June 1997

Renormalization of the Electroweak Standard Model

to All Orders ∗
arXiv:hep-th/9709154v1 20 Sep 1997

†
Elisabeth Kraus

Physikalisches Institut, Universität Bonn

Nußallee 12, D-53115 Bonn, Germany

Abstract

We give the renormalization of the standard model of electroweak interactions

to all orders of perturbation theory by using the method of algebraic renor-
malization, which is based on general properties of renormalized perturbation
theory and not on a specific regularization scheme. The Green functions of
the standard model are uniquely constructed to all orders, if one defines the
model by the Slavnov-Taylor identity, Ward-identities of rigid symmetry and
a specific form of the abelian local gauge Ward-identity, which continues the
Gell-Mann Nishijima relation to higher orders. Special attention is directed to
the mass diagonalization of massless and massive neutral vectors and ghosts.
For obtaining off-shell infrared finite expressions it is required to take into
account higher order corrections into the functional symmetry operators. It
is shown, that the normalization conditions of the on-shell schemes are in
agreement with the most general symmetry transformations allowed by the
algebraic constraints.

∗
to be published in Ann. Phys. (NY) 1997
†
Supported by Deutsche Forschungsgemeinschaft
Contents

1 Introduction 3

2 The tree approximation of the standard model 8

2.1 The gauge invariant part of the action . . . . . . . . . . . . . . . . . . . . 8
2.2 The gauge fixing and rigid transformations . . . . . . . . . . . . . . . . . . 14
2.3 BRS-invariance and Faddeev-Popov ghosts . . . . . . . . . . . . . . . . . . 17
2.4 The tree approximation: the Slavnov-Taylor identity . . . . . . . . . . . . 20

3 The construction of higher orders: The algebraic method 28

4 The algebraic characterization of the symmetry transformations 32

4.1 The general ansatz and discrete symmetries . . . . . . . . . . . . . . . . . 32
4.2 The vector-ghost sector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.3 The scalar sector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.4 The fermion sector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
4.5 The algebraic characterization of an abelian local Ward operator . . . . . . 43

5 The general local solution of the Slavnov-Taylor identity and rigid sym-
metries 46
5.1 The normalization conditions . . . . . . . . . . . . . . . . . . . . . . . . . 46
5.2 The symmetry transformations and the general action . . . . . . . . . . . . 50
5.3 The vector-scalar and fermion part of the action . . . . . . . . . . . . . . . 53
5.4 The gauge fixing and ghost sector . . . . . . . . . . . . . . . . . . . . . . . 62
5.4.1 The classical approximation . . . . . . . . . . . . . . . . . . . . . . 64
5.4.2 The solution of the ghost equations in higher orders . . . . . . . . . 70
5.5 Summary of the classical approximation . . . . . . . . . . . . . . . . . . . 72

6 The Callan-Symanzik equation 76

6.1 The soft breaking of dilatations . . . . . . . . . . . . . . . . . . . . . . . . 78
6.2 The dilatational anomalies – hard breakings . . . . . . . . . . . . . . . . . 80

1
7 Higher orders 88
7.1 The quantum numbers of higher orders breakings . . . . . . . . . . . . . . 88
7.2 The cohomolgy and the Adler-Bardeen anomaly . . . . . . . . . . . . . . . 91
7.3 The establishment of symmetries . . . . . . . . . . . . . . . . . . . . . . . 92
7.4 Induction to all orders . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97

8 Conclusions and outlook 100

A The quantum numbers of fields 103

2
1. Introduction

The standard model of electroweak interaction has been tested in the last few years
with precision experiments of remarkable accuracy [1]. Theoretical predictions are based
on the consistent perturbative formulation of the standard model of electroweak interac-
tion as a renormalizable and unitary quantum field theory, which allows the derivation of
unambiguous results for physical scattering processes order by order in perturbation the-
ory. In order to match the level of accuracy given by experiments, it is also necessary to
take into account also higher order quantum corrections to the different processes consid-
ered. Conversely, present experiments enable the standard model to be tested beyond tree
approximation. Higher order corrections to the electroweak processes have been computed
and evaluated quite systematically by several groups. (For a review see [2] and references
listed below.) The agreement between experimental results and theoretical predictions is
quite impressive and by now there is no evidence — either from theoretical arguments or
from experiments — for physics beyond the standard model.
The evaluation of higher order corrections in the standard model is quite an involved
task. First, one has to remove the divergencies which appear in the naive perturbative
expansion of Green functions in the course of renormalization and one has to establish the
defining symmetries of the theory. At the same time, the independent parameters of the
standard model have to be specified and fixed by normalization conditions in such a way
that the remaining undetermined constants, such as masses and the coupling strength,
can be taken as input parameters from experiment. Finally, there remains the explicit
evaluation of higher order loop diagrams. Nearly all calculations have been carried out in
the framework of dimensional regularization. There the one-loop order has been studied
quite systematically (see [2]) and the computations have reached a high field-theoretic
standard. However, there is no abstract approach which analyses the renormalization of
the electroweak standard model to higher orders. With the present article we fill this
gap, with special attention being paid to the symmetries, normalization conditions and
infrared-finiteness of off-shell Green functions. In particular, the analysis does not refer
to invariance properties of a scheme, but is based on properties of finite renormalized
perturbation theory. (For a review of algebraic renormalization see [3].) The purpose of
a systematic analysis is twofold. First, it is evident that such an analysis will support
explicit calculations by allowing symmetries to be established and possible breakings of
the symmetries to be characterised quite systematically. In particular, if one wants to
take into account higher order corrections in theoretical predictions, either by summing
up one loop induced higher order corrections or by explicit evaluation of some higher
order diagrams, it has to be ensured that the defining symmetries are not violated at

3
any stage of the calculations and that Green functions exist to higher orders, once they
are specified in 1-loop order. Higher order existence of Green functions can be destroyed
in the standard model due to off-shell infrared-divergencies, whenever a photon mass
counterterm is enforced by symmetries and by lower order normalization conditions. In
fact it appears that infrared finiteness and the establishment of symmetries cannot be
considered as separate from each other. Apart from these practical reasons the analysis
is also important in its own right. Since the standard model has been so successful by
now, we are convinced that electroweak interaction can only be embedded into a more
complete theory of fundamental interactions, once one understands their structure in its
quantized version as prescribed by the standard model.
From the algebraic point of view, the abstract approach to the quantized version of
the standard model is similar to the construction of the classical Lagrangian [4, 5, 6]. If
one takes the charged currents of weak interactions as given in the lepton sector, it is seen
that the algebra is closed, when one includes the weak and electromagnetic currents into
the group structure. Coupling these currents to the gauge bosons of weak interactions
and to the photon, and requiring local SU(2) × U(1) gauge invariance, the algebra, and
at the same time the classical action, is uniquely determined and the transformation of
all further fields is restricted [4]. Introducing a complex scalar doublet with one physical
Higgs field, one generates all masses by the mechanism of spontaneous symmetry breaking
and the final standard model Lagrangian is invariant under spontaneously broken SU(2)×
U(1) symmetry, which is a natural algebraic generalization of unbroken symmetry. The
implementation of symmetries is also the main ingredient of abstract renormalization.
At an early stage it was observed in the framework of dimensional regularization that
gauge theories are renormalizable [7], in the sense that divergencies can be absorbed
into a redefinition of coupling constants, mass parameters and fields. If one uses the
renormalizable ‘t Hooft gauges, the divergence structure of a spontaneously broken theory
is seen to be no worse than that of unbroken theories [8, 9].
The main advances in the systematic definition of renormalizable gauge theories oc-
curred, when it was observed that gauge theories, including the gauge fixing and Faddeev-
Popov part [10], are invariant under nonlinear symmetry transformations, the Becchi-
Rouet-Stora (BRS) transformations [11, 12]. It is then possible to derive and postulate
the Slavnov-Taylor identities, which are the functional version of BRS-symmetry, as ex-
pressing the defining symmetries of gauge theories in the quantized version. In particular,
the program of algebraic renormalization has been applied to the abelian Higgs-Kibble
model [13] and spontaneously broken gauge theories with semisimple gauge groups [14].
With the help of the action principle in its quantized version [15, 16], and algebraic consis-

4
tency it was shown that the Green functions are completely characterized by normalization
conditions on the mass and coupling constants and the Slavnov-Taylor identity. If the
Adler-Bardeen anomalies [17, 18, 19] are absent, the Slavnov-Taylor identity can be estab-
lished to all orders for off-shell Green functions. Then one is finally able to prove unitarity
of the physical S-matrix, i.e. compensation of unphysical fields in physical scattering pro-
cesses and gauge parameter independence of the physical S-matrix [13, 14, 20]. Algebraic
renormalization yields finite Green functions by requiring invariance under symmetry
transformations instead of defining them by an invariant scheme. In gauge theories with-
out parity violation a specification by an invariant scheme is quite satisfactory. However,
if anomalies are not forbidden for reasons of symmetry, then the algebraic renormaliza-
tion becomes important, if one wants to formulate the theory consistently to all orders of
perturbation theory.
In the early papers only gauge theories with a semisimple gauge structure were con-
sidered. Later on the renormalization procedure was extended to non-semisimple groups
with several abelian factors [21]. But the analysis, as it is carried out there, is not im-
mediately applicable to the standard model, due to the restriction to massive fields and
due to the fact that the Green functions are not specified for on-shell fields. However, we
shall use some technical components of this paper such as the form of the Slavnov-Taylor
identity and the use of the Callan-Symanzik operator for solving the cohomology.
In the remainder of this introduction we shall outline the procedure of renormalizing
the standard model, as it is presented in the paper. As the first step we have to specify
all the symmetry transformations which characterize the tree approximation and higher
order Green functions. It is important to note that the weak hypercharges are determined
by requiring electromagnetic current conservation according to the Gell-Mann Nishijima
formula. In the procedure of quantization electromagnetic gauge invariance is replaced
by BRS-symmetry and the Slavnov-Taylor identity. For deriving the analog of the Gell-
Mann Nishijima relation, we have to establish the local U(1) Ward identity in addition to
the Slavnov-Taylor identity. For specifying the abelian factor, however, it is necessary to
have invariance under rigid SU(2) × U(1) Ward identities. This constraint restricts order
by order the independent parameters of the gauge fixing functions, but rigid invariance
is immediately established on the matter and Yang-Mills parts of the action. Only if
one includes all these symmetry transformations, are the finite Green functions uniquely
specified as being those of weak and electromagnetic interactions.
The symmetry invariants are free parameters and have to be fixed by normalization
conditions. Here, the abstract analysis benefits from the fact, that different parameter-
izations have been considered for one-loop calculations and have been discussed quite

5
extensively in the past (for a review see [22]), since their definition also enters the the-
oretical predictions of higher orders. It has been pointed out that those schemes are
adequate, which allow the computation of different processes without switching to differ-
ent parameter sets [23]. On-shell schemes which specify the mass parameters as physical
masses on the 2-point functions [24, 25, 26, 27, 29, 28, 30, 31, 32] are certainly a safe
choice, because all S-matrix elements are computed without adjusting further parame-
ters when taking the LSZ-limit. Throughout this paper, we adopt an on-shell definition
for the masses and in particular require mass diagonalization for massive/massless parti-
cles on-shell. In the abstract approach, such on-shell conditions are crucial, not only for
physical particles but also for unphysical fields, when one finally wants to prove unitarity
of the physical S-matrix [14, 20]. As far as the residua are concerned, we remain quite
general in the construction, and do not specify special conditions. We finally see that
some of the normalization conditions of residua can be eliminated by requiring a simple
form of rigid Ward identities, but this is not essential at any stage of the procedure. The
critical point in the analysis is the observation that on-shell conditions indeed fix more
parameters than there are naive invariants. Requiring symmetries in their explicit tree
form, one is unable simultaneously to adjust the W -mass and to diagonalize the neutral
mass matrix at the mass of the Z-boson and at p2 = 0. These normalization conditions
are also deeply connected with off-shell infrared divergencies to higher orders. It has
been pointed out already in [28] that complete on-shell schemes are compatible with the
Slavnov-Taylor identity, and there are scattered remarks in the literature that on-shell
schemes are in agreement with the symmetries if the transformations are themselves sub-
ject to renormalization (see e.g. [33]). But neither the Slavnov-Taylor identity nor rigid
or local Ward-identities have been given in an explicit form valid for the Green functions
of the standard model. The Slavnov-Taylor identity in its homogeneous form as given
in [28] is not quite an adequate choice for the SU(2) × U(1)-symmetry of the standard
model, since one has to split off the abelian factor explicitly as done in [21]. In terms of
on-shell fields, all symmetry transformations depend on the weak mixing angle in the tree
approximation, and it is seen, that due to off-shell infrared divergencies, the symmetry
operators have to be modified order by order in perturbation theory. For this reason we
start the analysis by characterizing the symmetry transformations by algebra and field
content, and find in this way all higher order deformations which are compatible with the
symmetries. These general symmetry operators finally allow us to construct unique Green
functions in the on-shell schemes, without introducing off-shell infrared divergencies.
For the present paper, we restrict ourselves to a diagonal quark mass matrix, because
we are mainly interested in the renormalization of the vector sector. Apart from this we
stay quite general and proceed as far as possible along the lines of concrete calculations. In

6
particular, we use the general Rξ -gauges, although in a modified form with an auxiliary
field which couples to the gauge fixing functions. Particular attention is paid to the
solution of the classical approximation, which gives the local four-dimensional invariants
of symmetry transformations. In the higher order construction of finite Green functions
we use the BPHZL scheme [34, 35]. In this scheme, massless particles are treated quite
systematically by establishing those normalization conditions in the scheme which are
necessary for the computation of finite Green functions to all orders. These normalization
conditions are essentially the conditions for mass diagonalization of massless/massive field
at p2 = 0 (i.e. for the Z-boson and photon and the respective Faddeev Popov fields) and
are established in the above-cited on-shell schemes by adjustment of counterterms.
The plan of this paper is as follows: In section 2, we give the classical action in
renormalizable gauges compatible with rigid symmetry and local U(1)-gauge symmetry.
We also present the symmetry transformations of the tree approximation in a functional
form. These are the Slavnov-Taylor identity, rigid Ward-identities and the local U(1)
Ward identity. In section 3, we outline the method of algebraic renormalization. In
section 4 we solve the algebra of symmetry operators and obtain the general consistent
symmetry operators of the standard model. Section 5 is devoted to solving the symmetries
for the local four-dimensional field polynomials. This analysis allows us to give the free
parameters of the model and also to list the invariant counterterms of higher orders.
In section 5.4 a complete treatment of the ghost equations is also included. In section
6, we derive the Callan-Symanzik equation of 1-loop order. By means of symmetric
differential operators, it is possible to characterize symmetric nonlocal contributions of
higher orders and in particular to determine the independent parameters of the theory in
a scheme-independent way. In section 7, we proceed to higher orders and prove that Green
functions can be constructed in agreement with the infrared normalization conditions to
all orders, if one takes into account the modifications of the symmetry operators to higher
orders as suggested by the tree approximation.

7
2. The tree approximation of the standard model

2.1. The gauge invariant part of the action

The standard model of electroweak interactions is a non-abelian gauge theory with the
non-semisimple gauge group SU(2) × U(1). The gauge structure is essentially determined
in the matter sector: It is seen, that the matter currents of weak interactions, the charged
µ
current JCC and the neutral current JNµ C , together with the electromagnetic current jemµ

form a closed representation with respect to SU(2) × U(1) [4]. In order to embed these
currents into a gauge theory, one groups the fermions into left-handed doublets, which
transform under the fundamental representation of SU(2) × U(1), and right handed sin-
glets, which only transform with respect to the abelian subgroup. The decomposition of
the Dirac spinors into left and right handed fields is defined by the following projections:

f L = 12 (1 − γ5 )f f R = 21 (1 + γ5 )f
f L = f¯1 (1 + γ5 )
2
f R = f¯1 (1 − γ5 )
2
(2.1)

The fermions appear in families: Each family consists of a neutrino νi , a charged

lepton ei with electric charge Qe = −1, and the up and down-type quarks ui and di with
charge Qu = 23 and Qd = − 13 . For simplicity we suppress the colour index of the quarks
throughout the paper. The lepton doublets FlLi and quark doublets FqLi , i = 1, 2, 3, are
given by
       
νiL νeL νµL ντL
FlLi =   =       (2.2)
eLi eL
µ L
τ L
      
L 
uLi  
uL   cL   tL 
Fqi = = (2.3)
dLi dL sL bL

The singlets only comprise the charged fermions:

fiR = eR R R
i , u i , di (2.4)

The SU(2) and U(1) gauge transformations (α = +, −, 3):

ταT L
ǫα (x)δα FδLi = iǫα (x) F ǫα (x)δα fiR = 0
2 δi
Yδ
ǫ4 (x)δ4 FδLi = −iǫ4 (x) W FδLi ǫ4 (x)δ4 fiR = −iQf fiR (2.5)
2

8
give rise to the matter currents of electroweak interactions
1X L µ 1X L µ
J+µ = − F γ τ− FδLi J−µ = − F γ τ+ FδLi
2 δi δi 2 δi δi
1X L µ 1X δ L µ L X
J3µ = − F γ τ3 FδLi J4µ = Y F γ Fδi + Qf fiR γ µ fiR (2.6)
2 i δi 2 δi W δi fi

with δ = l, q. If one identifies out of the neutral currents the electromagnetic current
X
µ
jem = Qf f¯i γ µ fi = J4µ − J3µ (2.7)
i,f

the weak hypercharge and the electric charge are related according to the Gell-Mann
Nishijima formula:
1
(τ3 + YW ) = Q (2.8)
2
which means explicitly
1
YWl = −1 and YWq = (2.9)
3
In (2.5) and (2.6) τα , α = +, −3, denote the generators of the charged fundamental
representation of SU(2). They are defined by
 √     
0 2  0 0  1 0 
τ+ =  τ− =  √ τ3 =  (2.10)
0 0 2 0 0 −1

and satisfy the following commutation relations:

[τα , τβ ] = 2iǫαβγ τγT (2.11)

The structure constants ǫαβγ are imaginary and completely antisymmetric in all three
indices:
ǫ+−3 = −i (2.12)
According to the Noether construction of gauge theories the matter action consists of the
kinetic terms and the currents coupled to a SU(2)-triplet of vector fields Wαµ , α = +, −, 3,
and an abelian vector field W4µ
NF Z 
X
Γmatter = iFlLi ∂/FlLi + iFqLi ∂/FqLi + ifiR ∂/fiR (2.13)
i=1

− g2 (W+µ Jµ− + W−µ Jµ+ + W3µ Jµ3 ) − g1 W4µ Jµ4
NF Z 
X 
≡ FlLi iD/FlLi + FqLi iD/FqLi + fiR iD/fiR
i=1

9
The covariant derivatives are therefore given by:
τα L µ Yδ
D µ FδLi = ∂ µ FδLi − ig2 Fδi Wα + ig1 W FδLi W4µ δ = l, q (2.14)
2 2
µ
D µ fiR µ R R
= ∂ fi + ig1 Qf fi W4

The gauge transformations of the vectors are uniquely determined from gauge invariance
of the matter action:
ǫα (x)δα Wβµ = (I˜αβ g12 ∂ µ + Wγµ I˜γγ ′ ǫγ ′ βα )ǫα (x) ǫ4 (x)δ4 Wαµ = 0
(2.15)
ǫα (x)δα W4µ = 0 ǫ4 (x)δ4 W4µ = g11 ∂ µ ǫ4 (x)

The matrix I˜αβ is the charge conjugation matrix:

 
0 1 0 0
 

 1 0 0 0 
 I˜+− = I˜−+ = I˜33 = I˜44 = 1
I˜ =   (2.16)

 0 0 1 0 
 I˜αβ = 0, else
0 0 0 1

From (2.15) the Yang-Mills part which involves the kinetic terms of the vectors is deter-
mined
Z  
1 ˜
ΓY M = − Gµν
α Iαα′ Gµνα′ + F
µν
Fµν (2.17)
4
with

Gµν
α = ∂ µ Wαν − ∂ ν Wαµ + g2 I˜αα′ ǫα′ βγ Wβµ Wγν (2.18)
F µν = ∂ µ W4ν − ∂ ν W4µ (2.19)

The bosons of weak interactions as well as the charged fermions are massive. The
mass terms break chiral gauge invariance and have to be generated by the spontaneous
breaking of the gauge symmetry. In the standard model all the masses are generated by
introducing one complex scalar doublet Φ and its complex conjugate Φ̃:
   
φ+ (x) √1 (H(x) − iχ(x))
Φ≡ 1  Φ̃ ≡ iτ2 Φ∗ =  2
−
 (2.20)
√ (H(x) + iχ(x)) −φ (x)
2

φ± are charged, H and χ neutral scalar fields. The doublet transforms under the fun-
damental representation and includes in its transformation a constant shift v into the
direction of the neutral component of the scalar doublet:
ταT
ǫα (x)δα Φ = iǫα (x) (Φ + v)
2
Ys
ǫ4 (x)δ4 Φ = −iǫ4 (x) W (Φ + v) (2.21)
2

10
with  
0
v=  (2.22)
√1 v
2

The weak hypercharge is determined from (2.8)

YWs = 1 (2.23)

As a response to the transformation (2.21), the gauge invariant parts of the action Γscalar
and ΓY uk , depend on the shift:
Z  
1 m2H † † 2
Γscalar = (D µ Φ)† Dµ Φ − (Φ Φ + v †
Φ + Φ v) (2.24)
√ 2 v2
N
X Z
F
2
ΓY uk = − (mei FlLi (Φ + v)eR
i
i v

+ mui FqLi (Φ + v)uR L R

i + mdi Fqi (Φ̃ + ṽ)di + h.c.) (2.25)

The Yukawa interaction contains via the shift all mass terms of the fermions mfi . We
have chosen the couplings of the Yukawa interactions in such a form, that the mass terms
are parametrized by the mass of the respective fermions. For the purpose of this paper we
forbid mixing between different families and especially assume CP-invariance throughout
the paper.
The scalar part consists of the kinetic terms of the scalars and the scalar potential,
which includes the mass of the Higgs field m2H . In order to have a proper particle inter-
pretation, we have arranged the terms such that the contributions linear in H(x) drop
out. Via the covariant derivative
τα YWs
Dµ Φ = ∂µ Φ − i(g2 Wµα − g1 Wµ4 )(Φ + v) (2.26)
2 2
the masses of the gauge fields are generated by eating up the massless Goldstone bosons
φ+ , φ− and χ:
1 g22 v 2 g2 g1 v 2 µ 1 g 2v 2
(2W+µ Wµ− + W3µ Wµ3 ) + W3 Wµ4 + 1 W4µ Wµ4 (2.27)
2 4 4 2 4
Physical fields are constructed by diagonalizing the mass matrix with an orthogonal ma-
trix:  
1 0 0 0
 
 0 1 0 0 
 
Wαµ = Oαa (θW )Vaµ Oαa (θW ) =   (2.28)
 0 0 cos θW − sin θW 
 
0 0 sin θW cos θW
The fields which are generated by the rotation are the physical on-shell fields Vaµ =
(W+µ , W−µ , Z µ , Aµ ). Throughout the paper roman indices a, b, c are reserved to on-shell

11
indices a, b, c = +, −, Z, A, whereas Greek indices α, β, γ denote group indices of SU(2)
and U(1) α, β, γ = +, −, 3, 4. In the tree approximation one calculates the following
relations between the ratio of the gauge parameters and the weak mixing angle θW :
g1
= tan θW (2.29)
g2
W±µ are the charged bosons of weak interactions with mass MW
2
, Z µ is the neutral boson
with mass MZ2 , Aµ the massless photon:
2 g22v 2 g22 v 2
MW = MZ2 = (2.30)
4 4 cos2 θW
If one eliminates the parameters θW and v in favour of the masses MW and MZ , one
arrives at the on-shell parameter set
MW , MZ , mfi , mH (2.31)
which specifies the particles by their masses and electric charge. The weak mixing angle
is then defined by the mass ratio of the W- and Z-mass
MW
cos θW ≡ . (2.32)
MZ
If one chooses the on-shell set for parametrizing the free parameters of the standard model,
then one remains with one coupling constant, which in the QED-like parametrizations is
taken to be the coupling of the electromagnetic current to the photon:
e = g2 sin θW (2.33)

The gauge invariant part of the classical action ΓGSW [4, 5, 6] is given by the sum of
the gauge invariant parts (2.13) (2.17) (2.24) and (2.25):
ΓGSW = ΓY M + Γscalar + Γmatter + ΓY uk (2.34)
It is completely specified by the gauge transformations, the masses of the interacting
particles, their electric charge and the electromagnetic coupling.
We want to summarize the gauge transformations of the on-shell fields within the QED-
like on-shell parameter set (2.31) and (2.33). In the spirit of the subsequent considerations
we express the gauge transformations thereby in a functional operator acting on ΓGSW :
 sin θW µ δ 
−w+ − ∂ ΓGSW = 0 (2.35)
e δW−µ
 sin θW µ δ 
−w− − ∂ ΓGSW = 0 (2.36)
e δW+µ
 sin θW µ δ δ 
−w3 − ∂ (cos θW µ − sin θW µ ) ΓGSW = 0 (2.37)
e δZ δA
 cos θW µ δ δ 
w4 − ∂ (sin θW µ + cos θW µ ) ΓGSW = 0 (2.38)
e δZ δA

12
The functional operators of SU(2)-transformations are given by (α = +, −, 3)

δ
wα = I˜αα′ Vbµ Obβ
T
(θW )ε̂βγα Oγc (θW )I˜cc′ µ
δVc′
→
− ←−
τα′ δ δ τα′
+i(Φ + v)† †
−i (Φ + v)
2 δΦ δΦ 2
NF X  →
− ←
− !
X
L τα
′ δ δ τα′ L 
+ iFδi − i L Fδi (2.39)
i=1 δ=l,q 2 δFδL δFδi 2
i

The transformation of the on-shell vectors depends on the weak mixing angle:



ε̂Z+,− = −i cos θW
T
Obβ (θW )εβγα Oγc (θW ) ≡ ε̂bc,α = ε̂A+,− = i sin θW (2.40)



ε̂+−,3 = −i

The abelian Ward operator is given by

→
− ←
−
i † δ i δ
w4 = (Φ + v) − (Φ + v)
2 δΦ† 2 δΦ
NF
X X  →
− ←− 
δ i L δ i δ L
+ YW Fδi L − F
i=1 δ=l,q 2 δFδ 2 δFδLi δi
i
→
− ←
− !
X  δ δ 
R
− Qf ifiR R − i R fi (2.41)
f R δfi δfi

In the notation we understand summation over all fermion singlets and doublets. The
Ward operators satisfy the local SU(2) × U(1)-algebra:

[wα (x), wβ (y)] = δ(x − y)εαβγ I˜γγ ′ wγ ′ (x) (2.42)

[wα (x), w4 (y)] = 0

It is obvious that ΓGSW is also invariant with respect to rigid transformations which
are obtained by taking the infinitesimal parameters ǫα as constants or, equivalently, by
integrating the local Ward operators (α = +, −, 3, 4):
Z
Wα ΓGSW = 0 and Wα = wα (2.43)

Rigid symmetries can be established for off-shell Green functions to all orders of pertur-
bation theory in a modified form and turn out together with the abelian Ward identity
to be important ingredients for defining the standard model in its quantized version.

