.
12.3 Jordan Curve Theorem
In this section we will prove the Jordan curve theorem.
Theorem 12.9 (Jordan) If J ⊂ C∗ is a (closed) Jordan curve then C∗ \ J has exactly

two components, each of which is simply-connected. Moreover, J is the boundary of each
component.
The proof of Theorem 12.9 is divided into four lemmas.
Lemma 12.10 If J is a (closed) Jordan curve, and if U is a component of C∗ \ J, then

∂ U = J.
Jn
ζ
zn U
αn z0
Jn
Figure 12.5 Boundary of a component.
ζ1 J1 ζ2
z2 σ
ζ0 z1
D0
J2
Figure 12.6 A modification of J.
Proof Suppose ζ ∈ J and z0 ∈ U. Because J is the homeomorphic image of the unit circle,
given n < ∞ we can find Jordan arcs Jn, Jn± ⊂ J with Jn ∪ Jn± = J, ζ ∈ Jn, ζ ² ∈ Jn± and Jn ⊂ Dn,
where Dn = {z : |z − ζ | < 1/n}. By Corollary 12.8, Jn± does not separate ζ and z0. So there is
a polygonal curve σ n from z0 to ζ such that σn ∩ Jn± = ∅ . See Figure 12.5.
Let zn be the first intersection of σn with ∂ Dn . The subarc αn ⊂ σn from z0 to zn does
not intersect Jn± and does not intersect Jn ⊂ Dn , and hence does not intersect J. Since αn is
connected, we have αn ⊂ U, and hence zn ∈ U. But limn zn = ζ ∈ / U, so ζ ∈ ∂ U. This
shows that J ⊂ ∂ U. If ζ ∈ ∂ U, then ζ does not belong to any component of C∗ \ J since the
components are open. Thus ζ ∈ J, and we have shown J = ∂ U.
Next we will construct a modification σ of the curve J and prove that C∗ \ σ has exactly
two components, and then use this result to prove that C∗ \ J has exactly two components.
Take ζ 0 ∈ C∗ \ J. Then there is a straight-line segment [ζ 0 , ζ1 ] with ζ1 ∈ J and [ ζ0 , ζ 1) ∩ J =
∅. Morever, we can choose another line segment [ζ0, ζ2 ] with ζ2 ∈ J and [ζ0, ζ2 ) ∩ J = ∅ and
[ζ0 , ζ 1] ∩ [ζ0 , ζ 2] = {ζ0 }. Otherwise J would be contained in a half-line from ζ1 to ∞ , which
is impossible. Write J = J1 ∪ J2, where Jj are Jordan arcs with J1 ∩ J2 = {ζ 1, ζ2 }. Switching
ζ1 and ζ 2 if necessary,
σ = J1 ∪ [ζ1, ζ0 ] ∪ [ζ0 , ζ2]

is a (closed) Jordan curve. In other words, σ is the modification of J found by replacing J2
with the union of two intervals. See Figure 12.6.
Lemma 12.11 The set C∗ \ σ has exactly two components.
Proof If D0 is an open disk centered at ζ 0 with J ∩ D0 = ∅, then D0 \ ([ζ1 , ζ 0] ∪ [ζ 0, ζ2 ])

consists of two connected open circular sectors. By item (a) in Section 12.2, each sector must
be contained in a component of C∗ \ σ and so, by Lemma 12.10, there can be at most two
components in C∗ \ σ .
Take z1 and z2 in distinct sectors of D0 \ ([ζ 1, ζ0 ] ∪ [ζ 0, ζ2 ]). Then (σ \ D0 ) ∪ ∂ D0 does not

separate z1 and z2. Moreover, σ ∩ ((σ \ D0 ) ∪ ∂ D0) = σ \ D0, which is connected. So if σ also
does not separate z1 and z2 then, by Janiszewski’s lemma, E = σ ∪ (σ \ D0) ∪ ∂ D0 does not
separate z1 and z2 . But clearly ∂ D0 ∪ [ζ 1, ζ0 ] ∪ [ζ0 , ζ 2] ⊂ E does separate. This contradiction
proves that z1 and z2 are in distinct components of C∗ \ σ , proving Lemma 12.11.
Lemma 12.12 The set C∗ \ J has at least two components.
Proof ∈ J1 \ {ζ1, ζ2 } and define

Take ζ
α = J2 ∪ [ ζ2 , ζ 0] ∪ [ζ 0, ζ1 ].
Let D be a disk centered at ζ such that D ∩ α = ∅. By Lemma 12.10, ζ ∈ ∂ G1 ∩ ∂ G2 ,

ζ ζ
where G1 and G2 are the two components of the complement of σ . Take w 1 ∈ G1 ∩ D and ζ
w 2 ∈ G2 ∩ D . The points w 1 and w2 are separated by σ , but not by α . Note that J ∩ α = J2 is

ζ
connected so if J also does not separate w 1 and w2 then, by Janiszewski’s lemma, J ∪ α does
not separate w1 and w2 . But this is a contradiction since σ ⊂ J ∪ α . We conclude that J must
separate w1 and w 2 and hence, C∗ \ J has at least two components.
Lemma 12.13 The set C∗ \ J has no more than two components.
Proof Suppose H1 and H2 are the components of C∗ \ J containing w1 and w2 from the
proof of Lemma 12.12. If C∗ \ J has another component H3, then, by Lemma 12.10 again, we
can find w3 ∈ H3 ∩ Dζ . But w3 ∈/ σ and hence w 3 ∈ G1 or w3 ∈ G2. If w3 ∈ G1 then w1 and
w 3 are not separated by σ and are not separated by α because w1 , w3 ∈ Dζ and Dζ ∩ α = ∅.
But σ ∩ α = [ζ 1, ζ0 ] ∪ [ζ 0 , ζ2 ] is connected, so that, by Janiszewski’s lemma, w1 and w3 are
not separated by σ ∪ α. But this contradicts the assumption that J ⊂ σ ∪ α separates w1 and
w 3. A similar contradiction is obtained if w3 ∈ G2, proving Lemma 12.13.
Proof of Theorem12.9 Theorem 12.9 now follows from Lemmas 12.12, 12.13, and 12.10,
and the observation that the complement in C∗ of one component of C∗ \ J is equal
to the closure of the other component, and hence is connected. So each component is
simply-connected.
We can use the method of Exercise 5.1 and the Jordan curve theorem to show that two of
the displayed points in Figure 5.3 are in the bounded component of the complement of the
Jordan curve, and the other point is in the unbounded component. So, there is an arc in the
complement which connects two of the points, and there is another arc in the complement
connecting the third point to a point far from J. This exercise can be done without assigning a
direction to the curve.
12.4 Carathéodory’s Theorem
In the course of proving the Schwarz–Christoffel formula in Section 8.4 we proved that a
conformal map of the disk onto a region bounded by a simple polygon extends to be a
homeomorphism of the closed disk onto the closure of the polygonal region. This is a special
case of a more general theorem due to Carathéodory.
A Jordan regionis a simply-connected region in C∗ whose boundary is a Jordan curve.
Theorem 12.14 (Carathéodory) If ϕ is a conformal map of D onto a Jordan region ², then

ϕ extends to be a homeomorphism of D onto ². In particular, ϕ (eit ) is a parameterization of
∂ ².
For an application of Theorem 12.14, see the first paragraph of Section 8.1.
Proof Using an LFT, we may suppose ² is bounded. First we show ϕ has a continuous
extension at each ζ ∈ ∂ D. Let 0 < δ < 1 and set
γδ = D ∩ {z : |z − ζ | = δ}.
The idea of the proof is that the image curve ϕ(γδ ) cuts off a region Uδ whose closure shrinks
to a single point as δ → 0. See Figure 12.7. It is not hard to show that the area of Uδ decreases
to 0, but this is not enough. We use a “length-area principle” to show that the diameter of the
boundary of Uδ , and hence the diameter of Uδ , tends to 0.
The open analytic arc ϕ (γδ ) is the homeomorphic image of the interval (0, 1) with length
µ
L(δ ) = |ϕ ± (z)||dz|.
γδ
By the Cauchy–Schwarz inequality,

µ
2
L (δ ) ≤ πδ |ϕ ± (z)|2 |dz|,
γδ
so that, for r < 1,

µ r µ µ
L 2(δ )
dδ ≤π |ϕ± (z)|2 dx dy
0 δ D ∩B(ζ ,r) (12.1)
= π area (ϕ (D ∩ B(ζ , r)) < ∞
since ² is bounded.
Therefore there is a decreasing sequence δ n → 0 such that L(δ n) → 0. When L(δn ) < ∞,
the curve ϕ(γδn ) has endpoints αn , βn , and both of these endpoints must lie on ∂ ², because ϕ
is proper (see Exercise 3.7). Furthermore,
|αn − βn| ≤ L(δn) → 0. (12.2)
ζ αn βn V
Un
γδn ϕ(γ δn) ϕ(0)
0
Figure 12.7 Crosscuts γδn and ϕ( γδn ).

Let γ be a homeomorphism of ∂ D onto ∂ ². Write αn = γ (ζn ) and βn = γ ( ψn ). Because γ

is uniformly continuous, given ε > 0 there is a δ > 0 so that, if |ζ n − ψn | < δ, then for ζ in the
smaller arc of ∂ D between ζn and ψn, we have | γ (ζ ) − γ (ζ n)| < ε. But γ −1 is also uniformly
continuous so there exists η > 0 such that if |αn − βn| < η then |ζ n − ψn | < δ. Thus if σn is
the closed subarc of ∂ ² of smallest diameter with endpoints αn and βn, then, by (12.2),
diam(σn ) → 0.
By the Jordan curve theorem, the curve σn ∪ ϕ (γδn ) divides the plane into two (connected,
open) regions, and one of these regions, say Un , is bounded. The unbounded component V of
the complement of ∂ ² is connected so if z ∈ Un ∩ V then there is a polygonal arc γz from z
to ∞ contained in V . See Exercise 4.2(a). Because ϕ( D) ∩ V = ∅ and σn ⊂ ∂ ², γz does not
meet ∂ Un, contradicting the boundedness of Un. Thus Un ∩ V = ∅, and hence Un ∩ V = ∅ .
By the Jordan curve theorem applied to ∂ ², Un ⊂ ². Since
( )
diam (∂ Un ) = diam σn ∪ ϕ(γ δn ) → 0,
we conclude that
diam (Un ) → 0. (12.3)
Set Dn = D ∩ { z : |z − ζ | < δn }. Then ϕ(Dn ) and ϕ(D \ Dn ) are connected sets which
do not intersect ϕ (γδn ). Since ϕ maps onto ², either ϕ(Dn ) = Un or ϕ(D \ Dn) = Un. But
diam ( ϕ(D \ Dn )) ≥ diam (ϕ(B(0, 1/2)) > 0 and diam (Un) → 0 by (12.3) so that ϕ(Dn ) = Un
¶
for n sufficiently large. Since δ n is decreasing, ϕ (Dn+1 ) ⊂ ϕ (Dn) and so ϕ(Dn ) consists of
a single point. Thus ϕ has a continuous extension to D ∪ {ζ }.
Let ϕ also denote the extension ϕ : D → ². If zn ∈ D converges to ζ ∈ ∂ D then we can
find z±n ∈ D → ζ so that ϕ(zn ) − ϕ(z±n ) → 0. By the continuity of ϕ at ζ , we must have
ϕ (zn) → ϕ (ζ ), and we conclude that ϕ is continuous on D.
Because ϕ(D) = ², ϕ maps D onto ². To show that ϕ is one-to-one, suppose ϕ (ζ1 ) = ϕ( ζ2 )
but ζ1 ² = ζ 2. Because ϕ is proper, ϕ(∂ D) ⊂ ∂ ² and so we can assume ζj ∈ ∂ D, j = 1, 2. The
Jordan curve
{ϕ (rζ1 ) : 0 ≤ r ≤ 1} ∪ {ϕ(r ζ2 ) : 0 ≤ r ≤ 1}
bounds a bounded region W. See Figure 12.8.
Arguing exactly as above, replacing Un , Dn and D \ Dn with W and the two components of
· ¸
D \ {rζ1 : 0 ≤ r ≤ 1} ∪ {rζ2 : 0 ≤ r ≤ 1} ,
ζ1 ϕ(ζ1)
ϕ(0) W
0
ζ2
Figure 12.8 Proof that ϕ is one-to-one.

we conclude that W ⊂ ² and ϕ−1 (W) must be one of these two components. Because
−1
ϕ (∂ D) ⊂ ∂ ² and ϕ is proper on ϕ (W), we conclude that
ϕ (∂ D ∩ ∂ ϕ−1 (W)) ⊂ ∂ ² ∩ ∂ W = {ϕ(ζ1)}.

Thus ϕ is constant on an arc of ∂ D. It follows that ϕ − ϕ(ζ 1) ≡ 0, by the Schwarz reflec-
tion principle and the uniqueness theorem. This contradiction shows that ϕ must be one-
to-one.
The idea of applying the Cauchy–Schwarz inequality to prove (12.1) is known as the length-
area principle. It led to the development of extremal length, an important tool in geometric
function theory.
Corollary 12.15 If h : ∂ D → C is a homeomorphism then h extends to be a homeomorphism

of C onto C.
Proof Suppose f and g are conformal maps of the interior and exterior of the disk onto the
bounded and unbounded regions in C \ h(∂ D), with g(∞ ) = ∞. For | ζ | = 1, set
⎧
⎨ f (rf −1 (h(ζ ))), for r ≤1
F(r ζ ) =
⎩
g(rg−1(h(ζ ))), for r ≥ 1.
Corollary 12.15 fails in higher dimensions. Alexander’s horned sphere is a homeomorphic

image of the ball in R3 . The boundary is the homeomorphic image of the sphere, but the
unbounded component of the complement is not “simply-connected” in the sense that there
are curves in the complement that cannot be continuously deformed to a point. In particular,
the homeomorphism cannot be extended to a homeomorphism of R3.
The Jordan curve theorem and Carathéodory’s theorem yield the following useful result.
Corollary 12.16 If J ⊂ C is a closed Jordan curve then J can be oriented so that n(J, z) = 1
for z in the bounded component of the complement of J and n(J, z) = 0 for z in the unbounded
component of the complement of J.
Proof Let ϕ be a conformal map of D onto the bounded component U of the complement
of J. By Carathéodory’s theorem, ϕr (z) = ϕ(rz) converges uniformly on ∂ D to ϕ as r ↑ 1.
Fix w 1 ∈ U, where U is the bounded component of C \ J. Set γ (t) = ϕ(eit ) and γr (t) =
ϕ (re ). As in Lemma 4.31, cover J = γ by finitely many disks of the form Bj = {z : |z −
it
γ (tj )| < dist(γ , w 1)} such that γ ([tj −1, t j ]) ⊂ B j for a finite partition { tj } of [0, 2π ]. Then, for
r sufficiently close to 1, γr ([t j−1 , tj ]) ∪ γ ([tj−1 , tj ]) ⊂ Bj . Then, by adding and subtracting line
∑
segments between the endpoints of these arcs, we can write γ − γr = βj , where each βj is
a closed curve contained in Bj . Thus for r near 1
µ µ ¹µ
dw dw dw
w − w1
− w − w1
= w − w1
= 0,
γ γr βj
j
12.5 Exercises 205
³
where the integral γ dw/(w − w1) is defined as in Theorem 4.32. By the argument principle,
the number of zeros of ϕ(z) − w1 in rD = {ζ : |ζ | < r} is equal to
µ ± (z) dz
ϕ
= n( γr , w1 ) = n(γ , w1 ),
∂ rD ϕ (z) − w1 2 π i
where ∂ D has the usual counter-clockwise orientation. Thus ϕ (eit ), t ∈ [0, 2π ], is a parame-
terization of J so that n(J, w) = 1 for each w ∈ U. By property (b) in Section 5.2, n(J, w) = 0
for each w in the unbounded component of the complement of J.
See Exercise 12.10 for the finitely-connected version of Corollary 12.16.
12.5 Exercises
12.1 Prove items (a)–(c) in Section 12.2. Hint: If Ua is a component of an open set U then
both Ua and U \ Ua are open.
12.2 Suppose G ⊂ C∗ is a connected open set and suppose J is a (closed) Jordan curve,
then let H1 and H2 be the components of the complement of J. If ∂ G ⊂ H1 then either
G ⊂ H1 or H2 ⊂ G.
12.3 (a) Prove that if E is connected, if Ea is connected for each index a, and if E a intersects
E for all indices a, then the union of E and all E a is connected;
(b) Prove that if E is connected then its closure E is connected;
(c) Prove that if E is compact and connected, and if Ua is a component of C∗ \ E, then
Ua is simply-connected.
(d) Suppose U is simply-connected and suppose S is connected and S ∩ (C \ U) ² = ∅.
Prove that each component of U ∩ (C \ S) is simply-connected.
12.4 Prove that if U is a simply-connected region then ∂ U ⊂ C∗ is connected in C∗ . Hint:
Use the conformal map of D onto U.
12.5 Suppose ϕ is a conformal map of a region ²1 onto a region ²2 , where both ²1 and
²2 are bounded by finitely many disjoint Jordan curves. Prove that ϕ extends to be a
homeomorphism of ²1 onto ²2 .
12.6 A compact set K ⊂ C is locally connectedif for each ε > 0 there exists δ > 0 so
that if z, w ∈ K with | z − w| < δ, then there exists a compact connected set L ⊂ K
with z, w ∈ L and diam(L) < ε . Suppose f is a conformal map of D onto a bounded
simply-connected region ² which extends to be continuous on D. Prove that ∂ ² is
locally connected.
12.7 Let ² be a bounded simply-connected region. Prove that ∂ ² is locally connected if
and only if a conformal map ϕ : D → ² extends continuously to D. See Exercise 12.6.
Hint: Apply the proof of Carathéodory’s theorem to the map of D onto ².
12.8 (a) Let ² be a simply-connected region and let σ ⊂ ² be a crosscut, i.e., an (open)
Jordan arc in ² having endpoints in ∂ ². Prove that ² \ σ has two components,
²1 and ²2, each simply-connected.
(b) Let ² be simply-connected, let ψ : ² → D be a conformal map onto D, and fix
z ∈ ² and ζ ∈ ∂ ². Assume ζ and z can be separated by a sequence of crosscuts
σn ⊂ ² such that length(σn ) → 0. Let Un be the component of ² \ σn such that
z ∈/ Un. Prove that ψ (Un) → α ∈ ∂ D.
12.9 Find and read a construction of Alexander’s horned sphere and the proof of the claims
in the paragraph after the proof of Corollary 12.15. Show also that the notion of
simply-connected in R3 given there is not the same as having a connected complement.
12.10 Suppose ² is a region bounded by finitely many pairwise disjoint closed Jordan curves
{Jk }m1 ⊂ C. If ² is unbounded, then we can orient ∂ ² so that n(∂ ², z) = 0 for all
z ∈ ² and n(∂ ², z) = 1 for all z ∈/ ². If ² is bounded, then we can orient ∂ ² so that
n(∂ ², z) = 1 for all z ∈ ² and n(∂ ², z) = 0 for all z ∈
/ ².
12.11 The main spiral from Figure 12.2 terminates at a point z0 . If γ is an arc in the comple-
ment terminating at z0, then either arg(γ (t) − z0 ) → +∞ or arg(γ (t) − z0 ) → −∞.
Construct a curve σ which has the property that if γ is an arc in the complement ter-
minating at z0 then lim sup arg(γ (t) − z0) = +∞ and lim inf arg(γ (t) − z0 ) = −∞. A
point with this property is called a MacMillan twist point.
13 The Dirichlet Problem
In this chapter we will treat the Dirichlet problem on arbitrary regions. Let C(∂ ±) denote the
set of continuous functions on the boundary in C∗ of a region ±. The Dirichlet problemon
± for a function f ∈ C(∂ ±) is to find a harmonic function u on ± which is continuous on
∗
± ⊂ C and equal to f on ∂ ±. If u exists then it is unique by the maximum principle, but
it is not always possible to solve the Dirichlet problem. If f = 0 on ∂ D and f (0) = 1, then
f ∈ C(∂ ±) where ± = D \ {0}. But, by Lindelöf’s maximum principle, Theorem 7.15, if u is
harmonic and bounded on ± with u = 0 on ∂ D then u(z) = 0 for all z ∈ ±. Thus u extends to
be continuous at 0, but u(0) ± = f (0).
There are several approaches to this material. See for example Garnett and Marshall [10].
We will use the elegant Perron process because it is the fundamental method underlying the
proof of the uniformization theorem given in Chapter 15.
13.1 Perron Process
Recall that a subharmonic function v on ± is continuous as a map of ± into [−∞ , +∞) and
satisfies the mean-value inequality on all sufficiently small circles. See Definition 7.2. There
are several key properties of subharmonic functions:
(a) A subharmonic function v on a region ± satisfies the maximum principle: if there exists
z0 ∈ ± such that v(z0 ) = supz ∈± v(z) then v is constant on ±.
(b) If v1, v2 are subharmonic then v1 + v2 and max(v1 , v2 ) are subharmonic.
(c) If v is subharmonic on a region ± and v > −∞ on ∂ D, where D is a disk with D ⊂ ±,
then the function vD which equals v on ± \ D and equals the Poisson integral of v on D is
also subharmonic. In other words, if D = {z : | z − c| < r}, we can replace v on D by
± 2π 1 − |(z − c)/r |2 dt
vD (z) = |e − (z − c)/ r|
it 2
v(c + reit ) ,
2π
0
and still be subharmonic.
See Exercise 13.1. For example, if f is analytic then | f |p , for p > 0, log | f | and log(1 +| f |2 )
are subharmonic. See Exercise 13.4.
Definition 13.1 A family F of subharmonic functions on a region ± is called a Perron

family if it satisfies:
(i) if v1, v2 ∈ F then max(v1 , v2 ) ∈ F ,
(ii) if v ∈ F and D is a disk with D ⊂ ±, and if v > −∞ on ∂ D, then vD ∈ F and
(iii) for each z ∈ ± there exists v ∈ F such that v(z) > −∞ .
See property (c) of subharmonic functions for the definition of vD .