13
2.2. The gauge fixing and rigid transformations

The perturbative construction of Green functions and finally the S-matrix starts with
the specification of the free fields and their respective propagators. In the standard model
the scalars φ± and χ are unphysical fields being absorbed into the longitudinal polarization
of the massive vectors W± and Z. Eliminating them by a gauge transformation, however,
leads to propagators with a bad ultraviolet behaviour, and renormalizability by power
counting is not evident anymore. For a systematic treatment of higher orders one better
uses the renormalizable gauges as the Rξ -gauges. If one constructs the off-shell Green
functions in the renormalizable gauges, one is able to refer to power counting properties
of renormalized perturbation theory and, especially, to the quantum action principle. In
the end one has then to prove unitarity of the physical S-matrix, i.e. it has to be shown
that the unphysical fields, as the scalar component of the vectors and the Goldstone
bosons, do not appear in physical scattering processes.
The free field propagators are calculated from the bilinear parts of the gauge invariant
action ΓGSW and the gauge fixing part Γg.f. :
(bil)
Γ(bil) = ΓGSW + Γg.f. (2.44)

The gauge fixing in the Rξ -gauges is given by:

Z
1 1 1
Γg.f. = − F+ F− − FZ FZ − FA FA (2.45)
ξW 2ξZ 2ξA
with

F± ≡ ∂µ W±µ ∓ iMW ζW φ±
FZ ≡ ∂µ Z µ − MZ ζZ χ (2.46)
FA ≡ ∂µ Aµ

The free field propagators are seen to have a good UV-behaviour which guarantees renor-
malizability by power counting:

Gϕk ϕl −→ p−2(2−dϕk ) if p2 → ∞ (2.47)

where dϕα is the mass dimension of the field ϕk . They also have good infrared behaviour,
i.e. they diverge for the massless particles not stronger than p−2 as for the photon field.
p−4 infrared divergent terms are removed by introducing mass terms for the would-be
Goldstone fields into the gauge fixing functions.
Adding such a gauge fixing with arbitrary gauge parameters to the action one does
not keep any knowledge about the SU(2) × U(1) structure of the standard model in the

14
free field propagators, but treats the bilinear action as if it were composed of several U(1)-
(bil)
factors. But as a consequence of the gauge construction ΓGSW has definite transformation
properties under rigid unbroken SU(2) × U(1): 4-dimensional terms are invariants; the 3-
dimensional terms, which are the fermion mass terms and the mixed scalar-vector terms,
together with their variations transform as a vector under unbroken SU(2) × U(1). The
mass terms of the vectors are composed with their variations and second variations to a
second rank tensor. In order not to spoil these transformation properties by the gauge
fixing part, one has to choose:

ξ ≡ ξW = ξZ = ξA and ζ ≡ ζZ = ζW (2.48)

Instead of requiring the complicated transformation behaviour of the mass terms one
can introduce an external scalar field Φ̂ and its complex conjugate, which couples to the
masses and their variations (see also [36]):
! !
φ̂+ † φ̂−
Φ̂ = Φ̂ = (2.49)
√1 (Ĥ + iχ̂) √1 (Ĥ − iχ̂)
2 2

Under rigid transformations it transforms in the same way as the scalar doublet Φ, but
includes a different shift parameterized by ζv into the transformation (ǫα , ǫ4 = const.):

ταT
ǫα δα Φ̂ = iǫα (Φ̂ + ζv)
2
Ys
ǫ4 δ4 Φ̂ = −iǫ4 W (Φ̂ + ζv) (2.50)
2
Algebraically this is the same procedure as one carries out if one introduces the scalar
doublet and the Higgs mechanism for generating the masses of the fermions and vectors,
but the external field is required to be non-propagating and does not have physical mean-
ing. The gauge fixing functions can be enlarged by the external field in such a way that
they transform as a vector under the adjoint representation:
e  τT τT 
Fa → Fa = ∂µ Vaµ − i (Φ̂ + ζv)† a (Φ + v) − (Φ + v)† a (Φ̂ + ζv) (2.51)
sin θW 2 2
The gauge fixing part is then invariant under the rigid transformations, if one includes
the transformations of the external fields:
Z
1
Γg.f. = − Fa I˜ab Fb ǫα δα Γg.f. = 0 (2.52)
2ξ
Va , a = +, −, Z, A are the on-shell fields, and the respective representation matrices τa are
obtained by acting with the orthogonal matrix O(θW ) (2.28) on τ3 and G1:
T T
τa (G) = Oaα (θW )τα + Oa4 G1 (2.53)

15
Explicitly they read:

τZ (G) = cos θW τ3 + G sin θW 1

τA (G) = − sin θW τ3 + G cos θW 1 (2.54)

The abelian parameter G is not fixed by rigid invariance. Choosing it

sin θW
G=− (2.55)
cos θW
one obtains for vanishing external fields the original gauge fixing with the parameters
according to (2.48). The masses of the would-be Goldstones are generated by the shift of
the external field. The transformation properties of the trilinear and the mass terms are
now governed by the transformation properties of the external field Φ̂.
Modifying the functional operators of rigid transformations (2.43) by the transforma-
tions of the external field according to (2.50)
Z  →
− ←− 
˜ † τα′ δ δ τα′
Wα → Wα + Iαα′ i(Φ̂ + ζv) − i ( Φ̂ + ζv) α = +, −, 3 (2.56)
2 δΦ† δ Φ̂ 2
→
− ←
−
i † δ i δ
w4 → w4 + (Φ̂ + ζv) − (Φ̂ + ζv)
2 δ Φ̂† 2 δ Φ̂
we write the invariance properties of ΓGSW + Γg.f. in functional form:

Wα (ΓGSW + Γg.f. ) = 0 (2.57)

Furthermore it is seen that the gauge transformation of the abelian subgroup is broken
linearly in propagating fields:
e δ δ  1
( w4 − ∂ µ (sin θW µ + cos θW µ ) (ΓGSW + Γg.f. ) = − ✷(sin θW FZ + cos θW FA )
cos θW δZ δA ξ
(2.58)
For this reason it is possible to extend and interpret (2.58) as a Ward identity for Green
functions.
This construction of the gauge fixing sector is essential if one wants to proceed to higher
orders perturbation theory. Especially it is seen that we need a local Ward identity of
the form (2.58) for the Green functions in order to fix the weak hypercharge and electric
charge in a scheme independent way.
Finally we want to mention that choosing the parameter G according to (2.55) is
arbitrary and not related to any symmetries. It turns out that this parameter as well as
an additional abelian gauge parameter are renormalized in higher orders of perturbation
theory.

16
2.3. BRS-invariance and Faddeev-Popov ghosts

The linear Rξ -gauges break also in their covariant form gauge invariance and especially
bring about that the unphysical fields, the scalar components of the vectors and the
would-be Goldstones, interact with the physical fields violating thereby unitarity of the
physical S-matrix. For this reason one has to introduce the Faddeev-Popov ghosts ca , a =
+, −, Z, A with ghost charge 1 and the respective antighosts c̄a , a = +, −, Z, A with
ghost charge -1. They are anticommuting scalars with negative norm and compensate
the unphysical degrees of freedom introduced by the gauge fixing, if one adds the ghost
action in such a way, that the complete action is invariant under BRS-transformations.
There a several approaches to introduce the Faddeev-Popov fields [10] into the per-
turbative formulation of gauge theories. One way to proceed is to consider BRS-transfor-
mations in a first step as an alternative way to characterize the Lie algebra of the gauge
group. This approach is close to the algebraic analysis which we carry out in the higher
order construction, and therefore we outline the procedure in the following: Starting from
the gauge transformations of the fields as summarized in functional form in (2.39) and
(2.41) one translates the infinitesimal parameters ǫα (x) into anticommuting parameters
cα (x), α = +, −, 3, 4. Considering the gauge transformations on the on-shell fields (2.35)
one is lead to carry out the orthogonal transformation O(θW ) (2.28) on the ghosts as well

cα = Oαa (θW )ca ca = (c+ , c− , cZ , cA ) (2.59)

In this procedure is quite some arbitrariness, which has to be exploited in higher order
perturbation theory for a proper definition of massless ghost propagators (see section
5.4). The BRS-transformations [13] on the vector bosons Vaµ = (W+ , W− , Z, A), the
scalar doublet Φ and the fermion doublets and singlets read in the physical on-shell
parameterization:
e ˜
sVµa = ∂µ ca + Iaa′ fa′ bc Vµb cc
sin θW
e τa (Gs )
sΦ = i (Φ + v)ca
sin θW 2
e τa (Gδ ) L
sFδLi = i Fδi ca δ = l, q (2.60)
sin θW 2
sin θW R
sfiR = −ieQf f cZ − ieQf fiR cA
cos θW i
The matrices τa , a = +, −, Z, A are given in (2.54) and satisfy the algebra

[τa (G), τb (G)] = ifabc I˜cc′ τc′ (G) (2.61)

17
with the structure constants

 f+−Z = −i cos θW
T T
fabc = Oaα (θW )Obβ (θW )ǫαβγ Oγc (θW ) = (2.62)
 f+−A = i sin θW

The abelian parameter G appearing in the BRS-transformations is related to the weak

hypercharge according to the Gell-Mann Nishijima relation, explicitly:



 1 for the scalar (k = s)
sin θW
Gk = −YWk YWk = -1 for the lepton doublets (k = l) (2.63)
cos θW 

 1
3
for the quark doublets (k = q)

The algebra of the functional operators (2.42), which contains the complete informa-
tion about the group structure, is translated into the BRS-transformation of the ghosts
e
sca = − I˜aa′ fa′ bc cb cc (2.64)
2 sin θW
The representation equations and also the Jacobi identities are now encoded in the nilpo-
tency of the BRS-transformations:

s2 ϕk = 0 with ϕk = Vµa , Φ, FδLi , fiR , ca (2.65)

From the construction it is obvious that the gauge invariant part of the action (2.34) is
BRS-invariant:
sΓGSW = 0 (2.66)

The gauge fixing (2.52) breaks gauge invariance; having introduced the anticommuting
fields ca this breaking is absorbed into the transformation of the antighosts:
Z
1 !
− Fa I˜aa′ sFa′ − sc̄a I˜aa′ sFa′ = 0 (2.67)
ξ
Therefrom one obtains:
1
sc̄a = − Fa (2.68)
ξ
and Z
s(Γg.f. + Γghost ) = 0 with Γghost = − c̄a I˜ab sFb (2.69)
The ghost action contains kinetic terms for the Faddeev-Popov fields, which allows to
introduce them as dynamical fields into the theory.
The BRS-transformation of the anti-ghosts is not nilpotent. To remedy this situation
one reformulates the gauge fixing part of the action by introducing the auxiliary fields
Ba , a = +, −, Z, A
Z
1
Γg.f. = ξBa I˜ab Bb + Ba I˜ab Fb (2.70)
2

18
It can be transformed into the usual form of the Rξ gauges by eliminating the Ba -fields
via their equations of motions:
δΓ 1
= I˜ab (ξBb + Fb ) = 0
∗ ∗
=⇒ Ba = − Fa (2.71)
δBa ξ
Therefore the propagators of vectors and scalars are not changed, but in addition one has
mixed propagators between Ba -fields and vectors and Ba -fields and scalars. The ghost
action is likewise determined from (2.69), but the BRS-transformations turn out to be
nilpotent also on the antighosts:

sc̄a = Ba sBa = 0 (2.72)

For the algebraic characterization it is useful to have nilpotency of BRS-transformations

throughout and we refer to this form of the gauge fixing in the algebraic proof of renor-
malizability in higher orders. Invariance under rigid transformation is maintained, if one
transforms the Ba -fields according to the adjoint representation (ǫα , ǫ4 = const.)

ǫα δα Bb = Bc I˜cc′ ε̂c′b,α (θW )ǫα ǫ4 δ4 Bb = 0 (2.73)

The tensor ǫ̂bc,α (θW ) is defined in eq. (2.40)

The gauge fixing functions Fa depend on the external scalar doublet Φ̂ and we have
to assign to them also definite transformation properties under BRS-transformations.
Transforming Φ̂ into an external anticommuting scalar doublet q̂ with ghost charge 1

sΦ̂ = q̂ sq̂ = 0 (2.74)

does this job and allows to distinguish the propagating and external scalar fields alge-
braically.
Explicitly the ghost action is given by
Z 
e
Γghost = −c̄a ✷I˜ab cb − c̄a fabc ∂(Vb cc ) (2.75)
sin θW
e
+i (q̂† τa (Gs )(Φ + v) − (Φ + v)† τa (Gs )q̂)c̄a
2 sin θW
e 
− (Φ̂ + ζv)† τa (Gs )τb (Gs )(Φ + v)
4 sin θW
 
†
+ (Φ + v) τb (Gs )τa (Gs )(Φ̂ + ζv) c̄a cb

Gs is related to the weak hypercharge of the scalar doublets according to (2.63). Via
the shift of the external and the quantum scalar fields the charged ghosts as well as

19
the neutral Z-ghost become massive, whereas the ghost associated with the photon field
remains massless. The bilinear part of the ghost action
Z  
(bil)
Γghost = −c̄a ✷I˜ab cb − ζMW
2
(c̄+ c− + c̄− c+ ) − ζMZ2 c̄Z cZ (2.76)

gives rise to free field propagators for the Faddeev Popov fields.
The ghost action is seen to be invariant under rigid transformations if one assigns the
following transformations under SU(2) × U(1) (ǫα , ǫY =const.)

ǫα δα c̄b = c̄c I˜cc′ ε̂c′ b,α (θW )ǫα ǫα δα ca = cc I˜cc′ ε̂c′ b,α (θW )ǫα
(2.77)
ǫ4 δ4 c̄a = 0 ǫ4 δ4 ca = 0
(bil) (bil)
In particular Γghost transforms covariantly in the same way as ΓGSW .

2.4. The tree approximation: the Slavnov-Taylor identity

In the last sections we have derived the classical action of the standard model

Γcl = ΓGSW + Γg.f. + Γghost (2.78)

in a way that is invariant under BRS-transformations

sΓcl = 0 (2.79)

Spontaneously broken rigid SU(2) × U(1)-symmetry has been established by introducing

an external scalar doublet Φ̂ into the gauge fixing part of the action.
In order to quantize the model in perturbation theory one has to construct the Green
functions of the interacting theory according to the Gell-Mann Low formula.

Gϕi1 ...ϕin (x1 , ..., xn ) = hT ϕi1 (x1 )...ϕin (xn )i (2.80)

(o) (o) (o)
hT ϕi1 (x1 )...ϕin (xn )eiΓint (ϕk ,Φ̂,q̂)
i
= R (o)
iΓint (ϕk ,Φ̂,q̂) Φ̂=0
hT e i q=0

where ϕk denotes the propagating fields of the standard model


Vaµ , Ba , ca , c̄a


 a = +, −, Z, A
ϕk = φ± , H, χ (2.81)



νiL , ei , ui , di i = 1...NF
Γint includes all the interactions and the field polynomials depending on the external fields
and is obtained by splitting off from the classical action the bilinear part:

Γcl = Γ(bil) + Γint (2.82)

20
with
(bil) (bil)
Γ(bil) = ΓGSW + Γg.f. | Φ̂=0 + Γghost (2.83)
q̂=0

The index (o) stands for free fields.

The formal expansion of the exponential yields the Green functions of the interacting
theory in expressions of time ordered vacuum expectation values of free fields. These
expressions are decomposed into a sum of products of free field propagators and certain
vertex factors according to Wick’s theorem. The combinatorics and the vertex factors are
summarized graphically in the Feynman rules. The free field propagators are determined
from Γ(bil) . The Feynman rules of the standard model are listed in the literature and are
given e.g. in [37] according to the conventions we have adopted.
Due to the well-known ultraviolet divergencies the formal expansion of the Gell-Mann
Low formula is not meaningful in higher orders of perturbation theory and has to be
rendered meaningful in the course of renormalization. (This is the sense of R in eq. (2.80).)
In the lowest order, the tree approximation, the Green functions are well-defined and it
has to be shown, that the physical S-matrix, which is constructed from these Green
functions according to the LSZ reduction formula, is unitary in the lowest order. This
means, that one has to verify that unphysical particles do not contribute in physical
scattering processes, and that they are canceled among each other. This cancellation
mechanism is governed by the Slavnov-Taylor identity, which expresses consequences of
the classical BRS-symmetry for the off-shell Green functions. In order to derive the
Slavnov-Taylor identity in the tree approximation we introduce the generating functional
of Green functions:

Z(jaµ , jaB a , ̄a , J, J † , ηi , η̄i ) (2.84)

D  Z  E
= T exp i dx I˜ab (jaµ Vµb + jaB Bb + ̄a cb + c̄a b ) + Φ† J + J † Φ + f¯i ηi + η̄i fi

In (2.84) we understand summation over on-shell field indices a, b = +, −, Z, A and sum-

mation over all fermions fi = νi , ei , ui , di . The source functions are commuting (jaµ , jaB , J)
and anticommuting (, ̄, ηi ) test functions. Electric and φπ-charge is assigned in such
a way that the generating functional is neutral. The Green functions are obtained by
differentiation with respect to the respective source functions.
Although the Green functions are the basic objects of the theory, for the purpose of
renormalization one better refers to the building blocks composing them. These are the
connected Green functions and the one-particle-irreducible (1PI) Green functions. The
generating functional of connected Green functions

Zc (jk ) ≡ Zc (jaµ , jaB , a , ̄a , J, J † , ηi , η̄i ) (2.85)

21
is defined by
Z(jk ) = eiZc (jk ) (2.86)
and one can show that the differentiation with respect to the sources yields the connected
Green functions in the diagrammatic expansion. The generating functional of 1PI Green
functions is obtained from Zc (jk ) by Legendre transformation. For this purpose one
introduces the classical fields
δZc (ji )
ϕcl
k (x, ji ) = ϕcl
k (x, 0) = 0 (2.87)
δjk (x)
and defines the generating functional of 1PI Green functions Γ(ϕcl
k ) according to

cl † cl ¯cl
Zc (jaµ , a , ̄a , J, J † , ηi , η̄i ) = Γ(Vµa
cl
, Bacl , ccl cl cl
a , c̄a , Φ , Φ , fi , fi ) (2.88)
Z  
cl † ¯cl
+ dx I˜ab (jaµ Vµb
cl
+ jaB Bb + ̄a ccl cl † cl cl
b + c̄a b ) + Φ J + J Φ + fi ηi + η̄i fi

Here the sources jk are understood as solutions of (2.87)

jk = jk (x, ϕcl
k) jk (x, 0) = 0 (2.89)

The 1PI Green functions are obtained by differentiating the generating functional with
respect to the classical fields ϕcl
k , and one can show, that they correspond to the 1PI
diagrams in the diagrammatic expansion according to the Feynman rules.
The Slavnov-Taylor identity of the tree approximation can be derived most simply on
the generating functional of 1PI Green functions, because its lowest order is seen to be
the classical action:
Γ(ϕcl cl
k ) = Γcl (ϕk )| Φ̂=0 + O(h̄) (2.90)
q=0

Therefore we are able to write down the Ward identity of BRS-transformation as

XZ δΓcl (ϕcl
i )
dx sϕcl
k (x) cl
=0 (2.91)
ϕcl
δϕk (x)
k

which is a well defined expression in the tree approximation. The BRS-transformations

are non-linear symmetry transformation in propagating fields and it is seen that the non-
linear symmetry transformations become insertions into (connected) Green functions, if
one carries out the Legendre transformation. Roughly speaking one has to replace

sϕcl
k −→ [sϕk ] · Zc (jk ) (2.92)

where [sϕk ]·Zc (jk ) is the generating functional of BRS-inserted connected Green functions.
The Green functions with insertions are defined according to

Gsϕk ;ϕi1 ...ϕin (x; x1 , ..., xn ) = hT : sϕk (x) : ϕi1 (x1 )...ϕin (xn )i (2.93)
(o) (o) (o) (o)
iΓint (ϕk )
hT : sϕk (x) : ϕi1 (x1 )...ϕin (xn )e i
= R (o)
iΓint (ϕk )
hT e i

22
and are summarized in the functional of BRS-inserted (connected) Green functions ac-
cording to the above definitions. For setting up the Slavnov-Taylor identity for off-shell
Green functions one does not only have to consider the ordinary Green functions but
also the ones with BRS-insertions. For defining the BRS-inserted as well as the ordinary
Green functions consistently one enlarges the classical action by the external field part
and couples the non-linear BRS-transformations to external fields:

Γcl (ϕk ) −→ Γcl (ϕk , Υk ) = Γcl (ϕk ) + Γext.f (ϕk , Υk ) (2.94)

Z 
Γext.f. = ρµ+ sWµ,− + ρµ− sWµ,+ + ρµ3 (cos θW sZµ − sin θW sAµ ) (2.95)

+ σ+ sc− + σ− sc+ + σ3 (cos θW scZ − sin θW scA )

+ Y † sΦ + (sΦ)† Y
NF
X 
L R
+ (ΨR L
δi sFδi + ψfi sfi + h.c.)
i=1

The external fields ρµα and σα , α = +, −, 3, are anticommuting and commuting SU(2)-
triplets with ghost charge −1 and −2 respectively. The external field Y is a complex
anticommuting scalar doublet with ghost charge −1, ψfLi denotes external left-handed
spinor singlets with ghost charge −1

ψfLi ≡ ψeLi , ψuLi , ψdLi (2.96)

whereas ΨR
δi denotes external right-handed spinor doublets
   
ψR
 νi 
ψR
 ui 
ΨR R R
δi ≡ Ψli , Ψqi ΨR
li = ΨR
qi = (2.97)
ψeRi ψdRi

The Green functions with insertions (2.93) are defined via the external field part:

δ
Gsϕk ;ϕi1 ...ϕin (x; x1 , ..., xn ) = hT ϕi1 (x1 )...ϕin (xn ) exp{iΓext.f }i (2.98)
δΥk (x) Υi =0

The generating functional of Green functions

D  Z  E
Z(jk , Υk ) = T exp i dx jk ϕk + Γext.f. (ϕk , Υk )

includes ordinary and BRS-inserted Green functions, which are obtained by differentiation
with respect to the external fields Υk . ϕk jk symbolically denotes the sum over the quantum
fields coupled to their sources as explicitly given in (2.84). Therefrom the connected Green
functions are obtained according to (2.86) and the 1PI Green functions by a Legendre

23
transformation of the propagating fields as given in (2.87) and (2.88), where the classical
fields depend on the sources and external fields:
Z
Zc (jk , Υk ) = Γ(ϕcl
k , Υk ) + dx ϕcl
k jk (2.99)