Definition 13.2 If F is a Perron family on a region ± then we define
uF (z) ≡ sup v(z).

v ∈F
Theorem 13.3 If F is a Perron family on a region ± then uF is harmonic on ± or uF (z) =

+∞ for all z ∈ ±.
The “bootstrap technique” for climbing a cliff is to put your foot above your head, then pull
yourself up by your bootstraps or shoelaces. Repeat. The proof of Theorem 13.3 is a classic
bootstrap proof.
Proof Fix z0 ∈ ±. Find v1, v2 , · · · ∈ F such that limj vj (z0 ) = uF (z0). Set
v²j = max(v1 , v2, . . . , vj ).
By (i) and induction, v²j ∈ F , v²j ≤ v²j+1 and lim v²j (z0 ) = uF (z0). Suppose D is a disk with
z0 ∈ D ⊂ D ⊂ ±. By (iii), the continuity of subharmonic functions and (i), we may suppose
that v²j is continuous and v²j > −∞ on ∂ D. Let v²²j equal v²j on ± \ D and equal the Poisson
integral of v²j on D. Then, by (ii), v²²j ∈ F . Moreover, v²j ≤ v²²j by the maximum principle. The
Poisson integral on D of the non-negative function v²j+1 − v²j is non-negative so that v²²j ≤ v²²j +1.
Set V = limj v²²j . By Harnack’s principle, Theorem 7.20, either V ≡ +∞ in D or V is harmonic
in D. Note also that V (z0) = uF (z0) because of the choice of vj and the maximality of uF (z0).
Now take z1 ∈ D, wj ∈ F , with limj wj (z1 ) = uF (z1). Set
w²j = max(v²²j , w1 , w2 , . . . , w j )
and let w²²j ∈ F equal w²j on ± \ D and equal the Poisson integral of w²j on D. As before,
w²j ≤ w²²j ≤ w²²j+1 . Set W = limj w ²²j . As before, either W is harmonic in D or W ≡ +∞ in D,
and W(z1 ) = uF (z1 ). Because v²²j ≤ w²j ≤ w²²j , we must have that V ≤ W. But also
uF (z0 ) = V (z0) ≤ W(z0 ) ≤ uF (z0)
so that V (z0) = W(z0 ). If uF (z0) < ∞ then V −W is harmonic on D and achieves its maximum
value 0 on D at z0 . By the maximum principle, it must equal 0 on D. Because z1 was arbitrary
in D, there are two possibilities, either uF ≡ +∞ on D or uF = V on D and hence is
harmonic on D. Since z0 is an arbitrary point in ±, {z : uF (z) = +∞} is then both closed and
open in ±. Since ± is connected, either uF ≡ +∞ on ± or uF is harmonic on ±.
Note that the proof of Theorem 13.3 only used local properties of the family F , the mean-
value property and the maximum principle on small disks contained in ±. We will use this
observation in Chapter 15.
If ± ⊂ C∗ is a region, and if f is a real-valued function defined on ∂ ± (a compact subset of
C∗ ) with | f | ≤ M < ∞ on ∂ ±, set
Ff = {v subharmonic on ± : lim sup v(z) ≤ f (ζ ), for all ζ ∈ ∂ ±}. (13.1)
z∈ ±→ζ
Then Definition 1.13(iii) holds since the constant function −M ∈ Ff . So Ff is a Perron family.
Moreover, each v ∈ Ff is bounded by M by the maximum principle. By Theorem 13.3,
uf (z) ≡ sup v(z) (13.2)
v∈Ff
is harmonic in ±. The function uf is called the Perron solution to the Dirichlet problem on
± for the function f . It is a natural candidate for a harmonic function in ± which equals f on
∂ ± . In the next section we will explore to what extent uf has boundary values equal to f .
13.2 Local Barriers
In this section we develop a criterion for the Perron solution to the Dirichlet problem on ± for
a function f to have boundary value f (ζ ) at ζ ∈ ∂ ±. Our account is based on Ransford [21].
Throughout this section, if ζ ∈ C then B(ζ , ε) is the open disk centered at ζ with radius
ε > 0. If ζ = ∞ ∈ C∗ then B(ζ , ε ) is the neighborhood of ζ given by { z : |z| > 1/ε }∪{∞} ⊂
C∗ . We will refer to B(∞, ε) as a disk centered at ∞, because B(∞, ε ) corresponds to a small
disk on the Riemann sphere centered at the north pole.
Properties (i) and (ii) in Definition 13.1 make it somewhat easier to build subharmonic
functions than harmonic functions.
Definition 13.4 If ± ⊂ C∗ is a region and if ζ 0 ∈ ∂ ± then b is called a local barrier atζ0

for the region± provided
(i) b is defined and subharmonic on ± ∩ D for some open disk D with ζ0 ∈ D,
(ii) b(z) < 0 for z ∈ ± ∩ D and
(iii) limz ∈±→ζ0 b(z) = 0.
The function b approaches its supremum on ± ∩ D as z tends to ζ0 and it can also approach
this value elsewhere on ∂ (± ∩ D). Note that if b is a local barrier for the region ± at ζ0 , defined
on ± ∩ D, then it is also a local barrier on ± ∩ D1 for any smaller disk D1 with ζ0 ∈ D1 ⊂ D.
Definition 13.5 If there exists a local barrier at ζ 0 ∈ ∂ ± then ζ0 is called a regular point
of ∂ ±. Otherwise ζ0 ∈ ∂ ± is called an irregular point of∂ ±. If every ζ ∈ ∂ ± is regular,
then ± is called a regular region.
If ζ 0 is an isolated point of ∂ ±, then there cannot be a local barrier at ζ0 for ± by the

Lindelöf maximum principle. The next result says that if ζ 0 belongs to a non-trivial component
of the boundary, then there is a local barrier at ζ0 .
Theorem 13.6 Suppose ± ⊂ C∗ is a region and suppose ζ 0 ∈ ∂ ±. If C ⊂ ∂ ± is closed

in C∗ and connected, with ζ 0 ∈ C but C ± = {ζ 0}, then ζ0 is regular. Every simply-connected
region is regular.
Proof We claim that each component ±1 of C∗ \ C is simply-connected. To see this, we

may assume ∞ ∈ C \ {ζ0 }. Let γ be any closed polygonal curve contained in ±1 . Then
n(γ , ζ ) = 0 for all ζ ∈ C \ {∞} because C is connected and unbounded, and hence contained
in the unbounded component of the complement of γ . If β ∈/ ±1 and n(γ , β ) ± = 0, let L be a
line segment from β to some ζ ∈ C. Then L must intersect ∂ ±1 before it intersects γ because
γ ⊂ ±1. Because ∂ ±1 ⊂ C, this intersection gives a point ζ1 ∈ C with n(γ , ζ 1) = n(γ , β ) ± =
0. This contradiction proves that every closed polygonal curve in ±1 is homologous to 0, and
so, by Theorem 5.7, ±1 is simply-connected.
If ζ2 ∈ C \ {∞} with ζ 2 ± = ζ 0, then
² ³
z − ζ0
f (z) = log
z − ζ2
can be defined to be analytic on each component ±1 of C∗ \ C by Corollary 5.8, and there-
fore on ±. Then b(z) = Re(1/f (z)) is a local barrier at ζ 0, defined on ± ∩ D, where
D = { z : | (z − ζ 0)/(z − ζ2 )| < 1/2}. The same function works if we only assume ± is
simply-connected. See also Exercise 12.4.
The local barrier constructed in the proof of Theorem 13.6 can approach 0 along sequences
tending to boundary points other than ζ0 . For example, if C1 = {rei/ (1−r) : 0 ≤ r < 1} is a
spiral from the origin to ∂ D then D \ C 1 is simply-connected. The local barrier constructed in
the proof of Theorem 13.6 with ζ 2 = 0 and ζ0 = 1 will tend to zero at each point of the unit
circle in |(z − 1)/z| < 1/2 because the imaginary part of the logarithm is unbounded.
Theorem 13.7 Suppose ζ 0 ∈ ∂ ± is regular. If f is a real-valued function defined on ∂ ±,

with | f | ≤ M < ∞ on ∂ ±, and if f is continuous at ζ 0 , then
lim uf (z) = f (ζ0 ),
z ∈±→ζ0
where uf is the Perron solution to the Dirichlet problem for the function f .
See Exercise 13.2 for a partial converse to Theorem 13.7.
Corollary 13.8 If ± is a regular region and if f ∈ C(∂ ±) then uf is harmonic in ± and

extends to be continuous on ± with uf = f on ∂ ±.
The proof of Theorem 13.7 requires two lemmas. If v is subharmonic and not harmonic then
−v is not subharmonic. So it does not easily follow that the map f → uf is linear. However,
the next lemma is not difficult to prove.
Lemma 13.9 uf ≤ −u−f on ±.

Proof If v ∈ Ff , as defined in (13.1), and if w ∈ F−f , then
lim sup (v(z) + w(z)) ≤ f (ζ ) + ( −f (ζ )) = 0.
z ∈±→ζ ∈ ∂ ±
By the maximum principle, v + w ≤ 0 on ±. Taking the supremum over all v ∈ Ff and

w ∈ F−f shows that uf + u−f ≤ 0.
Lemma 13.10 (Bouligand) Suppose ζ 0 is a regular point of ∂ ± and suppose b is a local

barrier at ζ 0 defined on ± ∩ B(ζ0 , ε). Given 0 < δ < ε , there exists bδ , which is subharmonic
on all of ± satisfying
K L
L
Ω
ζ0
L B
K L
Figure 13.1 Proof of Bouligand’s lemma.
(i) bδ < 0 on ±,
(ii) bδ ≡ −1 on ± \ B(ζ0 , ε),
(iii) lim inf z∈±→ζ0 bδ (z) ≥ −δ .
Condition (iii) is weaker than the corresponding condition (iii) in Definition 13.4, but the
function bδ is defined on all of ±, not just a neighborhood of ζ0 .
Proof Set B = B(ζ0, r) for some r < ε. Then

∞
´
∂B ∩±= Ik ,
k=1
where each Ik is an open arc. Choose n < ∞ and compact subarcs Jk ⊂ Ik , k = 1, . . . , n, so

that if K = ∪nk=1Jk and L = (∂ B ∩ ±) \ K, then the total length of L is less than π δ r. See
Figure 13.1.
Let ω = PI B( χ L ) be the Poisson integral of the characteristic function of L on B. Then ω is
harmonic on B and 0 < ω < 1 on B. Because L is open in ∂ B, ω(z) → 1 as z ∈ B → η ∈ L,
by the remark after the proof of Schwarz’s Theorem 7.5. Because b is continuous and b < 0
on ± ∩ B(ζ0 , ε ), and because K ⊂ ± ∩ B( ζ0 , ε) is compact, there exists −m < 0 so that
b ≤ −m on K. So if η ∈ K ∪ L then
² ³
b(z)
lim sup − ω (z) ≤ −1. (13.3)
z ∈B→η m
Set
⎧ ² ² ³ ³
⎨ max 2 b(z) − ω (z) , −1 , on ± ∩ B
bδ (z) = m
⎩ −1, on ± \ B.
By (13.3), bδ ≡ −1 in a neighborhood of K ∪ L in ±. Because 2( mb − ω) is subharmonic

on B ∩ ±, (13.3) and property (b) in Section 13.1 show that bδ is subharmonic in ±. More-
µ
over, bδ < 0 on ±. Finally, note that ω( ζ0 ) = L |dz| /(2π r) = |L |/(2π r) < δ/2. Thus
lim infz ∈±→ζ0 bδ (z) = −2ω(ζ 0) ≥ −2δ/2 = −δ .
Proof of Theorem13.7 Choose ε > 0 so that | f (ζ ) − f (ζ0 )| < δ for ζ ∈ B ∩ ∂ ±, where

B = B(ζ 0, ε). Let bδ be the function produced in Bouligand’s Lemma 13.10. The idea is to
build a subharmonic function which is approximately f (ζ0 ) near ζ 0, but smaller than f away
from ζ0 , using bδ , which is small near ζ0 and large negative away from ζ0 . Set
= f (ζ0) − δ + (M + f (ζ0))b (z).

v(z) δ
Then v is subharmonic on ± and for ζ ∈ ∂ ± ∩ B
lim sup v(z) ≤ f (ζ0 ) − δ ≤ f (ζ ),

z ∈±→ζ
because M + f (ζ 0) ≥ 0 and b δ < ∈ ∂ ± \ B then

0. Also if ζ
lim sup v(z) = f (ζ0 ) − δ − (M + f (ζ 0)) ≤ f (ζ ),
z∈±→ζ
because bδ = −1 on ± \ B. Thus v ∈ Ff by (13.1). By Lemma 13.10(iii),

lim inf uf (z) ≥ lim inf v(z) ≥ f (ζ 0) − δ + (M + f (ζ 0 ))(−δ ).
z∈ →± ζ0 z∈ → ± ζ0
Since δ > 0 is arbitrary, lim infz ∈±→ζ0 uf (z) ≥ f (ζ 0 ). Replacing f with −f , we also have that
lim infz→ζ0 u−f (z) ≥ −f (ζ 0). By Lemma 13.9,
lim sup uf (z) ≤ lim sup −u−f (z) = − lim inf u−f (z) ≤ f (ζ0 ).
z ∈±→ζ0 z ∈±→ζ0 z ∈±→ζ0
This proves Theorem 13.7.
13.3 Riemann Mapping Theorem Again
One application of Theorems 13.6 and 13.7 is a proof of the Riemann mapping theorem that
avoids the use of normal families as well as the geodesic algorithm.
Theorem 13.11 (Riemann mapping theorem)If ± ⊂ C is a simply-connected region not

equal to all of C, then there exists a conformal map ϕ of ± onto D.
As mentioned before, we adopt the common phrase “conformal map” to mean a one-to-one
analytic function. See Proposition 6.5 for the uniqueness portion of the Riemann mapping
theorem.
Proof Repeating the first paragraph of the proof of Theorem 8.11, using LFTs and a square
root, we may suppose ± is bounded and 0 ∈ ±. Then f (ζ ) = log | ζ | ∈ C(∂ ±). Let uf be
the Perron solution to the Dirichlet problem for the boundary function f . By Theorems 13.6
and 13.7, uf is harmonic on ± and extends to be continuous on ± and equal to f on ∂ ±. By
Theorem 7.10, there is an analytic function g on ± such that Reg = uf . Set
= ze−g(z) .
ϕ (z)
Then ϕ is analytic in ±, ϕ(0) = 0 and ϕ(z) ± = 0 if z ± = 0. Moreover, |ϕ(z) | = elog |z|− u (z) → 1
f
as z ∈ ± → ∂ ±. Thus ϕ maps ± into D by the maximum principle.

Fix ε > 0 and let K = ϕ−1 (|w| ≤ 1 − ε). Because |ϕ| → 1 as z → ∂ ±, the set K is
ε ε
a compact subset of ±. As in the proof of Runge’s Theorem 4.23, we can construct a closed
curve γ ⊂ ± which winds once around each point of Kε . The winding number n(ϕ(γ ), w) is
constant in each component of C \ ϕ(γ ) and |ϕ| > 1 − ε on γ so that if |w| < 1 − ε then
13.3 Riemann Mapping Theorem Again 213
n(ϕ( γ ), w) = n(ϕ (γ ), 0). By the argument principle, the number of zeros of ϕ − w must equal
the number of zeros of ϕ . But by construction ϕ = 0 only at one point, namely 0. Letting
ε → 0, we conclude that each value in | w| < 1 is attained exactly once. Thus ϕ maps ±
one-to-one and onto D.
The function uf (z) − log |z| = − log | ϕ(z)| constructed in the proof of Theorem 13.11 is
called Green’s function. See Exercise 13.5.
Another application of Theorems 13.6 and 13.7 is the doubly-connected version of the
Riemann mapping theorem. A region ± is doubly-connected if C∗ \ ± = E1 ∪ E2 , where E1
and E 2 are disjoint, connected and closed in C∗ .
Theorem 13.12 If ± is doubly-connected then there is a conformal map f of ± onto an

annulus A = {z : r1 < |z| < r2 }, for some 0 ≤ r1 < r2 ≤ ∞ .
See Exercise 8.14 for a corresponding uniqueness statement.
Proof Write C∗ \ ± = E1 ∪ E2 , where E1 and E 2 are disjoint, connected and closed in C∗. If
E 1 and E2 each consist of one point, then an LFT will map ± onto A with r1 = 0 and r2 = ∞.
If E1 contains more than one point and E2 consists of a single point, then, by the Riemann
mapping theorem, there exists a conformal map ϕ of the complement of E1 in C∗ onto the
unit disk with ϕ(E 2) = 0. Thus ϕ(±) = A with r1 = 0 and r2 = 1.
If neither E1 nor E2 consist of a single point, then, by the Riemann mapping theorem, we
can find a conformal map g1 of C∗ \ E 1 onto D. We can then find a conformal map g2 of
C∗ \ g1(E 2) onto D. The image g2( ∂ D) is then an analytic curve in D. So we may suppose that
∂ ± = ∂ D ∪ E1 , where E 1 ⊂ D is an analytic Jordan curve.
By Theorems 13.6 and 13.7, we can find a harmonic function ω on ± which is continuous
on ± and equal to 1 on ∂ D and equal to 0 on E1 . By the Schwarz reflection principle applied
to ω − 1 and ω after a conformal map, ω extends to be harmonic in a neighborhood of ±.
Let L be a line segment in ± connecting E1 to ∂ D. Then ± \ L is simply connected so that
ω = Reg for some g analytic on ± \ L. Moreover, g ² (z) = ωx − iωy is analytic on ±. By
Cauchy’s theorem
± ±
g² (z)dz = g² (z)dz ≡ ic.
γ ∂ D
for all curves γ ⊂ ± with γ ∼ ∂ D. We claim c > 0. If so, then Im2πg/c increases by 2π
as z traces γ and thus f (z) = e2πg/ c is continuous in ± and analytic on ± \ L. By Morera’s
theorem, f is analytic on ±. Moreover f ² /f is analytic on ± and hence so is f . Set A = {z :
1 < |z| < e2π/c }. Then f (∂ ±) ⊂ ∂ A because | f | = e2πω/c . If ∂ ± is given the usual positive
orientation with respect to ±, then the cycle f (∂ ±) winds exactly once around each z ∈ A and
does not wind around any point C \ A. By the argument principle, f is a conformal map of ±
onto A.
To see that c > 0, note that
± ±
ic = i (ωx dy − ωy dx) = i ωr dθ .
∂ D ∂ D
µ
But 0 < ω < 1 on ± and ω = 1 on ∂ D so ωr ≥ 0 on ∂ D. If ∂ D ωr dθ = 0, then ωx −
iωy ≡ 0 on ∂ D because ωθ also = 0. By the uniqueness theorem, ωx − i ωy ≡ 0 in ± which
is impossible. This proves c > 0 and Theorem 13.12 follows.
It is known that if ± ⊂ C is a region whose complement consists of finitely many disjoint

compact sets then there is a conformal map of ± onto a region whose boundary consists of
finitely many circles and points. It is an interesting open problem, known as the Koebe con-
jecture, to prove that every region in the plane is conformally equivalent to a region bounded
by circles and points.
13.4 Exercises
13.1 Prove items (a)–(c) at the beginning of Section 13.1.

13.2 Prove the following partial converse to Theorem 13.6. Suppose ± ⊂ C∗ is a region
with ζ0 ∈ ∂ ±, ζ 0 ± = ∞ . Set f (ζ ) = max{−|ζ − ζ 0 |, −1} for ζ ∈ ∂ ±. Then f ∈ C(∂ ±).
Let uf be the Perron solution to the Dirichlet problem on ± with boundary function f .
If limz ∈±→ζ0 uf (z) = 0 and if uf ± ≡ 0 then uf is a local barrier for ± at ζ 0.
13.3 Prove the following alternative way to solve the Dirichlet problem on an annulus
A = {z : 0 < r < | z| < R < ∞}. If f is bounded on ∂ A, set g(z) = f (ez ) for
z on the boundary of a vertical strip. Let u1 be the solution to the Dirichlet problem
in the simply-connected strip with boundary values g. This can be accomplished by
using a conformal map and the Poisson integral formula. Prove u(z) = u1 (log z) is well
defined and harmonic on A. Show that if f is continuous at ζ ∈ ∂ A then u extends to be
continuous and equal to f at ζ .
13.4 Prove that if f is analytic on ± with f (±) ⊂ ±1 and if v is subharmonic on ±1 then v◦f is
subharmonic on ±. Note that subharmonic functions are not necessarily differentiable
so we cannot use the Laplacian to prove this. Hint: First assume f is one-to-one and
solve a Dirichlet problem. Then try it for f (z) = zm on |z| < r.
13.5 Suppose ± ⊂ C is a bounded region and suppose z0 ∈ ±. Then f (ζ ) = log |ζ − z0| ∈
C(∂ ±). Let uf be the Perron solution to the Dirichlet problem on ± with boundary
function f . If uf ± ≡ +∞ then g(z, z0) = uf (z) − log | z − z0 | is called Green’s function
on ± with pole at z0 . The point z0 is not really a pole in the usual sense, but by (b)
below limz →z 0 g(z, z0 ) = +∞, so the graph of g(z, z0 ) is tangent to a vertical line in
R3. If uf ≡ +∞ we say that Green’s function does not exist. If Green’s function exists,
prove the following:
(a) g(z, z0 ) is harmonic in ± \ {z0};
(b) g(z, z0 ) + log | z − z0 | is bounded in a neighborhood of z0 ;
(c) g(z, z0 ) > 0 for z ∈ ±;
(d) ζ 0 ∈ ∂ ± is a regular point if and only if lim z→ζ0 g(z, z0 ) = 0;
13.4 Exercises 215
(e) if ± is regular, and if h satisfies (a), (b), mutatis mutandis, and lim z→∂ ± h(z) = 0,
then h(z) = g(z, z0);
(f) gx − igy is a meromorphic function on ± with a simple pole with residue = 1 at z0
and no other poles;
(g) if we remove the hypothesis that ± is bounded then f may no longer be bounded on
∂ ±. As in the proof of the Riemann mapping theorem, if ∂ ± contains a connected
set which is not a single point then we can conformally map ± onto a bounded
region ±1 using LFTs and a square root. We can then use Green’s function on ±1
to construct Green’s function on ±.
Green’s function is probably the most important function in the study of harmonic
functions on a region, if it exists.
13.6 (a) Suppose ± is a region such that ∂ ± consists of finitely many analytic curves. Let
g(z) = g(z, z0) denote Green’s function, as in Exercise 13.5. Prove that if f is
analytic on ± and continuous on the closure of ± then
±
1
f (z0 ) = − f (ζ )(gx − igy )(ζ )dζ .
2π i ∂±
(b) Prove that if f ∈ C(∂ ±) and if ± is bounded by finitely many analytic curves then
define ±
u(z) = f (ζ )
∂ g(ζ , z0) |dζ | ,
∂± ∂η 2π
where η (ζ ) is the unit inner normal to ∂ ± at ζ . Then u is harmonic in ±, continuous
in ± and u = f on ∂ ±. Hints: Let uf be the Perron solution to the Dirichlet problem
with boundary function f . Extend g to be harmonic in a neighborhood of ∂ ± and
apply Exercise 7.5(a) to g and uf on the region {z : g(z, z0) − ε > 0 and |z − z0 | > ε}
then let ε → 0.
(c) Prove that if ± is a region bounded by continuously differentiable Jordan curves
such that |∇ g| is continuous on ± then the conclusion in (a) still holds.
(d) Show that if ± = D then 21π ∂ g(∂ζη,z0 ) is the Poisson kernel in D for z0. For this reason
the normal derivative of g is sometimes called the Poisson kernel on±.
13.7 Prove that if Green’s function g(z, z0) exists on ± then g(z, z1) exists, for z0 , z1 ∈ ±.
Moreover, g(z0, z1 ) = g(z1 , z0). Hint: See Exercise 7.5(a).
14 Riemann Surfaces
14.1 Analytic Continuation and Monodromy
∑
The root test tells us that each convergent power series ∞ 0 a n(z − b) has an open disk
n
centered at b as its natural region of definition. The series converges in the disk and diverges
outside the closed disk. We also found that the series is analytic on its disk of convergence,
in the sense that, given any other point c in the disk, the power series can be rearranged to
be a series converging in a disk centered at c. However, it is possible that the largest disk of
convergence centered at c also contains points outside the disk centered at b. This allows us
to extend our original series as an analytic function on a larger region. For example, the series
∑∞ n
0 z converges in the unit disk, but the natural region of definition as an analytic function
is C \ { 1}, because it agrees with the function 1 /(1 − z) on the unit disk. Though the process
would be laborious, this function can be “discovered” by continually rearranging power series
on more and more disks. This is the idea behind analytic continuation.
Suppose U1 and U2 are regions such that U1 ∩ U2 is connected. If fj is analytic on Uj ,
j = 1, 2, and if f1 = f2 on an infinite set with a cluster point in U1 ∩ U2, then by the uniqueness
theorem f 1 = f2 on U1 ∩ U2. Thus
±
f1 (z), if z ∈ U1
f3 (z) = (14.1)
f2 (z), if z ∈ U2
is a well-defined analytic function on U1 ∪ U2 . In this way, f2 provides an analytic extension

of f 1 on U1 to f3 on U1 ∪ U2.
This process can be repeated.
Definition 14.1 Suppose fj is analytic on Uj , j = 0, . . . , n, such that fj = f j+1 on Uj ∩ Uj +1,

j = 0, . . . , n − 1. Then fn is called a direct analytic continuationof f0 to Un .
By (14.1) we obtain analytic functions on Uj ∪ Uj+1 , but not necessarily an analytic function
on all of ∪j Uj . For example, we can define log z on a sequence of disks whose union contains
the unit circle, giving analytic continuations around the unit circle, but there cannot be an
analytic function in a neighborhood of the unit circle which equals the definition of log z on
each disk. The difficulty is that the continuation around the circle increases or decreases by
2π i, depending on the direction. The continuation of f0 on U0 to f n on Un depends not only
on the final region Un, but also on how the intermediary regions Uj are chosen. One way to
treat this difficulty is to define analytic continuation along a curve.
Definition 14.2 If γ : [0, 1] → C is a curve and if f0 is analytic in a neighborhood of γ (0),

then an analytic continuation off0 along γ is a finite sequence f1 , . . . , fn of functions where
0 = t0 < t1 < · · · < t n+1 = 1 is a partition of [0, 1] and fj is defined and analytic in a
neighborhood of γ ([tj , t j+1]), j = 0, . . . , n, such that f j = fj+1 in a neighborhood of γ (tj+1 ),
j = 0, . . . , n − 1.
Lemma 4.31 and Theorem 4.32 are sometimes useful for constructing analytic contin-
uations. The function log(z) can be analytically continued along all curves γ in C \ {0}
using
²
1
dz (14.2)
γ z
by subdividing γ into arcs, as in Theorem 4.32. But, as mentioned above, the analytic
continuation does not agree with an analytic function defined on all of C \ { 0}.
Another way that an analytic continuation arises is the following. If u is harmonic in a
region ±, and if D ⊂ ± is a disk, then we can find an analytic function f on D so that
Ref = u. For instance, we can use the Herglotz integral and a linear change of variables to
define f as in Corollary 7.7. Moreover, any two analytic functions which have the same real
part on a region differ by only a purely imaginary constant because the difference is not open.
Fix a point b ∈ ± and an analytic function f0 defined on a disk D0 containing b with Ref = u.
If γ is a curve in ± with γ (0) = b, then by Lemma 4.31 we can cover γ with a finite union
of disks Dj so that γ ([tj , tj +1]) ⊂ Dj . If fj is analytic with Refj = u on Dj , then, by adding an
appropriate imaginary constant to each fj , j ≥ 2, we will create an analytic continuation of
f along γ . Thus f can be continued along all curves in ±. It is not always true, though, that
u = Ref for some analytic function f on all of ±.
A central question for this chapter is as follows. Suppose we can find an analytic continua-
tion of f0 along all curves in a region ±. When can we find a single analytic function on all of
± which agrees with each of these continuations?
Here are some observations:
(i) An analytic continuation along γ : [0, 1] → C determines a well-defined function on