Differentiation with respect to external fields on Γ(ϕcl

k , Υk ) reproduces BRS-insertions into
1PI Green functions. Especially one verifies

δΓ(ϕcl
i , Υi ) δZc (ji , Υi )
= (2.100)
δΥk (x) δΥk (x)

With the help of the external field part one is now able to derive the Slavnov-Taylor
identity in a way, that the non-linear BRS-transformations are properly defined as inser-
tions into Green functions. Taking the external field as being invariant under classical
BRS-transformations
sΥk = 0 (2.101)
the enlarged classical action (2.94) is BRS-invariant due to the nilpotency of BRS-trans-
formations.
sΓcl (ϕk , Υk ) = 0 (2.102)
The lowest order of Γ(ϕcl
k , Υk ) is the classical action (cf. (2.90)) and the Ward identity
of BRS-transformations (2.91) can be rewritten into the Slavnov-Taylor (ST) identity of
1PI Green functions in the tree approximation:
Z  
δΓcl δΓcl 
S(Γcl ) = + cos θW
(sin θW ∂µ cZ + cos θW ∂µ cA ) sin θW (2.103)
δZµ δAµ
δΓcl  δΓcl δΓcl  δΓcl  δΓcl δΓcl 
+ µ cos θW − sin θW + cos θW − sin θW
δρ3 δZµ δAµ δσ3 δcZ δcA
δΓcl δΓcl δΓcl δ δΓcl δΓcl δΓcl δΓcl δΓcl δΓcl δΓcl δΓcl
+ µ + µ + + + +
δρ+ δWµ,− δρ− δWµ,+ δσ+ δc− δσ− δc+ δY † δΦ δΦ† δY
NF 
X 
δΓcl
Γcl δ δΓcl Γcl δ
+ R
+ + h.c.
i=1 δψfi δfi
L
δΨR
δi
δFδLi

δΓcl δΓcl δΓcl †
+Ba + q̂ + q̂ = 0
δc̄a δ Φ̂ δ Φ̂†
There we have dropped the index classical for the classical fields appearing in the gener-
ating functional of 1PI Green functions

Γ(ϕk , Φ̂, q, Υ) = Γcl (ϕk , Φ̂, q, Υ) + O(h̄) (2.104)

Nonlinear BRS-transformations are now obtained by differentiating with respect to the

external fields. We have included the external fields Φ̂ into the definition of the generating

24
functional in order to be able to derive Ward identities of rigid symmetry for the Green
functions. They produce by differentiation the mass insertions and their variations under
rigid symmetry of Goldstone fields.
The algebraic properties of BRS-transformations are transfered to nilpotency proper-
ties of the Slavnov-Taylor operator:

sΓ S(Γ) = 0 for any functional Γ (2.105)

sΓ sΓ = 0 if S(Γ) = 0

The operator sΓ is the linearized version of the ST identity and is defined by

Z   δ δ 
sΓ = (sin θW ∂µ cZ + cos θW ∂µ cA ) sin θW + cos θW (2.106)
δZµ δAµ
δ δ δ †
+ Ba + q̂ + q̂
δc̄a δ Φ̂ δ Φ̂†
 δΓ δ 
X δΓ δ 
+ uk +
ϕk ,Υk δΥk δϕk δϕk δΥk

The sum is over all external and corresponding propagating fields which gave rise to
bilinear appearance of Γ in the ST identity, uk denotes the respective coefficients as
cos θW , sin θW and 1.
By Legendre transformation one is immediately able to give the ST identity for the
functional of connected Green functions Zc ≡ Zc (jk , Φ̂, q, Υk ) in the tree approximation:
Z  
δZc δZc 
S(Zc ) = (sin θW ∂µ jZµ + cos θW ∂µ jAµ ) sin θW + cos θW (2.107)
δZ δA
  δZ   δZ
c c
+ cos θW jZµ − sin θW jAµ + cos θW Z − sin θW A
δρµ3 δσ3
δZc δZc δZc δZc δZc δZc
+jµ+ µ + jµ− µ + + + − + J† † + J
δρ+ δρ− δσ+ δσ− δY δY
NF 
X δZc δZc δZc L δZc R 
+ ηiL + ηiR + η + R ηi
i=1 δψfLi δψδRi δψfLi i δψδi

δZc δZc δZc †
+̄a B + q̂ + q̂ = 0
δja δ Φ̂ δ Φ̂†

The ST identity of the connected Green functions is linear in contrast to the one of the
1PI Green functions. It is the starting point for proving unitarity of the physical S-matrix
[13, 14, 20] . Although the proof of unitarity is beyond the scope of the paper we want
to indicate how the cancellation mechanism works: Eliminating the Ba -fields and their
sources in (2.107) by
δ 1˜  δ δ δ 
−→ − I ′
aa aF ′ , , , Φ̂ (2.108)
δjaB ξ δjbµ δJ δJ †

25
with
 δ δ δ  δ e  T T 
˜ † τb δ δ † τb
Fb , , , Φ̂ = I ′
bb µ∂ − i ( Φ̂ + ζv) ( + v) − ( + v ) ( Φ̂ + ζv)
δjaµ δJ δJ † δjbµ′ sin θW 2 δJ δJ † 2
(2.109)
it is seen that the ST identity indeed relates the Green functions of ghosts to the ones with
longitudinal vector propagators and would-be Goldstones at the external legs. (Due to
linear contributions the Green functions of external fields include 1-particle reducible ghost
propagators.) Applying the S-matrix operator the corresponding unphysical contributions
have to be shown to cancel in physical scattering processes.
Renormalization concerns the 1PI Green functions. Having these well-defined the
connected Green functions exist and are also well-defined and can be obtained to all
orders by the Legendre transformations (2.88). For this reason we are able to restrict all
the further considerations to 1PI Green functions.
In the procedure of renormalization the ST identity is the defining symmetry of the
theory, because it yields unitarity of the physical S-matrix as indicated above. Due to
the abelian subgroup, however, the ST identity is not sufficient to fix uniquely the Green
functions of higher orders. In addition we have to take into account the Ward identities of
rigid SU(2)×U(1) invariance and especially the local U(1) Ward identity for being able to
fix the electric charges of the fermions. In the tree approximation the Ward identities of
rigid symmetry are immediately derived according to the construction of the gauge fixing
sector (cf. (2.43), (2.57), (2.73) and (2.77)). For consistency we have to assign definite
transformation properties under rigid transformation to the external fields Υk in such a
way, that the external field part Γext.f. (2.94) is rigid invariant. It is obvious, that the
fields ρα and σα transform under the adjoint representation, whereas Y and ΨR under the
fundamental representation of SU(2). We thus arrive at

Wα Γcl (ϕk , Υk , Φ̂, q̂) = 0 and W4 Γcl (ϕk , Υk , Φ̂, q̂) = 0 (2.110)

where Γcl is understood to be the lowest order of the generating functional of 1PI Green
functions.
Γcl = ΓGSW + Γg.f. + Γghost + Γext.f (2.111)
The Ward operators of rigid SU(2)-transformations include all the propagating and ex-
ternal fields we have introduced:
Z
δ
Wα = I˜αα′ Vbµ Obβ
T
(θW )ε̂βγα Oγc (θW )I˜cc′ µ + {ca , Ba , c̄a } (2.112)
δVc′
δ
+ρµβ εβγα′ I˜γγ ′ µ + {σα }
δργ ′

26
 −
→ ←−
† τα′ δ δ τα′
+i(Φ + v) − i (Φ + v) + {Y, Φ̂ + ζv, q̂}
2 δΦ† δΦ 2
NF X  →
− ←
− !
X τ ′ δ δ τ ′

α α L R
+ iFδLi − i L Fδi + {Ψδi }
i=1 δ=l,q 2 δFδL δFδi 2
i

The abelian Ward operator comprises the doublets and right-handed fermions together
with the external fields coupled to their BRS-variations.
Z →
− ←
−
i † δ i δ
W4 = (Φ + v) − (Φ + v) + {Y, Φ̂ + ζv, q} (2.113)
2 δΦ† 2 δΦ
NF  X
X  →
− ←
− 
δ i L δ i δ L R
+ YW Fδi L − Fδ + {Ψ δ }
i=1 δ=l,q 2 δFδi 2 δFδLi i i

→
− ←
− !
X  
R δ δ R R
− Qf ifi − i R fi + {ψfi }
f δfiR δfi

The Pauli matrices τα are defined in (2.10), the antisymmetric tensor εαβγ in (2.12). The
Ward operators of rigid symmetry satisfy the SU(2) × U(1) algebra:

[Wα , Wβ ] = εαβγ I˜γγ ′ Wγ ′ (2.114)

[Wα , W4 ] = 0

In connection with the abelian Ward identity of rigid symmetry there exists also a
local version (cf. (2.58)), which reads in Ba -gauges:

1  δΓcl δΓcl  1
w4 Γcl − cos θW sin θW ∂ + cos θW ∂ = cos θW (sin θW ✷BZ + cos θW ✷BA )
e δZ δA e
(2.115)
The local operator w4 is defined by dropping the integration from the rigid operator:
Z
W4 = w4 (2.116)

The ST identity (2.103), the Ward identities of rigid symmetry (2.110) and the local
abelian Ward identity (2.115) are the algebraic symmetries of the standard model in the
tree approximation. It has to be shown, that these symmetries can be continued to higher
orders and that they together with appropriate normalization conditions uniquely define
the Green functions of the standard model to all orders.

27
3. The construction of higher orders: The algebraic method

In the procedure of renormalization one has to make meaningful the undefined expres-
sions, which are obtained in the formal expansion of the Gell-Mann Low formula according
to Wick’s theorem (cf. (2.80) and (2.93)). As we have already mentioned renormalization
concerns the 1PI Green functions summarized in the generating functional (2.99)

Γ(φk , Υ, Φ̂, q̂) = Γ(Va , Φ, fi , ca , c̄a , ρα , Y, ψi , σα , Φ̂, q̂) (3.1)

It depends on the external fields and the “classical” fields defined by the Legendre trans-
formation (2.87). For simplification we have dropped the index ‘classical’. The 1PI Green
functions are divergent according to their degree of divergencies:
X X
dΓ = 4 − dE − (di − 4) (3.2)
ext.legs vertices

Here dE is the ultraviolet (UV) dimension of the fields appearing at the external (am-
putated) legs: They include propagating as well as external fields. di denotes the UV-
dimension of the vertices. The UV-dimensions of the fields are listed in the appendix.
There are different schemes, which remove the divergencies consistently. For practical
calculations it is convenient to use dimensional regularization in connection with a pre-
scription for subtracting the D-dimensional poles in the limit to 4 dimensions [38]. For
abstract renormalization one better refers to the momentum subtraction scheme in the
version of BPHZL [34, 35, 39].
For higher orders the Gell-Mann Low formula has to be modified taking in the inter-
action part not only the vertices of the tree approximation but also the counterterms of
higher orders. In the QED-like on-shell schemes the counterterms are power series in the
electromagnetic coupling e. All terms are collected in a Γef f :
(int)
Γef f = Γcl + O(h̄) = Γ(bil) + Γef f (3.3)

The bilinear parts are defined in (2.83). At first Γef f contains all field polynomials in
external and quantum fields which are compatible with the power counting analysis of
renormalizable quantum field theory, i.e. they have UV-dimension less than of equal to
4. The explicit form of Γef f depends on the renormalization scheme one has used to
remove the divergencies. Therefore rather than relying on properties of an explicit Γef f
and a subtraction scheme, one deals in the construction of 1PI Green functions with finite
renormalized Green functions and their properties with respect to the symmetries of the
standard model. (For an introduction to algebraic renormalization see [3]

28
 In the last sections we have given the tree approximation and the symmetries of
the tree approximation, the Slavnov-Taylor identity (2.103), the Ward identities of rigid
symmetry (2.110) and the local U(1)-Ward identity (2.115). Having readily defined the
lowest order, the Green functions of the 1-loop approximation are calculated with a spe-
cific renormalization scheme leading to a finite result Γren . Different schemes differently
dispose of the local contributions of the next order, whereas the non-local contributions
are uniquely defined. Therefore after subtraction the Green functions are well-defined up
to local contributions. In order to determine these local contributions one has to adjust
those which break the symmetry, in a way that the symmetries of the lowest order are
restored in the 1-loop order. The remaining (symmetric) ones have to be fixed by normal-
ization conditions. Then one is able to proceed to higher orders by induction repeating
the above steps from order n to order n + 1.
The symmetries of the lowest order can be also violated by anomalies. Anomalies arise
from non-local contributions and cannot be removed by adjusting local contributions.
They have then explicitly to be proven to be absent to all orders of perturbation theory.
In the standard model restoration of symmetries and the setting of proper normal-
ization conditions are deeply connected with each other: It has to be shown that one
is able to impose normalization conditions on the 2-point functions in such a way that
the 2-point Green functions have one particle properties in the LSZ limit (apart from the
problem of unstable particles). Thereby special attention has to be paid to the massless
particles: In order not to introduce off-shell infrared divergent diagrams to the next order
the 2-point functions of massless particles as well as also the mixed 2-point functions of
massive and massless particles have to be required to vanish at p2 = 0 to all orders of
perturbation theory:

ΓZA (p2 = 0) = ΓAA (p2 = 0) = 0 (3.4)

Γc̄A cZ (p2 = 0) = Γc̄Z cA (p2 = 0) = Γc̄AcA (p2 = 0) = 0

These normalization conditions have to be proven to be in accordance with the symmetries

of the standard model and will be shown to lead to higher order corrections of the weak
mixing angle in the ST identity and the Ward identities.
The 1PI Green function of the standard model summarized in the generating functional
are defined in order n by
(n) (n)
Γ(≤n) = Γ(≤n)
ren + Γinv + Γbreak (3.5)
and have to be shown to have well defined normalization properties and to satisfy the
symmetries  (≤n)  (≤n)
S(Γ) =0 Wα Γ =0 (3.6)

29
and a local U(1) Ward identity. The ST operator and the Ward operators are thereby
(n)
established via their algebraic characterization (2.105) and (2.114). In (3.5) Γinv and
(n)
Γbreak denote purely local field polynomials. They depend on propagating and external
fields introduced in the classical approximation and constitute a complete basis of field
polynomials with UV-dimension less than or equal 4. In a specific scheme the local
contributions are governed by the counterterms appearing in a Γef f . Discrete symmetries
are not affected by renormalization, we are therefore able to restrict the analysis to field
polynomials which are neutral with respect to electric and Faddeev-Popov charge and
are CP-even, due to the fact, that we did not introduce family mixing in the classical
approximation. The quantum numbers of the fields under the discrete symmetries are
listed in the appendix.
As indicated by the notation (3.5) local contributions are algebraically divided into two
classes: the invariant and non-invariant field polynomials. The invariant field polynomials
appearing in Γinv constitute together with Γcl the general classical solution Γgencl , i.e. the
general field polynomials, which are solutions of the Slavnov-Taylor identity and are rigid
invariant
∞
X (n) S(Γgen
cl ) = 0
Γgen
cl = Γcl + Γinv gen (3.7)
n=1 Wα (Γcl ) = 0
The free parameters of Γgen
cl are determined by the normalization conditions and the local
U(1) Ward identity order by order in perturbation theory.
The non-invariant field polynomials Γbreak are used to remove the breakings of the
symmetries, which have been introduced by an implicit scheme dependent adjustment
of finite counterterms in the subtraction procedure. The abstract construction of Γbreak
is carried out with the algebraic method which is based on the action principle in its
quantized version valid for off-shell Green functions [16, 39]. It is most easily formulated
on the general functional of the respective Green functions and relates variations with
respect to sources or classical fields, respectively, and external fields to insertions with a
well-defined UV and IR-degree. Especially the action principle states that the symmetries
of the tree approximation are broken at most by a integrated field polynomial in 1-loop
order and proceeds to higher orders by induction:
(n)
(S(Γ))(≤n−1) = 0 (S(Γ))(≤n) = ∆brs
=⇒ (3.8)
(Wα Γ)(≤n−1) = 0 (Wα Γ)(≤n) = ∆(n)
α

The breakings have well-defined properties with respect to the discrete symmetries. For
example ∆brs has φπ-charge 1, is neutral with respect to electric charge and even under
CP, if the classical action is CP-invariant. Furthermore they have a well-defined ultra-
violet and infrared degree. Up to this point the analysis has been completely scheme

30
independent just being founded on properties of renormalized perturbation theory, but in
classifying the breakings according to their UV- and IR-degree we assume that the renor-
malized Green functions Γren have been constructed within the BPHZL scheme. In the
BPHZL scheme the normalization conditions (3.4), which otherwise have to be established
by hand, are immediately implemented by the subtraction procedure, guaranteeing, that
nowhere infrared divergent contributions are introduced by the subtraction scheme. In-
frared divergent contributions are detected by a pure power counting analysis. Especially
counterterms of IR dimension less than 4 are forbidden in the BPHZL-scheme because
they would destroy the normalization conditions (3.4). Therefore we adopt the UV and
IR degrees of fields as given in the appendix. They are uniquely determined by the be-
haviour of the free-field propagators for p2 → ∞ and p2 → 0, respectively. Then by an
analysis of the ST identity and the Ward operators it is derived that the breakings of the
symmetries have the following UV and IR degrees:

dimU V ∆brs ≤ 4 dimIR ∆brs ≥ 3

dimU V ∆α ≤ 4 dimIR ∆α ≥ 2

and it is seen, that all symmetries have to be carefully constructed concerning the infrared.
Apart from the IR and UV degree we do not refer to further properties of the BPHZL
scheme as the Γef f , but classify the breakings by the algebra of symmetry transformations,
the nilpotency of BRS-transformations and the algebra of Wα . (For details see section 7.)
Especially we have to show that all the breakings of the ST identity are variations of the
linear ST operator (2.106) and that they can be absorbed into local contributions of Γbreak
without spoiling the normalization conditions especially (3.4):

sΓcl Γbreak = −∆brs (3.9)

Via equs. (3.7) and (3.9) the local contributions are uniquely fixed.
The construction as outlined above is not only interesting from an abstract point of
view for having properly defined the standard model but it passes through all the steps,
which have been carried out in explicit calculations, too. Especially the construction of
the symmetries in 1-loop order including the rigid Ward identity is essential if one wants
to proceed to higher orders of perturbation theory. Dimensional regularization makes the
analysis of Γbreak easier, because it is an invariant scheme for parity conserving gauge
theories, but also there it is well-known that the normalization conditions and especially
the infrared conditions have to be established by an explicit adjustment of naively non-
invariant counterterms, which lead to corrections of the weak mixing angle in the ST
identity and the Ward identities.

31
 According to the procedure we have outlined in this section we will proceed for con-
structing the 1PI Green functions of the standard model as follows:

1. We construct the most general ST operator and Ward operators of rigid symmetry,
which are in accordance with the algebraic characterization. (Section 4)

2. We impose normalization conditions according to the on-shell schemes which allow

to define proper 2-point functions in the LSZ-limit apart from the problem of unsta-
ble particles and derive the most general classical solution, which is in accordance
with the normalization conditions and the symmetries. (Section 5)

3. We classify the breakings according to the symmetries and show that they can be
absorbed into local contributions to the 1PI Green functions. (Section 7)

4. The algebraic characterization of the symmetry

transformations

4.1. The general ansatz and discrete symmetries

In section 2.4 we have derived the symmetries of the standard model for the 1PI-
Green functions in the tree approximation: the Slavnov-Taylor identity (2.103), the Ward
identities of rigid symmetries (2.112) and the local U(1)-Ward identity (2.113). The
functional operators depend in lowest order explicitly on the weak mixing angle θW ,
which is in the on-shell schemes defined by the ratio of the W- and Z-mass [26]

MW
cos θW ≡ (4.1)
MZ
It is obvious and can be seen also from explicit 1-loop calculations that the lowest order
gets higher order corrections. These higher order corrections depend on the normalization
conditions, one has chosen for fixing the 2-point Green functions. Since the standard
model includes massless particles, especially the photon and the corresponding φπ-ghosts,
it is even not possible to define the Green functions of higher orders by the symmetries
as given in lowest order. For this reason we construct in a first step towards quantization
the symmetry operators in a most general form and characterize them by their algebraic

32
properties. We restrict the analysis to the generating functional of 1PI Green functions
as defined in (2.99), which depends on the classical fields as well as on the external fields:

Γ ≡ Γ(Va , Ba , Φ, fi , ca , c̄a , ρα , Y, ψi , σα , Φ̂, q̂) (4.2)

The vectors, φπ-ghost and the B-fields have on-shell field indices a = +, −, Z, A

Vaµ = (W+µ , W−µ , Z µ , Aµ ) ca = (c+ , c− , cZ , cA )

(4.3)
Ba = (B+ , B− , BZ , BA ) c̄a = (c̄+ , c̄− , c̄Z , c̄A )

Since the theory is spontaneously broken, it is more adequate to introduce the scalars as
a 4-vector with indices a = +, −, H, χ

φa = (φ+ , φ− , H, χ) φ̂a = (φ̂+ , φ̂− , Ĥ, χ̂)

(4.4)
Ya = (Y+ , Y− , YH , Yχ ) qa = (q+ , q− , qH , qχ )

The external fields ρα and σα are three component fields

ρa = (ρ+ , ρ− , ρ3 ) σa = (σ+ , σ− , σ3 ) (4.5)

The charged vectors and scalars are complex fields with

ϕ∗+ = ϕ− , (4.6)

whereas the neutral vectors and scalars are real fields

ϕ∗a = ϕa a = Z, A, H, χ, 3 . (4.7)

We group the fermions into a vector according to

fiL = (νiL , eLi , uLi , dLi ) ψfRi = (ψνRi , ψeRi , ψuRi , ψdRi )
(4.8)
fiR = (eR R R
i , u i , di ) ψfLi = (ψeLi , ψuLi , ψdLi )

i = 1, ..., NF denotes the family index. The quantum number of fields are summarized in
the appendix.
The algebraic properties of the functional operators acting on Γ are the nilpotency of
the Slavnov-Taylor operator (2.105)

sΓ S(Γ) = 0 for any functional Γ (4.9)

sΓ sΓ = 0 if S(Γ) = 0

and the algebra of rigid operators (2.114)

[Wα , Wβ ] = ε̂αβγ I˜γγ ′ Wγ ′ (4.10)

33
with α, β, γ = +, −, 3, 4 and ε̂αβγ denotes the structure constants of SU(2) × U(1), which
are taken as completely antisymmetric in all 3 indices

 ε̂+−3 = −i
ε̂αβγ = (4.11)
 ε̂+−4 = 0

In addition to the algebra the Ward operators are specified by their transformation with
respect to complex conjugation:

W+† = W− W3† = W3
(4.12)
W−† = W+ W4† = W4

which is assigned in agreement with the tree approximation.

Since the functional Γ will be constructed as a simultaneous solution of rigid trans-
formations and the ST identity the respective functional operators have to satisfy the
consistency relation

Wα S(Γ) − sΓ Wα Γ = 0 for any functional Γ (4.13)

Discrete and global unbroken symmetries, we want to impose on the functional Γ, can
be imposed to all orders and are not affected by renormalization. These symmetries are
electric and φπ-charge neutrality:
Z 
δ δ δ δ δ δ
Wem = −i dx W+ − W− + B+ − B− + c+ − c−
δW+ δW− δB+ δB− δc+ δc−
δ δ δ δ δ δ
+ c̄+ − c̄− + ρ+ − ρ− + σ+ − σ−
δc̄+ δc̄− δρ+ δρ− δσ+ δσ−
δ δ δ δ
+ φ+ − φ− + Y+ − Y−
δφ+ δφ− δY+ δY−
δ δ δ δ
+ φ̂+ − φ̂− + q+ − q−
δ φ̂+ δ φ̂− δq+ δq−
NF 
X δ δ δ δ
− − ei + ψ̄ei
Qe (ēi − )
i=1 δēi δei δ ψ̄ei δψei ψei
δ δ δ δ
+Qu (ūi − ui + ψ̄ui − ψu )
δūi δui δ ψ̄ui δψui i
δ δ δ δ 
¯
+Qd (di ¯ − di + ψ̄di − ψd )
δ di δdi δ ψ̄di δψdi i
Z 
δ δ δ δ δ δ
Wφπ = dx ca − c̄a − ρα − 2σα − Ya + qa (4.14)
δca δc̄a δρα δσα δYa δqa
NF 
X  
δ δ
− ψ̄mi + ψmi
i=1 δ ψ̄mi δψmi

34
Since fermions and quarks cannot be mixed and since we forbid explicitly family mixing
we have some further symmetries which correspond to lepton and quark family number
conservation:
Z 
δ δ δ δ
Wli = i dx ēi − ei + ψ̄ei − ψe (4.15)
δēi δei δ ψ̄ei δψei i

L δ δ L δ δ R
+ νi L − L νi + ψνi R − R ψνiR
δνi δνi δψνi δψνi
Z 
δ δ δ δ
Wqi = i dx d¯i ¯ − di + ψ̄di − ψd (4.16)
δ di δdi δ ψ̄di δψdi i

δ δ δ δ
+ ūi − ui + ψ̄ui − ψu
δūi δui δ ψ̄ui δψui i

The functional Γ is to all orders invariant under these global symmetries by construction:

Wem Γ = 0 Wli Γ = 0
(4.17)
Wφπ Γ = 0 Wqi Γ = 0

In addition we have also colour SU(3)-invariance, which we do not consider explicitly. All
the functional operators when applied on Γ are seen to be restricted with respect to these
global symmetries, especially

[Wem , W+ ] = −iW+ [Wem , W3 ] = 0

(4.18)
[Wem , W− ] = +iW− [Wem , W4 ] = 0

and
[Wli , Wα ] = 0 [Wqi , Wα ] = 0 (4.19)

The functional Γ we consider in this paper is also invariant with respect to CP-trans-
formations. Therefrom it is derived that the rigid operators as well as the Slavnov-
Taylor operator have definite transformation properties with respect to CP. Explicitly it
is possible to characterize them by their transformation as given in the tree approximation:
CP CP
W+ −→ −W− W3 −→ −W3
CP CP (4.20)
W− −→ −W+ W4 −→ −W4

whereas the ST operator is CP-even.