[0, 1], but it does not necessarily yield a well-defined function on (the image) γ ⊂ ±, as
indicated by the logarithm example.
(ii) If f1 , . . . , fn is an analytic continuation of f 0 then fn−1, . . . , f0 is an analytic continuation
of f n.
(iii) If f1 , . . . , fn is an analytic continuation of f0 for the partition 0 = t0 < t1 · · · < tn+1 = 1,
then we can refine the partition by choosing s with t j < s < tj+1 and using the function
fj on γ ([tj , s]) and on γ ([s, tj+1 ]). So if g1, . . . , gm is another analytic continuation along
γ of f0 , then we can choose a common refinement so that the two sequences of analytic
functions are defined on the same partition 0 = u0 < u1 < · · · < uk+1 = 1. But
f1 = f0 = g1 in a neighborhood of γ (u1 ), so by the uniqueness theorem f1 = g1 on a
neighborhood of γ ([u1, u2 ]), and by induction gj = fj on a neighborhood of γ ([uj , uj+1 ]).
In this sense, analytic continuation along a curve is unique.
218 Riemann Surfaces
(iv) Suppose f is an analytic function on a region ±. If f1, . . . , fn is an analytic continuation

of f 0 along a curve γ ⊂ ±, where f0 is the restriction of f to a neighborhood of γ (0), then
by the uniqueness theorem and induction fj (γ (t)) = f (γ (t)) for tj ≤ t ≤ tj+1 , 0 ≤ j ≤ n.
(v) Suppose that f 1, f2, . . . , fn is an analytic continuation of f0 along γ with partition
0 = t0 < · · · < t n+1 = 1. We can choose ε > 0 so that if σ is another curve such
that | σ (t) − γ (t)| < ε, for all 0 ≤ t ≤ 1, then f 1, f2, . . . , fn is an analytic continuation of
f0 along σ . Indeed, if ε > 0 is sufficiently small then fj is defined and analytic in a
neighborhood of σ ([t j , tj+1 ]) and fj = fj +1 in a neighborhood of σ (tj +1), j = 0, . . . , n.
To extend (v) to curves that are not uniformly close we introduce a concept similar to, but
different than, homology. Suppose γ0 and γ1 are curves in a region ± that begin at b and end
at c. We say γ0 is homotopic in± to γ1 if there exists a collection of curves γs : [0, 1] → ±,
0 < s < 1, so that γs (t), as a function of (t, s), is uniformly continuous on the closed unit
square [0, 1] × [0, 1], with γs (0) = b and γs (1) = c, 0 ≤ s ≤ 1. If γ0 is homotopic to γ1, we
write γ0 ≈ γ1 . See Figure 14.1.
The function H(t, s) ≡ γs (t) is called a homotopy in± from γ0 to γ1 . Note that γ1 ≈ γ0
using γ1−s , so we say simply that γ0 and γ1 are homotopic in ±. It is not hard to prove
transitivity: if γ1 ≈ γ2 and γ2 ≈ γ3 then γ1 ≈ γ3 . The reader is encouraged to do Exercise 14.3
at this point.
The next lemma says that sufficiently close curves are homotopic and homologous.
Lemma 14.3 If γ : [0, 1] → ± is a curve in a region ± then there is an ε > 0, depending

on the region ± and the curve γ , such that, if σ : [0, 1] → ± is a curve with | γ (t) − σ (t)| < ε
for all t ∈ [0, 1], and σ (0) = γ (0) and σ (1) = γ (1), then σ ≈ γ and γ − σ ∼ 0.
Proof Choose disks Bj ⊂ ± and a partition 0 = t0 < t1 · · · < tn < tn+1 = 1 so that
γj = γ ([tj , tj+1 ]) ⊂ B j . If ε > 0 is sufficiently small then σj = σ ([tj , t j+1]) ⊂ Bj , for each j.
Then
γs (t) = (1 − s)γ (t) + sσ (t)
is a homotopy in ± from γ to σ .
Let Lj ⊂ B j ∩ Bj−1 denote the line segment from γ (tj ) to σ (t j), and set L 0 = { γ (0)} and
Ln+1 = {γ (1)}. Then αj ≡ γj + L j+1 − σj − L j is a closed curve contained in B j ⊂ ± and
∑
hence is homologous to 0 in ±. Thus γ − σ = nj=0 αj is also homologous to 0.
Corollary 14.4 If γ0 ≈ γ1 in a region ± then γ0 − γ1 ∼ 0 in ±.

1
H γ1
c
γs
s
γ0
0
b
Figure 14.1 Homotopy of γ0 to γ1 .
Proof If γs (t) is a homotopy of γ0 to γ1 then we can cover [0, 1] with finitely many open
intervals Jk so that if r, s ∈ Jk then γr − γs ∼ 0 by Lemma 14.3. Thus γ0 − γ1 ∼ 0 by
transitivity.
The converse to Corollary 14.4 fails. See Figure 5.4 and Exercise 14.5. (The reader may
wish to defer doing Exercise 14.5 until after reading Example 14.13.) However, homotopy
does give us another characterization of simple connectivity.
Theorem 14.5 A region ± ⊂ C is simply-connected if and only if every closed curve

contained in ± is homotopic to a constant curve.
Proof If γ is homotopic to a constant curve then γ ∼ 0 by Corollary 14.4. So if all curves in

± are homotopic to constant curves then, by Theorem 5.7, ± is simply-connected. Conversely,
if ± is simply-connected and if γ is a closed curve in ± ± = C beginning and ending at z0, and
if f is a conformal map of ± onto D with f (z0) = 0, then f (γ ) is a closed curve in D beginning
and ending at 0. But then γs (t) = f −1 (sf ( γ (t))) is a homotopy of the constant curve z0 to γ0 .
If ± = C, we can use z − z0 instead of f .
If a closed curve γ is homotopic to a constant curve, then the constant must be γ (0). Some
books use the notation γ ≈ 0.
Theorem 14.6 Suppose γs (t), 0 ≤ s, t ≤ 1, is a homotopy from γ0 to γ1 in a region ±.

Suppose f0 is analytic in a neighborhood of b = γ0(0) = γ1 (0) and suppose f0 can be analyt-
ically continued along each γs . Then the analytic continuation of f0 along γ0 agrees with the
analytic continuation of f 0 along γ1 in a neighborhood of c = γ0 (1) = γ1(1).
Proof By (iii), the analytic continuation of f0 along each γs is unique, 0 ≤ s ≤ 1. By (v), for
each s ∈ [0, 1], the analytic continuation of f0 along γs agrees with the analytic continuation
of f0 along γu in a neighborhood Us of c if |u − s| < ε for some ε = ε(s). By compactness, we
can cover [0, 1] with finitely many such open intervals (sj − εj , sj + εj ), for 1 ≤ j ≤ m. Then
the analytic continuations of f0 along each γs agree on ∩m j=1 Usj .
Corollary 14.7 (monodromy theorem) Suppose ± is simply-connected and suppose f0 is

defined and analytic in a neighborhood of b ∈ ±. If f0 can be analytically continued along
all curves in ± beginning at b then there is an analytic function f on ± so that f = f0 in a
neighborhood of b.
Proof If c ∈ ± and if γ0 is a curve in ± from b to c, let fn be the analytic continuation

of f0 along γ0 to a neighborhood of c, and define f (c) = fn(c). If γ0 and γ1 are curves in ±
beginning at b and ending at c then γ0 ≈ γ1 by Theorem 14.5 and Exercise 14.3(e). So, by
Theorem 14.6 the definition of f (c) does not depend on the choice of the curve γ0 . By (v),
f = fn in a neighborhood of c, so that f is analytic at c. Thus f is defined and analytic in ±
and f = f0 in a neighborhood of 0.
The monodromy theorem can be used to give another proof of the second statement in
Theorem 7.10, that a harmonic function u on a simply-connected region ± is the real part
of an analytic function. If f is analytic on a ball B ⊂ ± with Ref = u on B, then, as noted
after Definition 14.2, f can be continued along all curves in ±. By the monodromy theorem,
because ± is simply-connected, there is an analytic function f on all of ± with Ref = u.
The monodromy theorem is typically used to find a global inverse of an analytic function
with non-vanishing derivative. Suppose f is analytic in ± with f ² ± = 0 on ±. If c ∈ ±, then
there is a function g analytic in a neighborhood of f (c) so that g( f (z)) = z in a neighborhood of
c. If g can be analytically continued along all curves in f (±), and if f (±) is simply-connected,
then by the monodromy theorem there is a function G which is analytic on f (±) satisfying
G( f (z)) = z for z ∈ ±. See Exercises 14.6 and 14.7.
One final remark: analytic continuation really only depends upon the continuity of the func-
tions and the uniqueness theorem on disks, so that the monodromy theorem holds for much
more general classes of functions. For example, if two harmonic functions agree on a small
disk in a region, then they agree on the entire region. So if we replace “analytic” with “har-
monic” or “meromorphic” in our definition of continuation along a curve and in the statement
of the monodromy theorem, the theorem remains true.
14.2 Riemann Surfaces and Universal Covers
Analytic continuation along curves arose in the process of trying to understand the inverse
of an analytic function. Riemann had a more geometric resolution which is closer to the idea
of direct analytic continuation. Instead of considering the function ez , for example, we can
consider the function f (z) = (Re ez , Im ez , Im z) = (ex cos y, ex sin y, y) which maps C into R3.
Each horizontal line in C is mapped to a half-line parallel to the (x, y, 0) plane, whose height
increases with Imz. The image R = f ( C) is an infinite “parking-lot” ramp that spirals around
the line L = {(0, 0, t) : t ∈ R}. (The line L is not in the image.) See Figure 14.2.
The function f is then one-to-one on C and its inverse maps the ramp R onto the plane. We
can think of a direct analytic continuation of log z as determining regions on the ramp, which
are mapped into the plane by this inverse map. The image of one “sheet” or “floor” of the ramp
L
B∗4 R
f B∗3
B∗2
π
f −1 B∗1
\{ 0}
B 0
Figure 14.2 Parking-lot ramp surface.

is a horizontal strip of height 2π . A similar analysis for the function z2 on C \ {0} does not
quite work, but it is possible to use a map into R4 instead. Nevertheless, we informally think of
many analytic functions with non-zero derivative as maps of regions into R3 where the image
is somehow gradually lifted in the third dimension so that the map becomes one-to-one, and
thus we can picture its inverse.
Definition 14.8 A Riemann surfaceis a connected Hausdorff space W, together with a

collection of open subsets Uα ⊂ W and functions zα : Uα → C such that
(i) W = ∪Uα ,
(ii) zα is a homeomorphism of U α onto the unit disk D and
(iii) if Uα ∩ Uβ ± = ∅ then zβ ◦ z−
α
1 is analytic on z (U ∩ U ).
α α β
The functions zα are called coordinate functions or maps , and the sets Uα are called
coordinate charts or disks . We can extend our collection of coordinate functions and disks
to form a base for the topology on W by setting Uα,r = z− α (rD) and zα ,r = r zα on Uα ,r , for
1 1
r < 1. We can also compose each zα with a Möbius transformation of the disk onto itself so
that we can assume that for each coordinate disk Uα and each p0 ∈ Uα there is a coordinate
function zα with zα (p0 ) = 0. For the purposes of the following discussion, we assume that
our collection {zα , Uα } includes all such maps. Note that a Riemann surface W is pathwise
connected since the set of points that can be connected to p0 is both open and closed for each
p0 ∈ W.
These concepts have been extended to more general surfaces called differentiable mani-
folds, where the functions zβ ◦ z− α
1 are required only to be differentiable. However, in that
generality, it is necessary also to assume that the surface is second countable. One conse-
quence of the uniformization theorem, which we will prove in Chapter 15, is that all Riemann
surfaces are second countable.
More informally, a Riemann surface is a set which locally looks like the plane. Many of the
concepts introduced in this book depend only on the local behavior of functions and so can be
defined on Riemann surfaces. A function f : W → C is called analytic if, for every coordinate
function zα , the function f ◦ z−
α
1
is analytic on D. Harmonic, subharmonic and meromorphic
functions on W are defined in a similar way. This means that any theorem about regions in
the plane whose proof depends solely on the local behavior of functions is also valid for Rie-
mann surfaces. The key fact needed is that the result or concept should be invariant under the
transition mapszβ ◦ z− 1
α , where zα and zβ are coordinate maps. By (ii) and (iii), the transition
maps are one-to-one and analytic. For example, the characterizations of zeros and poles of
analytic functions is the same. The maximum principle can be stated in the following form:
if an analytic or harmonic function achieves its maximum on a Riemann surface, then it is
constant. Lemma 4.31 holds for Riemann surfaces where disks are replaced by coordinate
charts, and can be used to create analytic continuations. Differentiation, however, presents a
problem, since, if zα is a coordinate map, the derivative of f ◦ z−
α
1 will depend on the choice of
zα . However, if both f and g are analytic on a Riemann surface then f ² ◦ z−α (z)/g ◦ zα (z) does
1 ² −1
not depend on the choice of zα by the chain rule. Similarly, the property of (real) differentia-
bility does not depend on the choice of coordinate map (though the individual derivatives do).
Integration of continuous functions has a similar problem. We can transport an integral over
a curve contained in a coordinate chart to the disk, but the result will depend on the choice of
the coordinate function.
A Riemann surface does not necessarily lie in some Euclidean space, so we cannot speak
of the complement of a Riemann surface or of its boundary. This requires us to use an alter-
nate definition of some concepts such as “simply-connected.” A Riemann surface W is called
simply-connected if every closed curve in W is homotopic to a constant curve. Equivalently,
any two curves in W that begin at the same point and end at the same point are homotopic
(see Exercise 14.3). This definition agrees with our earlier definition of a simply-connected
region in the plane by Theorem 14.5. The monodromy theorem, Corollary 14.7, holds for
simply-connected Riemann surfaces because the proof only depends on the local properties
of analytic functions and the fact that any two curves with the same initial and terminal points
are homotopic. Consequently, every harmonic function on a simply-connected Riemann sur-
face has the form u = Ref for some analytic function f defined on W. The Perron process also
works on a Riemann surface.
Example 14.9 (Riemann sphere) If π is the stereographic projection of C into the unit
sphere S 2 as given in Section 1.3, then coordinate charts on S 2 are given by π (B), where B
is a disk in C. The corresponding coordinate map is π −1 followed by a linear map of B onto
D. A coordinate chart containing the north pole is given by π (|z| > r) together with the north
pole. The corresponding coordinate map is given by π −1 followed by r /z, declaring the value
of this map at the north pole to be 0.
Example 14.10 (Torus) A torus T can be constructed from a rectangle R = {(x, y) : 0 <
x < a, 0 < y < b} by identifying points on opposite sides of the boundary. We can define an
equivalence relation on C by z ∼ w if and only if z − w = ka + inb for some integers k, n. The
quotient map π : C → C/ ∼ associated with this equivalence relation identifies the opposite
sides of R. We can define charts on the quotient space by π (B), where B is a Euclidean ball.
The corresponding coordinate maps are given by π −1 followed by a linear map of B onto
D. It is an exercise to check that the quotient space is a Riemann surface. The torus can be
visualized by first forming a cylinder by bending the rectangle in R3 until the top and bottom
edges meet so that the vertical sides of R become circles. The cylinder is then bent until the
two circles meet so that the line corresponding to the top and bottom edges of R becomes a
circle. It looks like a donut.
In terms of the rectangle R, a small ball centered at a point in the interior of R corresponds to
a basic open set in T . The image by the map π of a small disk centered at ic, with 0 < c < b,
is the same as the image of two half-disks in R, one centered at ic and the other centered at
a + ic, together with the vertical line segment on the boundary of each (which is identified
to a single line segment in T ). A similar description holds for the open horizontal lines in R.
¯
The image of a disk centered at a corner of R is also the image of four quarter-circles in R.
These sets in T form the charts on the torus constructed from R, and the chart maps are linear
on each portion in R.
Example 14.11 (Riemann surface of an analytic function) If f is analytic on a region

± ⊂ C with f ± = 0 on ±, then we can construct a Riemann surface ±f associated with f
²
by declaring charts to be the images f (B(z, r)) of disks B(z, r) on which f is one-to-one. The
associated chart maps are f −1 composed with a linear map of B(z, r) onto D. More formally,
³∞
we write ± = ∪∞ j =1Bj , where f is one-to-one on each Bj . Set U = j =1 f (B j ), the disjoint
union of the sets f (Bj ). We then identify w ∈ f (Bi ) and w ∈ f (Bj ) if and only if w = f (z) for
some z ∈ B i ∩ B j . In other words, we identify the copies of f (B i ∩ B j ) in the two images f (B i )
and f (Bj ). The corresponding quotient space ±f is a Riemann surface. The function f can be
viewed as a one-to-one map of ± onto ±f , and f −1 becomes a well-defined function on ±f .
See Exercise 14.4 for the extension of this construction to functions whose derivative vanishes
at points of ±.
The idea behind analytic continuation can be used to create what is called a “covering
surface” of a region contained in the plane. Returning to the parking-lot ramp example, Fig-
ure 14.2, if B is a disk in C \ {0} then the vertical cylinder B × R cuts out countably many
¯
topological disks B∗j from the ramp. Each B∗j is mapped onto B by the vertical projection
∗ ∗
π : Bj → B. If B has center c and radius r, then zB∗ = (π − c)/r maps Bj homeomorphically
∗ j
onto D. Then {(zB∗j , Bj ) : −∞ < j < ∞} are coordinate maps and charts that make R into a
Riemann surface. Let p = (1, 0) ∈ C and p∗ = (1, 0, 0) ∈ R, so that π (p∗ ) = p. Then we can
find the center of B∗j , c∗j ∈ π −1 (c), by following a path γ ∗ ⊂ R starting at p∗ . The vertical
projection γ = π (γ ∗ ) is a curve in C \ {0} from p to c. Note that if σ ∗ is another curve from
p∗ to the center of B∗j then γ ∗ (σ ∗ )−1 is a closed curve on the ramp. See Exercise 14.3 for the
definition of the product of two curves (the first followed by the second). The surface R is
homeomorphic to C by the map f (z) so that any closed curve in R is homotopic to a constant
curve. Indeed, a homotopy in C transplants to a homotopy in R using the homeomorphism f .
Thus γ ∗ ≈ σ ∗ . Applying the projection map π to the homotopy, we obtain γ ≈ σ , where
∗
σ = π (σ ). In fact, a curve α is homotopic to γ in C \ { 0} if and only if we can find a curve
α ⊂ R which begins and ends at the same places as γ ∗ with π (α ∗ ) = α. So we can make an
∗
identification between the equivalence classes of curves and points of the ramp R.
We can use this idea to create a Riemann surface that “covers” a region ± ⊂ C. Fix a point
b ∈ ±. Let [γ ] be the equivalence class under homotopy of a curve γ ⊂ ±. Let ±∗ be the
collection of equivalence classes
∗ = {[γ ] : γ is a curve in ± with γ (0) = b}.
±
Define a projection map π : ±∗ → ± by π ([γ ∗ ]) = γ (1). We can make ±∗ into a Riemann

surface by describing coordinate maps and charts. If c ∈ ±, let B = B(c, r) be a disk in ±
centered at c with radius r. Let γ be a curve in ± from b to c. For any point d ∈ B, let σd be a
curve in B from c to d. Let
B∗
= {[γ σd ] : d ∈ B} (14.3)
be the equivalence classes of all curves γ σd , for all d ∈ B. Since all curves in B beginning at
c and ending at d are homotopic, [γ σd ] does not depend on the choice of σd . By the definition
of π , π ([γ σd ]) = d so that π is a one-to-one map of B∗ onto B. Note that if γ1 ≈ γ , then
we obtain the same set of equivalence classes B∗ using γ1 instead of γ . See Exercise 14.3(c).
Thus B ∗ ⊂ π −1 (B) is uniquely determined by the equivalence class of γ , namely a point in
B ∗ , and the disk B. Set zB∗ = (π − c)/r. Then we can give a topology on ±∗ by declaring each
set B∗ to be open, for all disks B(c, r) ⊂ ± and all equivalence classes [γ ] of curves γ ⊂ ±
from b to c. Indeed, the “disks” B∗ form a basis for this topology. Equivalently, we give ±∗
the topology required to make each zB∗ a homeomorphism.
Theorem 14.12 Suppose ± ⊂ C is a region. Then

(i) the surface ±∗ is a simply-connected Riemann surface with coordinate functions zB∗ and
charts B∗ ;
(ii) if B∗1 and B ∗2 are coordinate charts with π (B∗1 ) = π (B∗2 ) then either B∗1 ∩ B∗2 = ∅ or
B ∗1 = B∗2 ;
(iii) if γ ⊂ ± is a curve beginning at b and if b∗ ∈ ±∗ with π (b∗ ) = b, then there is a unique
curve γ ∗ ⊂ ±∗ , called a lift of γ , beginning at b∗ with π (γ ∗ ) = γ ;
(iv) a curve γ ∗ ⊂ ±∗ is closed if and only if γ = π (γ ∗ ) is homotopic to a constant curve
in ±.
Because Theorem 14.12(ii) holds, we say that π evenly covers±. It says that each z ∈ ± is
contained in a disk B such that π −1 (B) is the disjoint union of sets homeomorphic to B by the
map π . An analytic function f with f ² ± = 0 on a region W does not necessarily evenly cover
f (W).
Proof By construction, Definitions 14.8(i) and (ii) hold. The transition functions are just
linear maps and hence analytic. The set ±∗ is pathwise connected, for if [γ ] ∈ ±∗ , where
γ : [0, 1] → ±, let γ x (t) = γ (xt), for 0 ≤ x, t ≤ 1. Then γ x : [0, 1] → ±, and it is not
hard to verify that γ ∗ (x) ≡ [γ x ] defines a continuous curve in ±∗ from [ {b}] to [γ ], where
[{b}] is the equivalence class containing the constant curve {b}. To see that ±∗ is Hausdorff,
suppose p∗ = [γp], q∗ = [γq] ∈ ±∗ , where γp is a curve from b to p = π (p∗ ) ∈ ± and γq is
a curve in ± from b to q = π (q∗ ) ∈ ±. Suppose p∗ ± = q∗. If p ± = q, choose disjoint disks Bp
and Bq centered at p and q, and form the “disks” B∗p and B∗q as in (14.3). Then B∗p ∩ B∗q = ∅
since π (B∗p ) ∩ π (B∗q ) = ∅. If p = q, then let B ⊂ ± be a disk centered at p. Form the sets B∗p
and B ∗q using the curves γp and γq as in (14.3). Since p∗ ± = q∗ , the curves γp and γq are not
homotopic, though they begin at b and end at p = q. If r∗ ∈ B∗p ∩ B ∗q then r∗ = [γp σd ] for
some curve σ d ⊂ B, and similarly r ∗ = [γqσ e], where σ e ⊂ B. But this implies γpσd ≈ γqσe .
But σd and σe are contained in the simply-connected region B; they begin at p = q and end at
the same point in B so that σd−1 ≈ σe−1 . By Exercise 14.3,
γp ≈ γp σd σd−1 ≈ γq σe σe−1 ≈ γq .
This contradicts p∗ ± = q∗ , and so B ∗p ∩ B ∗q = ∅. Thus ±∗ is Hausdorff and we have proved ±∗
is a Riemann surface. This also proves (ii).
Now suppose γ : [0, 1] → ± is a curve beginning at b, and suppose b∗ = [σ ] ∈ ±∗ with
∗ ∗
γ (0) = b = π (b ) = σ (1). If γ is the portion of γ from γ (0) to γ (x) then γ (x) ≡ [σ γ ],
x x
0 ≤ x ≤ 1, defines a lift of γ to ±∗ . Let B ∗ be a coordinate disk containing b∗ . Because π

is a homeomorphism of B∗ onto B, if γ x ⊂ B then there is only one lift of γ x from ± to ±∗
beginning at b∗ . Cover γ by finitely many balls Bj . If γ x has a unique lift (γ x )∗ beginning at
b∗ , and if αj = γ ([x, y]) ⊂ Bj , then αj has a unique lift beginning at (γ x )∗(1). Repeating this
lifting finitely many times proves that the lift of γ to the curve γ ∗ is the unique lift beginning
at b∗ , proving (iii).
Suppose γ ∗ ⊂ ±∗ is a closed curve. Then γ = π (γ ∗ ) is a closed curve in ±. If [σ ] = γ ∗ (0),

then, as above, γ ∗ (x) = [σ γ x ]. Thus γ ∗ (1) = [σ γ ]. Since γ ∗ (0) = γ ∗ (1), we have that
[ σ ] = [σ γ ]. It follows that γ is homotopic to the constant curve {σ (1) = γ (0)}.
Conversely, if γs (t) is a homotopy of the constant curve {γ (0)} to γ then [σ γs ] = [σ ] for
all s ∈ [0, 1]. The lift γs∗ of σ γs to a curve in ±∗ beginning at [σ ] ends at [σ γs ] and thus is a
closed curve. This proves (iv), but also it shows that if γ ∗ is a closed curve in ±∗ then γs∗ gives
a homotopy of the constant curve {γ ∗ (0)} to γ ∗. Thus ±∗ is simply-connected, completing
the proof of Theorem 14.12.
We remark that the same construction works if we replace ± with any Riemann surface W,
since it depends only on the local structure of the surface. In this case, disks in ± are simply
replaced by coordinate charts. The constructed surface W ∗ is called a universal covering
surface of W and π is called a universal covering map. If zα is a coordinate function on
coordinate chart Uα ⊂ W then zα ◦ π is a corresponding coordinate function on Uα∗ .
The parking-lot ramp example, Figure 14.2, can be thought of as pasting a copy of the
upper half-plane to the lower half-plane along the positive reals. Then another copy of the
lower half-plane is pasted to the previous copy of the upper half-plane along the negative
reals. This process of pasting half-planes along half-lines is repeated as much as possible. We
can make a similar construction for the twice-punctured plane.
Example 14.13 (Covering surface of C \ {0, 1}) Suppose ± = C \ {0, 1}. Let I1 , I2 , I3 be
the three open intervals in R \ {0, 1 }. We can create the universal cover and covering map for
± as follows. Let Hu = {z : Imz > 0 } be the upper half-plane and let Hl = {z : Imz < 0 }
be the lower half-plane. Let A = Hu ∪j Ij and let B = Hl ∪j Ij . Attach one copy of A to
one copy of B by identifying the interval I1 on A with the interval I1 on B. The resulting set
has unidentified intervals on A and B. For each such interval on A (respectively B) attach a
new copy of B (respectively A) along the corresponding interval. Repeat this process with all
unidentified intervals on all copies of A and B resulting in new unidentified intervals. Repeat
the process indefinitely. The maximal such surface S∗ has all edges I j on every copy of A
(respectively B) identified with a corresponding Ij on a copy of B (respectively A). Let π be
the projection which maps a point on a copy of A or B to the corresponding point in ±. If
D = {z : |z − c| < r} is a disk in ± then each component D∗ of π −1 (D) lies in the union
of a copy of A and a copy of B. The corresponding chart map associated with a chart D∗ is
( π − c)/r. It is not hard to verify that S∗ is then a Riemann surface.
If γ is a curve in C \{ 0, 1}, then we can “lift” γ to S∗ by passing from one half-sheet, a copy
of A or B, to another as γ crosses an interval Ij ∈ R \ {0, 1}. If γ ⊂ ± is a curve beginning,
say, at a point γ (0) ∈ ± then its homotopy class is determined by which half-sheet it ends up
on (and the point γ (1)). The curve γ is homotopic to a constant curve if and only if its lift to
S∗ is closed. This gives a one-to-one map of the universal covering surface of C \ {0, 1 } onto
S∗ . We will construct another version of this surface in Section 15.1.
This same construction can be done for C \ E, where E is a compact subset of R, where we
identify half-planes along the open intervals in R complementary to E. If E = {0} then the
corresponding S∗ is homeomorphic to the parking-lot ramp constructed earlier.
14.3 Deck Transformations
There are natural maps of a universal covering surface to itself, called deck transformations,
which can be used to recover a Riemann surface W from its universal cover W∗ . Let π be a
universal covering map of W∗ onto W. Fix a point b ∈ W and write W∗ as the collection of
equivalence classes of curves beginning at b. If σ is a closed curve beginning and ending at b,
we can define a map M[σ ] : W ∗ → W ∗ by
M[σ ] ([γ ]) = [σ γ ].
Note that this map is one-to-one and onto and does not depend on the choice of the curve
in [σ ]. If B∗ is a coordinate disk centered at [γ ] then M[σ ](B∗ ) is a coordinate disk centered
at [σ γ ] with the same projection as B∗ . So the map M[σ ] is a homemorphism of W ∗ onto
W ∗. These maps M[σ ] are called the deck transformations. The deck transformations form a
group G under composition. If b∗ = [{b}] is the equivalence class of the constant curve {b},
then Mb∗ is the identity map in this group G.
Lemma 14.14 If p ∗, q∗ ∈ W ∗ , then π (p∗ ) = π (q ∗) if and only if there is a deck

transformation M with M(p∗ ) = q∗.
Proof Recall that π ([γ ]) = γ (1). If p∗ = [γ ], q∗ = [α ] ∈ W ∗ have the same projection

γ (1) = α (1), then, setting σ = αγ
−1 , we have that σ is a closed curve beginning at b and
σ γ ≈ α so that M[σ ] ([γ ]) = [ α]. Conversely, if σ ⊂ W is a closed curve beginning and ending
at b then π ([σ γ ]) = γ (1) = π ([γ ]), so that M[σ ] ([γ ]) and [γ ] have the same projection.
The group G gives an equivalence relation on W ∗, where p∗ ∼ q∗ if and only if there is a

deck transformation M with M(p∗ ) = q∗ . The quotient space W∗ /G with the quotient topol-
ogy is a Riemann surface. The coordinate charts on W∗ /G are just the images by the quotient
map of the coordinate charts on W ∗ . Lemma 14.14 says that the map π induces a one-to-one
map of W ∗ /G onto W. As a group, G is isomorphic to the group of equivalence classes under
homotopy of all closed curves in W beginning at b, which is called the fundamental group
of W at b. See Exercise 14.3. For this reason, the set of deck transformations is sometimes
also called the fundamental group of W.
Definition 14.15 If W1 and W 2 are Riemann surfaces and if f is a map of W1 into W2 , then
we say that f is analytic provided
wβ ◦ f ◦ z−1
α
is analytic for each coordinate function zα on W1 and wβ on W2 wherever it is defined.
For example, if Wb∗ and Wc∗ are the universal covering surfaces of W based on curves begin-
ning at b ∈ W and c ∈ W, respectively, then let σ ⊂ W be a curve from c to b. Define
N[σ ] : Wb∗ → Wc∗ by N[σ ]([γ ]) = [σ γ ]. Then N [σ ] is a one-to-one analytic map of Wb∗
onto Wc∗ .
If f is meromorphic on region W ⊂ C, then f can also be viewed as a map into the
extended plane C∗, or via stereographic projection into the Riemann sphere S2 . Such maps f
14.4 Exercises 227
are then analytic in the sense of Definition 14.15. So some care must be taken when speak-
ing of analytic functions on a Riemann surface such that both the domain and range surfaces
are understood. We shall make the tacit assumption, unless stated otherwise, that an analytic
function maps into the plane.
Corollary 14.16 The deck transformations form a group G of one-to-one analytic maps of
W ∗ onto W ∗ with the property that each p∗ ∈ W∗ has a neighborhood B∗ so that M(B ∗) ∩
B ∗ = ∅ for all deck transformations M not equal to the identity map. The projection map
∗ ∗
π : W → W induces a one-to-one analytic map of W /G onto W.
We say that a group with the property in Corollary 14.16 is properly discontinuous. The
proof that the deck transformations are properly discontinuous follows from Theorem 14.12.
See also Exercise 14.8.
14.4 Exercises
14.1 Create a direct analytic continuation from an analytic continuation along a curve in C. It
is sometimes harder to work with direct analytic continuations because the intersection
of two regions can have many components. This can be avoided by considering only
disks, which are convex and have convex intersections, but in the abstract setting of
Riemann surfaces we do not have this Euclidean geometry.
14.2 (a) Prove that the radius of convergence of power series expansions for an analytic
function in a region ± is Lipschitz continuous: if r(z) is the radius of convergence
of the series at z then |r(z1 ) − r(z2)| ≤ |z1 − z2 |, provided r(zj ) is finite, j = 1, 2.
(b) Prove that f0 has an analytic continuation along γ if and only if there are convergent
power series
∞
´
ft (z) = an (t)(z − γ (t))n
n=0
such that ft (γ (s)) = a0(s) for s sufficiently close to t.

(c) Prove that the coefficients an(t) in part (ii) are continuous.
14.3 All curves in this exercise are contained in a region ± and all homotopies are assumed
to be within ±.
(a) Prove that if γ1 ≈ γ2 and γ2 ≈ γ3 then γ1 ≈ γ3 .
(b) Prove that if σ is a reparameterization of γ (in the same direction) then σ ≈ γ .
(c) If σ begins where γ ends, then we define the product (γ followed byσ ) by
±
γ σ (t) = γ (2t), ∈ [0, 21 ]for t
σ (2t − 1), ∈ [ 21 , 1].
for t
Prove that if γ1 ≈ γ2 and if σ1 ≈ σ 2 then γ1 σ1 ≈ γ2σ 2. Here we assume that σj
begins where γj ends, j = 1, 2.
(d) Define γ −1 by γ −1 (t) = γ (1 − t). Prove that γ γ −1 ≈ {γ (0)}, where {γ (0)} is the
constant curve equal to γ (0) for all t ∈ [0, 1]. Similarly, γ −1γ ≈ {γ (1)}.
(e) Prove γ0 ≈ γ1 if and only if γ0γ1−1 ≈ {b}, where γ0, γ1 begin at b and end at c,
and where {b} is the constant curve at b. Note that in one case the homotopy uses
curves with endpoints b and c, and in the other case the homotopy uses curves that
begin and end at b. Hint: γ1−1 γ1 ≈ {c} by (d).
Homotopy defines equivalence classes of curves beginning at a fixed point
b ∈ ±. The product in (c) is not always defined, so we pick a point b ∈ ± and con-
sider all closed curves that begin and end at b. Then homotopy defines equivalence
classes of curves beginning and ending at b.
(f) Prove that the equivalence classes of curves which begin and end at b ∈ ± form a
group under multiplication with the identity equal to the constant curve at b. This
group, written as π1(±, b), is called the fundamental group of±.
(g) Prove that π 1(±, b) is isomorphic to π1( ±, c), for b, c ∈ ±.
14.4 If f is analytic on ± then Example 14.11 gives a Riemann surface associated with f
on ± \ {z : f ² (z) = 0}. If f (z) = zn on ± = C \ {x ≥ 0} then we can visualize the
surface in R3 using the map z → (Rezn, Imzn, arg z) similar to the ramp created for
ez . But we run into difficulties if ± = C \ {0} because the top edge of this ramp at
height 2π n must be identified with the bottom edge at height 0. We obtain a Riemann
surface by the construction in Example 14.11, but we cannot complete the ramp in R3.
Give coordinate maps and charts for the Riemann surface associated with the map zn
on C \{0} explicitly. We can extend this surface ±zn to include one more point, namely
{0}, by declaring the image of a ball centered at 0 of radius r by the map zn on ±zn
together with 0 to be a coordinate disk. The associated coordinate map is then z1/ n.
Show that this is a Riemann surface. The added point is called a branch point. Give
a similar construction for zeros and poles of a meromorphic function on a region ±.
Hint: If f ² (z0 ) = 0, write f (z) − f (z0) = [(z − z0 )g(z)]k , where g(z0 ) ± = 0.
14.5 Prove that the curve illustrated in Figure 5.4 (p. 68) is homologous to 0 but is not homo-
topic to a constant curve. One elementary approach to this exercise involves mapping
an equilateral triangle to the upper half-plane and then applying the Schwarz reflection
principle enough times. Another way would be to use Example 14.13, after filling in
all of the details. The first approach does not create a universal covering surface for
C \ {0, 1}.
14.6 Suppose p is a polynomial and D is a disk with the property that if p(z) ∈ D then z ∈ D
and p² (z) ± = 0. Prove that there is an analytic function f defined on D so that p( f (z)) = z
for all z ∈ D. This exercise is used in the field of complex dynamics. Be careful that
your reasoning cannot be used to contradict Exercise 14.7(e).
14.7 Set
² z
h(z) =
2
ew dw,
0
where the integration is along any curve in C from 0 to z.

14.4 Exercises 229
(a) Prove that h is well defined and entire.

(b) Find h² and note that it is an even function which is never 0.
(c) Prove that h maps C onto C,
(d) Prove that for every point b ∈ C and each c with h(c) = b there is a neighborhood
U of b and an analytic function g on U so that g(b) = c and h(g(z)) = z for z ∈ U.
(e) Show that every curve γ : [0, 1] → C can be covered by finitely many neighbor-
hoods as in (d), but there is no analytic function G on C such that h(G(z)) = z for
all z ∈ C.
(f) Find where one of the hypotheses of the monodromy theorem fails. Hint: Look at
µ∞ 2
the preimage of i[0, c] where c = 0 e−x dx.
14.8 Suppose G is a group of properly discontinuous LFTs mapping D onto D. Define an
equivalence relation on D by z ∼ w if and only if there exists g ∈ G with g(z) = w.
Give coordinate charts and maps on the quotient space D/G making it into a Riemann
surface. Show that D is the universal covering surface of D/G and that the quotient
map is the universal covering map. Here we identify two spaces if there is a one-to-one
analytic onto map between them. Do the same for a properly discontinuous group of
linear maps on C. This exercise will be used in Chapter 15.
14.9 Prove that the analytic functions (in the sense of Definition 14.15) mapping the
Riemann sphere into itself correspond to the rational functions via stereographic pro-
jection. Prove that the number of zeros equals the number of poles. The one-to-one
analytic functions are the LFTs.
14.10 Prove that if f is continuous and non-vanishing on a simply-connected Riemann surface
W (no assumptions about analyticity), then there is a continuous function g so that
f (z) = eg(z) on W. Hint: If | f (z) − f (z0 )| < | f (z0 )| on a disk B then we can find such
a function g on B. Consider all possible functions g of this form on all coordinate
charts in W. For example, it follows from this exercise that a continuous non-vanishing
function on D has a continuous logarithm.
14.11 Suppose W is a region in C∗ such that C∗ \ W consists of n compact connected
sets. Find n − 1 deck transformations such that the smallest group containing these
transformations is the set of deck transformations for W on W ∗ .
14.12 Suppose f is an analytic function on a Riemann surface V with values in another Rie-
mann surface W. If W ∗ is the universal cover of W and π : W ∗ → W is the universal
covering map, prove there is an analytic function g : V → W∗ so that f = π ◦ g.
15 The Uniformization Theorem
In this chapter we prove a generalization of the Riemann mapping theorem called the uni-
formization theorem which says that the only universal covering surfaces are the disk, plane
and sphere, up to conformal equivalence. This will give us a more concrete realization of all
Riemann surfaces using groups of LFTs on these three spaces. The proof illustrates the power
of the Perron method and the maximum principle. As is standard, the hyperbolic case, D,
is proved by constructing Green’s function. The non-hyperbolic cases are treated in a very
similar manner by constructing the dipole Green’s function.
15.1 The Modular Function
We begin the chapter with an example for motivation. The universal covering surface of
C \{0, 1} in Example 14.13 has another realization using the Schwarz reflection principle. The
conjugate analytic map R(z) = c + r2 /(z − c) fixes points on the circle C = {z : |z − c| 2 = r2}
and maps the disk D bounded by C to C∗ \ D and vice versa. The map R is called reflection
about C. If ϕ is a conformal map of H onto D, then ϕ −1 ◦ R ◦ ϕ(z) = z, the usual reflection
about R. The map R(z) is an LFT, so it preserves angles between curves, and maps (gen-
eralized) circles to (generalized) circles. Because z → z preserves the magnitude of angles
between curves, but reverses the direction, and maps circles to circles, we conclude that R(z)
also preserves the magnitude of angles between curves, but reverses the direction, and maps
circles to circles.
Proposition 15.1 Suppose C is a circle orthogonal to the unit circle and suppose D 1 and
D2 are the two regions in D \ C. Then reflection R about the circle C maps D1 onto D2 and
D2 onto D1 . If E is any circle orthogonal to ∂ D then R(E) is a circle orthogonal to ∂ D. The
map τ (z) = R(z) is an LFT of D onto D, given by reflection about C followed by reflection
about R.
Proof Reflection about C is the identity on C and preserves orthogonality of circles. So

it maps the unit circle to a circle which is orthogonal to C and has the same two points of
intersection with C. There is only one such circle, so that reflection about C maps the unit
circle to itself, and it must interchange the portion of ∂ D outside C with the portion inside C.
The statement about the regions D1 and D2 then follows because the map is the identity on C.
In particular, R(D) = D. By the formula for R given before the statement of Proposition 15.1,
τ is an LFT.
15.1 The Modular Function 231
Figure 15.1 Reflections of a circular triangle T .
Let C be the circle with center 1 + i and radius 1. Note that C meets the unit circle orthog-
onally at 1 and at i. Let T be the region in D bounded by three arcs: the interval (−1, 1), the
circular arc C1 = C ∩ D and the reflection of C1 about the imaginary axis. We call T a circular
triangle because it is bounded by three circular arcs. See Figure 15.1.
If we reflect T about one of its boundary circles, then by Proposition 15.1 we obtain an
adjacent region contained in D bounded by three orthogonal circles, another circular triangle,
with vertices on ∂ D. By Carathéodory’s theorem, a conformal map of T onto the upper half-
plane is a homeomorphism of T onto the closure of H in C∗ . See also Exercise 15.2 for a more
elementary proof of this fact. Composing with an appropriate LFT, we obtain a conformal map
π which maps −1, 1 and i to 0, 1 and ∞, respectively. Each boundary arc of T is mapped
to one of the intervals ( −∞, 0], [0, 1] or [1, ∞). By the Schwarz reflection principle, we can
extend π to be a conformal map of the reflection of T about any one of its boundary circles
onto the lower half-plane. By repeated reflections, we can extend π to be a locally conformal
map of the union of all repeated reflections of T and their boundaries onto C \ { 0, 1}. The
process of reflection across arcs of R in Example 14.13 is exactly the same as reflection about
these orthogonal arcs in D. In this way we create a new version of the covering surface of
C \ {0, 1} contained in D.
This gives another proof of Picard’s little theorem.
Corollary 15.2 If g is an entire function which omits the values 0 and 1, then g is constant.
Proof Choose z0 so that g(z0 ) ∈ H. The function π −1 ◦ g can be defined as an analytic

function in a neighborhood of z0. By the Schwarz reflection principle, this function can
be continued along all paths in C beginning at z0. By the monodromy theorem, it gives
an entire function which is bounded and hence constant by Liouville’s theorem. Thus g is
constant.
The next proposition may see obvious, but nevertheless we give a proof.
Proposition 15.3 The repeated reflections of T (and its boundary) fill the unit disk.
Proof After finitely many reflections of T ∩ D, we obtain a region in D which is bounded

by finitely many circles orthogonal to ∂ D, and hence the complement in D is the closure of a
finite union of disjoint disks intersected with D. Note that reflection of this larger region about
any of its bounding arcs is the union of reflections of individual copies of T. If ± ⊂ D is the
region obtained by all possible reflections, then any p ∈ D \ ± is contained in the intersection
of a sequence of disks whose boundaries are orthogonal to ∂ D. But then there is a maximal
disk D with D ∩ D ⊂ D \ ± containing p. Since ∂ D is orthogonal to ∂ D, it must be the limit of
orthogonal arcs which come from reflections of ∂ T. But reflection about an arc near ∂ D will
send the origin to a point much closer to ∂ D, and hence inside D, which is a contradiction. In
fact, by Exercise 15.1(b), the reflection of the origin about a circle which is orthogonal to ∂ D
at a and b is the average, (a + b)/2.
It follows that π is an analytic map of D onto C \ { 0, 1} which is locally one-to-one. As in

Example 14.11, we can view the map π as a a one-to-one conformal map of the unit disk D
onto the covering surface of C \{0, 1} constructed in Example 14.13, because this surface was
constructed by exactly these reflections of H. Thus the unit disk is conformally equivalent to
the universal covering surface of C \ {0, 1 }. The uniformization theorem will yield the same
conclusion for every plane region with at least two points in the complement.
The function λ(z) = π ((i − z) /(i + z)), the transplant of π to the upper half-plane, is called
the modular function. The preimage of T is the region between the vertical lines Rez = 0
and Rez = 1 and above the circle orthogonal to R at 0 and 1.
15.2 Green’s Function
One of the most important functions on a region or Riemann surface is Green’s function.
Green’s function has many uses, but we will focus in this chapter only on using it to prove the
uniformization theorem.
Suppose W is a Riemann surface. Fix p0 ∈ W and let z : U → D be a coordinate function
such that z(p0 ) = 0. From now on we assume that we have extended our collection of charts
on W to include a local base as described after Definition 14.8 and that each coordinate chart
Uα has compact closure U α ⊂ W, by restricting our collection. Moreover, we can assume that
the coordinate map zα is a homeomorphism U α onto D.
Let Fp0 be the collection of subharmonic functions v on W \ p0 satisfying
v = 0 on W \ K, for some compact K ⊂ W with K ±= W (15.1a)
and
lim sup (v(p) + log |z(p)|) < ∞ . (15.1b)
p→p0
Note that v ∈ Fp0 is not assumed to be subharmonic at p0 , and indeed it can tend to +∞ as
p → p0 . Set
gW (p, p0 ) = sup{v(p) : v ∈ Fp0 }. (15.2)
Condition (15.1b) does not depend on the choice of the coordinate function provided it van-
ishes at p0 . For, if zα is another coordinate map with zα (p0 ) = 0, then lim p→p0 |zα (p)/z(p)| =
|(zα ◦ z−1)² (0)| ± = 0, ∞. The collection Fp0 is a Perron family on W \ {p0}, so one of the
following two cases holds by Harnack’s theorem:
case 1: gW (p, p0 ) is harmonic in W \ {p0}, or
case 2: gW (p, p0 ) = +∞ for all p ∈ W \ {p0 }.
In the first case, gW (p, p0) is called Green’s function onW with pole (or logarithmic singu-
larity) at p0. In the second case, we say that Green’s function with pole at p0 does not exist
on W.
The primary tool we will use in this section and the rest of this chapter is the maximum
principle or Lindelöf’s maximum principle, Theorem 7.15, which allows us to deal with an
isolated point like p0 . We remark that it is difficult to apply the maximum principle directly
to a harmonic or subharmonic function because W is not necessarily contained in a larger
Riemann surface so that we cannot consider the “boundary” of W. One possibility is to replace
lim ζn→∂ W with the requirement that {ζ n } is eventually outside every compact subset of W.
However, there are examples of surfaces W and sequences that are eventually outside of every
compact set of W yet gW does not tend to 0 on these sequences. We avoid these difficulties by
considering Perron families of functions which are equal to 0 off a compact set. We remind the
reader that our subharmonic functions are assumed to be continuous as maps into [ −∞, ∞ ),
so by continuity the functions are also 0 on the boundary of the appropriate compact set. To
bound such a function we need only to use the maximum principle or Lindelöf’s maximum
principle on the interior of the compact set. Once we obtain a uniform estimate for all members
of the family, we can then take the supremum over the family. We shall use this idea many
times in this chapter.
Lemma 15.4 Suppose p0 ∈ W and suppose z : U → D is a coordinate function such that

z(p0 ) = 0. If gW (p, p0) exists, then
gW (p, p0 ) > 0 for p ∈ W \ {p0}, and (15.3)

gW (p, p0 ) + log | z(p)| extends to be harmonic in U. (15.4)
Proof The function

±
v0 (p) = − log |z(p)|, for p ∈ U
0, for p ∈ W \ U
is in Fp0 by Definition 13.1(i). Hence, gW (p, p0 ) ≥ 0 and gW (p, p0 ) > 0 if p ∈ U. By the
maximum principle applied to −g in W \ { p0 }, (15.3) holds.
If v ∈ Fp0 then, by (15.1b) and Lindelöf’s maximum principle,
sup (v + log |z|) = sup v ≤ sup gW < ∞.