We therefore make for the ST operator the ansatz:
Z  
δΓ δΓ 
S(Γ) = Z4 (sin Θg3 ∂µ cZ + cos Θg3 ∂µ cA ) sin ΘV4
+ cos ΘV4 (4.21)
δZµ δAµ
δΓ  δΓ δΓ  δΓ σ  δΓ δΓ 
+ µ z ρ cos ΘV3 − sin ΘV3 + z cos Θg3 − sin Θg3
δρ3 δZµ δAµ δσ3 δcZ δcA

35
 δΓ δΓ δΓ δΓ δΓ δΓ δΓ δΓ δΓ ˜ δΓ
+ µ + µ + + + Iaa′
δρ+ δWµ,− δρ− δWµ,+ δσ+ δc− δσ− δc+ δYa δφa′
NF 
X 
δΓ Γδ δΓ Γδ
+ R
+ + h.c.
i=1 δψfi δfi
L
δψiR δfiL

δΓ δΓ
+Ba ĝab + qa =0
δc̄b δ φ̂a
It is neutral with respect to electric charge, commutes with the operators Wli and Wqi
and arises φπ charge by one unit. Defining its linear version sΓ as given in (2.106) one
immediately checks the nilpotency properties (4.9). The three independent angles ΘV3 , ΘV4
and Θg3 describe, how vectors and ghosts are rotated with respect to the abelian subgroup.
In the writing of (4.21) we have already anticipated that we are able to absorb constants
in external fields at will. The coefficients z ρ as well as z σ could be reabsorbed into the
external fields ρ3 and σ3 , but we take them as arbitrary for a proper adjustment later
on. For similar reasons we also keep Z4 in the ansatz. The matrix ĝab is an arbitrary
neutral matrix, it can be introduced into the ST identity without spoiling nilpotency and
rigid symmetry (see (4.40)). Its explicit form will be considered, when we give the general
classical solution of the gauge-fixing ghost sector.
For the Ward operators of rigid symmetry we take the most general ansatz, which is
linear in fields:
Z 
δ ˜ δ + cb âg ′ I˜cc′ δ + c̄b âḡ ′ I˜cc′ δ
Wα = I˜αα′ dx Vbµ âVbc,α′ I˜cc′ µ + Bb âB bc,α′ Icc′ bc,α bc,α
δVc′ δBc′ δcc′ δc̄c′
δ δ δ δ
+ φb bφbc,α′ I˜cc′ + Yb bYbc,α′ I˜cc′ + φ̂b bφ̂bc,α′ I˜cc′ + qb bqbc,α′ I˜cc′
δφc′ δYc′ δ φ̂c′ δqc′
δ δ
+ vcα′ I˜cc′ + v̂cα′ I˜cc′
δφc′ δ φ̂c′
δ δ
+ ρβ aρβγ,α′ I˜γγ ′ + σβ aσβγ,α′ I˜γγ ′
δργ ′ δσγ ′
NF 
X δ δ fi † L ˜
+ fiL hffif ′ ,α′ + hf ′ f,β fi Iβα′
i=1 δf ′ Li δf ′ Li
δ δ
+ ψfRi hψf fi ′ ,α′ + R hfψ′if,β †
ψfRi I˜βα′
δψfL′ δψf ′
i i

δ δ fi † R ˜
+ fiR h̃ffif ′ ,α′ + h̃ ′ f Iβα′
δfi′R δfi′R f f,β i

δ δ ψi † L ˜ 
+ ψfLi h̃ψf fi ′ ,α′ + L h̃f ′ f,β ψfi Iβα′ (4.22)
δψfL′ δψf ′
i i

The coefficients are restricted by the prescription for complex conjugation (4.12) and by
electric charge conservation (4.18).

36
 In the notation these properties are taken into account by having neutral index struc-
ture throughout and changing + and − by complex conjugation. Well-defined transfor-
mation properties under CP (4.20) yields furthermore that âbc,α and aβγ,α as well as h̃f f ′ ,α
and hf f ′ ,α are imaginary. Similar restrictions are derived for the coefficients bbc,α . Family
mixing as well as lepton quark mixing are forbidden according to (4.19).
In the following we will solve the algebra (4.10) as well as the consistency equation
(4.13) in all generality. Since we construct the Green functions in perturbation theory,
it would be also sufficient to start from the tree approximation and consider its possible
perturbations. Such a treatment, however, would disguise the simple algebraic structure
of the final solution.

4.2. The vector-ghost sector

Evaluating the algebra of rigid operators (4.10) for the vectors, φπ-ghosts, the B-fields
and the external fields σα and ρα yields the following representation equations for the
coefficients:

âϕα I˜ âϕβ − âϕβ I˜ âϕα = −ε̂αβγ I˜γγ ′ âϕγ′ ϕ ≡ Vaµ , ca , c̄a , Ba

aΥ ˜ Υ Υ˜ Υ ˜ Υ
α Iaβ − aβ Iaα = −εαβγ Iγγ ′ aγ ′ Υ ≡ ρα , σα (4.23)

Here we have introduced a matrix notation: (âα )bc = âbc,α denotes 4 × 4 matrices and
(aα )βγ = aβγ,α 3 × 3 matrices. Due to CP-invariance the non-trivial solutions of (4.23)
are uniquely related to the adjoint representation:

âϕα ∼ ε̂α âΥ

α ∼ εα (4.24)

with (ε̂α )βγ = ε̂βγα defined by the structure constants of SU(2)×U(1) (4.11) and (εα )βγ =
εβγα by the structure constants of SU(2). From this special solution one obtains the
general solution by the following equivalence transformations:

−1
(âα )bc = ŷbβ ε̂βγα ŷγc
(aα )βγ = yββ ′ εβ ′ γ ′ α (y −1 )γ ′ γ (4.25)

The matrices y and ŷ have to be chosen in accordance with the discrete symmetries.
When we consider the general classical solution of the standard model it is seen that the
equivalence transformations are related to field redefinitions. Therefore we parameterize

37
them in the following way:
 
ẑW 0 0  



 zW 0 0
 0 ẑW 0 0   
(ŷ T )αa ≡ ẑαa =   yβγ ≡ zβγ =  0 zW 0 


 0 0 ẑZ cos θZ −ẑA sin θA 
 
0 0 z3
0 0 ẑZ sin θZ ẑA cos θA
(4.26)
We have suppressed the field indices, but one has to keep in mind, that the algebra allows
independent field redefinitions for each field we consider.
In the tree approximation the representation matrices are given by

â(0) T (0)
α = O (θW )ε̂α O(θW ) and aα = εα (4.27)

for all fields in question, i.e. one has in perturbation theory for the propagating fields
ϕ
ẑαb = O(θW )αa (1 + δẑ ϕ )ab with (δẑ ϕ )ab = O(h̄) (4.28)

The matrix O(θW ) is the orthogonal matrix, which transforms the SU(2) × U(1) gauge
fields into on-shell fields (2.28). The matrix ẑ is however not completely specified by the
representation matrices, indeed it is seen that equivalence transformations with diagonal
matrices ẑinv leave the adjoint representation invariant:
−1 −1
ε̂α = ẑinv ε̂α ẑinv εα = zinv εα zinv (4.29)

if  
ẑ2 0 0  



 z2 0 0
 0 ẑ2 0 0   
ẑinv ≡   zinv = 
 0 z2 0  (4.30)
 0 0 ẑ2 0 
 
0 0 z2
0 0 0 ẑ1

There are several possibilities to parameterize the remaining parameters. A symmetric

parameterization, which is well adapted to treat higher order corrections of the vectors,
is given by
zZ cos θZ zA sin θA
rA = rZ = (4.31)
zW cos Θ zW sin Θ
1
cos Θ = √
1 + tan θZ tan θA
In this parameterization the general solution of (4.23) reads explicitly
 
0 0 0 0
 
 0 0 −1 −1
−irZ cos Θ irA sin Θ 
 
â+ =   â3 = ε̂3 (4.32)
 0 irZ cos Θ 0 0 
 
0 −irA sin Θ 0 0

38
  
0 0 irZ−1 cos Θ −irA−1 sin Θ
 
 0 0 0 0 
 
â− =   â4 = 0 (4.33)
 −irZ cos Θ 0 0 0 
 
irA sin Θ 0 0 0
One has to determine three independent parameters for each field in the rigid Ward
identity. They are fixed by the normalization conditions imposed on the 2-point Green
functions. Vice versa it is seen that for the vectors we could also choose rA = 1 and
rZ = 1 replacing two normalization conditions by the Ward identities of rigid symmetry.
Such a choice corresponds to the minimal on-shell scheme [31].
Finally the consistency equation between the Ward operators of rigid symmetry and
the ST operator (4.13) relates the angles appearing in the ST operator to the parameters
of rigid Ward operators. In the parameterization (4.31) one gets
V
rA V
rZ
tan ΘV3 = V
rZ
tan ΘV tan ΘV4 = V
rA
tan ΘV
r
z ρ r3ρ = 1
V )2
(rZ
cos2 ΘV + 1
V )2
(rA
sin2 ΘV (4.34)

and similar equations for the parameters of the ghosts and σ-fields
v
u
g rAg u 1 1
tan Θ3 = g tan Θg z σ r3σ = t
2 cos2 Θg + 2 sin2 Θg (4.35)
rZ (rZg ) (rAg )
It has to be proven, that these relations can be consistently maintained to all orders of
perturbation theory. In the tree approximation they are obviously fulfilled. Furthermore
it is seen, that the normalization constants z σ and z ρ can be fixed by the Ward identity
of rigid symmetry. Requiring that the external fields transform to all orders just as in the
tree approximation
aρβγα = εβγα aσβγα = εβγα (4.36)
the parameters z σ and z ρ are uniquely determined.
In order to determine the transformation matrices of the B-fields âB α , it has to be
observed, that the gauge fixing is linear in propagating fields. Differentiating the func-
tional of 1PI Green functions with respect to Ba therefore yields a local expression to all
orders of perturbation theory, which allows to fix the normalization of the B-fields on the
longitudinal parts of the vectors:
δΓ
= ξab Bb + I˜ab ∂ µ Vµb + rbc,a φ̂b φc + wca φc + ŵca φ̂c (4.37)
δB
Applying the Ward operators of rigid symmetry on this local expression it is seen, that
the transformation of the Ba -fields is completely governed by the vectors:
T
(aB V
α ) = −(aα ) (4.38)

39
which reads for the parameters introduced above (4.31)
1 1
rAB = rZB = tan ΘB = tan ΘV (4.39)
rAV rZV
One is able to establish rigid symmetry quite trivially on the B-dependent part of the
generating functional. Accordance with rigid symmetry directly restricts the independent
parameters appearing in (4.37). The explicit form is given in section 5.4.
Finally the consistency condition (4.13) relates the matrix ĝab to the rigid transforma-
tions of anti-ghosts:
−1T
âḡbc,α = −ĝbb
T V −1T
′ ab′ c′ ,α ĝc′ c = −(ĝz V )Tbβ εβγα (ĝz V )γc (4.40)

From rigid invariance it is therefore allowed to introduce an arbitrary matrix into the
BRS-transformation of ghosts. From (4.40) it is obvious that such a general ansatz is
related to different field redefinitions of B-fields and anti-ghosts and, finally, vectors and
anti-ghosts.

4.3. The scalar sector

The algebra for the coefficients of the scalar fields has the same form as the one for
the vectors

bsα I˜ bsβ − bsβ I˜ bsα = −ε̂αβγ I˜γγ ′ bsγ ′ s ≡ φa , Ya , φ̂a , qa (4.41)

with (bα )bc = bbc,α . The solution, however, is distinguished from the one of the vector
representation equations due to a different transformation behaviour of scalars with re-
spect to CP: The general solution of the scalar representation equations (4.41) is the
fundamental representation with its equivalence class:

bsα ∼ t̂α (4.42)

with
   
0 0 0 0 0 −i 0 0
   
1
 0 0 i −1 
 1
 i 0 0 0 

t̂+ =   t̂3 =  
2
 0 −i 0 0 
 2
 0 0 0 1 

0 1 0 0 0 0 −1 0
   
0 0 −i −1 0 −i 0 0
   
1
 0 0 0 0  s Gs 
 i 0 0 0 
t̂− =   G t̂4 =   (4.43)
2
 i 0 0 0  2 
 0 0 0 −1 

1 0 0 0 0 0 1 0

40
It involves in the abelian component an undetermined parameter for each field. The 4-
dimensional representation we have chosen here is equivalent to the complex 2-dimensional
representation, which is usually assigned to the scalars in the tree approximation and
which we have introduced in section 2. The general solution is obtained from the special
solution (4.43) by an equivalence transformation. Because the Ward operators have to
be CP-odd, the transformation matrices have to be real and diagonal. This means, that
mixing between Higgs- and the neutral would-be Goldstone is forbidden in a CP-invariant
theory:

(bα )bc = zbb′ (t̂α )bc zc−1

′c (4.44)

with  
z+ 0 0 0
 
 0 z+ 0 0 
 
zab =   (4.45)
 0 0 zH 0 
 
0 0 0 zχ
We have again suppressed the scalar field indices. The dependence of the representation
matrices on these parameters is quite simple, it is seen that the representation matrices
only involve the ratios
z+ zχ
r+ = rχ = . (4.46)
zH zH
As in the vector sector, rigid symmetry allows independent field redefinitions for each
scalar field. Likewise one can fix the field redefinitions of the charged and CP-odd com-
ponents by the Ward identity of the tree approximation:

r+ = 1 + δr+ rχ = 1 + δrχ with δra = O(h̄) (4.47)

Finally the consistency condition (4.13) relates the transformation of the external fields
Ya to the transformation of the propagating fields φa , and the transformation of qa to the
transformation of φ̂a :
bYbc,α = bφbc,α bqbc,α = bφ̂bc,α (4.48)
which reads for the free parameters involved

raY = raφ GY = Gφ
raq = raφ̂ Gq = Gφ̂

Because we are free to dispose over the external fields at will, as long as we do not find
any restrictions in the procedure of quantization, we restrict the transformation of the
fields φ̂a to be the same as the one of the quantum fields.

bφbc,α = bφ̂bc,α (4.49)

41
 The representation equations of the shifts (ṽbα ≡ vbα , v̂bα )

ṽbα I˜bb′ bφb′ cβ − ṽbβ I˜bb′ bφb′ cα = −ε̂αβγ I˜γγ ′ ṽcγ ′ (4.50)

are solved by
1 −1
v+− = −v−+ = i
2
−1
vr+ vχ3 = r v
2 χ
vχ4 = − 12 Gφ rχ−1 v
1 −1
(4.51)
v̂+− = −v̂−+ = i
2
−1
ζvr+ v̂χ3 = r ζv
2 χ
v̂χ4 = − 12 Gφ rχ−1 ζv
All the other components vanish according to charge neutrality and CP-invariance. The
free parameters are the shift of the quantum field v and the shift of the external field ζv:
MZ
v=2 cos θW sin θW + O(h̄) (4.52)
e

4.4. The fermion sector

The algebra for representation matrices of fermions has the following form:

hfαi hfβi − hfβi hfαi = −ε̂αβγ I˜γγ ′ hfγi′

h̃fi h̃fi − h̃fi h̃fi = −ε̂αβγ I˜γγ ′ h̃fi′
α β β α γ (4.53)

From the consistency equation with the Slavnov-Taylor operator it is seen, that the trans-
formation of external fields is governed by the transformation of propagating fields:

hfαi = hψαi h̃fαi = h̃ψαi (4.54)

The matrices (hfαi )f f ′ , f, f ′ = ν, e, u, d, are 4 × 4 matrices and (h̃fαi )f f ′ , f, f ′ = e, u, d, are

3 × 3 matrices. CP-invariance implies, that they are imaginary.
Lepton and quark number conservation enables one to treat quarks and leptons sepa-
rately and, actually, one only has to consider 2-dimensional representation matrices. The
non-trivial solution of the algebra is represented by the Pauli-matrices completed by the
unit matrix and its equivalence representations. We know from the tree approximation,
that left-handed fermions transform according to doublets, whereas right-handed fields
transform trivially under SU(2). This transformation behaviour cannot be spoiled in
perturbation theory. Therefore we assign in accordance with the tree approximation
τα
hδαi ∼ i h̃δαi ∼ 0 with α = +, −, 3 (4.55)
2
δi = li , qi is the index for quarks and leptons. The abelian component is not well-defined
by the algebra, but contains some free parameters. For the nontrivial solution one finds
always one undetermined parameter for left-handed leptons and quarks of each family
 
iGli 1 0 
hf4i =  (4.56)
0 iGqi 1

42
The singlet solution involves undetermined parameters for each right-handed fermion.
 
iGei 0 0
fi  
h̃4 = 
 0 iG ui
0 
 (4.57)
0 0 iGdi

Due to charge conservation and CP-invariance the equivalence transformations are carried
out by diagonal real matrices, which are related in the course of quantization to the field
redefinitions of right- and left-handed fields:
τα li −1 τα qi −1
hlαi = iz li (z ) hqαi = iz qi (z ) (4.58)
2 2
with    
νi ui
z 0  z 0 
z li =  z qi =  (4.59)
0 z ei 0 z di
The singlet representation is independent from field redefinitions. The charged compo-
nents of the doublet representation depend on the ratio of field redefinitions carried out
for left-handed up-and down type quarks and left-handed neutrinos and charged leptons,
respectively.
z ei z di
rli = νi rq i = u i (4.60)
z z
Having analysed the general structure of SU(2)×U(1) operators it is obvious, that rigid
symmetry does not restrict the number of independent field redefinitions. Therefore it is
allowed to impose independent normalization conditions for the propagating physical fields
as well as for the would-be Goldstones and φπ-ghosts without spoiling rigid symmetry.

4.5. The algebraic characterization of an abelian local Ward operator

The algebraic analysis of the last sections has shown that the SU(2)-components of
the rigid Ward operators are uniquely fixed up to equivalence transformations, which are
related to field redefinitions of the different fields in question. The abelian component W4 ,
however, involves several free parameters, which in higher orders appear as instabilities of
the abelian subgroup and have to be determined. If we assume now, that the instabilities
of the Ward operator are indeed the only breakings which appear in higher orders, then
one is able to fix some of the free parameters to all orders of perturbation theory. But
there are left the parameters which correspond to lepton and quark family conservation,
and it is obvious that they remain independent parameters of the abelian rigid Ward
operator.

43
 When one constructs the electroweak standard model from gauge invariance these
parameters are determined on the gauge transformation by the Gell-Mann Nishijima
relation, which ensures that the photon couples to the electromagnetic current. In the
course of renormalization the gauge symmetries are broken and the role of gauge symmetry
is taken over by BRS-symmetry and the Slavnov-Taylor identity. Via the nilpotency
properties it contains also the algebraic structure of the group in the external field part.
When solving the Slavnov-Taylor identity it is seen that one is lead to representation
equations for the BRS-transformations, which have the same form as the ones we have
solved for establishing rigid operators. It turns out, that the abelian component is also not
uniquely defined in the solution of the Slavnov-Taylor identity. In fact one finds the free
parameters which correspond to lepton and quark family conservation to be undetermined
as well. Leaving them as free parameters the photon will not couple properly to the
electromagnetic current but also on the currents associated with lepton and quark family
conservation. For this reason we have to use a local Ward identity in addition to the
Slavnov-Taylor identity for defining the gauge transformations of the abelian component
in an appropriate way. The local Ward identity of electromagnetic symmetry has non-
abelian components and does not exist in renormalizable gauges. Therefore we have to
use the abelian Ward operator for fixing the undetermined parameters continuing the
Gell-Mann Nishijima relation on a functional level to all orders of perturbation thery.
As we have already mentioned the Ward identities, which correspond to charge con-
servation and conservation of lepton and quark family number are not affected by renor-
malization. Therefore the identity
 NF
X 
Wem + (gli Wli + gqi Wqi ) Γ = 0 (4.61)
i=1

is valid to all orders of perturbation theory with arbitrary parameters gli and gqi . Adding
the general Ward operators W3 and W4 in a way that for vectors and scalars the electro-
magnetic Ward operator arises and the shifts vanish
1
W = W3 + W4 (4.62)
Gs
one gets by using (4.61) the following identity, when acting on the functional Γ:
Z  NF 
1 X δ qi R δ
WΓ = φ
i(Gui − Gqi − Gφ )uR
i + i(G di
− G )di R
G i=1 δuRi δdi
δ 
ei li φ
+ i(G − G − G )eR
i + h.c. Γ (4.63)
δeR
i

If one assumes, that these are the only breakings of the rigid Ward operators, which arise
in higher orders, then the coefficients appearing therein have to vanish to all orders of

44
perturbation theory, because they can be independently tested on non-vanishing vertices
of the classical action, namely on the scalar interaction, the fermion masses and the gauge
fixing:
Gu i = Gq i + Gφ Gdi = Gqi Gei = Gli (4.64)
These relations just state that the charges of leptons and quarks of each family differ by
one unit, which is determined by the charge of the W+ . The final abelian Ward operator
acting on Γ takes the form:
 NF
X 
W4 Γ = Wem − W3 + (gli Wli + gqi Wqi ) Γ (4.65)
i=1

with undetermined parameter gli and gqi . Here we have also chosen the overall normal-
ization appropriately, i.e. Gφ = 1. The problem of deriving a local Ward identity in
connection with the abelian subgroup is therefore not well-posed, but has to be restated
by requiring to have a local Ward identity in connection with the electromagnetic current.
Defining the local Ward operator connected with electromagnetic current conservation by
Z
w4Q ≡ wem − w3 with W3 − Wem = (w3 − wem ) (4.66)

it is seen that it is algebraically unique up to a total derivative acting on the differentiation

with respect to the abelian combination of vector fields. The operator
1 δ 1 δ
ŵ4Q = g1 w4Q − V
∂ sin ΘV − V ∂ cos ΘV (4.67)
rZ δZ rA δA
commutes with the Slavnov-Taylor operator and the Ward operators of rigid symmetry:
h i
ŵ4Q , Wα = 0 sΓ ŵ4 Γ − ŵ4Q S(Γ) = 0 for any Γ (4.68)

The final version of the abelian Ward identity we have to prove to all orders of perturbation
theory takes the form
!
1 δ 1 δ 1 1
g1 w4Q − V∂ sin ΘV − V ∂ cos ΘV Γ= V
✷BZ sin ΘV + V ✷BA cos ΘV (4.69)
rZ δZ rA δA rZ rA

with rZV , rAV and ΘV determined on the charged rigid SU(2) Ward identities (4.31). It
involves an overall normalization parameter, which depends on the parametrization one
has chosen and in higher orders on the normalization condition of the coupling. In the
QED-like on-shell schemes it is given by
e
g1 = + O(h̄) (4.70)
cos θW

The Ward identity (4.69) has to be established in higher orders of perturbation theory,
in order to fix the undetermined parameters appearing in the action as a consequence of

45
the instability of the abelian subgroup. These instabilities are connected with the fact,
that it is not possible to algebraically distinguish between gauging the electromagnetic
current and the currents associated with lepton and quark familiy conservation. If we
were not able to establish the abelian local Ward identity to all orders we had to impose
a normalization condition for one charged-fermion photon vertex of each family, but we
would loose thereby the control if the gauge symmetry is indeed the electromagnetic
symmetry and not the current associated with lepton and quark family conservation,
which one gauges in higher orders.
These observations have important consequences for the construction of the gauge
fixing and ghost sector: In order to identify the abelian Ward identity according to (4.66)
and (4.68) rigid SU(2) × U(1) Ward identities have to be established. The gauge fix-
ing sector has therefore to be constructed with the help of the external scalar fields as
introduced in section 2.2. In this procedure the number of independent gauge param-
eters is restricted. In order to avoid infrared divergent counterterms for the φπ-ghosts
one is forced to introduce in higher orders an independent ghost angle, which appears in
the Slavnov-Taylor identity and the Ward identities of rigid symmetry via different field
redefinitions of vectors and antighosts as derived in (4.40).

5. The general local solution of the Slavnov-Taylor identity and

rigid symmetries

5.1. The normalization conditions

As we have outlined in section 2, the construction of higher orders proceeds by proving,

that it is possible to adjust local contributions in such a way, that the functional of
1PI Green functions is invariant under the ST identity and the Ward identities of rigid
symmetry. Local contributions are algebraically separated into two classes: invariants of
the symmetry and non-invariant contributions (cf. (3.5)). The coefficients of the invariants
have to be fixed by appropriate normalization conditions and vice versa it has to be shown,
that the normalization conditions one wants to impose for a proper particle interpretation
only dispose of invariant coefficients.
We impose for all physical fields on-shell conditions as given in the literature (see

46
e.g. [28]). For vectors and scalars they read on the 2-point functions

ReΓT+− (p2 )|p2 =M 2 = 0 ReΓTZZ (p2 )|p2 =M 2 = 0

W Z
(5.1)
ΓTAA (p2 )|p2 =0 = 0 ReΓHH (p2 )|p2 =m2 = 0
H

The photon and the Z-boson are not distinguished by quantum numbers and mix from
1-loop order onwards. Therefore they have to be separated on-shell:

ΓTZA (p2 )|p2 =0 = 0 ReΓTZA (p2 )|p2 =M 2 = 0 (5.2)

Z

The Higgs and the neutral would-be Goldstone are distinguished by their transformation
properties under CP. For this reason the respective conditions for scalars are valid by
construction in a CP-invariant theory. The transversal part of the vector 2-point functions
is defined according to
pµ pµ T pµ pµ L
ΓVaµ Vbν ≡ −(η µν − )Γ ab − Γ (5.3)
p2 p2 ab
For the unstable particles the counterterms are fixed by the requirement that the real
part of the 2-point functions is vanishing. This prescription has to be reanalysed [40], if
one constructs the S-matrix and especially wants to prove gauge parameter independence
of physical quantities. For the construction of finite Green functions it is certainly a well-
defined normalization condition, which continues the tree approximation of the on-shell
scheme to higher orders in a proper form.
Because the residua are finally canceled when constructing the S-matrix, there is quite
some arbitrariness involved in the respective normalization conditions. In the complete
on-shell scheme the residua of all physical particles are fixed at the pole position. In
order to avoid on-shell infrared divergencies we modify the complete on-shell scheme by
introducing a normalization point κ2a for each vector and impose for the transversal part
of the vectors:

Re∂p2 ΓT+− (p2 )|p2 =κ2 = 1 Re∂p2 ΓTZZ (p2 )|p2 =κ2 = 1 Re∂p2 ΓTAA (p2 )|p2 =κ2 = 1 (5.4)
W Z A

In this form they allow to switch between different normalization conditions by adjusting
κ2a . As we have already mentioned two of these normalization conditions can be replaced
by the Ward identities of rigid symmetry, which corresponds to the minimal on-shell
scheme.
Finally one has to specify normalization conditions for the residua of the scalars,
which is carried out similarly as above. As it will be seen from the general solution of
the ST identity also the residua of unphysical Goldstone bosons are not fixed by the ST
identity, but could be fixed on the Ward identities of rigid symmetry. Because they are not

47
considered in external legs in physical scattering processes, the divergencies appearing in
dimensional regularization are often subtracted according to the MS scheme. In order to
remain quite general in the construction we impose normalization conditions on arbitrary
normalization points µ2a :

Re∂p2 Γ+− (p2 )|p2 =µ2 = 1 Re∂p2 ΓHH (p2 )|p2 =µ2 = 1 Re∂p2 Γχχ (p2 )|p2 =µ2χ = 1 (5.5)
W H

The normalization conditions for fermions are listed in the literature quite generally
also for the case, that there is CP-violation via the CKM matrix [28]. They simplify
considerably, if one assumes lepton and quark family conservation. Decomposing the
fermion 2-point functions according to

Γf¯i fi = p/ΓLfi (p2 ) 21 (1 − γ5 ) + p/ΓR 2 1 m 2

fi (p ) 2 (1 + γ5 ) − mfi 1Γfi (p )

≡ p/ − mfi + p/ΣLfi (p2 ) 21 (1 − γ5 ) + p/ΣR 2 1 m 2

fi (p ) 2 (1 + γ5 ) − mfi 1Σfi (p ) (5.6)

the on-shell conditions read

Re(ΓLfi (p2 ) − Γm 2
fi (p ))|p2 =m2 = 0 Re(ΓR 2 m 2
fi (p ) − Γfi (p ))|p2 =m2 = 0 (5.7)
fi fi

They impose pole conditions on the Dirac spinors and forbid parity violation for the on-
shell propagators. On-shell residua are endangered by infrared divergencies as it is the
case in QED with a massless photon. We therefore introduce off-shell conditions:

Re∂p2 (p2 ΓLfi (p2 ) + p2 ΓR 2

fi (p ))|p2 =κ2 = 1 (5.8)
i

Just by construction of vertex functions it is ensured that

ΓH = 0 (5.9)

which forbids to introduce linear Higgs field terms into the local contributions of higher
order corrections.
For proving unitarity of the physical S-matrix we have also to impose normalization
conditions on the unphysical fields. The poles of propagators of the longitudinal parts of
the vectors, of the unphysical would-be Goldstones and the Faddeev-Popov ghosts are seen
to be related by the ST identity. The normalization conditions on the poles of unphysical
particles are most easily established on the Faddeev-Popov fields and read:

ReΓc̄+ c− (p2 )|p2 =ζW M 2 = 0 ReΓc̄Z cZ (p2 )|p2 =ζZ M 2 = 0 Γc̄A cA (p2 )|p2 =0 = 0 (5.10)
W Z

Furthermore one has to require on-shell separation for neutral ghosts

ReΓc̄Z cA (p2 )|p2 =ζZ M 2 = 0 Γc̄Z cA (p2 )|p2 =0 = 0

Z
(5.11)
ReΓc̄A cZ (p2 )|p2 =ζZ M 2 = 0 Γc̄A cZ (p2 )|p2 =0 = 0
Z

48
Finally we impose also normalization conditions on the residua of the ghost propagators:

Re∂p2 Γc̄+ c− (p2 )|p2 =κ2 = 1 Re∂p2 Γc̄Z cZ (p2 )|p2 =κ2 = 1 Re∂p2 Γc̄A cA (p2 )|p2 =κ2 = 1 (5.12)
W Z A

Solving the ST identity for the most general local action which is compatible with UV
dimension 4, the parameters, which are fixed order to order by normalization conditions,
have to be free parameters in terms of which all the other couplings are determined. The
local contributions which are fixed by the above normalization conditions are given by
Z 
Γgen
bil = − 14 (∂ µ Vaν − ∂ ν Vaµ )Zab
V
(∂µ Vνb − ∂ν Vµb )+ 21 Vaµ MVab Vµb

+ 12 ∂ µ φa Zab
S
∂µ φb − 21 MH2 H 2 (x)

+ iZfRi f¯iR ∂/fiR + iZfLi f¯iL ∂/fiL − Mfi (f¯iR fiL + f¯iR fiL )

g
− c̄a Zab ✷cb − c̄a Mgab cb (5.13)

The matrices and parameters are chosen in accordance with charge neutrality and CP-
S
invariance, especially Zab is a diagonal matrix. In perturbation theory the parameters are
order by order determined by the above normalization conditions:
V
Zab = I˜ab + δZab
V
ZfRi = 1 + δZfRi
(5.14)
S
Zab = I˜ab + δZab
S
ZfLi = 1 + δZfLi

and respective expressions for the Higgs mass and fermion masses

MH2 = m2H + δm2H Mfi = mfi + δmfi (5.15)

The vector mass matrix is non-diagonal and can be decomposed into an orthogonal matrix
and a diagonal matrix:
 
2 2
0 M+− 0 0 MW + δMWM+− =
 
 M+− 0 0 0  MZ2 + δMZ2MZZ =
MVab = O (θ) 
T
 with 
 O(θ)
 0 0 MZZ 0  0 MAA =
 
0 0 0 MAA 0 + δθ θ =
(5.16)
2
MW , MZ2 , m2H and mfi are the physical masses of the particles. The explicit form of
the local counterterms is of course dependent on the way one has constructed the finite
renormalized 1PI Green functions. The objects we are able to talk about in a scheme
independent way are the finite Green functions. Constructing them in accordance with the
symmetries, they are finally governed by the normalization conditions and are independent
of the scheme, one has used for subtracting the divergencies. Especially the conditions

49
for separating massless and massive particle at p2 = 0 (3.4)

ΓZA (p2 = 0) = ΓAA (p2 = 0) = 0 (5.17)

2 2 2
Γc̄A cZ (p = 0) = Γc̄Z cA (p = 0) = Γc̄AcA (p = 0) = 0

have to be established on the finite 2-point functions in order to be able to carry out in-
frared finite higher order calculations. In the BPHZL scheme, which treats massless par-
ticles quite systematically, these normalization conditions are implemented in the scheme.
One has therefore δθBP HZL = 0. In dimensional regularization these normalization con-
ditions have to be carefully implemented by adjusting e.g. δθdim .

5.2. The symmetry transformations and the general action

For finding the invariant counterterms, which are added order by order in perturbation
theory to the nonlocal contributions, we have to solve the ST identity and the Ward
identities of rigid symmetry for the most general local action Γgen
cl , which is compatible
with renormalizability by power counting (cf. (3.5) and (3.7)).

S(Γgen
cl ) = 0 Wα (Γgen
cl ) = 0 (5.18)

For solving these equations one could proceed in a perturbative expansion, but as for
the Ward operators of rigid symmetry such a treatment disguises the simple algebraic
structure of the final solution, hence we proceed differently.
For the ST operator and the Ward operators we take the general operators as they are
determined by consistency and by the SU(2) × U(1)-algebra in section 3 from the general
ansatz (4.21) and (4.22). The ST operator is written in the following form:
Z  
g g 1 δΓ 1 δΓ 
S(Γ) = (r4Z ∂µ cZ + r4A sin Θ
∂µ cA ) + cos Θ (5.19)
rZ δZµ rA δAµ
δΓ  1 δΓ 1 δΓ  δΓ 1  g δΓ g δΓ

+ µ cos Θ − sin Θ + r 4A − r 4Z
δρ3 rZ δZµ rA δAµ δσ3 det r g δcZ δcA
δΓ δΓ δΓ δΓ δΓ δΓ δΓ δΓ δΓ ˜ δΓ
+ µ + µ + + + Iaa′
δρ+ δWµ,− δρ− δWµ,+ δσ+ δc− δσ− δc+ δYa δφa′
NF 
X 
δΓ δΓ δΓ δΓ
+ R
+ + h.c.
i=1 δψfi δfi
L
δψfRi δfiL

δΓ δΓ
+Ba (r V )−1
aα δĝ αb + qa
δc̄b δ φ̂a
The Ward operators involve the representation matrices of the fundamental and adjoint
representation with their equivalence classes (cf. (4.24),(4.42) and (4.55)). Because the

50
abelian Ward operator is related to the nonabelian neutral Ward operator W3 and to
the operators of global unbroken symmetries Wem and Wli , Wqi according to equ. (4.65),
we only have to consider the non-abelian Ward operators for establishing rigid symmetry
(α = +, −, 3).
Z 
δ ˜ δ (5.20)
Wα = I˜αα′ dx Vbµ (r V )Tbβ ε̂βγα′ (r V )−1T ˜
γc Icc′
V −1 V
µ + Bb (r )bβ ε̂bc,α′ (r )γc Icc′
δVc′ δBc′
−1T ˜ δ ˜ δ
+ cb (r g )Tbβ εβγα′ (r g )γc Icc′ + c̄b (δg)−1
bβ ε̂βγα′ δgγc Icc′
δcc′ δc̄c′
−1 δ −1 δ
+ (rbS φb + δHb v)t̂bc,α′ rcS I˜cc′ + Yb rbS t̂bc,α′ rcS I˜cc′
δφc′ δYc′
−1 δ −1 δ
+ (rbS φ̂b + ζ̂vδHb )t̂bc,α′ rcS I˜cc′ + qb rbS t̂bc,α′ rcS I˜cc′
δ φ̂c′ δqc′
δ δ
+ ρβ εβγ,α′ I˜γγ ′ + σβ εβγ,α′ I˜γγ ′
δργ ′ δσγ ′
NF X 
X iτα′ δi −1 δ δ δi iτα′ δi −1 L
+ FδLi r δi (r ) − r (r ) Fδi
i=1 δ=l,q 2 δFδL δFδLi 2
i


δi −1 iτα ′
δi δ δ δi −1 iτα′ δi R
+ ΨR
δi (r ) r − (r ) r Ψ δ
2 δΨLδi δΨRδi 2 i

There we have parameterized the equivalence classes of rigid transformations by r V , r S , r g

and r li , r qi taking into account, that we are able to determine field redefiniton matrices
up to invariant matrices. According to (4.31) and (4.32) we define the matrix r V by
 
1 0 0 0
 
 0 1 0 0 
V  
rαa = 



(5.21)
 0 0 rZ cos Θ −rA sin Θ 
0 0 rZ sin Θ rA cos Θ

The equivalent transformations for scalars and fermions are chosen as in (4.46) and (4.60):
 
S
r+ 0 0 0  
 
 S
0 r+ 0 0  1 0 
 
rS =   r δi =  (5.22)
 0 0 1 0  0 rδi
 
0 0 0 rχS

Furthermore the vector transformations in the ST operator (5.19) are parametrized in

agreement with the relations gained from the consistency between the general Ward op-
erators and the ST operator (cf. (4.34) and (4.36)). The transformation matrix of ghosts

51
is defined by  
1 0 0 0
 
 0 1 0 0 
g  
rαa =  g g  (5.23)
 0 0 r3Z r3A 
 
g g
0 0 r4Z r4A
Here we have disposed over the invariant abelian parameter by taking the linear BRS-
transformation of vector fields into the transformation matrix. The parameters in the
nonlinear transformations of the neutral ghosts are then chosen in accordance with nilpo-
tency and the consistency relation. Finally we have splitted the matrix ĝab which governs
the BRS-transformations of antighosts into the rigid transformation matrices of B-fields
and antighosts:
ĝab = (r V )−1
aβ δĝβb (5.24)
The rigid transformations of antighosts are determined from the consistency condition as
related to δĝαb .
With these operators we have to act on Γgen
cl , which consists of all local field polynomials
compatible with UV-dimension 4. (See the appendix for quantum numbers.) For finding
the invariant counterterms we do not use a specific scheme for treating massless particles.
If the parameters of the bilinear action (5.13) are indeed free parameters, it is ensured, that
we are able to establish all normalization conditions, especially the ones for separating
massless and massive particles at p2 = 0. Further restrictions on Γgen cl are neutrality
with respect to electric and Faddeev-Popov charge (4.14) and lepton and quark family
conservation (4.15)
Wem Γgen
cl = 0 Wli Γgen
cl = 0
gen gen (5.25)
Wφπ Γcl = 0 Wqi Γcl = 0
According to lepton and quark family conservation one is able to restrict the analysis to
CP-invariant field polynomials. From formal unitarity it is required, that local contribu-
tions are hermitean:
gen †
Γgen
cl = Γcl (5.26)
The general ansatz for Γgencl is quite lengthy. For the purpose of the present paper we
explicitly give only the most general external field part. When solving (5.18) it is seen
that the solution is traced back to representation equations for the general couplings of
the external field part. Finally it has to be shown that these couplings as well as the ST
identity and the Ward operators of rigid symmetry are uniquely determined as functions
of those parameters of which one disposes by the normalization conditions (5.13). Vice
versa the bilinear part of the action cannot possibly be restricted by the ST identity and
rigid symmetry, because otherwise unitarity and particle interpretation of the field theory
is endangered.

52
 We start the presentation of the general classical solution in the vector, scalar and
fermion part of the action. Having established there the ST identity, rigid invariance
follows as a consequence with well determined coefficients of rigid transformations for
vectors, scalars and fermions. The gauge fixing sector can then be established in accor-
dance with rigid invariance and allows to compute the ghost sector and at the same time
the transformation parameters of ghosts and antighosts. The procedure for adjusting
symmetric local contributions as outlined here in the abstract approach has to be done
exactly the same way in practice when calculating order by order in perturbation theory.

5.3. The vector-scalar and fermion part of the action

In this section we present the general solution of the ST identity in the vector, scalar
and fermion part of the action, which can be solved in combination with the external field
part self-consistently. The bilinear part of the most general local action Γgen
cl is given in
(5.13). Because the parameters appearing therein are fixed by normalization conditions,
they should not be determined in the course of the calculations. For this reason we
redefine the vector and scalar fields in such a way, that the bilinear part takes a simple
form:
o V o
Vµa = zab Vµa fioR = z̃fi fiR
(5.27)
φoa = zaS φa fioL = zfi fiL
In a first step we have to show that these parameters are uniquely determined from the
parameters appearing in the bilinear part of the action Γgen
bil , we fix by the normalization
conditions. Due to CP-invariance all field redefinitions can be chosen real. Fermion
und scalar field redefinitions are determined on the kinetic parts up to a sign, which is
irrelevant in perturbation theory and can finally be adjusted in the tree approximation:
 
2
0 z+ 0 0
 
 2
z+ 0 0 0  ZfRi = z̃f2i
S  
Zab =   (5.28)
 2
0 0 zH 0  ZfLi = zf2i
 
0 0 0 zχ2

Because photon and Z-boson are not distinguished by any quantum numbers, the vector
redefinition matrix nondiagonal in the neutral sector. We parameterize this matrix as in
(4.26):
 
zW 0 0
 
 0 z 0 0 
V  W 
zab =   (5.29)
 0 0 zZ cos θZ −zA sin θA 
 
0 0 zZ sin θZ zA cos θA

53
 V
On the kinetic parts zab is determined up to an orthogonal matrix:
V V T˜ V V T ˜ V
Zab = zaa′ Ia′ b′ zb′ b = (O(θ)z )aa′ Ia′ b′ (O(θ)z )b′ b (5.30)
This remaining orthogonal matrix can be fixed on the vector mass matrix. Requiring Moab
to be diagonal
 
o 2
0 MW 0 0
 
 o 2
MW 0 0 0 
 
Moab = (z V )−1T V −1
aa′ Ma′ b′ (z )b′ b with Moab =  o2
 (5.31)
 0 0 MZ 0 
 
0 0 0 MAo 2
V
finally determines zab uniquely up to signs. Transforming likewise the masses of the
fermions and the Higgs into bare masses:
mofi = z̃f−1
i
zf−1
i
Mfi moH 2 = MH2 zH
−2
(5.32)
the bilinear part of the action is transformed into the standard form expressed in terms of
V
bare quantities, which depend by definition on the arbitrary field redefinitions zab , zaS , zfi
and z̃fi .
Z 
Γgen
bil = − 14 (∂ µ Vaν − ∂ ν Vaµ )Zab
V
(∂ µ Vνb − ∂ ν Vµb )+ 21 Vaµ MVab Vµb

+ 21 ∂ µ φa Zab
S
∂µ φb − 21 MH2 H 2 (x)

+ iZfRi f¯iR ∂/fiR + iZfLi f¯iL ∂/fiL − Mfi (f¯iR fiL + f¯iR fiL )
Z 
= − 41 (∂ µ Vaoν − ∂ ν Vaoµ )I˜ab (∂ µ Vνbo − ∂ ν Vµb
o
)+ 21 Vaoµ Moab Vµb
o

+ 21 ∂ µ φoa I˜ab ∂µ φob − 21 mo2 o2

H H (x)

+ if¯ioR ∂/fioR + if¯ioL ∂/fioL − mofi (f¯ioR fioL + f¯ioR fioL ) (5.33)

These redefinitions are carried out throughout in Γgencl by redefining also all the arbitrary
couplings appearing therein as we did it for the masses. At the same time we have to
transform the original fields into bare fields in the ST identity and the Ward identities of
rigid symmetry. The arbitrary field redefinitions appearing thereby are absorbed into a
redefinition of parameters and external fields:
−1
ρoµα = zW ρµα ψfoR = zf−1 ψfRi
S −1
i i
(5.34)
o
Ya = za Ya ψfoL
i
= z̃f−1
i
ψfLi
The bare parameters are defined via the representation matrices of rigid invariance:
(r oV )−1 ε̂α (r oV ) = z V (r V )−1 ε̂α r V (z V )−1
(r oS )−1 t̂α (r oS ) = z S (r S )−1 t̂α r S (z S )−1
(r oδi )−1 τα (r oδi ) = ziδ (r δi )−1 τα r δi (ziδ )−1 (5.35)

54
 We are now ready to apply the ST operator on Γgen cl . The computation is quite lengthy,
therefore we only quote the final result and the crucial equations. Most important for the
solution is the external field part of the general action:
Z 
1
Γgen
ext.f. = − σα fα,bc cb cc
2
oµ
+ρoµα (a′g µ ′ o ′ o ′
αb ∂ cb + âa,bc Vb cc ) + Ya (tab,c φb cc + vab cb )
NF
X
+ (ψ̄foR oL ′i ¯oL oR ′i
′ fi hf ′ f,a ca − fi ψf ′ hf f ′ ,a ca
i i
i=1

+ ψ̄foL oR ′i
′ fi
i
h̃f ′ f,a ca − f¯ioR ψfoL ′i
′ hf f ′ ,a ca )
i
(5.36)

For simplicity we have suppressed the interaction polynomials of external scalars φ̂a .
These polynomials are considered in the context of the gauge fixing and ghost sector.
The arbitrary coupling matrices are restricted by the global symmetries (5.25), com-
plex conjugation (5.26) and CP-invariance. We have already carried out the transforma-
tion into bare fields for vectors, scalars and fermions and the respective external fields
and have transformed at the same time the original couplings into primed couplings (see
(5.37)).
Via the ST identity the couplings of the vector, scalar and fermion part of the general
action are determined as functions of the coupling matrices appearing in the external field
part and of the parameters of the ST identity. Explicitly they depend on the following
combinations:
g−1 V −1 g−1
âoa,bc = (r oV )−1 ′ oV −1
aα âαbc′ z̃cc′ = zW (r )aα âαbc′ zb′ b z̃cc′
g−1 S−1 g−1
toab,c = t′ab,c′ z̃cc ′
S
= zaa′ ta′ b′ ,c′ zb′ b z̃cc′

o ′ g−1 S g−1
vac = vac ′ z̃c′ c = zaa′ va′ c′ z̃c′ c

g−1 g−1
hoi ′i −1 i
f f ′ c = hf f ′ c′ z̃c′ c = zf zf ′ hf f ′ c′ z̃c′ c
g−1 g−1
h̃oi ′i −1 i
f f ′ c = h̃f f ′ c′ z̃c′ c = z̃f z̃f ′ h̃f f ′ c′ z̃c′ c (5.37)
g
z̃ab denotes a ghost transformation matrix which arises from the linear part of vector
transformations and the matrix r oV : Vector transformations consist of the linear part ap-
g
pearing in the ST identity r4b and the linear part of the nonlinear vector transformations,
′g g
aαb = zW aαb , α = +, −, 3:
X
g
z̃ab = (r oV )−1 ˜ ′ oV −1 g
aα Iα′ α aαb + (r )a4 r4b (5.38)
α,α′ =
+,−,3

Evaluating the ST identity one finds, that the couplings defined in (5.37) have to
satisfy the following equations:

55
• On the part containing the 4-dimensional vector polynomials âoabc is determined to
be completely antisymmetric and is seen to be the solution of the Jacobi identity:

âoabc = −âobac = âobca (5.39)

âoabc I˜aa′ âob′ a′ c′ + âoacc′ I˜aa′ âob′ a′ b + âoac′ b I˜aa′ âob′ a′ c = 0

• On the scalar-vector part, which contains the bare mass matrix as defined in (5.31),
o
the matrix vab is determined in terms of the bare masses:

o 2
o
|v+−


 |2 = MW
Moab = I˜a′ b′ vao′ a vbo′ b o
=⇒ (vχZ )2 = MZo 2 (5.40)


 o
(vχA )2 = 0

Therefrom it follows that the mass of the photon has to vanish and is not an inde-
pendent parameter of the theory:

MAo 2 = 0 (5.41)

The matrices toab,c have to be antisymmetric in the scalar indices

toab,c = −toba,c (5.42)

and satisfy the following representation equations:

toba,b′ I˜aa′ toa′ c,c′ − toba,c′ I˜aa′ toa′ c,b′ = −âob′ c′ a′ I˜a′ a tobc,a (5.43)

and
toba,b′ I˜aa′ vao′ c′ − toba,c′ I˜aa′ vao′ b′ = −âob′ c′ a′ I˜a′ a vba
o
(5.44)