U \{p0 } ∂U ∂U
Taking the supremum over v ∈ Fp0 , we obtain
gW + log |z| ≤ sup gW < ∞

∂U
in U \ { p0 }. We also have that
+ log |z| ≥ v0 + log |z| = 0

gW (15.5)
for p ∈ U \ {p0}. Thus gW + log | z| is bounded and harmonic in U \ {p0 }. Using the Poisson
integral formula on zα(U), we can find a bounded harmonic function on U which agrees with
gW + log |z| on ∂ U. By Lindelöf’s maximum principle, gW + log |z| extends to be harmonic
in U.
The Green’s function for the unit disk D is given by

² ²
² 1 − az ²
gD (z, a) = ²
log ² ².
z− a ²
If g(z) = log |1|z−az|
−a| then, by Lindelöf’s maximum principle, each candidate subharmonic
function v in the Perron family Fa is bounded by g. Moreover, max(g − ε, 0) ∈ Fa, when
ε > 0. Letting ε → 0, we conclude g = g D (z, a).
The next theorem gives a large collection of Riemann surfaces for which Green’s function
exists. See also Exercise 15.3.
Theorem 15.5 Suppose W0 is a Riemann surface and suppose U0 is a coordinate disk whose
closure is compact in W0 . Set W = W0 \ U0. Then gW (p, p0) exists for all p, p0 ∈ W with
p ± = p 0.
Proof Fix p0 ∈ W and let U ⊂ W be a coordinate disk with compact closure and with
coordinate function z : U → D and z(p0 ) = 0. To prove that gW exists, we show that the
family Fp0 is bounded above. Fix r, with 0 < r < 1, and set rU = {p ∈ W : |z(p)| < r}. If
v ∈ Fp0 then, by (15.1b) and Lindelöf’s maximum principle,
v(p) + log |z(p)| ≤ max (v(q) + log |z(q)|) = max v(q),
q∈∂ U q∈∂ U
for all p ∈ U. Thus

max v(p) + log r ≤ max v(p). (15.6)
p∈∂ rU p∈∂ U
Equation (15.6) is a growth estimate on v in U. We can obtain another growth estimate

which uses the fact that we removed a disk from W0 by constructing a harmonic function in
W \ rU which is equal to 1 on ∂ rU and in some sense equal to 0 on ∂ W. More precisely, let F
denote the collection of functions u which are subharmonic on W \ rU with u = 0 on W0 \ K
for some compact set K ⊂ W0 , which can depend on u, and such that
lim sup u(p) ≤ 1 and lim sup u(p) ≤0
p→ ζ p→α
for ζ ∈ ∂ rU and α ∈ ∂ U0. See Figure 15.2.

Applying the maximum principle on the interior of K \ (U0 ∪ rU), we obtain u ≤ 1 on
W \ rU. Because F is a Perron family,
ω (p) = sup{u(p) : u ∈ F }
u ≤0
p0
u ≤1
rU U0
∂U
K
u=0
Figure 15.2 Harmonic measure of ∂ rU in W \ rU.

is harmonic in W \ rU. We can construct a local barrier at each point of ∂ U0 ∪ ∂ rU for the
region W \ rU by transporting the problem to a region in D via a coordinate map. Thus, by
exactly the same argument used for the Dirichlet problem, the harmonic function ω extends
to be continuous at each point of ∂ U 0 and each point of ∂ rU so that ω(p) = 0 for p ∈ ∂ U0
and ω (p) = 1 for p ∈ ∂ rU. In particular, this implies ω is not constant. Moreover, 0 ≤ ω ≤ 1
because 0 ∈ F and each candidate u ∈ F is bounded by 1. Since ω is not constant, the
maximum principle implies 0 < ω(p) < 1 for p ∈ W \ rU. The function ω is called the
harmonic measureof ∂ rU in the region W \ rU.
Returning to our function v ∈ Fp0 , we have that
³ ´
v(p) ≤ max v ω(p),
∂ rU
for p ∈ W \ rU, by the maximum principle applied to v − (max ∂ rU v)w. So

³ ´ ³ ´
max v ≤ max v max ω(p) ≤ max v (1 − δ ) (15.7)
∂U ∂ rU p∈ ∂ U ∂ rU
for some δ > 0. Adding inequalities (15.6) and (15.7) yields

1
δ max v ≤ log
∂ rU r
for all v ∈ F , with δ is independent of v. This implies that case 2 does not hold and hence
Green’s function exists.
Green’s function on a Riemann surface can be recovered from Green’s function on the
universal cover by the following Theorem 15.6. A consequence is Corollary 15.7, which plays
a critical role in the proof of the uniformization theorem.
Theorem 15.6 Suppose W is a Riemann surface for which gW with pole at p0 ∈ W exists.
Let W ∗ be the simply-connected universal covering surface of W, let π : W ∗ → W be the
universal covering map and suppose π (p∗0 ) = p0 . Then gW ∗ with pole at p∗0 exists and satisfies
µ
gW (π (p∗ ), π (p∗0 )) = gW∗ (p∗ , q∗ ). (15.8)
q∗:π (q∗)=π (p∗0 )
We interpret the infinite sum on the right-hand side of (15.8) to be the supremum of all
sums over finitely many q∗ .
Proof Suppose q∗1 , . . . , q∗n are distinct points in W ∗ with π (q∗j ) = p0 = π (p∗0 ). Suppose
vj ∈ Fq∗j , the Perron family for the construction of gW∗ (· , q∗j ). So vj = 0 off K ∗j , a compact
subset of W ∗ and
( )
lim sup vj (p∗ ) + log |z ◦ π (p∗ )| < ∞,
p∗ →q∗j
where z is a coordinate chart on W with z(p0 ) = 0. Recall that gW (p, p0 ) + log |z(p)| extends
to be finite and continuous at p0 , and hence
lim gW (π (p∗ ), p0) + log | z( π (p∗ ))|
p∗→q∗j
exists and is finite because π (q∗j ) = p0 . Thus

⎛ ⎞
n
µ
⎝ vj (p∗ )⎠ − gW (π (p∗ ), p0)
j =1
is bounded in a neighborhood of each q∗j , for j = 1, . . . , n, and is ≤ 0 off K ∗ = ∪j Kj∗ , by

(15.3). By Lindelöf’s maximum principle on the interior of K ∗ \ {q∗1 , . . . , q∗n}, it is bounded
above by 0. Taking the supremum over all such vj , we conclude that gW∗ (p∗ , q∗j ) exists and
⎛ ⎞
µn
⎝ gW ∗ (p∗ , q∗j )⎠ − gW (π (p∗ ), p0 ) ≤ 0.
j =1
Taking the supremum over all such finite sums, we have

µ
S(p∗ ) ≡ gW∗ (p∗ , q∗ ) ≤ gW (π (p∗ ), p0).
q∗: π (q∗ )=p0
Moreover, as a supremum of finite sums of positive harmonic functions, S(p∗ ) is harmonic on

W ∗ \ π −1 (p0 ) by Harnack’s principle. By (15.4),
S(p∗ ) + log | z(π (p∗ ))| (15.9)
extends to be harmonic in a neighborhood of each q∗ ∈ π −1(p0 ).

Now take v ∈ Fp0 , the Perron family used to construct gW (p, p0). Let U∗ be a coordinate
disk containing p∗0 , so that z ◦ π is a coordinate map of U ∗ onto D, vanishing at p∗0 . Then, by
(15.9) and Lindelöf’s maximum principle,
v(π (p∗ )) − S(p∗ ) ≤ 0
for p∗∈ U∗ . Taking the supremum over all such v, we obtain

gW (π (p∗ ), p0) ≤ S(p∗),
and thus (15.8) holds on U ∗ \ p∗0 and therefore on W ∗ .
The converse of Theorem 15.6 is not true. See Exercise 15.7.
Corollary 15.7 Suppose W is a Riemann surface for which Green’s function gW with pole
at p exists, for some p ∈ W, and suppose W ∗ = D. Then gW with pole at q exists for all q ∈ W
and
gW (p, q) = gW (q, p). (15.10)
We will remove the hypothesis that W ∗ = D in Section 15.3. See Corollary 15.9.
Proof As noted earlier, gD (a, b) = − log | (a − b)/(1 − ba)|. If τ is an LFT of the disk onto
the disk, then, by an elementary computation, gD (a, τ (b)) = gD (τ −1 (a), b) for all a, b ∈ D
with a ± = τ (b). Let G denote the group of deck transformations, a group of LFTs τ mapping
D onto D and satisfying π ◦ τ = π . Moreover, if π (q∗ ) = π (p∗ ) then there is a τ ∈ G such
that τ (p∗) = q∗ , by Lemma 14.14. Suppose g(p, p0) exists for some p0 ∈ W and all p ± = p0 .
Choose p∗0 ∈ D so that π (p∗0 ) = p0 . By Theorem 15.6 for p∗ ∈/ π −1(p0 ),
² ²
µ ² p∗ − τ (p∗) ²
gW (π (p∗ ), π (p∗0 )) =
²
− log ²² 0 ²
² (15.11)
τ ∈G
1 − τ (p0 )p∗ ²
∗
² ²
µ ² τ −1(p∗ ) − p∗ ²
= − log ²² ² 0 ²
². (15.12)
τ ∈G
1 − τ (p )p∗0 ²
− 1 ∗
Fix p∗ ∈ D \ π −1(p0 ). Each term in the sum

² ²
² −1 ∗ ∗²
− log ²²² τ (p−1) −∗ q ∗ ²²²
µ
S(q∗) =
τ ∈G 1 − τ (p )q
is a positive harmonic function of q∗ ∈ D \ τ −1 (p∗ ). Since the sum of these positive harmonic
functions converges when q∗ = p∗0 , the function S is harmonic in D \ {τ −1(p∗ ) : τ ∈ G},
by Harnack’s theorem. If v ∈ Fp, the Perron family for gW (q, p), where p = π (p∗ ), then,
by Lindelöf’s maximum principle, v ≤ S ◦ π −1 . Taking the supremum over all v ∈ Fp, we
conclude that gW (q, p) exists and gW (q, p) ≤ S ◦ π − 1(q), for all q ± = p. Thus
gW (π (p∗0 ), π (p∗ ))
≤ S(p∗0 ) = gW (π (p∗ ), π (p∗0 )).
Reversing the roles of p∗0 and p∗ proves (15.10) for q = p0 and all p ± = p0. Because the
Green’s function gW with pole at p then exists for every p ∈ W, (15.10) must hold for all p
and q.
15.3 Simply-Connected Riemann Surfaces
We will now consider simply-connected Riemann surfaces and prove the uniformization
theorem in case 1.
Theorem 15.8 (uniformization, case 1)If W is a simply-connected Riemann surface then

the following are equivalent:
gW (p, p0) exists for some p0 ∈ W, (15.13)

gW (p, p0) exists for all p 0 ∈ W, (15.14)
there is a one-to-one analytic map ϕ of W onto D. (15.15)
Moreover, if g W exists, then

gW (p1 , p0) = gW (p0 , p1), (15.16)
and g W (p, p0) = − log |ϕ (p)| , where ϕ(p0) = 0.

Proof Suppose there is a one-to-one analytic map ϕ of W onto D and let p0 ∈ W. By
composing ϕ with an LFT, we can assume that ϕ(p0) = 0. If v ∈ Fp0 then v = 0 off a
compact set K, so, by (15.1b) and Lindelöf’s maximum principle applied on the interior of
K \ { p0 },
v + log |ϕ| ≤ 0
on W. Taking the supremum over all such v shows that gW (p, p0 ) < ∞ and therefore (15.14)
holds. Clearly, (15.14) implies (15.13).
Now suppose (15.13) holds. By (15.4) there is an analytic function f defined on a coordinate
disk U containing p0 so that
Ref (p) = gW (p, p0 ) + log | z(p)|
for p ∈ U. Hence the function

ϕ (p) = ze−f (p)
is analytic in U and satisfies |ϕ(p)| = e−gW (p,p0) and ϕ(p0 ) = 0. On any coordinate disk Uα
with p0 ∈/ Uα , gW (p, p0 ) is the real part of an analytic function because it is harmonic. The
difference of two analytic functions with the same real part is constant, so that ϕ on U can
be analytically continued along all curves in W beginning at p0 . By the monodromy theorem,
there is a function ϕ, analytic on W, such that
|ϕ(p)| = e−g (p,p ) < 1.

W 0
We claim that ϕ is one-to-one. If ϕ(p) = ϕ(p0) = 0, then p = p0 because gW (p, p0 ) is finite

for p ± = p0 . Let p1 ∈ W, with p1 ± = p0. Then, by (15.3), |ϕ (p1 )| < 1 and
ϕ − ϕ (p 1)
ϕ1 ≡
1 − ϕ (p1 )ϕ
is analytic on W and |ϕ1| < 1. If v ∈ Fp , then, by (15.1) and Lindelöf’s maximum principle,
1
as argued above,
v + log |ϕ1| ≤ 0.
Taking the supremum over all such v, we conclude that gW (p, p1) exists and that
gW (p, p1) + log | ϕ1 | ≤ 0. (15.17)
Setting p = p0 in (15.17) gives
gW (p0 , p1 ) ≤ − log | ϕ1 (p0)| = − log | ϕ(p1)| = gW (p1 , p0).
Switching the roles of p0 and p1 gives (15.16). Moreover, equality holds in (15.17) at p = p0
so that, by the maximum principle, gW (p, p1 ) = − log |ϕ1(p)| for all p ∈ W \ {p1 }. Now, if
ϕ (p 2) = ϕ (p1 ), then, by the definition of ϕ 1, ϕ 1(p 2) = 0. Thus g W (p 2, p1 ) = ∞ and p 2 = p 1.
Therefore ϕ is one-to-one.
The image ϕ(W) ⊂ D is simply-connected, for if γ ⊂ ϕ(W) is a closed curve then
ϕ −1 (γ ) ⊂ W is closed and therefore homotopic to a constant curve. Applying the map ϕ
to the homotopy gives a homotopy in ϕ(W) of γ to a constant curve. If ϕ(W) ± = D then, by the
Riemann mapping theorem, we can find a one-to-one analytic map ψ of ϕ (W) onto D with
ψ (0) = 0. The map ψ ◦ ϕ is then a one-to-one analytic map of W onto D, with ψ ◦ ϕ (p 0) = 0,
proving (15.15).
We remark that the map ϕ1 in the proof of Theorem 15.8 is actually onto, as can be
seen by applying the “onto” argument in the proof of the Riemann mapping theorem. See
Exercise 15.5.
As promised, we now remove the hypothesis that W ∗ = D in Corollary 15.7.
Corollary 15.9 Suppose W is a Riemann surface for which Green’s function gW with pole
at p exists, for some p ∈ W. Then gW with pole at q exists for all q ∈ W and
gW (p, q) = gW (q, p). (15.18)
Proof If W is a Riemann surface such that gW with pole at some p ∈ W exists, then gW∗
exists by Theorem 15.6. But then, by Theorem 15.8, W ∗ is conformally equivalent to D.
Applying Exercise 15.4(a) and Corollary 15.7 yields (15.18).
Theorem 15.10 (uniformization, case 2)Suppose W is a simply-connected Riemann sur-

face for which Green’s function does not exist. If W is compact, then there is a one-to-one
analytic map of W onto C∗ . If W is not compact, then there is a one-to-one analytic map of W
onto C.
As seen above, Green’s function for the disk with pole at 0 is given by G = − log | z|. There
is no Green’s function on the sphere or the plane, but this same function G plays a similar role.
Instead of one pole, or logarithmic singularity, G has two poles on the sphere, with opposite
signs. We will call it a dipole Green’s function, because it is the two-dimensional analog of
an electric dipole, two charges with opposite signs. The next lemma says that a dipole Green’s
function exists for every Riemann surface. For a simply-connected Riemann surface without
Green’s function, the dipole Green’s function will be used to construct a conformal map to
the sphere or the plane in much the same way as Green’s function was used in case 1.
Lemma 15.11 Suppose W is a Riemann surface and, for j = 1, 2, suppose that zj : Uj → D

are coordinate functions with coordinate disks U j satisfying U1 ∩ U2 = ∅, and zj (pj ) = 0.
Then there is a function G(p) ≡ G(p, p1 , p2 ), harmonic in p ∈ W \ { p1 , p2}, such that
G + log |z1 | extends to be harmonic in U1, (15.19)

G − log |z2 | extends to be harmonic in U2 (15.20)
and
sup |G(p)| < ∞. (15.21)

p∈W \(U1 ∪U 2 )
Before proving Lemma 15.11, we will use it to prove the uniformization theorem in case 2,
since the proof is similar to the proof in case 1.
Proof of Theorem15.10 By Theorem 15.8, we may suppose that gW (p, p1 ) does not exist
for all p, p1 ∈ W. Because W is simply-connected, we can apply the monodromy theorem, as
in the proof of Theorem 15.8, to obtain a meromorphic function ϕ1 defined on W such that
|ϕ1(p)| = e−G(p,p ,p ),
1 2
where G is the dipole Green’s function from Lemma 15.11. Note that ϕ1 has a simple zero at
p1 , a simple pole at p2 and no other zeros or poles.
Let us prove ϕ1 is one-to-one. If p0 ∈ W \ { p1 , p2}, then ϕ1(p0 ) ± = 0, ∞ . Let ϕ0 be the
meromorphic function on W such that
|ϕ0(p)| = e−G(p,p ,p ) 0 2
and consider the function

H(p) =
ϕ 1 (p) − ϕ1 (p0) .
ϕ 0 (p)
Then H is analytic on W because its poles at p2 cancel and because ϕ0 has a simple zero at p0.
By (15.21) and the analyticity of H, |H | is bounded on W. But if v ∈ Fp1 , the Perron family
used to construct gW (p, p1 ), then, by Lindelöf’s maximum principle,
² ²
² H(p) − H(p1 ) ²
v(p) + log ²
² ² ≤ 0.
2 sup |H | ²
W
Because gW (p, p1) does not exist, sup{v(p) : v ∈ Fp } ≡ +∞ for every p ∈ W \ {p1 }, and
1
therefore
H(p) ≡ H(p1 ) = −ϕ1 (p0)/ϕ0(p1 ) ±= 0, ∞.
Since H ± = 0, we conclude that ϕ1(p) ± = ϕ1 (p0), unless ϕ0(p) = 0. But if ϕ0(p) = 0, then
p = p0 . Thus ϕ1 is one-to-one on W \ {p1 , p2}. But the only zero of ϕ1 is p1 and the only pole
of ϕ1 is p2, so that ϕ1 is one-to-one on W.
We have shown that ϕ1 is a one-to-one analytic map from W to a simply-connected region
∗ ∗
ϕ1 (W) ⊂ C . If C \ ϕ 1(W) contains more than one point, then, by the Riemann mapping
theorem, there is a one-to-one analytic map of ϕ1 (W), and hence of W, onto D. Since we
assumed that gW does not exist, this contradicts Theorem 15.8. Thus C∗ \ ϕ1 (W) contains at
most one point, and the last two statements of Theorem 15.10 are now obvious.
We now complete the proof of the uniformization theorem by proving Lemma 15.11.
Proof of Lemma15.11 If gW exists, then G(p) = g(p, p1 ) − g(p, p2 ) satisfies (15.19),

(15.20) and (15.21). The idea of the proof when there is no Green’s function is to remove
a small disk from W, giving a surface with Green’s function by Theorem 15.5 and therefore
a corresponding dipole Green’s function, then let the radius of the removed disk decrease to
zero. However, we need some bound on the difference to be able to pass to a limiting dipole
Green’s function. Because an approximating G is a difference, we cannot use the maximum
principle on the difference of two candidates for the corresponding Perron families. So we
reduce the size of each Green’s function by using g(p, p1 ) − g(p2 , p1) and g(p, p2) − g(p1 , p2).
By Corollary 15.9, this does not affect the difference, but we can estimate each on ∂ U 1 ∪ ∂ U2.
Suppose z0 is a coordinate function with coordinate chart U0 such that U0 ∩ Uj = ∅ for
j = 1, 2. Let p0 be the point in U0 such that z0(p0 ) = 0. Set tU0 = { p ∈ W : |z0(p)| < t} and
set Wt = W \ tU0. See Figure 15.3.
rU 1 U2
p1 p2
∂U 1
p0
tU0
Figure 15.3 The surface Wt .
By Theorem 15.5, gWt (p, p1 ) exists for all p, p1 ∈ Wt with p ±= p1 . Fix r, 0 < r < 1, and
set rU1 = {p ∈ W : |z1 (p)| < r}. By the maximum principle,
gWt (p, p1 ) ≤ M1 (t) ≡ qmax gWt (q, p1), (15.22)

∈ rU∂ 1
for all p ∈ Wt \ rU1, because the same bound holds for all candidates in the Perron family
defining gWt . The growth estimate (15.6) shows that
1
M1(t) ≤ max gWt (p, p1) + log . (15.23)
p∈∂ U 1 r
By (15.22), ut (p) ≡ M1(t) − gWt (p, p1 ) is a positive harmonic function in Wt \ rU1, and by
(15.23) there exists q ∈ ∂ U1 with ut (q) ≤ log 1r .
Riemann surfaces are pathwise connected so let γ be a curve in W \ (U1 ∪ U2 ) connecting
∂ U1 to ∂ U2 which does not pass through p 0 . Then, for t ≤ t0, K = ∂ U1 ∪ U2 ∪ γ ⊂ Wt \ rU1
is compact and connected. By Harnack’s inequality, there is a constant C < ∞ depending on
K and r but not on t so that, for all p ∈ K and t ≤ t0,
0 ≤ ut (p) ≤ C,
and
|gW (p, p1) − gW (p2 , p1)| = |ut (p2) − ut (p)| ≤ 2C.
t t
Likewise, if K ² = ∂ U2 ∪ U1 ∪ γ there is a constant C < ∞ so that
|gW (p, p2 ) − gW (p1, p2 )| ≤ C,

t t
for all p ∈ K ² and t ≤ t0.

By Corollary 15.9, gW (p1, p2 ) = gW (p2 , p1) and so the function
t t
Gt (p, p1 , p2) ≡ gW (p, p1) − gW (p, p2 )

t t
= (gW (p, p1) − gW (p2 , p1)) − (gW (p, p2 ) − gW (p1, p2 ))

t t t t
is harmonic in Wt \ {p1, p2 } and satisfies
|Gt (p, p1 , p2)| ≤ C,

for all p ∈ K ∩ K ² ⊃ ∂ U1 ∪ ∂ U2 and some finite C independent of t.
If v ∈ Fp , the Perron family for gW (p, p1 ), then v = 0 off a compact subset of Wt and
1 t
gWt > 0 so that, by the maximum principle,
¶ · ( ¶ ·)
sup v(p) − gWt (p, p2 ) ≤ max 0, sup v(p) − gWt (p, p2)
W t \ U1 ∂ U1
( ¶ ·)
≤ max 0, sup gWt (p, p1) − gWt (p, p2 ) ≤ C.
∂ U1
Taking the supremum over all such v yields

sup Gt (p, p1, p2 ) ≤ C.
p∈W t \ U1
Similarly,
inf
p∈ Wt \U 2
Gt (p, p1 , p2) =− sup −Gt (p, p1, p2 ) ≥ −C,
p∈W t \ U2
and so
|Gt (p, p1, p2 )| ≤ C
for all p ∈ Wt \ {U1 ∪ U2}. The function Gt + log |z1| extends to be harmonic in U1, so, by the
maximum principle, we have that
sup |Gt + log |z1 || = sup |Gt + log |z1 || = sup |Gt | ≤ C.
U1 ∂ U1 ∂ U1
Similarly,
sup |Gt − log |z2 || = sup |Gt − log |z2 || = sup |Gt | ≤ C.
U2 ∂ U2 ∂ U2
By normal families, there exists a sequence tn → 0 so that Gtn converges uniformly on com-
pact subsets of W \{p0, p1 , p2} to a function G(p, p1, p2 ) harmonic on W \{p0 , p1, p2 } satisifying
(15.19), (15.20) and (15.21). The function G(p, p1 , p2) extends to be harmonic at p0 because
it is bounded in a punctured neighborhood of p0. Indeed, we can transfer this problem to the
unit disk via the coordinate map, then use the Poisson integral formula to create a bounded
harmonic function on the disk with the same values on ∂ D. By Lindelöf’s maximum principle,
this is the harmonic extension.
Combining Theorems 15.8 and 15.10, we have proved Koebe’s uniformization theorem.
Theorem 15.12 (uniformization) Suppose W is a simply-connected Riemann surface.

(i) If Green’s function exists for W, then there is a one-to-one analytic map of W onto D.
(ii) If W is compact, then there is a one-to-one analytic map of W onto C∗.
(iii) If W is not compact and if Green’s function does not exist for W, then there is a one-to-
one analytic map of W onto C.
Since each of the spaces D, C and C∗ is second countable, we have the following
consequence.
Corollary 15.13 (Rado) Every Riemann surface satisfies the second axiom of countability.
Proof The universal covering map π sends a countable base on the universal covering
surface W to a countable base on W.
15.5 Exercises 243
We remark that the normal families argument used in the proof of Lemma 15.11 does not
require second countability. If we find a sequence t n so that Gtn converges on a disk B then,
on any compact set K containing B, the sequence Gtn has only one subsequential limit by the
uniqueness theorem. Thus the full sequence Gtn converges on K. Alternatively, we can deduce
Rado’s Corollary 15.13 using only Theorem 15.5, 15.6 and 15.8, avoiding the construction of
dipole Green’s function entirely.
15.4 Classification of All Riemann Surfaces
We can now give a classification of all Riemann surfaces up to one-to-one analytic maps
(conformal equivalence). By the uniformization theorem, the universal covering surface W ∗
of a Riemann surface W is, up to conformal equivalence, the disk, the plane or the sphere. By
Corollary 14.16, the deck transformations form a properly discontinuous group G of one-to-
one analytic maps of W ∗ onto W ∗, such that W is conformally equivalent to W ∗ /G. But, by
Exercise 14.9, the deck transformations are LFTs. See also Exercise 14.8.
Theorem 15.14 If U = C∗ , C or D, and if G is a properly discontinuous group of LFTs of U

onto U, then U /G is a Riemann surface. A function f is analytic, meromorphic, harmonic or
subharmonic on U /G if and only if there is a function h defined on U which is (respectively)
analytic, meromorphic, harmonic or subharmonic on U satisfying h ◦ τ = h for all τ ∈ G and
h = f ◦ π , where π : U → U /G is the quotient map. Every Riemann surface is conformally
equivalent to U /G for some such U and G.
Test your understanding of this result by checking all remaining details.

In Chapter 16, we will show that, up to conformal equivalance, the only Riemann surface
covered by the C∗ is C∗, and that the only surfaces covered by C are C, C \ {0} and tori. Any
Riemann surface not conformally equivalent to one of these is covered by the disk D.
15.5 Exercises
15.1 (a) Prove that an LFT M of D onto D is given by two reflections. Hint: If M = λ(z − c) /
(1 − cz), then | M(0)| = |M−1 (0)| . Let C be the circle orthogonal to ∂ D centered at
1/M(0) and let L be the diameter of D which bisects the angle determined by the
three points M(0), 0, M −1(0). Reflect about L then reflect about C.
(b) If a, b ∈ ∂ D give an explicit formula for reflection R about the circle orthogonal to
∂ D through a and b, show that R(0) = (a + b)/2.
15.2 In Section 15.1, we used Carathéodory’s theorem to conclude that a conformal map of
the circular triangle T onto the upper half-plane extends to be a homeomorphism of the
closure T onto the closure of the upper half-plane in C∗ . Give a more elementary proof
by using Corollary 8.18. Hint: Near a vertex v of T , open up the region by applying
C /(z − v) then ez to map a neighborhood of v in T to a neighborhood of 0 in H.
15.3 If E0 is a compact subset of a Riemann surface W0 such that there exists a local
barrier in W = W0 \ E0 for some ζ 0 ∈ E0 , prove gW exists. Then prove that
limp∈W→ζ0 gW (p, p0 ) = 0.
15.4 (a) Prove that if ϕ is a one-to-one analytic map of a Riemann surface W1 onto a
Riemann surface W2 , then Green’s function on W1 exists if and only if Green’s
function on W2 exists. Moreover, show that gW2 (ϕ(p), ϕ (p0 )) = gW1 (p, p0 ).
(b) Prove that if W1 ⊂ W2 are Riemann surfaces, and if gW2 exists, then gW1 exists.
The Riemann mapping theorem is used in two places in the proof of the uniformiza-
tion theorem. It is possible to avoid both by incorporating parts of the proof of the
Riemann mapping theorem, so that the Riemann mapping theorem becomes a conse-
quence of the uniformization theorem. The next two exercises outline the ideas. Fill in
the details.
15.5 At the end of the proof of Theorem 15.8, we obtain a conformal map ϕ of W onto a
possibly proper subset of D. If it is proper, apply the “onto” portion of the proof of
Theorem 10.11 to obtain a map σ mapping ϕ(W) into D such that
|σ ◦ ϕ (z)| > |ϕ(z)|. (15.24)
By the argument used to establish (15.14) and (15.16), gW + log | σ ◦ ϕ| ≤ 0. But

gW (p, p0 ) = − log | ϕ(p)|, so that log |σ ◦ ϕ| ≤ log | ϕ|, contradicting (15.24).
15.6 At the conclusion of the proof of Theorem 15.10, we obtained a conformal map of W
into C∗ . If two points were omitted, we obtained a conformal map ϕ of W onto D, then
obtained a contradiction. Another way to proceed is as follows: if there are at least two
points a, b ∈ C∗ \ ϕ1 (W) then use a construction from the proof of Theorem 8.11 to
map ϕ1 (W) into D, then apply Exercise 15.4(b) to obtain a contradiction.
15.7 Prove that if W = C \ {0, 1} then Green’s function gW does not exist. Prove also that
gW ∗ does exist.
15.8 (a) Prove that a positive harmonic function on a Riemann surface without a Green’s
function is constant. Hint: Follow the proof of Theorem 15.5, except instead of
requiring lim supp→α u(p) ≤ 0 for α ∈ ∂ U0 , require u to be 0 off some compact
set of W. Show ω(p) = supu u(p) is not identically equal to 1, as otherwise the
positive harmonic function will achieve its minimum on W at a point of ∂ rU.
(b) Prove that Green’s function with pole at p0 exists on a Riemann surface W if and
only if there exists a non-constant positive harmonic function on W \ {p0 }. Hint:
Use the proof of (a) and Lindelöf’s maximum principle.
15.9 Prove Montel’s theorem using the modular function. Some care is needed if a sequence
composed with the inverse of the modular function converges to a constant with abso-
lute value 1 since there are copies of C \ {0, 1} arbitrarily close to some points on the
unit circle.
15.10 (a) Suppose ± is a bounded region in C bounded by finitely many analytic Jordan
curves. Paste two copies of ± together along the boundaries. Give the resulting
surface a natural structure as a Riemann surface (called the “double” of ±) by
describing the coordinate charts and maps.
15.5 Exercises 245
(b) With ± as in part (a), suppose J is a finite union of open arcs on ∂ ± with pairwise
disjoint closures. Use a similar construction as in (a), attaching the two copies of
± along J to solve the mixed Dirichlet–Neumann problem: ²u = 0 in ±, u = f1
on ∂ ± \ J and ∇ u · n = 0 on J, where n = n(ζ ) is the unit normal to ∂ ± at ζ , and
f1 is continuous. Hint: See Exercise 8.9.
15.11 Cut out a finite number of equilateral triangles from a sheet of paper and paste them
together along their edges. The resulting surface in R3 may not sit in the plane if there
are fewer or more than six triangles meeting at a point. Construct a Riemann surface
on the union of the triangles which agrees on the interiors of each triangle with the
topology it inherits as a flat Euclidean triangle. If the surface is topologically a sphere
then describe the conformal image on the sphere: what kind of curves are the edges,
and how do they meet at a vertex?
15.12 Prove that there is no finite properly discontinuous group of LFTs mapping D onto D,
except the group containing just the identity map. Do the same for maps of C onto C.
Hint: See Exercise 15.1.
15.13 Suppose D1, . . . , Dn are disjoint open disks whose boundaries are orthogonal to R and
contained in {z : Rez > 0}. Let ± = {z : Rez > 0 and Imz > 0} \ ∪nj=1 Dj , let ϕ be a
conformal map of ± onto the upper half-plane H which fixes ∞ , and let E = ϕ(± ∩ R).
Reflect ± about each bounding circle (including the imaginary axis), creating a larger
region bounded by circles orthogonal to R. Then continue reflecting about the newly
created circles, and continue doing so as much as possible.
(a) Show that these reflections fill the upper half-plane.
(b) Show that H is conformally equivalent to the universal covering surface of C \ E
using the Schwarz reflection principle.
(c) Prove that the generators of the group of deck transformations are given by
reflection about the imaginary axis followed by reflection about ∂ Dj for j =
1, . . . , n.
(d) If E is a compact subset of R then R \ E = ∪∞ j =1Ij , where Ij are disjoint open
intervals. It is known that there is then a conformal map ψ of H onto H \ ∪Dj ,
where Dj are closed disks orthogonal to R such that ϕ(Ij ) = Dj ∩ H. Use this fact
to construct the universal cover and covering map of C \ E.
16 Meromorphic Functions on a Riemann
Surface
In this chapter we will study properly discontinuous groups of linear fractional transforma-
tions on the plane and on the disk, and give an introduction to the corresponding function
theory on surfaces covered by the plane and disk.
16.1 Existence of Meromorphic Functions
Not all Riemann surfaces admit non-constant analytic or harmonic functions. If W is a com-
pact Riemann surface then there are no non-constant analytic or harmonic functions on W by
the maximum principle. However, Theorem 16.1 shows that there are plenty of meromorphic
functions on a Riemann surface.
Theorem 16.1 If W is a Riemann surface with distinct points p1 , p2, p3 ∈ W then there is a
meromorphic function f on W with f (p1) = 0, f (p2) = 1 and f (p3 ) = ∞.
Proof Let G(p, p1 , p2 ) and G(p, p3, p2 ) be dipole Green’s functions on W as in Lemma 15.11.
If zα : Uα → D is a coordinate map, let u(ζ ) = G(z− α
1(ζ ), p , p ). Then u − iu is analytic
1 2 x y
on D1 = zα (Uα \ {p1 , p2}) with a simple pole at zα (p1) and residue −1, if p1 ∈ Uα , because
u(ζ ) + log | ζ − zα(p1 )| is harmonic in a neighborhood of zα (p1). Similarly ux − iuy has a simple
pole at zα (p2) with residue 1, if p2 ∈ Uα. Likewise, if v(ζ ) = G(z− α (ζ ), p 3, p 2 ) then vx − ivy
1
is analytic on D2 = zα (Uα \ {p3, p2 }) with a simple pole at zα(p3 ) and residue −1, if p3 ∈ Uα ,
and a simple pole at zα (p2) and residue 1, if p2 ∈ Uα . Then the function
F = uvx −− iuivy
x y
is analytic on zα (Uα \ {p1, p2 , p3}) with a zero of order at least 1 at zα (p1), a pole of order at
least 1 at zα(p3 ), and a removable singularity at zα(p2 ). In fact, the value at zα(p2 ) is 1 because
the residues of ux − iuy and vx − ivy are both 1 at zα(p2 ). Set f = F ◦ zα on Uα . Observe
that the definition of F does not depend on the choice of the chart map zα by the chain rule
because the transition functions are analytic. Thus f is a well-defined meromorphic function
on all of W.
16.2 Properly Discontinuous Groups on C∗ and C

As defined in Chapter 14, a group G of LFTs on W ∗ = C∗ , C or D is properly discontinuous
if each z ∈ W ∗ has a neighborhood B∗ so that τ (B ∗) ∩ B∗ = ∅ for all τ ∈ G \ G0, where
16.2 Properly Discontinuous Groups on C∗ and C 247
G0 contains only the identity map. By Theorem 15.14, every Riemann surface is given by a
properly discontinuous group of LFTs on C∗ , C or D.
Proposition 16.2 The only properly discontinuous group of LFTs on C∗ is the group G0
with one element, the identity map.
Proof If σ (z) = (az + b)/(cz + d) then there is a solution to σ (z) = z in C∗ . But if the group
is properly discontinuous then any element which is not the identity cannot have a fixed point.
Thus σ (z) ≡ z, and Proposition 16.2 follows.
For the remainder of this chapter we will assume G is a properly discontinuous group of
LFTs on C. Let Z denote the integers.
Theorem 16.3 The properly discontinuous groups G on C are

(i) the group with one element G0 = {z};
(ii) Gb = {z + nb : n ∈ Z }, where b ∈ C, b ± = 0;
(iii) Gb,c = {z + nb + mc : n, m ∈ Z}, where b, c ∈ C with b/c not real.
Proof If σ (z) = (az + b)/(cz + d) is a map of C onto C then it cannot have a pole and so
we can write σ (z) = az + b for some constants a, b. But if a ± = 1 then σ has a fixed point in
C. Since the group is properly discontinuous, we must have a = 1. So each σ (z) = z + b is
determined by its value at 0, since b = σ (0).
Suppose G ± = G0 . Set r = infσ ∈G\G0 |σ (0)|. If | σn(0)| → r, then σn(0) has a cluster point.
The values of elements of G at 0 form an additive group, so that the difference of two values at
0 is in this additive group. Because the group is properly discontinuous, this difference cannot
be arbitrarily small. Thus there is a b ∈ C so that z + b ∈ G and
|b| = r = σ
inf
∈ G\G0
|σ (0)|. (16.1)
If G ± = Gb = {z + nb : n ∈ Z}, then choose τ = z + c ∈ G with

|c| = |τ (0)| = inf{|α(0)| : α ∈ G \ Gb }. (16.2)
If σ (z) = z + b and τ (z) = z + c are chosen to satisfy (16.1) and (16.2), and if c/b ∈ R, then
|c/b − n| ≤ 1/2 for some integer n. But then |c − nb| ≤ |b|/2 and so δ = τ − nσ ∈ G satisfies
|δ(0)| < |σ (0)|. By our choice of σ , we must have c − nb = 0, contradicting our choice of c.
We conclude c/b is not real.
Now suppose α ∈ G, α (z) = z + d. Since c/b is not real, we can find real numbers t1 ,
t 2 so that d = t 1b + t2 c. Choose integers m, n so that | t1 − m| ≤ 21 and |t 2 − n| ≤ 12 . Set
a1 = d − (mb + nc). Then
|a1 | = |(t1 − m)b + (t2 − n)c| ≤ 21 |b| + 21 |c| ≤ |c|. (16.3)
But by (16.2), if a1 is not an integer multiple of b then | c| ≤ |a1| , so each of the inequalities in
(16.3) must be equalities. But then |b| = |c|, | t1 − m| = 21 and | t2 − n| = 21 . If t1 − m = 21 and
t 2 − n = 21 then a1 is the midpoint of the line segment between b and c, and hence | a1 | < | c|.
By our choice of b and c, a1 must be an integer multiple of b, in which case d ∈ Gb,c . The
248 Meromorphic Functions on a Riemann Surface
same conclusion holds if t 1 − m or t 2 − n is negative since then a1 is the midpoint of the

segment from either b or −b to either c or −c. We conclude that (iii) holds. This proves
Theorem 16.3.
If G = {z + nb : n ∈ Z}, let
S = {z : |z| < |z − nb| for all non-zero integers n}.
If S 1 = {z ∈ C; |Rez| < 1/ 2}, then S is the rotated and dilated strip bS 1. The analytic function
= e2 iz b
π (z)
π /
maps the plane onto the punctured plane C \ { 0}. It is a conformal map of S onto the slit plane
C \ (−∞, 0]. The map π identifies the two edges of S, mapping each of them to (−∞, 0].
Moreover, π (z) = π (w) if and only if (z − w)(2π/b) = 2π n if and only if z = τ (w) for some
τ ∈ G. The function π maps each translate of S by an integer multiple of b onto C \ (−∞, 0],
and maps the boundary of such translates onto (−∞, 0]. A closed curve γ ⊂ C \ {0} can
be lifted, using π −1 defined locally, to a curve γ ∗ ∈ C. The curve γ ∗ will be closed if and
only if γ is homotopic to a constant curve. Thus C is conformally equivalent to the universal
covering surface of C \ {0} and C/G is conformally equivalent to C \ {0}. The map π is the
projection map of the covering surface C onto the Riemann surface C \ {0}. In this context,
Theorem 15.14 says that g is entire and satisfies g(z + b) = g(z) for all z ∈ C if and only
if g(z) = f (e2π iz/b ) for some f which is analytic on C \ {0}. A similar statement holds for
meromorphic, harmonic and subharmonic functions.
Suppose now that Gb,c = {z + nb + mc : n, m ∈ Z}, where b, c ∈ C with b/c not real,
such that (16.1) and (16.2) hold. Let P be the parallelogram with vertices 0, b, c and b + c.
The elements of the group G translate P to cover the plane. The Riemann surface C/Gb,c is
topologically a torus obtained by identifying the opposite sides of P , and C is conformally
equivalent to the universal covering surface of C/Gb,c . See Exercise 14.8.
Theorems 16.3 and 15.14 yield the following.
Corollary 16.4 The plane is a universal covering surface of the plane, the punctured plane
and tori. The sphere is a universal covering surface of the sphere. Each Riemann surface not
conformally equivalent to one of these has the unit disk as a universal covering surface.
16.3 Elliptic Functions
In this section we will explore function theory on a torus. Fix a properly discontinuous group
G on C generated by σ (z) = z + b and τ (z) = z + c with c/b ∈/ R, such that (16.1) and (16.2)
hold. The parallelogram P with vertices 0, b, c and b + c is called a fundamental domainfor
the group G. For convenience in the following, we let P0 denote P together with two adjacent
edges:
P0 = {tb + sc : 0 ≤ t < 1 and 0 ≤ s < 1}.
Thus every point in C is equivalent in a point in P0 , and no two points in P 0 are equivalent.
If g is meromorphic in C and g(z + b) = g(z + c) = g(z), then g is called an elliptic
function. By Theorem 15.14, g is an elliptic function if and only if g ◦ π −1 is a meromorphic
16.3 Elliptic Functions 249
function on the torus C/G, where π is the quotient map. The zeros and poles of an elliptic
function are isolated, so there can be at most finitely many zeros and poles in P0 .
Lemma 16.5 If g is elliptic and has no poles, then g is constant.
Proof Every point in C is equivalent to a point in the compact set P . But g has a maximum
on P and therefore in C. By the maximum principle, g is constant.
Lemma 16.6 If g is elliptic then the number of zeros of g in P 0 equals the number of poles
of g in P 0, counting multiplicity. Thus g ◦ π −1 is an n-to-one (counting multiplicity) map of
C/ G onto C∗ , where n is the number of poles of g in P .
Proof Let Pε = P − ε(b + c). Then, for ε > 0 sufficiently small, g has no zeros or poles
on ∂ P ε . So Pε contains exactly one point from each equivalence class of zeros and poles.
Because g(z + b) = g(z + c) = g(z), it is also true that g² (z + b) = g² (z + c) = g²(z). So
±
g² (z)
dz = 0,
∂ Pε g(z)
because the integrals on opposite edges cancel. By the argument principle, the number of
zeros of g in P ε equals the number of poles of g in P ε . By the definition of P0 , this statement
holds for P0 as well.
Observe that if c ∈ C then the poles of g − c are exactly the poles of g, and hence the
number of zeros of g − c in P0 equals the number of poles of g in P0. This proves the last
statement of Lemma 16.6.
Exercise 14.9 is the analog of Lemma 16.6 for analytic functions mapping the sphere into
the sphere.
Lemma 16.7 Suppose that g is elliptic, that z 1, . . . , zn are the zeros of g in P0 and that
p1 , . . . , pn are the poles of g in P0 , where zeros and poles are repeated in the list according to
multiplicity. Then
n
²
(zj − pj ) = nb + mc ∈ {σ (0) : σ ∈ G},
j=1
for some n, m ∈ Z.
Proof Let Pε be the translate of P in the proof of Lemma 16.6. By the residue theorem,
±
zg² (z) ²n
dz = 2π i (zj − pj ). (16.4)
∂ Pε g(z)
j=1
Note that if I is an edge of ∂ P ε then g(I) is a closed curve with winding number n =
³ ² ² ² ²
I g (z)/g(z)dz/(2π i). But (z + b)g (z + b)/g(z + b) = zg (z)/g(z) + bg (z)/g(z). Similarly
² ² ²
(z + c)g (z + c)/g(z + c) = zg (z)/g(z) + cg (z)/g(z). Thus (16.4) equals
± ±
g² (z) g²(z)
−b g(z)
dz − c
g(z)
dz = b 2π in + c 2π im,
I J
where I and J are adjacent edges of ∂ Pε and n, m ∈ Z. Lemma 16.7 follows.
For example, if b/ c is not real and G = {nb + mc : n, m ∈ Z}, then Weierstrass’s P function,
1 ² 1
P (z) = + − ω12 ,
z2
ω ∈G \{0} (z − ω)2
satisfies P(z + b) = P (z + c) = P (z) and thus is elliptic. See Example 11.5. This function
has a double pole at the equivalents of 0 and no other poles in P0. By Lemma 16.6, P a is
two-to-one map of P0 onto C∗ , counting multiplicity.
Theorem 16.8 Suppose b/ c is not real and G = {nb + mc : n, m ∈ Z}. Then

´ µ z¶ + 21 z22
=z 1−
z
σ (z) eω ω
ω
ω ∈G\{0}
converges and is entire. Moreover, there exist t1 , t2 ∈ C such that
σ (z + b) = −σ (z)et (z+ 1
b
2) and σ (z + c) = −σ (z)e
t (z+ 2
c
2) . (16.5)
Proof In Example 11.5 we proved that there are at most Ck points ω ∈ G satisfying k ≤
|ω| < k + 1. Thus
² 1
< ∞.
∈ G\{0} | ω|
3
ω
By Theorem 11.11, the infinite product in Theorem 16.8 converges to an entire function.
Moreover,
²
σ (z) 1 ² · 1 1 z
¸
σ (z)
=z+ z−ω
+ ω + ω2
ω ∈ G\{0}
σ²
converges. But then −( σ )² = P by Example 11.5. By the invariance of P ,
· ² ¸² · ² ¸² · ² ¸²
(z + b) = (z + c) =
σ σ σ
(z).
σ σ σ
By an integration, it follows that

² ²
(z + b) = (z) + t1 ,
σ σ
σ σ
for some constant t1 , and hence
σ (z + b) = σ (z)det z , 1
for some constant d. We are allowed to rearrange the terms in the infinite product rep-
resentation of σ because it converges absolutely. Multiplying the terms involving ω and
−ω, if ω ±= 0, gives even functions. Thus σ is odd, and σ (b/ 2) = σ (−b/2)de−t1b/2 =
−σ (b/ 2)de−t1b/2 . Thus d = −et1b/2. This proves the first equality in (16.5), and the second
follows similarly.
16.4 Fuchsian Groups 251
The next theorem gives a complete characterization of elliptic functions.
Theorem 16.9 Suppose b/c is not real. Choose z1 , . . . , zn, p1 , . . . , pn

∑n ∈ C satisfying
j =1(zj − pj ) = 0. Let D be a constant, then
g(z) =D
ń
− zj )
σ (z
(16.6)
σ (z − pj )
j=1
is an elliptic function with poles {pj + kb + mc : k, m ∈ Z} and zeros {zj + kb + mc : k, m ∈ Z},

j = 1, . . . , n, and g(z + b) = g(z + c) = g(z), for all z ∈ C. Conversely, every elliptic function
satisfying g(z + b) = g(z + c) = g(z) for all z ∈ C can be written in this way.
Proof Using the notation of Theorem 16.8,
g(z + b) = D
−σ (z − zj )et (z−z + ) = g(z)et ∑(p −z ) = g(z).
ń 1 j
b
2
1 j j
t (z −p + )
j=1 −σ (z − p j )e
b
1 j 2
Similarly g(z + c) = g(z).

Now suppose h is elliptic with h(z + b) = h(z + c) = h(z). If the zeros of h in P0
are z1 , . . . , zn , and poles in P0 are p1, . . . , pn, repeated according to multiplicity, then, by
∑
Lemma 16.7, (zj − pj ) = kb + mc for some k, m ∈ Z. If we replace z1 by z1 − kb − mc, then
∑
(zj − pj ) = 0. Now form g as in (16.6). Then g has exactly the same zeros and poles as h so
that g/h is elliptic without zeros or poles. By Lemma 16.5, h = Dg for some constant D.
The Weierstrass function P is even so there exists α ∈ P0 with P (α) = 0 and P (b+ c−α ) =
P (−α ) = 0. It follows from Theorem 16.9 that the Weierstrass P function can be written in
the form
σ (z − α)σ (z + α)
P (z) = D σ (z)2
,
where σ is the infinite product in the statement of Theorem 16.8, and D is a constant. Since
z2 P (z) → 1 and σ (z)/z → 1 as z → 0, we must have 1 = Dσ (−α )σ (α). The first explicit
formula for α was found in 1981.
16.4 Fuchsian Groups
A properly discontinuous group of LFTs of D onto D is called a Fuchsian group. The metric
used for function theory on D is called the pseudohyperbolic metric, and it is given by
¹ ¹
¹ z− w ¹
ρ (z, w) = ¹ ¹
¹ 1 − wz ¹ .
The invariant form of Schwarz’s lemma says that an analytic function f mapping D into D is
a contraction in this metric: ρ ( f (z), f (w)) ≤ ρ (z, w), with equality if and only if f is an LFT
of D onto D. See Exercises 3.9 and 6.8 for more information on this metric.
If G is a Fuchsian group then the normal fundamental domainfor G is
F = {z ∈ D : ρ (z, 0) < ρ(z, σ (0)) for all σ ∈ G \ G0 },
where G0 is the identity map. Every w ∈ D can be written as w = σ (z) for some z ∈ F and
σ ∈ G because ρ (w, σ (0)) = ρ (σ
−1 (w), 0) by Exercise 3.9(a). If σ (0) is a closest equivalent of
−
0 to w, then z = σ (w) ∈ F . Since G is properly discontinuous, F contains a neighborhood
1
of 0. If ρ (z, 0) = ρ (z, σ (0)), then, by Exercise 3.9(b),

|1 − σ (0)z|2 = 1 − |σ (0)|2,
so that
¹ ¹2
¹ 1 ¹ 1
¹
¹ σ (0) − z¹¹ = |σ (0) |2 − 1.
º that z lies on the circle centered at 1/σ (0), the reflection of σ (0) about ∂ D, with
This means
|2 − 1. Thus the points equidistant between 0 and σ (0) in the pseudohyperbolic
1
radius |σ (0)
metric lie on a circular arc Cσ which is orthogonal to ∂ D. See Exercise 16.3 and Figure 16.1.
If the closed disk bounded by this circle is denoted by Dσ then
F = D \ ∪ ∈G\G0 Dσ σ .
If σ (z) = λ( 1z−−az
a
), with |λ| = 1 and | a| < 1, then σ (0) = −λa and σ −1(0) = a so that
|σ (0)| = |σ −1 (0)|. Thus the circles associated with σ (0) and σ −1 (0) have the same radius and
their centers have the same absolute value.
Note also ρ (z, 0) = ρ (z, σ −1 (0)) if and only if ρ (σ (z), σ (0)) = ρ (σ (z), 0) by Exer-
cise 3.9(a). Thus σ maps the circular arc C σ −1 onto the circular arc Cσ . By Exercise 15.1,
the map σ is then given by reflection about the diameter, which lies halfway between σ (0)
and σ −1(0), followed by reflection about the circular arc Cσ . This implies that a point of
intersection in D of the two circles is a fixed point of the map σ . Properly discontinuous maps
have no fixed points so the two circles cannot intersect in D. Circular arcs associated with
σ , τ ∈ G, with σ ± = τ −1, however, can intersect in D. The boundary of F then consists of a
subset of ∂ D together with subsets of circles orthogonal to ∂ D. Because the group is properly
discontinuous, there can be only finitely many σ (0) in |z| < r < 1, and hence the subsets of
these circles are subarcs, with at most finitely many arcs in | z| < r for each r < 1. The arcs
are identified in pairs by elements of G.
Note that the equalities σ (0) = −λa and σ −1 (0) = a show that we can recover the LFT
σ from subarcs of C σ and Cσ −1 by geometrically locating the centers of the corresponding
circles.
We have established the following result.
Figure 16.1 On C σ , ρ (z, 0) = ρ (z, σ (0)).