• In the fermion part of the action we find on the bare mass terms

h̃oi oi
f f A = hf f A (5.45)

for all massive fermions f = e, u, d. Because the kinetic terms of bare fields are
normalized, one has furthermore

f = ν, u
hoi oi
f f ′ + = −hf ′ f − (5.46)
f ′ = e, d

In addition one gets the following representation equations for each family:

hoi oi oi oi o ˜ oi
f f ′ ,b hf ′ f ′′ ,c − hf f ′ ,c hf ′ f ′′ ,b = −âbca′ Iaa′ hf f ′′ ,a (5.47)
h̃oi ′ h̃oi′ ′′ − h̃oi ′ h̃oi′ ′′ = −âo ′ I˜a′ a h̃oi ′′
f f ,b f f ,c f f ,c f f ,b bca f f ,a

56
 It is straightforward to solve these equations: Due to (5.39) âoabc are qualified as
structure constants of SU(2) × U(1) and are related to the structure constants ε̂αβγ by:

âoabc = g2oε̂αβγ Oαa (θW

o o
)Oβb (θW o
)Oγc (θW ) (5.48)
o
θW and g2o are at this stage two arbitrary parameters, which parameterize the two remain-
ing parameters of the coupling matrix âoabc . Therefore the representation equations (5.43)
and (5.44) are equivalent to equs. (4.41) and (4.50) and their solutions can be read off
from the solutions (4.51) and (4.44):

tobc,a = g2o t̂bc,α Oαa (θW

o
) o
vba o
= vbα o
Oαa (θW ) (5.49)

Here we have already used that antisymmetry (5.42) singles out from the equivalence class
the antisymmetric solution. Most important are the solutions of the shift equations, which
relate the remaining undetermined parameters to the masses of Z-boson and W-boson.
Inserting the relations (5.40) into (4.51) determines Gφ , θW
o
and v o :
o o
o MW sin θW 2 o
cos θW = Gφ = − vo = o
M cosθW (5.50)
MZo o
cos θW g2o Z
o
(The signs of cos θW and v o are chosen in accordance with the tree approximation.) Only
the nonabelian coupling g2o remains undetermined.
In the same way the representation equations of the fermion matrices (5.47) are equiv-
alent to equs. (4.53) we have solved in equs. (4.55), (4.56) and (4.57). The relation (5.46)
singles out the antisymmetric solution:
τ+ τ−
ho+ δi = ig2o ho− δi = ig2o (5.51)
2 2
τ 3 1
hoZ δi = ig2o (cos θW
o o
+ sin θW Gδi )
2 2
τ3 1
hoA δi = ig2o (− sin θW
o o
+ cos θW Gδi )
2 2
is the solution of the doublet representation and
 
iGei 0 0
 
h̃oi
Z
o o 
= ig2 sin θW  0 iGui 0  
di
0 0 iG
 
iGei 0 0
 
h̃oi
A
o o 
= ig2 cos θW  0 iGui 0   (5.52)
di
0 0 iG

is the solution of the singlet representation. There we have introduced a matrix notation
as in section 3.4 (δi = li , qi ). Although it is not relevant for perturbation theory, we want

57
to mention, that on the mass terms doublet and singlet representations are distinguished.
The ST identity is only solved, if we assign to one chirality the doublet representation
and to the second the singlet representation. Inserting furthermore the relation (5.45),
which relates the algebraically undetermined parameters of the abelian subgroup, yields
immediately the relations (4.64). With (5.50) they read
1 1 1
(− tan ΘoW + Gqi ) = Gui (tan ΘoW + Gqi ) = Gdi (tan ΘoW + Gli ) = Gei (5.53)
2 2 2
As expected we remain with one undetermined parameter for each lepton and quark
family. Parameterizing Gδi by the electric charge, Qe = −1, Qd = − 13 , and an remainder
gδi ,
Gqi = − tan θW
o
(2Qd + 1 + 2gqi ) Gli = − tan θW
o
(2Qe + 1 + 2gli ) (5.54)
it is seen that the free parameters correspond to coupling the Noether currents of lepton
and quark family conservation to the photon.
Finally the angle Θo , which appears as a free parameter in the ST identity is deter-
mined as function of the bare vector mass ratio by inverting the relation between âoabc and
â′αbc , α = +, −, 3:
1 1
âoZ+− = −ig2o cos θW
o
= o
cosΘoâ′3+− g (5.55)
rZ z̃++
1 1
âA+− = ig2o sin θW
o
= − o sin Θo â′3+− g (5.56)
rA z̃++
i.e.
rAo
tan Θo = o
tan θW (5.57)
rZo
On the ST identity it is not possible to fix rAoV and rZoV , but they are determined by using
rigid symmetry.
Having solved the above equations the action of vectors, scalars and fermions and the
external field part is determined from the ST identity

S(Γgen gen
GSW + Γext.f ) = 0 (5.58)

without having specified the ghost redefinition matrices and without using rigid or local
gauge symmetry. Ingredients are only nilpotency of the ST identity, the bare form of
the bilinear action, which states especially that there are massive vector bosons, and the
global symmetries as charge neutrality and lepton and quark family number conservation.
Explicitly we find as solution of the ST identity in the vector, fermion and scalar sector
the following general action expressed in terms of bare fields:

Γgen o o o o o o o o o o
GSW (Va , φa , fi ) = ΓY M (Va ) + Γscalar (φa , Va ) + ΓY uk (φa , fi ) + Γmatter (Va , fi ) (5.59)

58
with
Z
1 ˜
ΓY M = − Goµν o
α Iαα′ Gµνα′ (5.60)
4
Z  
1 mo2
H MWo
o 2
Γscalar = (D µ Φo )† Dµ Φo − g2o o2
(Φo† o
Φ + 4 H )
4 MW g2o
NF Z 
X 
Γmatter = FloL
i
iD/FloL
i
+ FqoL
i
iD/FqoL
i
+ fioR iD/fioR
i=1
NF Z 
X g2o
ΓY uk = − mofi f¯io fio + √ o
(moei FloL
i
Φo eoR
i
i 2MW

+ moui FqoL
i
Φo uoR o oL o oR
i + mdi Fqi Φ̃ di + h.c.)

For notational convenience we have rewritten the 4-vector of scalars φoa into the complex
scalar doublet Φo and Φ̃o . The structure of the individual ST-invariant terms is the same
as in the tree approximation. Therefore it is seen that the information on the SU(2)×U(1)
algebra is completely transfered to the ST identity. Because the bare form of the action
o
has been fixed, the covariant derivatives are immediately computed as functions of MW
and MZo . The weak mixing angle in its bare form θW o
is defined by the bare vector mass
ratio
Mo
o
cos θW = Wo (5.61)
MZ
and is not an independent parameter of the theory (cf. (5.50)). From the above construc-
tion it is obvious, that the broken theory is considered and characterized in its own right
and one never refers to the underlying symmetric theory.
Goµν
α = Oαa (θW o
)(∂ µ Vaoν − ∂ ν Vaoµ ) + g2o I˜αα′ ε̂αβγ Oβb (θWo
)Oγc (θWo
)V oµ V oν (5.62)
√ b c!
τ  2 0
α
Dµ Φo = ∂µ Φo − ig2o Oαa (θWo
) − tan θW o
O4a (θW o
) Vµao
(Φo + o o
)
2 g2 MW
 Gδi 
o τα
D µ FδoL = ∂ µ
− ig 2 ( O (θ
αa W
o
)V a
oµ
+ O4a (θWo
)Vaoµ ) FδoL δ = l, q
i
2 2 i

1
D µ fioR = (∂ µ + ig2o (tan θW o
+ Gδi ))fiR O4a (θW o
)Vaoµ fi = ei , di
2
1
D µ fioR = (∂ µ + ig2o (− tan θW o
+ Gqi ))fioR O4a (θW o
)Vaoµ fi = ui
2
g
The external field part depends on the ghost redefinition matrices a′g αb and r4b . In accor-
dance with rigid symmetry (5.23) we introduce the following notation
g ˜ g
a′g
αb = ẑW Iαβ rβb (5.63)
These parameters will be finally fixed in the ghost sector on the bilinear parts of the
ghosts:
Z  o
gen g g2 g g
Γext.f. = −ẑW σα ε̂αβγ rβb cb rγc cc
2

59
 g ˜ g
+ρoµα ẑW (Iαβ ∂ µ rβb cb + g2oε̂αβγ Oβb (θW
o
)Vboµ rγc
g
cc )
√ !
 
o o† τα g g 1 o g o 2 0
+g2 Y (i ẑW rαa − i tan θW r4a ))ca (Φ + o o
) + h.c. (5.64)
2 2 g2 MW
NF  X
X τα g g 1 g
+ Ψ̄oR o
δi ig2 ( ẑW rαa + Gδi r4a )ca FδoL
i=1 δ=l,q 2 2 i

1 g
+ ψ̄eoL ig2o (tan θW
o
+ Gli )eoR
i r4a ca
i
2
1 g
+ ψ̄doL ig2o (tan θW
o
+ Gqi )doR
i r4a ca
i
2
 
1 g
o
+ ψ̄uoLi ig2o (− tan θW + Gqi )uoR
i r4a ca + h.c.
2

The general form of the ST-invariant action is obtained by transforming the bare field
into the original fields according to (5.27). The parameters of the bilinear action remain
arbitrary as we did not to have to dispose on them when solving the ST identity. Only
the bare mass of the photon is determined as zero from the ST identity and is not an
independent parameter of the theory.
g
Besides the ghost redefinition matrix rαb (5.38) there remain undetermined the non-
abelian coupling g2 and the fermion couplings Gli and Gqi . In order to embed the structure
o

of quantum electrodynamics into the standard model they have to be fixed on the local
abelian Ward identity as given in (4.66) and (4.67), remaining with one free parameter
g2o , which can be finally adjusted to the fine structure constant in the Thompson limit.
For this reason the Ward identities of rigid invariance have to be established.
The solution of the ST identity Γgen
GSW in the vector, scalar and fermion sector as
given above is immediately seen to be invariant under rigid symmetry. Applying the
Ward operators Wα (5.20) on Γgen GSW determines uniquely the matrices of equivalence
transformations:
r oV = O(θW
o
) r oS = 1 r oδi = 1 (5.65)
For the parameter v o , which appears as a free parameter in the Ward operators, one gets
o
o 2MW
v = (5.66)
g2o
Inverting the relations (5.35) finally yields r S , r V , r δi as functions of masses and field
redefinitions:
zZ q
o o o
rZ = cos(θW + θZ ) 1 + tan(θW + θZ ) tan(θW + θA )
zW
zA q
o o o
rA = sin(θW + θA ) 1 + cot(θW + θZ ) cot(θW + θA )
zW

60
 q
o o
tan Θ = tan(θW + θZ ) tan(θW + θA ) (5.67)

and
1 S z ei z di
raS = z rli = rq i = (5.68)
zH a zνi zui
The shift parameter v as defined in (4.51) is determined on the general classical action to
o
2MW
v= (5.69)
g2ozH
The Ward identities of rigid symmetry hold then without further restrictions on Γgen
GSW
and Γgen
ext.f .
One is now ready to apply the local Ward operator wQ (4.67). Requiring the Ward
identity (4.69) to be valid on the Γgen gen
GSW + Γext.f
 
1 δ 1 δ
g1 w4Q − V∂ sin ΘV − V ∂ cos ΘV (Γgen gen
GSW + Γext.f ) = 0 (5.70)
rZ δZ rA δA

determines Gli and Gqi as functions of the electric charge of the charged leptons (Qe = −1)
and down-type quarks (Qd = − 31 ):

Gqi = − tan θW
o
(2Qd + 1) Gli = − tan θW
o
(2Qe + 1) (5.71)

The overall normalization constant g1 is determined on the general classical action as

function of the nonabelian coupling g2o , the wave function renormalization and the bare
masses: v
u o
u
o o t tan(θW + θA )
g1 = zW g2 tan θW o
(5.72)
tan(θW + θZ )
After having applied the local Ward identity there remains only one coupling g2o , which
is not fixed on the mass terms and by symmetries. In a QED-like parametrization this
coupling is determined by the fine structure constant, which measures the interaction
strength of the photon to the electromagnetic current in the Thompson limit:

ū(p)ΓeeAµ (p, p, 0)u(p)|p2 =m2e = ieū(p)γµ u(p) (5.73)

In the tree approximation this relation yields

e
g2o = + O(h̄) (5.74)
sin θW
Respectively one can parameterize the bare coupling g2o by the electromagnetic bare cou-
pling eo and the bare mass ratio of W - and Z-boson
eo
g2o = o
with with eo = e + δe (5.75)
sin θW

61
 Transforming the bare fields of the general action back to original fields and expanding
the free parameters in perturbation theory determines the symmetric local contributions
(n)
Γinv (3.5), which are in agreement with the ST identity, rigid symmetry and the local Ward
identity, as specified by algebra and nilpotency. Since some combinations of parameters
(5.67) explicitly enter as parameters the symmetry operators, also the explicit form of the
ST identity and the Ward identities is modified in higher orders of perturbation theory.
In concrete calculation it is wideley used that dimensional regularization is an invariant
scheme if parity is conserved. Under such circumstances only the symmetric counterterms
appearing in the general ST-invariant action would appear as counterterms in the Γef f ,
which governs the calculation of Green functions in the Gell-Mann Low formula. The
Feynman rules of the vector, scalar and fermion part derived from such a symmetric
Γef f are listed for example in [37] and have exploited in constructing ST-invariant 1-loop
Green functions in the physical sector. Furthermore, if one had an invariant scheme, the
parameters of the ST identity and the Ward identity would be related to the respective
counterterms in renormalized perturbation theory as derived in (5.67), (5.68) and (5.72).
Of course a necessary prerequisite of exploiting such relations is the complete construction
of the ST identity, the rigid symmetries and the local abelian gauge Ward identity order
by order in perturbation theory for all Green functions involved, especially also for the
ones of the ghost sector. In the next sections we outline the construction of the gauge
fixing and ghost sector in the classical approximation and in higher orders, taking care in
preserving the Ward identities of rigid and local gauge invariance.

5.4. The gauge fixing and ghost sector

Classically the ghost sector has been completely determined as BRS-variation of the
gauge fixing function (cf. (2.69)). The respective relation is immediately derived for the
generating functional of 1PI Green functions by differentiating the ST identity (5.19) with
respect to the B-fields: !
δΓ δΓ
sΓ = −(r V )−1
aα δĝαb (5.76)
δBa δc̄b
δΓ
In the linear gauges, moreover, δB a
is local (4.37), because there are no vertices which
could constitute loop diagrams with external B-legs. Then eq. (5.76) yields the linear
ghost equations, which have to be established to all orders of perturbation theory.
In the tree approximation we have constructed the gauge fixing sector to be invariant
under Ward identities of rigid symmetry by introducing the external scalar fields φ̂a (see
(2.51) – (2.57)). As the abelian fermion couplings are not well determined from the ST

62
identity, Ward identities of rigid symmetries have to be maintained for the generating
functional of Green functions. For this reason we have to choose the gauge fixing sector
as being invariant under rigid transformations as specified in the vector and scalar part
of the action. To the external scalars we assign the transformation behaviour of the
propagating scalars and derive for the most general rigid invariant gauge-fixing sector the
following expression:
Z 
B 1 ˜ 1ˆ ˜ µ
Γ = ξBa (r V )−1 V −1T V −1 V −1T
aα Iαβ (r )βb Bb + ξBa (r )a4 (r )4b Bb + Ba Iab ∂ Vµb (5.77)
2 2
X
+ gξ Ba (r V )−1 S S
aα (rb φb + δHb v)t̂bc,α (rc φ̂c + δHc ζ̂v)
α=
+,−,3

+ ĜBa (r V )−1 S S
a4 (rb φb + δHb v)t̂bc,4 (rc φ̂c + δHc ζ̂v))

V
The matrix rαa is the three parameter matrix, which parameterizes the rigid transforma-
tions of vectors (5.21) and rbS are the three parameters of the scalar rigid transformations
(5.22), v denotes the shift parameter of the scalar field as it appears as parameter in the
Ward operators. These parameters are determined by the normalization conditions on
the vector and scalar 2-point functions and one cannot dispose of them in the gauge fixing
sector anymore. ΓB as given in (5.77) holds for the B-dependent part of the generating
functional of 1PI Green functions in linear gauges to all orders. The free parameters of
the gauge fixing sector are
ˆ ζ̂, Ĝ, gξ
ξ, ξ, (5.78)
Finally a further free parameter is the overall normalization of the external scalar field
φ̂a , which can be used to fix the parameter gξ at will. In QED-like parameterizations it
is convenient to adjust
e
gξ = (5.79)
sin θW
(5.77) yields the linear ghost equations, which are valid in this form to all orders (see [3]
for details) (α = +, −, 3):
δΓ δΓ S ˜
∂µ µ + gξ r ′ Ib′ b t̂bc,α (rcS φ̂c + δHc ζ̂v)
δρα δYb′ b
δ
+(rbS φb + δHb v)t̂bc,α rcS q̂c = −δĝαb
δc̄b
g g δΓ S ˜
✷(r4Z cZ + r4A cA ) + gξ Ĝ r ′ Ib′ b t̂bc,4 (rcS φ̂c + δHc ζ̂v)
δYb′ b
δ
+(rbS φb + δHb v)t̂bc,4 rcS q̂c = −δĝ4b (5.80)
δc̄b
Using rigid symmetery the ghost equations are immediately integrated yielding that the
generating functional of 1PI Green functions depends on specific combinations between

63
the external fields ρµα and Ya and antighosts, whereas the remaining contributions are
local. Splitting off also the local B-dependent part the generating functional of 1PI
Green functions can be decomposed in the following way:
Γ(Vaµ , Ba , ca , φa , φ̂a , σα , ρµα , Ya , c̄a , fi , ψfi )
= Γnl (Vaµ , ca , φa , φ̂a , σα , ρ′µ ′ B µ
α , Ya , fi , ψfi ) + Γ (Ba , Va , φa , φ̂a )
X 
g
−c̄a (δĝ)−1
a4 r4b ✷cb − gξ c̄a (δĝ)−1 −1
aα t̂bc,α + (δĝ)a4 Ĝt̂bc,4 (φb + δHb v)q̂c (5.81)
α=
+,−,3

with
ρ′µ
α = ρµα + ∂ µ c̄a (δĝ)−1
aα
X 
Yb = Yb − c̄a gξ I˜bb′
′
(δĝ)−1 −1
aα t̂b′ c,α + Ĝ(δĝ)a4 t̂b′ c,4 (φ̂c + δHc ζ̂v) (5.82)
α=
+,−,3

For simplification we have absorbed the irrelevant scalar redefinitions, i.e. rbS = 1. All
non-local contributions are contained in the part Γnl . The proof that all breakings of the
ST identity can be absorbed by adjusting local contributions, can be finally restricted to
this functional (see section 7). The solution of the ghost equations is quite trivial in a
massless symmetric gauge. This is not the case for the standard model where the ghosts
are massive. There one not only has to solve the ghost equation, but also one has to
show, that the on-shell normalization conditions for the ghost 2-point functions and in
particular the infrared conditions
Γc̄AcZ (p2 = 0) = Γc̄Z cA (p2 = 0) = Γc̄A cA (p2 = 0) = 0 (5.83)
can be fulfilled by adjusting the parameters ζ̂, Ĝ and the BRS-transformation matrix δĝαb

5.4.1. The classical approximation

First we solve the ghost equations in the classical approximation, taking into account
that we impose normalization conditions on the ghost 2-point functions as specified in
(5.10), (5.11) and (5.12). The bilinear part of the ghost action is therefore fixed (see
g
(5.13)) and all parameters have to be determined as functions of Zab and Mgab , and of
the free parameters of the vector and scalar part of the action. We proceed therefore
as in the vector sector and define on the bilinear part bare ghosts by the following field
redefinitions
g g
coa = zab cb c̄oa = z̄ab c̄b (5.84)
Inserting these bare fields in the bilinear part of the ghost action and requiring the bare
action to have a standard form determines z g and z̄ g up to a diagonal matrix:
Z
Γgen g g
bil,ghost = − (c̄a Zab ✷cb + c̄a Mab cb )

64
 Z
g g
= − (c̄oa (z̄ g )−1T g −1 o o g −1T g −1 o
aa′ Za′ b′ ✷(z )b′ b cb + c̄a (z̄ )aa′ Ma′ b′ (z )b′ b cb )
Z
!
= − (c̄oa I˜ab ✷cob + c̄oa Mog o
ab cb ) (5.85)

Explicitly one gets a relation between the field redefinitions of ghosts and antighosts on
the kinetic part:
I˜ab′ (z g )b′ b = (z̄ g )−1T g
aa′ Za′ b (5.86)
The remaining undetermined matrix can be used to fix the ghost mass matrix as being
diagonal:  
o o2
0 ζW MW 0 0
 
 ζ o M o2 0 0 0 
og  W W 
Mab =   (5.87)
 0 0 ζ o
M o2
0 
 Z Z 
o2
0 0 0 MgA
and
g ˜
Mog g −1T g −1 g T
ab = (z̄ )aa′ Ma′ c Icc′ (Z )c′ b′ (z̄ )b′ b (5.88)
We have parameterized the W and Z-ghost mass by the masses of the W - and Z-boson and
o
independent parameters ζW and ζZo . In contrast to the vector mass matrix the ghost mass
matrix is not required to be symmetric in the neutral components. Indeed considering
the 1-loop diagrams it is seen, that the 2-point function Γc̄Z cA and Γc̄AcZ get different loop
corrections, because interactions of scalars and ghosts are unsymmetric in Z-ghosts and
A-ghosts in the tree approximation (cf. (2.75)). For diagonalizing an arbitrary matrix
the equivalence transformation has to be carried out by an invertible matrix. Therefore
diagonalization of the ghost mass matrix determines the wave functions renormalization
g
of anti-ghosts z̄ab up to a diagonal matrix. Taking for the BRS-transformation matrix
δĝαb the ansatz  
1 0 0 0
 
 0 1 0 0 
 
δĝαb =   (5.89)
 0 0 cos γZ − sin γA 
 
0 0 sin γZ cos γA
it is possible to diagonalize the ghost matrix by adjusting the arbitrary angles γZ and
γA , and at the same time the three undetermined parameters of the diagonal matrix are
fixed. We want to mention that in massless nonabelian gauge theories the antighost field
redefinitions are not determined on the bilinear part of the action and can therefore be
completely fixed by the ST identity.
In the classical approximation the external field vertices δΓgen µ gen
cl /δρα and δΓcl /δYa are
local field polynomials. They have been determined in the last section as functions of
V g g
vector field redefinitions zab , of scalar field redefinitions zaS and of the matrix rαb and ẑW

65
(5.63). One can either take the explicit form of Γgen ext.f. or better one goes back to the bare
form of the action as given in (5.64). For proceeding with the bare form, we have also to
transform the scalars and vectors in the gauge fixing part to bare fields. The parameters,
which appear by carrying out this transformation, are absorbed into a redefinition of B-
fields, into the overall redefinition of the external scalar fields and into a redefinition of the
arbitrary parameters into a bare form. One has to note that in the classical approximation
V
the matrices rαb and raS are determined as functions of field redefinitions and bare masses
(cf. (5.67) and (5.68)), especially one has in the classical approximation
r o +θ ) 
tan(θW
(r V )−1 o V
aα Oαb′ (θW )(z )b′ b = zW δab + zW Z
o +θ )
tan(θW A
− 1 δa4 δb4 (5.90)
(r S )−1 S
aα (z )αb = zH δab (5.91)

The bare form of the gauge fixing is then given by

Z 
1 1
Γgen
g.f. = ξ o Bao I˜ab Bbo + ξˆo(sin θW
o
BZo + cos θW
o
BAo )2 + Bao I˜ab ∂ µ Vµb
o
(5.92)
2 2
X
+ gξ Bao Oaα
T o
(θW )(φob + δHb v o )t̂bc,α (φ̂c + δHc ζ̂ ov o )
α=
+,−,3

+ Ĝo Bao Oa4
T o
(θW )(φob + δHb v o )t̂bc,4 (φ̂oc + δHc ζ̂ ov o ))

with the bare fields

zW
Bao = (z V )−1T
ab Bb φ̂oa = raS φ̂a (5.93)
zH
and bare parameters

2
ξ o = zW ξ (5.94)
 tan(θ o + θZ )  o
2 tan(θW + θZ ) ˆ
ξˆo = zW
2 W
o
−1 ξ+ zW o
ξ
tan(θW + θA ) tan(θW + θA )
zW
ζ̂ o = 2
ζ̂
zH
v
u o
u tan(θW + θZ )
Ĝo = t Ĝ
o
tan(θW + θA )

One has finally to transform the ghost fields and B-fields in the ST identity and in the
external field part (5.64) to bare fields, absorbing the field redefinition parameters into a
g g
redefinition of the by now undetermined parameters rαb and ẑW (5.63) and γZ , γA . The
bare transformation matrix δĝαb , which depends on the bare angles γZo and γAo as defined
in (5.89), is computed via
 
T o o
Oaα (θW )δĝαb = (z V ) O(θW
o
) (r V )−1 δĝβb′ (z̄ g )Tb′ b (5.95)
aβ

66
This equation determines also those three parameters of the antighost field redefinition
matrix (z̄ g )ab , which are not specified on the bilinear ghost part of the action.
Having transformed the general bilinear ghost action into its standard form the ghost
equations are solved quite simply. On the kinetic parts the matrix aogαb is related to the
angles of antighost transformations:
og
ẑW =1
og og
r3Z = cos γZo r3A = − sin γAo
og og (5.96)
r4Z = sin γZo r4A = cos γZo
The parameter ζ̂ o of the gauge fixing part is determined from the mass ratio of W -boson
and W -ghost:
gξ o o
ζ̂ = ζW (5.97)
g2o
Inserting this result yields on the neutral part the following equations:
o o2 0
ζW MW cos(γZo − θW o
) = cos θW cos γZo ζZo MZo2
o o2 0
ζW MW sin(γAo − θW o
) = cos θW sin γAo Mgo2A
0
−Ĝo ζW
o o2
MW cos(γZo − θW o
) = cos θW sin γZo ζZo MZo2
0
Ĝo ζW
o o2
MW sin(γAo − θW o
) = cos θW cos γAo Mgo2A (5.98)

They determine the BRS-transformations of antighosts, i.e. γZo , γAo , and the abelian pa-
rameter of the gauge fixing part Ĝo as functions of the vector boson mass ratio and the
ghost mass ratio. The mass of the photon ghost is seen to be not a free parameter of the
o o
model but has to vanish (cos θW ≡ MW /MZo ):
o o2
ζW MW cos γZo cos θW
o
o
= MgA =0 (5.99)
ζZo MZo2 cos(γZo − θWo
)

Ĝo = − tan γZo tan γAo = tan θW

o

The whole point in this calculation is the adjustment of the abelian coupling Ĝ via
the mass of the Z-ghost. For arbitrary Ĝ indeed one has to introduce the angle γZ into
the BRS-transformations of ghosts, as otherwise one is not able to keep the normalization
condition
Γc̄A cZ 2 = 0 (5.100)
p =0

which is crucial for infrared finite computations for off-shell Green functions. In the tree
approximation, of course it is possible to fix the ghost mass ratio equal to the vector mass
ratio by the normalization condition:

ReΓc̄+ c− (p2 )|p2 =ζM 2 = 0 ReΓc̄Z cZ (p2 )|p2 =ζM 2 = 0 (5.101)

W Z

67
Then the expressions (5.99) simplify to the ansatz we have taken in the classical approx-
imation:
MW
cos γZ = + O(h̄) (5.102)
MZ
and the BRS-transformation of antighosts is diagonal. For higher orders, however, the
normalization conditions (5.101) together with the infrared condition (5.100) does not
imply a diagonal transformation matrix for antighosts. Conversely requiring (5.101) and
a diagonal ghost transformation one has to introduce then counterterms c̄A cZ in order
to fulfil the ST identity. These counterterms produce in the next order off-shell infrared
divergencies. That the ghost mass ratio and the parameter Ĝ are indeed independent
parameters of the standard model, is already indicated by the computations we have
carried out in the classical approximation: Starting with the general bilinear ghost action
it is seen, that there is no parameter left, which could adjust the bare ghost mass ratio
to the bare vector mass ratio. Likewise when we transformed the gauge fixing part to
the bare form, we had to treat the parameter Ĝ as an independent parameter of the
theory. The Callan-Symanzik equation, we derive in the section 4, unambiguously allows
to determine the independent parameters of the model. There it is finally proven, that
the ghost mass ratio is a further independent parameter of the theory. The coupling Ĝ
and γZ are then determined order by order by normalization conditions, which fix the
mass of the Z-ghost and diagonalize the neutral ghost mass matrix at p2 = 0. Taking
therefore the ghost mass ratio as arbitrary also in the tree approximation as specified in
(5.10) we find
ζW
1− ζZ
cos2 θW
Ĝ = tan θW ζW + O(h̄)
ζZ
(1 − cos2 θW )
ζZ MZ2 − ζW MW 2
= q + O(h̄) (5.103)
ζW MW MZ2 − MW
2

This independent parameter does not only enter the gauge fixing part but also enters the
ST identity and the ghost interactions, and has, wherever it appears, to be differently
treated from the vector mass ratio, even if we fix the ghost mass ratio to be the same as
the vector mass ratio by the normalization conditions (5.101). The ghost interactions in
the classical approximation are immediately read off from (5.81):

Γgen o o gen o gen ′o ′o o o −1 o o

ext.f. (ρα , Ya ) + Γghost (c̄a ) = Γext.f (ρα , Ya ) − c̄a (δĝa4 ) (δĝ )4b ✷cb
X  o
MW
−gξ c̄oa (δĝ o)−1 o o −1 o
aα t̂bc,α − tan θG (δĝ )a4 t̂bc,4 (φb + δHb 2 o
g2
)q̂co (5.104)
α=
+,−,3

with
o o o −1
ρ′o
α = ρα + ∂c̄a (δĝ )aα (5.105)

68
 X  o
MW
Yb′o = Ybo − c̄oa gξ I˜bb′ (δĝ o)−1 o o −1 o
aα t̂b′ c,α − tan θG (δĝ )a4 t̂b′ c,4 (φ̂c + δHc ζ2 o
g2
)
α=
+,−,3

The interactions of the external fields with propagating fields are summarized in Γgen ext.f
(5.64). It is not modified by solving the ghost equations, but for bare ghosts the arbitrary
parameters therein are now specified to be related to δĝ o :
 
1 0 0 0
 
 0 1 0 0 
o  
δĝαb = 
 o o


(5.106)
 0 0 cos θG − sin θW 
o o
0 0 sin θG cos θW
o
For having simple notations we have introduced the ghost angle θG in analogy to the weak
o
mixing angle θW , which both are defined in the on-shell scheme by the vector mass and
the ghost mass ratio, respectively:
o
o MW
cos θW ≡
MZo
r
o2
MW
ζWo
MWo 1− o2
MZ
o
cos θG ≡ r (5.107)
ζZo MZo o M o2
ζW
1− W
o M o2
ζZ Z

Respective expressions are defined for the physical on-shell masses. It is quite instructive
to consider the ghost transformation matrix in the context of the classical field trans-
formations: When we introduced the ghosts in the classical approach by changing the
infinitesimal parameters of gauge transformations ǫα (x) to anticommuting parameters ca
(2.59), we have already mentioned that there is an arbitrariness in defining them. This
arbitrariness has now been exploited to construct a diagonal ghost mass matrix by the
transformation
ǫα −→ cα = δĝαb cb (5.108)

Finally it remains to solve the ST identity for the interactions, which depend on the
external scalar field we have suppressed up to now. Quite generally the dependence of
the generating functional of 1PI Green functions on the external scalars is governed by
the following equation, one derives from the ST identity:
!
δΓ δΓ
sΓ =− (5.109)
δqa δ φ̂a
Solving it in the classical approximation it turns out that one further parameter appears,
due to a field redefinition of the propagating scalars into propagating and external scalars.
The transformation ! !
φoa φoa − xo φ̂oa
−→ (5.110)
φ̂oa φ̂oa

69
is compatible with the ST identity and rigid symmetry if we redefine the external field
part by
Yao t̂abc (φob + v o δbH )coc −→ Yao t̂abc (φob − xo φ̂ob + v o δbH )coc − xo Yao I˜ab qbo (5.111)
As long as we do not want to interpret φ̂a as a background field, the normalization of x
is irrelevant, because finally Green functions are considered at φ̂a = 0. One can fix x by
the following normalization condition,
∂p2 ΓH Ĥ (p2 ) =x (5.112)
p2 =µ2H

For the purpose of this paper we choose x = 0, but nevertheless one gets nonlocal higher
order contributions between propagating and external scalars.

5.4.2. The solution of the ghost equations in higher orders

The purpose of this section is to prove, that the ghost equations can be indeed estab-
lished to all orders in accordance with the normalization conditions on the ghost 2-point
functions (5.10), (5.11) and (5.12). If one is able to implement the ghost normalization
conditions by adjusting the free parameters in the external field part, the gauge fixing
and the ST identity, then in the construction of higher orders one has only to consider
the non-local functional Γnl as defined in (5.81).
First we give the ghost equations in momentum space and test them with respect to
the mass normalization conditions. Introducing
Γρµα cb (p, −p) = −ipµ Γρα cb (p2 ) = −ipµ + O(h̄) (5.113)
the ghost equation of the charged ghost tested at p2 = ζW MW
2
reads:
2 2 2
ζW MW ReΓρ+ c− (ζW MW ) + iζ̂MW ReΓY+ c− (ζW MW )=0 (5.114)
The SU(2)-components of the ghost equation are tested at p2 = ζZ MZ2 and p2 = 0:
Re(ζZ MZ2 Γρ3 cZ (ζZ MZ2 ) − ζ̂MW ΓYχ cZ (ζZ MZ2 )) = 0 (5.115)
Re(ζZ MZ2 Γρ3 cA (ζZ MZ2 ) − ζ̂MW ΓYχ cA (ζZ MZ2 )) = − sin γA ReΓc̄A cA (ζZ MZ2 )
−ζ̂MW ΓYχ cZ (0) = cos γZ Γc̄Z cZ (0)
ζ̂MW ΓYχ cA (0)) = 0
The same test is carried out on the abelian component of the ghost equations:
g
ζZ MZ2 r4Z + Ĝζ̂MW ReΓYχ cZ (ζZ MZ2 )) = 0 (5.116)
g
ζZ MZ2 r4A + Ĝζ̂MW ReΓYχ cA (ζZ MZ2 ) = cos γA ReΓc̄AcA (ζZ MZ2 )
Ĝζ̂MW ΓYχ cZ (0) = sin γZ Γc̄Z cZ (0)
Ĝζ̂MW ΓYχ cA (0)) = 0

70
 In order to evaluate these equations one has to take into account that the vertex
functions Γρa cb (p2 ) and ΓYa cb (p2 ) are not independent from each other but are related
by the ST identity. We assume now, that we had already established the ST identity
and the Ward identities of rigid symmetry to order n − 1 for all Green functions and
to order n for all tests with respect to vectors and scalars. Having also applied the
normalization conditions on the 2-point functions and the one for fixing the nonabelian
coupling the external field part is determined up to local contributions of order n. These
local contributions are read off from the classical approximation (5.64). If the vertex
functions Γρα cb (p2 ) and ΓYa cb (p2 ) solve the ST identity then also the vertex functions
Γ′ρα cb (p2 ) and Γ′Ya cb (p2 ) solve the ST identity and they are related by
(n)
Γρ′(n)
α cb
(p2 ) = Γρ(n)
α cb
(p2 ) + aαb
′(n) (n) (n)
ΓY± c∓ (p2 ) = ΓY± c∓ (p2 ) ± iMW a++
′(n) (n) (n) (n)
ΓYχ ca (p2 ) = ΓYχ ca (p2 ) + MZ (cos θW a3a + sin θW a4a )ca (5.117)
(n) (n)
and a++ = a−− due to CP-invariance. Inserting in the ghost equations and taking ad-
vantage of the quantum action principle, which restricts the breakings of order n to local
expressions, it is seen that the ghost equations can be fulfilled, if we adjust the arbitrary
(n)
parameters aαb , the parameters of the gauge fixing part ζ̂ and Ĝ and the linear trans-
g(n)
formation parameters of the vectors r4a . In fact it is the same calculation as in the
classical approximation. In particular the diagonalization of the ghost mass matrix at
p2 = 0 (5.83) implies
Γc̄Z cZ (0)
ΓYχ cA (0) = 0 and ΓYχ cZ (0) = cos γZ (5.118)
ζ̂MW
and
Ĝ = − tan γZ (5.119)
Inserting the last relation into the solution of the ghost equation (5.81) we find, that the
condition
Γc̄A cZ (p2 = 0) = 0 (5.120)
is now fulfilled by construction, whereas

Γc̄Z cA (p2 = 0) = Γc̄A cA (p2 = 0) = 0 (5.121)

has to be established by requiring

ΓYχ cA (p2 = 0) = 0 (5.122)

We want to mention, that in the BPHZL-scheme this condition is implemented due to the
infrared degrees of cA and Yχ . Otherwise this condition has to be introduced in addition

71
to the usual ones in order to be able to protect internal ghost loops from off-shell infrared
divergencies.
Explicitly it is seen that the normalization conditions at p2 = 0 fix the counterterms
(n) g(n)
cos θW a3Z + sin θW r4Z
(n) g(n)
cos θW a3A + sin θW r4A (5.123)

On-shell conditions for the charged ghosts and Z-ghost and the separation of massive and
massless ghosts at p2 = ζZ MZ2 determine finally the parameters ζ̂, γZ , γA whereas a3Z , a3A
and a++ are fixed by the normalization conditions on the residua of ghosts.
The construction of higher orders can be therefore indeed restricted to Γnl as defined
in (5.81), but we have to take into account, that we are not able to dispose of the coun-
terterms Yχ cZ and Yχ cA , because these counterterms have to be adjusted for establishing
the ghost equations without introducing infrared divergencies. Thanks to the fact, that
the antighost transformations can be modified by introducing the angles γZ and γA , the
g g
coefficients appearing in the ST identity r4Z and r4A are at our disposal for absorbing
local breakings of the ST identity into corrections of the ST operator, even if the ghost
2-point functions are constructed with on-shell conditions.

5.5. Summary of the classical approximation

Because the approach we have chosen here for determining the invariant local counter-
terms is somewhat unconventional, we want to summarize the results of the last sections.
The general invariant action has been determined by requiring invariance with respect
to the ST identity and with respect to rigid SU(2)-symmetry,

S(Γgen gen
cl ) = 0 Wα Γcl = 0 (5.124)

The analysis is unconventional in so far as we did not prescribe the symmetry operators
explicitly as e.g. in the tree form, but we specified them by field content and algebraic
properties, i.e. nilpotency of the ST operator, algebra of rigid operators and the consis-
tency relation. This is the only form appropriate for the treatment of the standard model,
because the parameters of the tree approximation get higher order corrections as indi-
cated by the classical approximation, if one separates the massless and massive particles
at p2 = 0. This result can be read off from the explicit expressions for the vector 2-point
functions, which have been calculated in the on-shell scheme in the literature (see e.g. [37]
for a complete list). Actually, in the abstract approach one notices that one has to modify
the symmetry operators of the tree approximation, only if one classifies the higher order

72
breakings of the ST identity according to their infrared and ultraviolet degree, as we do it
in section 5. Indeed it will be seen there, that we have already solved the infrared part of
the problem by solving the classical approximation with the general symmetry operators.
A further important point in the treatment is the observation, that the ST identity
does not uniquely determine all parameters of the standard model. For gauging the
electromagnetic current rather than the currents of lepton and quark number conservation
the Gell-Mann Nishijima relation has to be extended for off-shell Green functions to higher
orders. For doing this one has to derive a U(1) Ward identity and specify therein the free
parameters as the ones of electromagnetic current conservation:

wQ = wem − w3 (5.125)

The general invariant action as solution (5.124) can be decomposed as in the tree
approximation into
Γgen gen gen gen gen
cl = ΓGSW + Γext.f. + Γghost + Γg.f. (5.126)
Apart from the mass of the photon and the photon ghosts, the parameters of the bilinear
action, i.e. masses, residua and nondiagonal mass matrix elements, are free parameters of
the theory, and one can dispose of them by the normalization conditions we have specified
in section 4.1. Masslessness of the photon and the photon ghost has to be proven to be
in agreement with the ST identity in higher orders. If we furthermore use the local U(1)-
Ward identity for determining the lepton and quark family couplings, then we remain with
one free parameter, the nonabelian coupling, which can be chosen to the electromagnetic
fine structure constant in the Thompson limit by the normalization condition on the
electron-photon vertex (5.73).
The general action can be written in the bare form by eliminating the non-diagonal
mass matrix elements of the general bilinear part and the arbitrary residua into a field
redefinition, which transforms the original fields to bare fields. The general expression
is obtained by undoing this transformations. Expressed in bare fields the general action
depends on the bare vector boson masses, the bare ghost masses, the bare fermion masses
and the bare Higgs mass and the the nonabelian coupling g2o or likewise eo .
o2 2 2 2 2
MW = MW + δMW mo2
H = mH + δmH
(5.127)
MZo2 = MZ2 + δMZ2 mo2 2 2
fi = mfi + δmfi

o o2 2 2 2
ζW MW = ζW MW + δζW MW + ζW δMW
(5.128)
ζZo MZo2 = ζZ MZ2 + δζZ MZ2 + ζZ δMZ2
We want to point out that δζZ and δζW are independent higher order corrections even if

73
we choose ζW = ζZ in the tree approximation. In a QED-like parameterization one has
MZ
g2o = e q + δg2 (5.129)
MZ2 − MW
2

Γgen
GSW has been given in (5.59), where we have to replace the couplings of lepton
and quark currents according to (5.71) in order ot fulfil the abelian Ward identity of
electromagnetic and weak current conservation. Taking for the gauge fixing function the
most general ansatz, which is compatible with rigid symmetry and is linear in propagating
fields (cf. (5.77) and (5.92)), the ghost equations relate Γgen
ghost to the external field part
according to (5.104). Using the notations for the ghost angle and weak mixing angle as
introduced in (5.106) and (5.107) the external field part as function of the bare vector
masses and bare ghost masses is given by
Z 
1
Γgen
ext.f. = − σao ε̂αβγ δĝβbo o o o
cb δĝγc cc (5.130)
2
+ρoµα (∂ µ I˜αβ δĝβb
o o o
cb + ε̂αβγ Oβb (θW )Vboµ δĝγc
o o
cc )
√ !
   
o† o τα o 1 o o o 2 0 o o
+ Y ig2 ( δĝαa − tan θW δĝ4a )(Φ + 2 o
)c a − xq̂ + h.c.
2 2 go MW
NF  X
X o τα Gδi o o oL
+ (Ψ̄oR
δi ig2 (
o
δgαa + δĝ )c F
i=1 δ=l,q 2 2 4a a δi
X 1
+ ψ̄foL ig2o (tan θW
o
+ Gδi ))fioR δĝ4a
o o
ca
f =e,d
i
2
1 
+ ψ̄uoLi ig2o (− tan θW
o
+G qi
)uoR o o
i δĝ4a ca + h.c.
2

Here we have again rewritten the scalars into complex doublets.

The ST operator and the Ward operators depend in the bare form also on the vector
mass ratio and ghost mass ratio:
Z  
δ δ 
S(Γgen
cl ) =
o
(sin θG o
∂µ cZ + cos θW ∂µ coA ) sin θW
o
+ cos Θ o
W (5.131)
δZµo δAoµ
δΓgen  δ δ  δΓgen  δ o δ

cl
+ oµ cos θW o − sin θW o + clo cos(θo1 −θo ) cos θW
o o o
− sin θ G
δρ3 δZµ δAµ δσ3 W G δcoZ δcoA
δΓgen δ δΓgen δ δΓgen δ δΓgen δ δΓgen
cl ˜ δ
+ cloµ o
+ cl
oµ o
+ cl
o o
+ cl
o o
+ o
Iaa′ o
δρ+ δWµ− δρ− δWµ+ δσ+ δc− δσ− δc+ δYa δφa′
NF 
X δΓgen
cl δ δΓgen
cl δ 
+ + + h.c.
i=1 δψfoL δfioR δψfoR δfioL
i i

74
  
1 0 0 0
  
 0 1 0 0  δ δ
 
+Bao   + qao Γgen
cl = 0
 0 o
0 cos(θG − θWo
) 0  δc̄b o
δ φ̂a
 
o o
0 0 sin(θG − θW ) 1 ab

The Ward operators of rigid symmetry are determined to (α = +, −, 3)

Z 
δ
Wα = I˜αα′ dx Vboµ Obβ
T o
(θW o ˜
)ε̂βγα′ Oγc (θW )Icc′ + {Bao } (5.132)
δVcoµ
′

˜ δ ˜ δ
+ cob (δĝ o)Tbβ εβγα′ (δĝ o)−1T
γc Icc′ o
+ c̄ob (δĝ o )−1 o
bβ ε̂βγα′ (δĝ )γc Icc′
δcc′ δc̄oc′
Mo ˜ δ + {Y o , φ̂o + ζ̂ o2 MW
o
+ (φob + δHb 2 gW o )t̂bc,α′ Icc′ o b b g
o
o , q̂a }
2 δφc′ 2

δ δ
+ ρoβ εβγ,α′ I˜γγ ′ o + σβo εβγ,α′ I˜γγ ′ o
δργ ′ δσγ ′
NF X 
X τα′ δ δ τα′ oL
+ FδoL i − F
i=1 δ=l,q
i
2 δFδoL δFδoL
i
2 δi
i

iτα′ δ δ iτα′ R 
+ ΨoR
fi − Ψ
2 δΨoR
f
δΨoR
δi 2 δi
i

Comparing the ST identity and the Ward operators of rigid symmetries expressed in terms
of bare fields, with the ones of the tree approximation, it is seen that they differ due to
the appearance of the ghost angle, which signals – apart from field redefinitions – different
corrections of the vector mass ratio and the ghost mass ratio in higher orders.
The local U(1) Ward identity
 
δ δ
g2o tan θW
o
w4Q − ∂ o
sin θW
o
− ∂ o
o
cos θW Γgen o o o o
cl = sin θW ✷BZ + cos θW ✷BA (5.133)
δZ δA
has been derived in the matter part of the action. It is broken only by gauge fixing the
longitudinal parts of the abelian vector field combination. It is valid in the ghost part of
the action as it is, if we take care to maintain rigid symmetry.
Transforming the bare fields to original fields yields the general form of the action
and the general form of the ST identity and Ward identities. It allows order by order to
determine the invariant local contributions by expanding the parameters in perturbation
theory. There it has to be noted that some of the parameters appear explicitly in the
symmetry operators. According to our conventions (cf. (5.21),5.22 and (5.23)) these
parameters are
rZV , rAV , ΘV , raS , rli , rqi , rαa
g
, γZ , γA (5.134)
As long as they are not needed for removing infrared divergent contributions, they could
be fixed in the symmetry transformations as it is e.g. for r S and rqi , rli . Furthermore

75
one has explicit dependence on the shift of the Higgs field v and the shift of the external
Higgs ζ̂v in the Ward operators of rigid symmetry. These parameters are fixed by on-shell
conditions on the mass of the W -boson and on the mass of charged ghosts,

MW , ζW MW (5.135)

The remaining parameters which are only specified on the general classical action are the
field redefinitions o
tan (θW + θA ) g
zW , o
, Z+− , zH , zνi , zui , z̃fi and xo (5.136)
tan (θW + θA )
the nonabelian coupling, which is fixed to the fine structure constant in QED-like schemes,
and the remaining masses of the standard model:

eo and MZo , moH , ζZo MZo , mofi (5.137)

These parameters, in any case, have to be fixed by normalization conditions on the finite
1PI Green functions.
The Callan-Symanzik equation, we derive in the next section, enables one to calcu-
late the asymptotic logarithmic corrections to the invariants. There it is seen, that the
invariants, which are connected with the parameters (5.136) and (5.137) get independent
logarithmic corrections to the next order.

6. The Callan-Symanzik equation

The Callan-Symanzik (CS) equation describes the response of the Green functions to
the scaling of all momenta by an infinitesimal factor. The dilatational operator
X Z
δ
W D
= (dϕk + x∂x )ϕk (x) dϕk = dimU V ϕk (6.1)
all fields δϕk (x)

acts on the 1PI-Green functions in the same way as the differentiation with respect to all
the mass parameters of the theory:

W D Γ = −m∂m Γ with (6.2)

NF X
X X
m∂m ≡ MW ∂MW + MZ ∂MZ + mH ∂mH + mfi ∂mfi + κa ∂κa
i=1 f κa

κa are the normalization points, which we have introduced to fix the residua of fields. In
concrete calculations one will introduce only one κ for fixing all the infrared divergent
residua of charged particles off-shell.