Figure 16.2 Normal fundamental domain.
Theorem 16.10 Suppose G is a Fuchsian group with normal fundamental domain F . Then
»
D= F ∩ D).
τ(
τ ∈G
Moreover, ∂ F ∩ D consists of arcs of circles orthogonal to ∂ D which are identified in pairs
C σ ∩ ∂ F , Cσ −1 ∩ ∂ F by elements σ ∈ G. The Riemann surface D\ G is conformally equivalent
to the Riemann surface obtained from F by this identification. See Figure 16.2.
If
G1 = {σ ∈ G : D ∩ C ∩ ∂ F ±= ∅},
σ
then the subgroup of G generated by G1 is still properly discontinuous and has the same
normal fundamental domain and thus corresponds to the same Riemann surface. Thus G is
the smallest group containing G1 . The normal fundamental domain gives us a geometric way
of visualizing the corresponding Riemann surface and the associated fundamental group (see
Figure 16.2).
16.5 Blaschke Products and Convergence Type
Theorem 15.14 shows that function theory on a Riemann surface W covered by D can be
transferred to function theory on D. In this section we will illustrate this principle by con-
structing a Green’s function, solving the Dirichlet problem and building analytic functions
using the corresponding Fuchsian group on D.
Not all Riemann surfaces covered by the unit disk have a Green’s function. For example,
W = C\{0, 1} does not have a Green’s function. Indeed, each v in the Perron family for g(z, p0 )
is bounded in | z| < r by sup{|z|= r} g(z, p0) for r sufficiently small by the maximum principle.
Thus gW (z, p0 ) extends harmonically to a neighborhood of 0. Similarly, g(z, p0) extends har-
monically to a neighborhood of 1 and ∞, and hence −g(z, p0) would have a maximum on
C∗ \ {p0}, which is impossible.
However, it is possible to tell if a Green’s function for a Riemann surface W exists directly
from the corresponding Fuchsian group.
Theorem 16.11 Suppose W is a Riemann surface with universal covering surface conform-
ally equivalent to D, and suppose π : D → W is a universal covering map. Let G be the
corresponding Fuchsian group. Then Green’s function gW exists if and only if
²
1 − |σ (0)| 2 < ∞. (16.7)
σ ∈G
If (16.7) holds then the Blaschke product
· ¸
B(z) =
|
´σ (0)| z − σ (0)
=
´ |σ −1 (0)| σ −1(z)
∈G −σ (0) 1 − σ (0)z
σ σ −1 ∈G σ −1 (0)
converges and
gW (π (z), π (0)) = − log |B(z)| .
If (16.7) holds, then we say that the Fuchsian group G is of convergence type. Otherwise,
we say that G is of divergence type.
Proof If gW exists, then, by Theorem 15.6,

¹ ¹
² ¹ z − σ (0) ¹
gW (π (z), π (0)) = − log ¹
¹ ¹. (16.8)
σ ∈G 1 − σ (0)z ¹
Because limx↑1 −1−log

x2
x
= 21 , the sum in (16.8) converges if and only if
¹ ¹
² ¹ z − σ (0) ¹2
1−¹
¹ ¹ < ∞. (16.9)
σ ∈G 1 − σ (0)z ¹
Setting z = 0, we obtain (16.7). By Theorem 11.19, the Blaschke product B(z) in the statement
of Theorem 16.11 converges, and gW (π (z), π (0)) = − log | B(z)|.
Conversely, if (16.7) holds, then, by Theorem 11.19, the Blaschke product B converges
uniformly and absolutely on compact subsets of D and vanishes precisely at {σ (0) : σ ∈ G}.
Thus G(z) = − log |B(z)| is harmonic in D \ { σ (0) : σ ∈ G}. Moreover, B ◦ σ has exactly the
same zero set as B so that | B ◦ σ | = |B| by Theorem 11.20. Thus G ◦ σ = G for all σ ∈ G.
Then H = G ◦ π −1 is a well-defined function which is harmonic and positive on W \ {π (0)}
with the property that H(w) + log |w − π (0)| is bounded in a neighborhood of π (0). So if v is
a subharmonic function in the Perron family for gW (w, π (0)) then v ≤ H. By Theorem 13.3,
gW exists.
We remark that, by Theorem 11.20, B ◦ σ = cσ B, where cσ is a constant of absolute

value 1. But because the composition will have different convergence factors for the linear
transformations in the product, and we cannot necessarily rearrange these convergence factors,
the constants cσ are not necessarily equal to 1.
For the second illustration we give an explicit solution to the Dirichlet problem for Riemann
surfaces contained in a larger Riemann surface. Theorem 16.12 below remains true under less
restrictive hypotheses, but we will be content with the statement below to illustrate the idea.
If τ (z) = λ(z − a)/(1 − az) then
² (z) − |a|2 .
τ
τ (z)
= (z −1a)(1 − az)
If z = eit , then
zτ ² (z)
= |1eit−−|aa||2 = Pa(eit ),
2
τ (z)
where P a is the Poisson kernel on ∂ D for the point a = τ −1(0) ∈ D. We can use this to give a
solution to the Dirichlet problem on nicely bounded Riemann surfaces.
Theorem 16.12 Suppose W is a Riemann surface such that W is a compact subset of a

Riemann surface W 1 with W1 \ W ± = ∅. Then D is a universal covering surface of W. Let π
be a universal covering map π : D → W and let G be the corresponding Fuchsian group.
Suppose that the boundary of the normal fundamental domain F consists of finitely many arcs
on circles orthogonal to ∂ D and finitely many arcs on ∂ D, none of which are a single point.
Suppose also that π extends to be a continuous map on F . If f is continuous on ∂ W then set
±
dt
u(z) = P z (eit )f (π (eit )) . (16.10)
∂ D 2π
Then u ◦ τ = u for all τ ∈ G and h(w) = u( π −1 (w)) is a well-defined harmonic function on
W which is continuous on W and satisfies h = f on ∂ W. Moreover, we can write
± ¼ ½
² dt
u(z) = Pσ (z) (e ) f ( π (eit ))
it
(16.11)
Fb
∈G σ
2π
²± dt
= Pσ (z) (eit )f (π (eit )) , (16.12)
σ ∈ G Fb 2π
where Fb = ∂ F ∩ ∂ D.
Equality (16.12) says that u(z) is the sum of the harmonic extensions of χ Fb ( f ◦ π ) at the
equivalents of z, where χ Fb is the characteristic function of Fb. If W ⊂ C∗ such that C∗ \ W
consists of finitely many compact sets, each of which contains more than one point, then,
by Exercise 14.11, the corresponding Fuchsian group is finitely generated and satisfies the
hypotheses of Theorem 16.12. See Exercise 16.4.
Proof Because W cannot be the sphere, the plane, the punctured plane or a torus, W must
be covered by the disk, by Corollary 16.4. A short computation using Exercise 3.9(b) shows
that
eit τ ² (eit )
Pz ( τ (eit ))
τ (eit )
= Pτ −1(z)(eit).
Changing variables eit = τ (ei ) in (16.10) yields
θ
±
eiθ τ ² (eiθ ) dθ
u(z) = P z (τ (eiθ ))f (π (eiθ ))
τ (ei θ ) 2π
±∂ D
= Pτ −1(z)(eiθ )f (π (eiθ )) 2dπθ = u(τ −1 (z)).
∂D
Thus there is a harmonic function h on W so that h( π (z)) = u(z). As shown in the proof of
Theorem 16.10, each σ ∈ G is given by reflection about the diameter equidistant between
σ (0) and σ −1(0), followed by reflection about the orthogonal circle Cσ . Since Fb consists
of finitely many arcs, none of which is a single point, the set E = ∪σ ∈G σ (Fb ) contains a
neighborhood of Fb in ∂ D. This implies that f ◦ π is continuous in a neighborhood of Fb . By
Schwarz’s theorem, u extends to be continuous and equal to f ◦ π on Fb . Since π (Fb ) = ∂ W,
h extends to be continuous on W with h = f on ∂ W.
We claim that the set E has length 2π . Note that if ζ ∈ σ (Fb) ∩ τ (Fb) with σ , τ ∈ G, then
ζ1 = σ
−1(ζ ) ∈ Fb and ζ2 = τ −1(ζ ) ∈ Fb , with α (ζ1 ) = ζ2 , where α = τ −1 ◦ σ ∈ G. But
this only occurs if ζ 1 is an endpoint of one of the finitely many intervals in Fb . Set
±
1
ω(z) = 2π
P z (eit )dt.
E
Then ω (τ (z)) = ω(z) for all z ∈ D by the change of variables argument used above. Moreover,
by Schwarz’s theorem, ω = 1 on a neighborhood of Fb in ∂ D, and hence ω ◦ π −1 (w) is
a well-defined harmonic function on W which equals 1 on ∂ W. By the maximum principle,
³ dt
ω ◦ π −1 (w) = 1 for all w ∈ W. In particular, ω (0) = E = 1. This proves that E has length
³ 2π
2π . In particular, ∂ D \E Pz (eit )f (π (eit ))dt = 0. Thus (16.11) holds, and (16.12) follows by the
change of variables argument presented above.
It is harder to represent analytic functions on a Riemann surface because of the diffi-

culty dealing with harmonic conjugates. However, we have the following. Differentiating the
Blaschke product B in Theorem 16.11 logarithmically,
B ² (z) ² τ ² (z) ² 1 − |τ (0)|2
= = . (16.13)
B(z)
τ ∈G τ (z) τ ∈G (z − τ (0))(1 − τ (0)z)
Locally on D \ {σ (0) : σ ∈ G} we can find an analytic function F such that ReF =
gW (π (z), π (0)). Equation (16.13) is an explicit expression for F ² (z) which is meromorphic
in D with simple poles at the equivalents of 0, and residue 1. If W satisfies the hypotheses in
Theorem 16.12, then the zeros of B, which are the equivalents of 0, do not accumulate on Fb,
so that B and B ² extend to be analytic across a neighborhood of Fb in ∂ D. See Exercise 11.4.
Moreover,
¹
zB ²(z) ¹¹ ²
B(z) ¹z =eit
= Pτ (0) (eit ) > 0
τ ∈G
on Fb.
Theorem 16.13 Suppose W, π and G satisfy the hypotheses in Theorem 16.12. If f is

bounded and analytic on D, then set
²
E( f )(z) = f ◦ τ (z)τ ² (z) τB(z) ²
1
(z) B (z)
. (16.14)
τ ∈G
Then E( f ) is analytic on D \ {z : B ²(z) = 0} and satisfies E( f ) ◦ τ = E( f ) for all τ ∈ G. The
function H ≡ E( f ) ◦ π −1 is a well-defined meromorphic function on W with poles only at the
critical points of gW (z, π (0)). Moreover, if h is meromorphic on W then E(h◦π ·f ) = h◦π ·E( f ),
and, since E(1) = 1, E(h ◦ π ) = h ◦ π . Finally,
lim sup |H(w)| ≤ lim sup | f (z)| .
w→∂ W z→∂ D
16.6 Exercises 257
Theorem 16.13 says that the map f → E( f ) ◦ π − 1 is a projection from the bounded analytic
functions on D onto the meromorphic functions on W bounded near ∂ W with poles only at
the critical points of gW .
Proof The analyticity of E( f ) follows from the infinite product representation of B given in
Theorem 16.11. The Blaschke product B is not necessarily invariant under the Fuchsian group
G, but the zeros of the Blaschke product B are invariant. So
B◦τ =c τ B, (16.15)
where cτ is a constant with absolute value 1. A short computation then proves that E( f ) ◦ τ =
E( f ) for all τ ∈ G. The critical points of gW (w, π (0)) are the places where its gradient van-
ishes. By the Cauchy–Riemann equations and Theorem 16.11, the critical points correspond
precisely to the zeros of B ² . The identity E(h ◦ π · f ) = h ◦ π E( f ) follows from the chain rule
and (16.15) because π ◦ τ = π . The identity E(1) = 1 follows from (16.13). To prove the final
statement of Theorem 16.13, let wn ∈ W → ζ ∈ ∂ W. Choose zn ∈ F with π (zn) = wn . Then
zn → α ∈ Fb. But the sum in (16.14) converges uniformly on Fb since B² and B are analytic
across Fb. Moreover, zτ ² (z)/τ (z) > 0 and B(z)/(zB ² (z)) > 0 on a neighborhood of Fb in ∂ D.
Thus
lim sup H(w) ≤ lim sup | f (z)| · E(1) = lim sup | f (z)|.
w→∂ W z →∂ D z →∂ D
Note that the zeros of B ² in Theorem 16.13 do not accumulate on Fb and hence gW has
only finitely many critical points in W. The function gW is not necessarily differentiable on
∂ W ⊂ W1 , but this comes from the possible non-differentiability of π , not B, under our
assumptions on W.
16.6 Exercises
16.1 Set G = {z = nb + mc : n, m ∈ Z}. Prove b, c satisfy (16.1) and (16.2) if and only
if − 21 ≤ Re(c/b) ≤ 12 and |c/b| ≥ 1. Draw a picture of the region in which c lies,
given b.
16.2 Find a normal fundamental domain for a torus with one point removed.
16.3 Prove that the circle Cσ in Section 16.3 is orthogonal to ∂ D. Hint: Use the converse of
Pythagoras’s theorem.
16.4 If W ⊂ C∗ such that C∗ \ W consists of finitely many pairwise disjoint compact con-
nected sets, each of which contains more than one point, then, by Exercise 14.11, the
corresponding Fuchsian group is finitely generated. Prove that W satisfies the hypothe-
ses of Theorem 16.12. Hint: First find a conformal map of W onto a region bounded by
finitely many analytic Jordan curves, by applying a finite sequence of conformal maps
of simply-connected regions onto the disk.
16.5 (a) Suppose that b/c is not real and that G = {nb + mc : n, m ∈ Z}. Prove that the
Weierstrass P function satisfies P (z) = z12 + Az2 + Bz4 + . . . near 0, where
² −4 and B = 5 ² ω−6 .
A =3 ω
ω ∈G \{0} ω ∈G\{0}
(b) Prove that (P² )2 − 4P 3 + 20A P = −28B. An equation of the form y2 = x3 + ax + b

is called an elliptic curve. Hint: Use (a) to show that the left-hand side is analytic
and elliptic.
(c) Use (b) to show that P is the inverse of an elliptic integral:
± P (z)
dζ
z − z0 = ¾ .
P (z0) 4ζ 3 − 20Aζ − 28B
The square root and path of integration from P(z0 ) to P (z) must be properly
chosen.
(d) Find the Schwarz–Christoffel integral for a conformal map of the upper half-plane
onto a rectangle such that ∞ is mapped to one of the vertices. Compare your
answer with (c).
16.6 (a) Prove that two tori are conformally equivalent if and only if there is a linear map
which maps one fundamental domain onto the other.
(b) Give a list of fundamental domains, exactly one in each conformal equivalence
class.
16.7 If W is a torus as in Section 16.2, find the dipole Green’s function in terms of the func-
tion σ given in Theorem 16.8. Hint: Use σ to build a function with the right logarithmic
singularities then add ax + by for some a, b with z = x + iy.
16.8 If G is a group of deck transformations associated with a torus W as in Section 16.2,
set F = {z : | z| < | z − σ (0)| : σ ∈ G, σ (z) ± ≡ z}.
(a) Prove that if w ∈ C, then there exists z ∈ F such that w = σ (z), for some σ ∈ G.
(b) Show that F is bounded by four or six line segments.
(c) Prove that the Riemann surface obtained by identifying opposite edges of F is
conformally equivalent to W.
Remark: F is the analog in C of the normal fundamental domains in D.

16.9 Suppose W satisfies the hypotheses of Theorem 16.13 and suppose f is bounded and
analytic on D with the property that it is invariant at the zeros of B² : if B² (z) = 0, then
f ◦ τ (z) = f (z) for all τ ∈ G. Prove that E( f ) is bounded and analytic on D and that
H(w) = E( f ) ◦ π −1(w) is a well-defined bounded analytic function on W.
16.10 Suppose ± ⊂ C is a region bounded by n < ∞ analytic Jordan curves.

(a) Prove that
±
u(z) =
∂ gW (ζ , z)
f (ζ )
|dζ |
∂± ∂ ηζ 2π
16.6 Exercises 259
is the harmonic extension of f ∈ C(∂ ±) to ±, where ηζ is the unit inner normal

at ζ ∈ ∂ ± and gW (z, b) is Green’s function with pole at b. Hint: Use Green’s
theorem or use ∂ gW /∂ s = 0, where s is the arc-length on ∂ ±, ∂ gW /∂ η > 0 and
zB² (z)/B(z) > 0 on Fb .
(b) Prove that, for each b ∈ W, ∇ gW (ζ , b) = (0, 0) has at most n − 1 solutions in ±.
(c) Prove that
±
∂ g W (ζ , b)
log
∂ gW (ζ , b) |dζ | = γ + ²
gw (ζ , b),
b
∂ ηζ ∂ ηζ 2π
∂±
∇ gw =0
ζ:
where ηζ is the unit inner normal at ζ ∈ ∂ ± and gW (z, b) is Green’s function

with pole at b. The constant γb comes from the expansion of g near b: g(z, b) =
− log |z − b| + γb + O(|z − b|), and is sometimes called the Robin constant.
16.11 Suppose G is a group of LFTs τ of D onto D with no fixed points: if τ ∈ G and
τ (a) = a for some a ∈ D, then τ (z) = z for all z ∈ D. Is G necessarily properly
discontinuous? (Prove or give a counter-example.)
Appendix
A.1 Fifteen Conditions Equivalent to Analytic
In the text and in the exercises we encountered several ways to determine when a function is
analytic. As a review, we list some here. For simplicity, we assume the region is the unit disk
D. These statements can be transferred to other disks using linear maps of the form az + b,
a, b ∈ C.
If f is continuous on D then the following are equivalent:
1. For each z0 ∈ D, f has a power series expansion centered at z0 , valid in a neighborhood

of z0. (Definition 2.6)
2. f has a power series expansion centered at 0 with radius of convergence at least 1.
(Theorem 2.7 and Corollary 4.15)
3. f ± (z) exists for each z ∈ D and f ± is continuous on D. (Corollary 4.15)
4. f ± (z) exists for each z ∈ D. (Goursat’s theorem, Exercise 4.12)
±
5. γ f (z)dz = 0 for all curves γ ⊂ D. (Cauchy’s Theorem 5.1 and Morera’s Theorem 4.19)
±
6. ∂R
f (z)dz = 0 for all closed rectangles R ⊂ D, with sides parallel to the axes. (Local
Cauchy’s theorem, Corollary 4.18, and Morera’s Theorem 4.19)
± ∂ϕ
7. D ∂ z fdxdy = 0 for all continuously differentiable functions ϕ with compact support
contained in D. (Weyl’s lemma, Exercise 7.13)
8. There exist rational functions rn with poles in {z : | z| ≥ 1} such that rn converges
uniformly on compact subsets of D to f . (Runge’s Corollary 4.28 and Weierstrass’s
Theorem 4.29)
9. There exist polynomials pn such that pn converges uniformly on compact subsets of D to
f . (Runge’s Theorem 4.27 and Weierstrass’s Theorem 4.29)
10. f = u + iv, where u, v are continuously differentiable real-valued functions satisfying
ux = vy and uy = −vx . (Cauchy–Riemann equations, Theorem 7.9)
11. f = g± for some g analytic on D. (Theorem 2.12 and Corollary 2.15)
12. f = ux − iuy for some u harmonic on D. (Theorem 7.10 and Corollary 2.15)
13. f is continuously differentiable (with respect to x and y) and f preserves angles, includ-
ing direction, between curves except at isolated points. (Section 3.2, Corollary 7.13 and
Corollary 5.10)
With a little more work, this list can be extended to include:

±
14. ∂S
f (z)dz = 0 for all closed squares S ⊂ D, with sides parallel to the axes.
Hint: If R ⊂ D is a closed rectangle with rational side ratio, subdivide R into squares. If
the side ratio is not rational, write ∂ R = ∂ R1 + ∂ R2 , where R1 and R2 are rectangles, R1
has rational side ratio and ∂ R2 has two sides with length < ε. By uniform continuity, the
integrals on the two long sides of R2 almost cancel.
15. For each r < 1 there exists M = M(r) < ∞ so that, for each triple z1 , z2, z3 with |zj | < r,
j = 1, 2, 3, there exists a polynomial p with p(zj ) = f (zj ), j = 1, 2, 3, and |p(z)| ≤ M on
|z| < r.
Hint: One direction follows from Exercise 4.13. For the other direction, find three poly-
nomials pj so that pj (zk ) = 0 if j ² = k, and pj (zj ) = 1. Truncate the power series for f to
∑
get a polynomial p0 which is uniformly close to f on |z| ≤ r. Then use p0 + cj pj with
|cj | small.
Analyticity is a local condition. But there are related global consequences. If f is analytic
on a region ± then
±
1. γ f (z)dz = 0 for all curves γ ∼ 0 (homologous to 0). (Cauchy’s Theorem 5.1)
±
2. γ f (z)dz = 0 for all curves γ ≈ {γ (0)} (homotopic to a constant curve). (Corollary 14.4)
3. There exist rational functions rn with poles in ∂ ± converging to f uniformly on compact
subsets of ±. (Runge’s Corollary 4.28)
4. If z0 ∈ ± then f has a power series expansion centered at z0 with radius of convergence at
least equal to the distance from z0 to ∂ ±. (Corollary 4.15)
Exercises
There are many ways to show a region is simply-connected. Locate results in the text for each
item below or give a proof. Suppose ± ⊂ C is a region, then the following are equivalent to
simple connectivity.
1. C∗ \ ± is connected.
2. The boundary of ± in C∗ is connected.
3. γ ∼ 0 for all closed curves γ ⊂ ±.
4. γ ∼ 0 for all closed polygonal curves γ ⊂ ±.
5. γ ≈ {γ (0)} for all closed curves γ : [0, 1] → ±.
6. γ ≈ {γ (0)} for all closed polygonal curves γ : [0, 1] → ±.
7. ± = C or there is a conformal map f of ± onto D.
8. Every f analytic on ± can be written as f = g± for some g analytic on ±.
9. Every non-vanishing f analytic on ± can be written as f = eg for some g analytic on ±.
10. Every non-vanishing f analytic on ± can be written as f = g2 for some g analytic on ±.
11. Every u harmonic on ± can be written as u = Ref for some f analytic on ±.
A.2 Program for Color Pictures
In this section we give a sample Matlab program for drawing the color pictures from
Chapter 6. A number of examples are also given to illustrate the Matlab notation. The
3D option gives a surface colored according to arg f (z) with height log | f (z)| above each
point z. The axes can be rotated using a mouse. The center and size of the domain can
also be changed with a mouse. For more advanced code, and better pictures, see [27] and
http://www.visual.wegert.com
262 Appendix
function fcn(action)
% Choices: (alternatives follow %)
domain=0; %plane=0, disk=1, ext(disk)=2, UHP=-1, RHP=-2
range=0; % plane=0, disk=1, ext(disk)=2, UHP=-1, RHP=-2
dimension=2; % 2d plot=2 3D plot =3
Size=8; %width of plot (domain)
resol=0.001; %relative grid size of domain (resolution)
modlines=12; % number of modulus lines -1.
mmod=1.8; %mod lines are in range:-mmod \le log|f| \le mmod
arglines=64; % number of argument lines
xcenter=0; % x coordinate of center of plot
ycenter=0 ; %y coordinate of center of plot
% end choices, except for the function, given below.
% The program runs faster with a bigger resolution size, but
% may result in jagged lines if it is too big.
% Warning: the function contour doesn’t work well with discont
% functions. Use NaN to remove pts where not cont. See below.
global Size
global xcenter
global ycenter
global resol
global range
global domain
global dimension
global mmod
global modlines
global arglines
if nargin < 1
fname = mfilename;
callbackStr=[fname ’(’’zoomin’’)’];
uicontrol(...
’Style’,’pushbutton’,...
’Units’,’normalized’,...
’Position’, [.70 .02 .1 .03],...
’Background’,’white’,...
’Foreground’,’black’,...
’String’,’ zoomin’,...
’tag’,’zoomin’,...
’Callback’,callbackStr);
callbackStr=[fname ’(’’zoomout’’)’];
uicontrol(...
’Position’, [.80 .02 .1 .03],...
’String’,’ zoomout’,...
’tag’,’zoomout’,...
callbackStr=[fname ’(’’quit’’)’];
uicontrol(...
’Position’, [.90 .02 .1 .03],...
’String’,’ quit’,...
’tag’,’quit’,...
callbackStr=[fname ’(’’2D/3D’’)’];
uicontrol(...
’Position’, [.60 .02 .1 .03],...
’String’,’ 2D/3D’,...
’tag’,’2D/3D’,...
elseif strcmp(action,’zoomin’)
[xcenter,ycenter]=ginput(1);
Size=Size/2;
elseif strcmp(action,’zoomout’)
Size=Size*2;
elseif strcmp(action,’quit’)
close;
return
elseif strcmp(action,’2D/3D’)
dimension=5-dimension;
end
deltax=resol*Size;
X = -Size/2.+xcenter:deltax:Size/2.+xcenter;
Y=X-xcenter+ycenter;
[x,y]=meshgrid(X,Y);
z = x+1i*y;
%
%FUNCTION CHOICE:
w=(z-3).^2./(z.^2+4);
% Some other functions to try: (put % in front of w in line
% above then remove a % on a line below)
264 Appendix
% you can also change your choice for domain, size, etc.
% w=z;
% w=(z-1i)./(z+1i); %Cayley Transform
% w=(z-1).*(z+2). *(z-i);
% w=exp(z);
% w=log(z);
% w=log(z)+2*pi *1i;
% w=z.^1.5
% w=z.^4;
% w=z.^(1/4);
% w=-z.^3;
% w=exp(z.^3);
% w=tan(z);
% w=(z+1./z)/2;
% w=exp((z+1./z)/2);
% w=.5*z.*(1+sqrt(1-1./z.^2));%
% w=1i*(1i*z).^0.5-.5* 1i;
% w=exp(z+1./z);
% w=(exp(z)+exp(-z))/2;
% w=(w+1i)./(w-1i);
% w=(besselj(0,z).^2-besselj(1,z).^2).* z./sin(2* z);
% w=(besselj(1,z).^2-besselj(3,z).^2).*z./sin(2*z);
% w=1./z-1./sin(z);
% w=besselj(0,z);
%Joukovski
% w=z.* (1+sqrt(1-1./z.^2));
% w=1./z;
% w=z+1i*sqrt(1-z.^2);% add NaN statement below
% w=z-1i*sqrt(1-z.^2);% add NaN statement below
% w(imag(z) < 0)=NaN;
% w(imag(z)==0 & real(z).^2 > 1)=NaN;
% ---Kill where mod is not cont.
% NOTE: if extraneous contours occur, decrease resol.
% Same map as sequence of compositions,
% w=-1i* sqrt((1-z)./(1+z));w=(1+w)./(1-w);
% w(imag(z)==0 & real(z).^2 > 1)=NaN;
% ---Kill where mod is not cont.
% in addition to get arccos using J map:
% w=-1i*log(w);
% w=exp(15./z);
% w=(z-1i/3)./(1+1i*z/3);
% w=sin(z)-z;
% w=cos(z);
% wz=z./(z-(1+1i));
% w=((z.^4-1)./(z.^4+1));
% w=sqrt(z.^2+1);
% w=z./(1-z);
% w=(z.*sqrt(1+1./z.^2));
% w=-1i*log(z+1i*sqrt(1-z.^2));
% w=exp((z+1)./(z-1));
% M.C. Escher:
% the push forward of a polar grid
% by the map exp((1+i)z/2) is the same as the
%pull back of the polar grid by the inverse map
%(1-i)log(w). The push forward of a rectangular grid
% is the same as the pull back of the polar grid by the map
% exp((1-i)logw)= w^{1-i).
% w=exp((1-1i)*log(z));
% w=z.^(1-1i);w(imag(z)==0 & real(z) < 0 )=NaN;
% ----Kills pts where mod is not continuous.
%If we color the plane with a rectangular grid instead
% of polar, then w would have the coloring of:
% w=exp(z);
% w=sqrt(1i* z+.25)-0.5;%exterior of parabola
% w=1i* w;
% w=exp(z);%strip to half disk
% w(abs(w)>1)=NaN;
% w(imag(z)>pi)=NaN;
% w(imag(z)<0)=NaN;
% w=z;w(imag(z)<0)=NaN;
% use formula only if z in domain, w in range
if range==1
w(abs(w)>1)=NaN;
elseif range==2
w(abs(w)<1)=NaN;
elseif range==-1
w(imag(w)<0)=NaN;
elseif range==-2
w(real(w)<0)=NaN;
end
if domain==1
w(abs(z)>1)=NaN;
elseif domain==2
w(abs(z)<1)=NaN;
elseif domain==-1
w(imag(z)<0)=NaN;
elseif domain==-2
w(real(z) < 0)=NaN;
end
ww=angle(w)/pi;
266 Appendix
% fix red color at discontinuity:

ww(imag(w)==0 & real(w) < 0)=-1;
% NEED: axis equal before imagesc; axis square after imagesc
axis equal
if dimension==2
ww(isnan(ww))=2; %NaNs are white
imagesc(X,Y,ww);
view(0,90)
rotate3d off
elseif dimension==3
surf(x,y,log(abs(w)),ww); %NaNs are automatically white
rotate3d on
view(-37.5,30)
end
shading interp
set(gca,’YDir’,’normal’);
caxis([-1,2 *modlines/arglines+1])%color=[-1,1] gray=[1,last]
axis square
colormap([hsv(arglines);gray(modlines)])
hold on
ww2=log(abs(w));
%flip order so small=white:
ww3=(mmod-ww2)* modlines/(mmod*arglines)+1;
if dimension==2
contour(x,y,ww3,1:2/arglines:2* modlines/arglines+1,...
’LineWidth’,1);
contour(x,y,abs(w),[1 1],’-k’,’LineWidth’,2)
elseif dimension==3
contour3(x,y,ww2,[0 0],’-k’)
xlabel(’real’), ylabel(’imaginary’), zlabel(’log-modulus’)
end
hold off
end
Bibliography
[1] Lars V. Ahlfors. Complex Analysis: An Introduction to the Theory of Analytic Functions of One
Complex Variable, 3rd edn. McGraw-Hill Book Co., New York, 1979.
[2] Lars V. Ahlfors. Conformal Invariants: Topics in Geometric Function Theory. McGraw-Hill Series
in Higher Mathematics. McGraw-Hill Book Co., New York, 1973.
[3] Robert B. Burckel. An Introduction to Classical Complex Analysis, Vol. 1. Pure and Applied
Mathematics, vol. 82. Academic Press, New York, 1979.
[4] Henri Cartan. Elementary Theory of Analytic Functions of One or Several Complex Variables.
Éditions Scientifiques, Hermann, Paris; Addison-Wesley Publishing Co., Reading, 1963.
[5] James Clunie, Alexandre Eremenko and John Rossi. On equilibrium points of logarithmic and
Newtonian potentials. J. London Math. Soc., s2-47(2):309–320, 1993.
[6] John B. Conway. Functions of One Complex Variable, 2nd edn. Graduate Texts in Mathematics,
vol. 11. Springer, New York, 1978.
[7] Bart de Smit, Hendrik W. Lenstra, Jr., Douglas Dunham and Reza Sarhangi. Artful mathematics:
the heritage of M. C. Escher. Notices Amer. Math. Soc., 50(4):446–457, 2003.
[8] Alexandre Eremenko, Jim Langley and John Rossi. On the zeros of meromorphic functions of the
∑
form f (z) = ∞ k=1 ak / (z − z k ). J. Anal. Math., 62:271–286, 1994.
[9] Theodore W. Gamelin. Complex Analysis. Undergraduate Texts in Mathematics. Springer-Verlag,
New York, 2001.
[10] John B. Garnett and Donald E. Marshall. Harmonic Measure. New Mathematical Monographs,
vol. 2. Cambridge University Press, 2008. (Reprint of the 2005 original.)
[11] Jeremy Gray. On the history of the Riemann mapping theorem. Rend. Circ. Mat. Palermo (2)
Suppl., 34:47–94, 1994.
[12] Robert E. Greene and Steven G. Krantz. Function Theory of One Complex Variable, 3rd edn.
Graduate Studies in Mathematics, vol. 40. American Mathematical Society, Providence, 2006.
[13] Einar Hille. Analytic Function Theory, Vol. 1. Introductions to Higher Mathematics. Ginn and Co.,
Boston, 1959.
[14] Einar Hille. Analytic Function Theory, Vol. II. Introductions to Higher Mathematics. Ginn and Co.,
Boston, 1962.
[15] Otto Hölder. Ueber den Casus Irreducibilis bei der Gleichung dritten Grades. Math. Ann.,
38(2):307–312, 1891.
[16] Morris Marden. The Geometry of the Zeros of a Polynomial in a Complex Variable. Mathematical
Surveys, no. 3. American Mathematical Society, New York, 1949.
[17] Tristan Needham. Visual Complex Analysis. Oxford University Press, New York, 1997.
[18] University of Washington Mathematics Department. Complex Analysis Preliminary Examinations
1953–2017. University of Washington, Seattle, 2017.
[19] Roger Penrose and Wolfgang Rindler. Spinors and Space-Time, Vol. 1: Two-Spinor Calculus
and Relativistic Fields. Cambridge Monographs on Mathematical Physics. Cambridge University
Press, 1984.
268 Bibliography
[20] Christian Pommerenke. Univalent Functions. With a Chapter on Quadratic Differentials by Gerd
Jensen. (Studia mathematica Band XXV.) Vandenhoeck & Ruprecht, Göttingen, 1975.
[21] Thomas Ransford. Potential Theory in the Complex Plane. London Mathematical Society Student
Texts, vol. 28. Cambridge University Press, 1995.
[22] Walter Rudin. Principles of Mathematical Analysis, 3rd edn. International Series in Pure and
Applied Mathematics. McGraw-Hill Book Co., New York, 1976.
[23] Walter Rudin. Real and Complex Analysis, 3rd edn. McGraw-Hill Book Co., New York, 1987.
[24] Donald Sarason. Complex Function Theory, 2nd edn. American Mathematical Society, Providence,
2007.
[25] Steve Smale. The fundamental theorem of algebra and complexity theory. Bull. Amer. Math. Soc.
(N.S.), 4(1):1–36, 1981.
[26] Kenneth Stephenson. Introduction to Circle Packing: The Theory of Discrete Analytic Functions.
Cambridge University Press, 2005.
[27] Elias Wegert. Visual Complex Functions: An Introduction with Phase Portraits.
Birkhauser/Springer Basel AG, Basel, 2012.
Index
Abel’s limit theorem, 39 chart, coordinate, 221

absolute value, 4 chordal metric, 10
analytic function “circle”, 10, 83
at a point, 20 circle, generalized, 83
between Riemann surfaces, 226 circular triangle, 231
in an open set, 20 complex number
weakly, 121 absolute value, |z|, 4
analytic continuation argument, arg z, 4
along a curve, 217 complex conjugate, z, 5
direct, 216 imaginary part, Imz, 4
annulus, 72 i, (0,1), 4
arc, 43 modulus, |z| , 4
analytic, 125, 138 real part, Rez, 4
geodesic, 125 complex numbers, C, 3
Jordan, 195 addition, 3
one-sided, 134 multiplication, 3
two-sided, 134 components
arc length integral, 46 closed, 60
arg, argument, 4 open, 60
argument principle, 76 conformal map, 93
Arzela–Ascoli theorem, 158 continuous functions
on boundary of region ± , C(∂ ±), 207
barrier, local, 209 on a region ±, C(± ), 157
Blaschke product, 183 convergence
Bloch’s constant, 167 locally uniform, 157
Bloch’s theorem, 167 non-tangential, 39
bootstrap technique, 208 normal, 157
Bouligand’s lemma, 210 quadratic, 29
boundary Harnack inequality, 117 convergence type, 254
branch point, 228 convergent series, 7
Brennan conjecture, 137 coordinate chart, 221
coordinate disk, 221
C, complex numbers, 3 coordinate function, 221
C(∂ ±), 207 coordinate map, 221
C(±), 157 cover-up method, 17
C∗ , extended plane, 9 covering lemma, 59
Carathéodory’s theorems, 130, 202 crosscut, 206
Cauchy integral, 109 curve, 43
Cauchy principal value integral, 146 closed, 43
Cauchy product, 23 Jordan, 195
Cauchy transform, 109 length, ²(γ ), | γ |, 46
Cauchy–Riemann equations, 111 lift, 224
Cauchy–Schwarz inequality, 7 piecewise continuous differentiable, 43
Cauchy’s estimate, 51 simple, 43
Cauchy’s integral formula, 64 sum or union, 45
Cauchy’s theorem, 63 cycle, 46
270 Index
∂D, unit circle, 5 H, upper half-plane, 32

D, unit disk, 5 Hadamard’s three-circles theorem, 114
deck transformation, 226 Hadamard’s three-lines theorem, 116
³, Laplacian, 105 harmonic conjugate, 112
dipole Green’s function, 239 harmonic function, 105
Dirichlet problem, 207 harmonic measure, on a Riemann surface, 235
Perron solution, 209 Harnack’s inequality, 116
Dirichlet–Neumann problem, mixed, 139 Harnack’s principle, 118
“disk”, 83 Harnack’s theorem, 116
disk Herglotz integral, 109
coordinate, 221 Herglotz kernel, 109
hyperbolically strictly convex, 126 Hölder, O., 6
disk-chain, 127 Hölder’s inequality, 116
closed, 127 holomorphic function, 47
dist(∂ ±, K), 54 homologous, to 0, 67
divergence type, 254 homologous curves, 67
divergent series, 7 homotopic curves, 218
dog bone contour, 155 homotopy, 218
double, of a Riemann surface, 244 Hurwitz’s theorems, 129
doubly-connected region, 213 hyperbolic disk, convexity, 126
hyperbolic geodesic, 125
elliptic curve, 258 hyperbolic geometry, 99
elliptic function, 138, 248 hyperbolic metric, 41, 99
elliptic integral, 138
enclosed point, 76 i, complex number (0,1), 4
entire function, 50 identity theorem, 23
equicontinuity index, 65
at a point, 158 inner normal, 68
on a set, 158 integers,
± Z , 247
uniformly, 158 f (z)dz, 44, 59
± γ
error, round-off, 30 γ
f (z)| dz|, 46
essential singularity, 75 integral
Euler’s constant, 184 on a continuous curve, 59
evenly covered set, 224 on a piecewise continuous differentiable curve, 44, 46
exp, exponential function, 28 irregular point, 209
extended plane, C∗ , 9 isolated singularity, 74
ez , exponential function, 28 at ∞, 75
F , 251 Janiszewski’s lemma, 198

Fb , 255 Jensen’s theorem, 181
Fourier series, 110 Jordan arc, 195
Fourier sine series, 110 Jordan curve, 195
Fourier transform, 145 with positive area, 195
Fuchsian group, 251 Jordan curve theorem, 199
fundamental domain, 248 Jordan region, 202
fundamental group, 226, 228 Joukovski map, 89
fundamental theorem of algebra, 15 jump theorem, 109
Jørgensen’s theorem, 125
´(z),184
gamma function, 184 keyhole contour, 148
genus, of a product, 178 Koebe conjecture, 214
geodesic, 99, 125 Koebe’s theorem, 167
geodesic arc, 125
geodesic zipper algorithm, 123 L, Laplace transform, 150
Green’s function, 213, 214 Lakes of Wada, 197
dipole, 239 Landau’s constant, 167
on a Riemann surface, 233 Landau’s theorem, 167
Index 271
Laplace transform, L, 150 P , Weierstrass P function, 174

inverse, 151 Painlevé’s theorem, 71
Laplace’s equation, 105, 112 parking-lot ramp, 220
Laurent series, 72 partial fractions, 16
Legendre’s duplication formula, 190 Perron family, 207
length of a curve, 46 Perron solution, 209
length–area principle, 202 π1 (±, b), fundamental group, 228
LFT, 82 Picard’s great theorem, 166
lift of a curve, 224 Picard’s little theorem, 166, 231
Lindelöf’s maximum principle, 115 Poisson integral, 108
linear fractional transformation, 82 Poisson kernel, 108, 215
Liouville’s theorem, 35 polar grid, standard, 81
local barrier, 209 pole
locally bounded family of functions, 160 of a rational function, 55
locally conformal function, 34 order of, 74
locally connected set, 205 simple, 75
log, logarithmic function, 70 polygonal curve, 60
polygonal tree, 137
MacMillan twist point, 206 polynomial, 13
Marty’s theorem, 162 degree of, 13
maximum principle, 31, 106 linear, 13
mean-value inequality, 106 zero of, 15
positive orientation, 68
mean-value linear theorem fails, 30
mean-value property, 31, 105 power series
Mellin transform, 148 convergent, 18
uniqueness of, 22
Mercator projection, 101
prevertices, 135
meromorphic function, 75
prime number theorem, 191
mesh, 46
principal value integral, 146
Mittag–Leffler’s theorem, 170
product
Möbius transformation, 82
absolute convergence of, 176
modular function, 232
convergence of, 176, 177
modulo 2 π , 5
genus of, 178
modulus, 4
of two curves, 227
monodromy theorem, 219
proper function, 41
Montel’s theorem, 165
properly discontinuous group, 227
Morera’s theorem, 52
pseudohyperbolic metric, 41, 251
multiple-valued function, 70
R, real line, 5
n(γ , a), winding number, 65 radius of convergence, 19
Neumann problem, 139 Rado’s theorem, 242
Newton iterate, 29 rational function, 16, 24
Newton’s method, 29 pole of, 55
approximate, 30 real line, R, 5
non-Euclidean geometry, 99 reflection
non-tangential cone, 38 about a circle, 230
normal family across ∂ D, 137
chordal metric, 160 across R, 5
Euclidean metric, 156 reflection principle, 133
normal fundamental domain, F , 251 region, 22
north pole, 9 doubly-connected, 213
regular point, 209
one pole in each hole, 56 regular region, 209
one-dimensional Hausdorff measure removable set, 72
zero, 71 removable singularity, 74
one-to-one function, 33 reparameterization, 44
one-sided boundary arc, 134 Resa , residue, 141
open map, 33 residue, Resa , 141
272 Index
residue theorem, 141 stereographic projection, 8

Riemann hypothesis, 186 Stirling’s formula, 190
Riemann mapping theorem Stolz angle, 38
via the Dirichlet problem, 212 subharmonic, 106
via Harnack’s principle, 132 symmetric about R, 132
via normal families, 163
Riemann sphere, S2 , 9, 222 torus, 222
Riemann surface, 221 transition map, 221
evenly covered, 224 triangle inequality, 6
of an analytic function, 222 trigonometric functions, 91
simply-connected, 222 inverse, 92, 93
Riemann zeta function, 61, 186 two-sided boundary arc, 134
Riemann’s removable singularity theorem, 70 two-to-two function, 33
for harmonic functions, 120
Robin constant, 259
uniformization theorem, 242
root test, 19
case 1, 237
Rouché’s theorem, 76
case 2, 239
Runge’s theorems, 54, 57
uniqueness theorem, 23
unit circle, ∂ D, 5
S2, Riemann sphere, 9 unit disk, D, 5
Schwarz reflection principle, 133
universal covering map, 225
Schwarz’s lemma, 35
universal covering surface, 225
for harmonic functions, 120
upper half-plane, H , 32
invariant form, 36
Schwarz’s theorem, 107
Schwarz–Christoffel theorem, 136 walking the dog, 14, 31
separates, 198 weakly-analytic function, 121
series Weierstrass P function, 174
convergent, 7 Weierstrass M-test, 18
absolutely convergent, 7 Weierstrass’s product theorem, 177
divergent, 7 Weierstrass’s theorem, 57
radius of convergence, 19 welding homeomorphism, 140
simply-connected region, 68 Weyl’s lemma, 121
and homology, 68 winding number, n(γ , a), 65
and homotopy, 219 winds, around a point, curve, 66
Riemann surface, 222 world’s greatest function, 38, 97
singular part, 170
singularity Z, 247
essential, 75 Zalcman’s lemma, 164
isolated, 74 zero, 22
pole, 74 order of, 74
removable, 74 simple, 75
spherical derivative, 161 ζ (z), 61, 186
spherical distance, 162 zeta function, ζ (z), 61, 186
standard polar grid, 81 zipper algorithm, 123

Marshall

Uploaded by

Copyright:

Available Formats

Marshall

Uploaded by

Document Information

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

Marshall

Uploaded by

Copyright:

Available Formats

What is the book about?

What is the book about?

What approach does the book take to complex analysis?

What approach does the book take to complex analysis?

Complex Analysis

This user-friendly textbook introduces complex analysis at the beginning graduate or

Donald E. Marshall is Professor of Mathematics at the University of Washington. He received

Cambridge Mathematical Textbooks is a program of undergraduate and beginning graduate-level text-

John B. Conway, George Washington University

A complete list of books in the series can be found at www.cambridge.org/mathematics

Recent titles include the following:

Cambridge University Press is part of the University of Cambridge.

List of Figures page viii

3 The Maximum Principle 31

4 Integration and Approximation 43

5.5 The Argument Principle 76

7 Harmonic Functions 105

8 Conformal Maps and Harmonic Functions 123

9 Calculus of Residues 141

10 Normal Families 156

11 Series and Products 170

11.4 The Gamma and Zeta Functions 184

12 Conformal Maps to Jordan Regions 195

13 The Dirichlet Problem 207

14 Riemann Surfaces 216

15 The Uniformization Theorem 230

16 Meromorphic Functions on a Riemann Surface 246

1.1 Cartesian and polar representations of complex numbers. page 3

6.13 Region between the branches of a hyperbola. 96

Wov’n through carefully chosen words

10. Continuous function, uniformly continuous function.

1.1 Complex Numbers

(a, b) + (c, d) = (a + c, b + d),

(a, b) · (c, d) = (ac − bd, bc + ad)

Figure 1.1 Cartesian and polar representations of complex numbers.

Figure 1.2 Addition.

Figure 1.3 Multiplication.

Theorem 1.1 (triangle inequality)

Figure 1.4 Triangle inequality.

|z| = |z + w + (−w) | ≤ |z + w| + | − w| = |z + w| + |w|.

Theorem 1.2 (Cauchy–Schwarz inequality)

Using the identity | A − B |2 = |A|2 + |B|2 − AB − AB, with A = aj bi and B = ai bj , we obtain

The above proof also gives the error

Corollary 1.3 If f and g are continuous complex-valued functions deﬁned on [a, b] ⊂R

This corollary can also be proved directly by expanding

1.3 Stereographic Projection

Figure 1.5 Stereographic projection.

z∗ = (x1, x2, x3 ) = (0, 0, 1) + t[(x, y, 0) − (0, 0, 1)].

So we will consider a line in C as just a special kind of “circle.”

Cauchy–Schwarz inequality, and also by directly multiplying it out using complex

be the average of f over the interval [a, b].

1.10 The equation

f (x, y) = (u(x, y), v(x, y)),

translates, or shifts, the plane. If a is a (ﬁxed) complex number then

|p(z)| = |z|n and arg p(z) = n arg z mod 2 π .

Each pie slice

{z : |z| < rn } \ ( −r n, 0).

p(z) = p(z0) + b1(z − z0) + b2(z − z0)2 + . . . + bn (z − z0)n . (2.2)

Figure 2.1 The power map.

2.2 Fundamental Theorem of Algebra and Partial Fractions