76
 The CS equation [41] is of utmost interest in the abstract approach as well as for con-
crete calculations. In the abstract approach the β-functions and anomalous dimensions,
which describe the breaking of dilatations by anomalies in higher orders, allow to deter-
mine the independent parameters of the theory in a scheme independent way. When we
solved the ST identity and the Ward identities of rigid symmetries for the local contribu-
tions, we gave a list of free parameters, which were not fixed by the symmetries and can be
adjusted by normalization conditions. The CS equation singles out from these parameters
the ones which get independent logarithmic corrections in the asymptotic region, where
all momenta are much larger than the masses of the theory. To be specific we consider as
an example the ghost mass ratio: It can, of course, be fixed by a normalization condition
to the vector boson mass ratio. However, from the CS equation it is immediately derived,
that the vertices, which depend on the ghost mass ratio, get different logarithmic cor-
rections in higher orders from the ones, which depend on the vector mass ratio. Similar
statements are true for other mass ratios, one would like to relate to the vector boson
mass ratio in lowest order, as it is required for example in the context of noncommutative
geometry [42]. With the help of the CS equation it is at least in massless theories possible
to find relations between independent parameters, which are compatible with renormaliz-
ability, using the principle of reduction of couplings [43, 44]. It is, however, not obvious,
how reduction has to be applied to spontaneously broken theories, due to the fact that
the β-functions do not only depend on the perturbative expansion parameters but also
on the mass ratios. This property has also the consequence, that the β-functions of the
CS equation depend on mass logarithms from 2-loop order onwards and that they differ
from the ones of the symmetric theory [45].
For the purpose of the present paper concerning the renormalization of the standard
model, the CS equation, especially the dilatational operator, is also of special interest,
because it allows to simplify the algebraic cohomology as carried out in the next section
considerably. One can derive that all algebraic anomalies of the ST identity and the Ward
identities are restricted to 4-dimensional field polynomials, 3-dimensional breakings are
immediately seen to be variations of terms with quantum numbers of the action [21]. In
this context we also want to mention that the CS equation plays an important role if
one wants to prove the nonrenormalization theorems for the Adler-Bardeen anomaly as
it appears in the ST identity (see [46] and references therein).
For concrete calculations the most important outcome of the CS equation is the de-
termination of higher orders leading logarithms. If there are large asymptotic logarithms
in 1-loop order, the CS equation allows consistently to determine those large leading
logarithms of higher orders, which are induced by the 1-loop logarithms of the lowest
order.

77
 In this section we construct the symmetric operators of the CS equation, which define
the β-functions and anomalous dimensions of 1-loop order. It is important to note, that
we are able to classify the dilatational anomalies also if we have not proved that the
ST identity and Ward identities of rigid symmetries are established in 1-loop. The only
ingredients are the symmetries of lowest order and the linear gauge fixing. The soft
breakings, however, can be only consistently classified when we have established rigid
symmetry of 1-loop order. Then the classification works as we present here in the tree
approximation.

6.1. The soft breaking of dilatations

In the standard model dilatations are already broken in the tree approximation by all
terms with mass dimensions less than 4, especially by the mass terms of the fields. Due to
the spontaneous symmetry breaking all the masses of the physical fields are generated by
the shift of the Higgs field. According to the construction of the gauge fixing sector using
rigid symmetry the ghost masses and the masses of would-be-Goldstones are generated
by the shift of the external Higgs (cf. (2.51),(2.52) and (2.75)). In the tree approximation
one reads off the breakings of dilatations by applying m∂m on the classical action. For
proceeding to higher orders it is unavoidable to characterize also the soft breakings by
their symmetries. For this reason we do the same in the tree approximation and solve
thereby also the problem of constructing the soft breakings of higher orders, if the ST
identity and rigid symmetry are established. We have in the tree approximation
m∂m Γcl = ∆m (6.3)
where ∆m is an integrated field polynomial with mass dimension less than 4, CP-even
and neutral with respect to electric and φπ charge. Commuting the operator m∂m with
the ST operator it is seen that ∆m is sΓcl -invariant.
sΓ m∂m Γ − m∂m S(Γ) = 0 =⇒ sΓcl ∆m = 0 (6.4)
The dilatational operator does not commute with the Ward identities of rigid symmety:
h i h Z 
δ δ i
Wα , m∂m = Wα , v + ζ̂ (6.5)
δH δ Ĥ
This implies Z 
δ δ 
Wα ∆m = Wα v + ζ̂ Γcl (6.6)
δH δ Ĥ
Noting that the differentiation with respect to the Higgs and the external Higgs is a
BRS-variation quite generally (5.109)
!
δΓ δΓ δΓ
sΓ YH = and sΓ = (6.7)
δH δ q̂H δ Ĥ

78
the polynomial ∆m can be decomposed into
Z 
δΓcl δΓcl 
∆m = v + ζ̂v + ∆inv (6.8)
δH δ̂H
and
sΓcl ∆inv = 0 Wα ∆inv = 0 (6.9)
Inspecting all 2- and 3-dimensional field polynomial it is seen that there is only one rigid
and sΓcl - invariant polynomial:
Z
∆inv ≡ (2φ+ φ− + χ2 + H 2 + 2vH) (6.10)

This invariant field polynomial we couple to an external scalar field ϕ̂o with UV dimension
2 and IR-dimension 2 and add it to the classical action
Z
Γcl → Γcl + ϕ̂o (2φ+ φ− + χ2 + H 2 + 2vH) . (6.11)

The enlarged classical action is ST-invariant and rigid invariant if we assign:

sϕ̂o = 0 δα ϕ̂o = 0 (6.12)

Finally we are able to write ∆m as a field operator acting on the classical action:
Z  δ δ m2 δ 
m∂m Γcl = v + ζ̂ + H Γcl (6.13)
δH δ Ĥ 2v δ ϕ̂o
and v is determined in the QED-like parameterization to
MZ MW
v=2 sin θW cos θW + O(h̄) cos θW ≡ (6.14)
e MZ
The soft breaking of dilatations, and in particular the breaking of the tree approximation,
is therefore completely characterized by its properties under the symmetry transforma-
tions and one can proceed in higher orders as in the tree approximation, if the ST identity
and rigid symmetries are established. We want to mention, that also at this stage rigid
symmetry plays an important role for classifying the soft breakings of dilatations. If one
breaks rigid symmetry in the gauge fixing and ghost sector it is not possible to derive an
unambiguous expression for the soft breakings in higher orders.
Because we have rewritten the dilatational breaking of lowest order into a sum of field
differentiations we are able to proceed immediately to the next order. When one applies
the symmetric operator
Z  δ δ m2 δ 
D
Wsym ≡ WD − v + ζ̂ + H (6.15)
δH δ Ĥ 2v δ ϕ̂o

79
on the functional of 1PI Green functions, all breakings of 1-loop order are now known
to be local due to the action principle. But in the construction of the ST identity and
rigid Ward identities we have also to consider Green functions which include the invariant
external field ϕ̂o . For the invariant counterterms as derived in the last sections the only
linear dependence on ϕ̂o enters via the interaction with the 2-dimensional scalar invariant
(6.11), which reads for the bare scalar fields φ̃oa = φoa + xo φ̂oa (cf. (5.110)):
Z
Γgen
ϕ̂o = ϕ̂oo (2φ̃o+ φ̃o− + χ̃o2 + H̃ o2 + 2v oH̃ o ) (6.16)

The field renormalization of the external scalar field ϕ̂o can be fixed by setting the coef-
ficient of the field differentiation with respect to ϕ̂o in the CS equation.

6.2. The dilatational anomalies – hard breakings

In higher orders the dilatations are not only broken by the soft mass terms but also
by hard terms, the dilatational anomalies. The importance of the CS equation is founded
in the fact, that these anomalies can be absorbed into differential operators, which are
connected with the anomalous dimensions and β-functions.
The dilatational anomalies of one-loop order are normalization point independent,
but the differential operators introduced depend on the special parameterization and
the specific form of the breaking mechanism. They are essentially characterized by the
symmetries of the tree approximation. Because we want finally to use the CS operator
for classifying the breakings of the ST identity we only assume that the ST identity
is established in lowest order and so for the rigid symmetries. However, for deriving
the CS equation we cannot completely stick to the tree approximation as given in the
section 2, but we have to treat the ghost mass ratio as an independent parameter. The
ST identity and Ward identities of the tree approximation have then the form as given in
(5.131) and (5.132), replacing all fields and parameters by ordinary fields and parameters.
Therefore the symmetry operators depend in the tree approximation on the vector mass
ratio, i.e. cos θW , and ghost mass ration, i.e. cos θG (5.107).
Using the action principle in its quantized version one derives that all symmetries of
the tree approximation are broken in 1-loop order by integrated field polynomials
 (≤1)  (≤1)
(1)
S(Γ) = ∆brs Wα Γ = ∆(1)
α (6.17)
 (≤1)
D
Wsym Γ = ∆(1)
m

These field polynomials are restricted according to the UV-degree and IR-degree of the

80
classical operators:
(1) (1)
dimU V ∆brs ≤ 4 dimIR ∆brs ≥ 3
dimU V ∆(1)
α ≤ 4 dimIR ∆(1)
α ≥ 2 (6.18)
UV (1) IR (1)
dim ∆m ≤ 4 dim ∆m ≥ 2
We want to mention already here that breakings with infrared dimension 2 are alarming,
because they do in general not exist as integrated insertions in higher orders. They
potentially contain mass insertions for massless particles, which are seen to be infrared
divergent off-shell. In particular, absence of mass insertions for massless particles, i.e.
Aµ Aµ and c̄A cA , has to be verified in the CS equation by a test with respect to the
respective normalization conditions to all orders.
Because the functional of 1PI Green functions is invariant under the global charge
symmetries (4.14) and (4.15) and CP-transformations, all breakings have a well-defined
transformation behaviour under these symmetries: ∆brs is CP-even, has φπ charge 1 and
is neutral with respect to electric charge, whereas ∆α is CP-odd, neutral with respect to
φπ charge and has electric charge ±1 for α = +, − and is neutral for α = 3. The im-
portant point is that ∆m in fact has the same quantum numbers as the general classical
action: It is neutral with respect to electric and φπ-charge and CP-even. Up to linear
field polynomials the most general basis for the integrated insertion ∆m is equivalent to
the general renormalizable action, from which one has constructed the general invariant
solution in section 5. For this reason we are able to make use of the symmetric dilata-
tional operator, when we consider the algebraic cohomology problem, for classifying the
breakings of the ST identity.
D
For proceeding we take advantage from the property that the operator Wsym (6.15)
commutes with the ST operator and the Ward operators of rigid symmetry. Therefrom
one derives the following consistency relations:
(1)
D
Wsym ∆(1) (1)
α = −Wα ∆m
D
Wsym ∆brs = −sΓ ∆(1)
m (6.19)

We decompose now the local insertions ∆α , ∆m and ∆brs according to their transformation
D
under Wsym
4
X
∆(1)
op = ∆kop with D
Wsym ∆kop = (k − 4)∆kop (6.20)
k=1

It is proven immediately that this decomposition is unique, because it is nothing else but
classifying field polynomials according to their mass dimension and taking the Higgs, the
external Higgs and the neutral scalar ϕ̂o in a shifted version
m2H
H +v, Ĥ + ζ̂v , ϕ̂o + (6.21)
4

81
The consistency relations split up into equations for any of the ∆kop and give different
informations as long as we did not establish the ST operator and rigid symmetry in 1-
loop order. The 4-dimensional breakings ∆4m are seen to be sΓcl -symmetric and rigid
symmetric

0 = −Wα ∆4m 0 = −sΓcl ∆4m (6.22)

These equations are used to construct the hard anomalies of the CS equation, which
are related to β-functions and anomalous dimensions. The further equations for lower
dimensional field polynomials state that the breakings ∆≤3 ≤3
brs and ∆α can be immedi-
ately written as sΓcl and Wα variations, respectively, of integrated field polynomials with
quantum numbers of the action.
For constructing the hard breakings of the CS equation we have therefore the task
to find all independent field polynomials satisfying the above constraints and to express
them in form of symmetric differential operators. The first problem has been already
solved in the last section, because ∆m has the same quantum numbers as the general
renormalizable action. We have only to fix the parameters, which appear explicitly in the
symmetry operators (5.134) to their tree values and have to expand the remaining ones
(5.136) and (5.137) to the next order, singling out the polynomials which contain only
lower dimensional field polynomials.
∞
X (n)
S(Γgen
cl )(θW , θG ) = S(Γcl + Γinv )(θW , θG )
n=1
 ∞
X 
(n)
= S(Γcl ) + sΓcl Γinv (θW , θG ) = 0 (6.23)
n=1

For finding the invariant operators corresponding to the invariant field polynomials we
construct in the usual way symmetric operators, which commute with the lowest order
rigid Ward operators and the lowest order ST operator and this all done we identify the
operators with invariant field polynomials. It is well-known that the invariant field polyno-
mials are separated into two classes: The first one contains all invariant field polynomials,
which are sΓcl -variations. These invariants are generated by acting with symmetric field
differentiation operators on the classical action and are connected with field redefinitions
and anomalous dimensions. The second class comprises the invariants which are gener-
ated by differentiation with respect to independent parameters of the theory. These field
polynomials are sΓcl -invariants without being variations.
First we give a list of all symmetric field differentiation operators. Because the vectors,
B-fields and ghosts are rotated from the SU(2) and U(1)-fields to mass eigenstates by the
matrix Oαa (θW ) and δgαb , the field differentiation operators are not purely leg counting

82
operators, but mix massless and massive neutral fields. In the vector ghost sector we find
the following invariant field differentiation operators
Z 
δ δ 1 δ δ 
NV = Va − ρα + cos(θW −θG )
(sin θG cZ + cos θW cA )(sin θW + cos θG )
δVa δρα δcZ δcA
Z 
δ δ
N̂V = (sin θW Z + cos θW A)(sin θW + cos θW )
δZ δA
δ δ 
+ cos(θW1 −θG ) (sin θG cZ + cos θW cA )(sin θW + cos θG )
δcZ δcA
Z 
δ δ 
NB = Ba + c̄a (6.24)
δBa δc̄a
Z 
δ δ
N̂B = (sin θW BZ + cos θW BA )(sin θW + cos θW )
δBZ δBA
δ δ 
+ cos(θW1 −θG ) (sin θW c̄Z + cos θG c̄A )(sin θG + cos θW )
δc̄Z δc̄A
Z 
δ δ δ δ
Nc = c+ + c− + cos(θW1 −θG ) (cos θG cZ − sin θW cA )(cos θW − sin θG )
δc+ δc− δcZ δcA
δ δ δ 
−σ+ − σ− − σ3
δσ+ δσ− δσ3
The symmetric field differentiation operators in the fermion sector have to commute also
with the operators of lepton and quark family conservation. They are also not leg-counting
operators for massive fermions, but involve the γ5 . It is convenient to split these operators
into the ones for left-handed and right handed fields, which are both invariant operators:
Z  
δ δ δ δ
NFLδ = FδLi − ΨR + F L
− Ψ R
δ = l, q (6.25)
i
δFδLi δΨRd,i
δi
δFδLi δi δΨR δi
δi

Z  
δ δ δ δ
NfRi = fiR R − ψiL L + R fiR − L ψiL fi = ei , di , ui
δfi δψi δfi δψi

The invariant scalar field differentiation operators comprise the ones of the propagating
scalars and of external scalars. They are symmetric with respect to the rigid operators,
if one includes the shift of the Higgs and external Higgs in the transformation.
Z Z 
δ δ δ δ 
NS + v = φa +v − Ya (6.26)
δH δφa δH δYa
Z Z  Z
δ δ δ δ 
NŜ + ζ̂v = φ̂a + ζ̂v + q̂a
δ Ĥ δ φ̂A δ Ĥ δqa

When acting on Γcl the invariant differentiation operators summarized in (6.24), (6.25)
and (6.26) are in one to one correspondence with the field redefinition parameters listed

83
in (5.137):
NS Γcl ←→ zH
(NV − NB ) Γcl ←→ zW
tan θW +θZ
NFLl Γcl ←→ zνi
(N̂V − N̂B ) Γcl ←→ tan(θ W +θA )
i (6.27)
g NFLqi Γcl ←→ zui
Nc Γcl ←→ Z+−
NfRi Γcl ←→ z̃fi

The operators NB , N̂B and NŜ correspond to field redefinitions of B-fields and external
scalars, which are, however, fixed in the gauge fixing part to the ones of vectors, propagat-
ing scalars and coupling redefinitions (cf. (5.93)). Taking them as independent operators
in the CS equation their coefficients are determined quite simply by a test on the local
gauge fixing polynomial.
The invariant field polynomial, which corresponds to the field redefinition of the pro-
pagating into the external scalar, i.e. to the parameter xo (5.110), is generated by the
mixed field differentiation operator:
Z Z  Z
δ δ δ 
ÑS + ζ̂v = φ̂a + ζ̂v (6.28)
δH δφa δH
It is symmetric with respect to rigid symmetry but not with respect to the ST operator.
The sΓcl -invariant insertion is then given by
Z Z
δ
(ÑS + ζ̂v )Γcl + qa I˜ab Yb (6.29)
δH
Now we turn to the non-variations among invariant field polynomials. From the general
classical symmetric solution they are read off by expanding the independent parameters
(5.137) in perturbation theory. Equivalently they are generated by differentiating the
classical action with respect to the independent parameters: These are the coupling e,
which is the perturbative expansion parameter, and furthermore the mass ratios, M W
MZ
, for
m
the weak interactions, m H
MZ
f
for the scalar interaction and MZi for the Yukawa interactions.
At this stage it is unavoidable to treat θG i.e. the ghost mass ratio as an independent
parameter, because its differentiation corresponds to an independent insertion in the gauge
fixing and ghost sector. Similarly it turns out that also the differentiation with respect to
the both gauge parameters, ξ and ξ, ˆ has to be included. The differentiation with respect
to parameters, which do not appear in the ST identity and the rigid Ward operators of
the tree approximation immediately correspond to respective invariant field polynomials:
ˆ
mH ∂mH , mfi ∂mfi , ξ∂ξ , ξ∂ (6.30)
ξ̂

The differentiation with respect to the coupling e is not a rigid invariant, but has to be
symmetrized by including the shift: the operator
Z  Z 
δ δ  2 δ δ 
e∂e − e∂e v + ζ̂ = e∂e + MZ sin θW cos θW + ζ̂ (6.31)
δH δ Ĥ e δH δ Ĥ

84
is sΓcl - and rigid symmetric. Without using the local U(1)-Ward identity there would
be seemingly invariant field polynomials corresponding to the lepton and quark family
coupling Gδi . They are however singled out by deriving quite in analogy to (6.19) and
(6.22), that the field polynomials ∆4m are U(1)-gauge invariant. This result finally relates
also the β-functions of the coupling e and the mass ratio M
MZ
W
to the anomalous dimensions
(see (6.39)).
In order to find the sΓcl -invariants of the mass ratios M
MZ
W
and ζζWZ M
MW
Z
it is not sufficient
to expand the mass ratios only in the general symmetric classical action, but one has to
take into account, that such a mass expansion concerns also the ST operator and Ward
operators of rigid symmetry. This subtlety comes in, because these mass ratios take a
twofold role: They appear in the field transformation matrices, which are introduced for
constructing mass eigenstates (cf. (2.28) and (5.106)) , and they take at the same time
the role of the abelian gauge coupling and the abelian coupling Ĝ, respectively (cf. (2.29)
and (5.103)). Expanding the bare mass ratios in perturbation theory
o o
cos θW = cos(θW + δθW ) cos θG = cos(θG + δθG ) (6.32)

one finds from the general symmetric solution:

(0) (1)
sΓcl Γinv (δθW , δθG ) + (δS)(1) (Γcl )(δθW , δθG ) = 0 + O(h̄2 ) (6.33)

with
S(Γ) =(S (0) + δS (1) )(Γ) + O(h̄2 ) (6.34)
Corresponding to these expressions differentiation with respect to θW as well as θG are not
(1)
sΓcl -invariant operators, because their action on Γcl produces only Γinv . In addition one
has to enlarge them both with mixed massless – massive field differentiation operators for
being sΓcl -invariant operators:
Z 
δ δ δ δ 
∂˜θW ≡ ∂θW + A −Z + BA − BZ
δZ δZ δBZ δBA
Z  δ 
δ
+ cos(θW1 −θG ) cA + sin(θW − θG )
δcZ δcZ
Z   δ
− cos(θW1 −θG ) c̄Z + sin(θW − θG )c̄A (6.35)
δc̄A
Z  δ δ 
∂˜θG ≡ ∂θG − cos(θW1 −θG ) cZ + sin(θW − θG )
δcA δcZ
Z   δ
+ cos(θW1 −θG ) sin(θW − θG )c̄Z + c̄A (6.36)
δc̄Z

The operator ∂˜θG is a symmetric operator with respect to rigid symmetry, whereas ∂˜θW
has to be enlarged by the contributions from the shift, because it acts on v already in the

85
lowest order. The operator
Z  Z 
δ δ  ˜ 2 δ δ 
∂˜θW − ∂θW v + ζ̂ = ∂θW − MZ cos2θW + ζ̂ (6.37)
δH δ Ĥ e δH δ Ĥ
is then also rigid symmetric.
Acting with the symmetric operators (6.24), (6.25), (6.26), (6.30), (6.31), (6.36) and
(6.37) on the classical action one produces together with the polynomial (6.29) a complete
basis for the hard breakings of the symmetric dilatational operator (6.15) in 1-loop order.
Therefore it is possible to give the dilatational anomalies in the form of a CS equation,
i.e. as a linear combination of differential operators. Writing all the soft breakings pro-
duced by symmetrization with respect to the shift on the r.h.s we get the CS equation of
the standard model in 1-loop order:
 NF X
X
m∂m + βe e∂e − βMW ∂˜θW + βmH mH ∂mH + βmfi mfi ∂mfi (6.38)
i=1 f

− γV (NV − NB + 2ξ∂ξ + 2ξ∂ ˆ + sin θG cos θG ∂˜θ ) − γc Nc

ξ̂ G

ˆ ) − γS NS − γ N − γ̃S ÑS
− γ̂V (N̂V − γ̂B N̂B + 2(ξ + ξ)∂ ξ̂ Ŝ Ŝ
NF
X 
− (γFli NFLl + γFqi NFLqi + γei NeRi + γui NuRi + γdi NdRi ) Γ(≤1)
i ϕ̂o =0
i=1
Z   δΓ δΓ  δΓ δΓ m2 δΓ 
= (1 + βe e∂e − βMW ∂θW )v + ζ̂ + v(γS + γ̃S ) + ζ̂vγŜ + H
δH δ Ĥ δH δ Ĥ 2 δ ϕ̂o
Z
+ γ̃S q̂a I˜ab Yb + ∆≤3
m

On the right hand side we have also collected all local lower dimensional 1-loop breakings,
which are not classified by the lowest order symmetries, into ∆≤3 m . When finally the ST
identity and Ward identities are established, we are able to prove that ∆≤3 m is vanishing.
The CS equation takes then essentially this form to all orders, in particular the number of
independent operators we have introduced is exhausted by the list giving above. It is only
the explicit form of higher order operators, which is changed due to the symmetrization
with respect to the general Ward operators and the generalized ST identity.
Calculating the commutator of the CS operator and the local U(1)-Ward operator
yields the abelian relation between the β-functions and anomalous dimensions.
sin θW
βe = βM + γV + γ̂V (6.39)
cos θW W

In the CS equation we have already inserted the result, which comes out from the test
on the local B-dependent part of the action (5.77) and (5.81). In addition one derives

86
for the anomalous dimension of the external scalars γ̂S in the QED-like parameterization
(5.79)
cos θW
γŜ = βe + βMW + γV − γS (6.40)
sin θW
whereas γ̃S is an independent anomalous dimension and can be determined on the mixed
external – propagating scalar 2-point functions. This function is only important, when
one wants to interpret φ̂a as a background field.
We want to give here the results for the β-functions and anomalous dimensions of
the vector-scalar sector. A complete list of the β-functions can be found in [47]. The
β-functions of the electromagnetic coupling and the vector mass ratio are determined to
e2  64

βe = − 42 − 3
NF
24 · 4π 2
e2  
βMW = − (43 − 8NF ) − (42 − 64
3
NF ) sin2 θW . (6.41)
4 · 24π 2 sin θW cos θW
The remarkable point is, that the β-function of the electromagnetic coupling is indeed
QED-like in the sense, that it does not involve mass ratios, as does βMW . In contrast
to QED it involves the nonabelian contributions of the charged vectors. This has as a
consequence that the sign of βe is negative, if one considers the standard model without
fermions or includes only one family. The anomalous dimensions of vectors in general
gauges compatible with rigid symmetry are given by
e2  6ξ − 25 1 
γV = 2 + NF (6.42)
4π 2 sin θW 24 3
e2  6ξ − 25 1 −3 + 8 sin2 θW 
γ̂V = − 2 + + 2 NF , (6.43)
4π 2 24 sin θW 24 cos2 θW 9 sin θW cos2 θW
The CS-functions (6.41) and (6.42) are seen to fulfil the abelian relation (6.39).
Because the anomalous dimension γV is nonvanishing, the coefficient of the ghost mass
differentiation ∂θG is also nonvanishing and the ghost angle has therefore to be treated as
an independent parameter. This proves finally that the ghost mass ratio gets independent
higher order corrections. A choice compatible with renormalizability is to set all ghost
masses equal, i.e. θG = 0. Such a choice, however, is connected with nondiagonal vector-
scalar propagators and not adequate for concrete calculations. Similarly it is seen that
also the abelian gauge parameter ξˆ is an independent parameter of the theory.
Before we turn to the higher order breakings of the ST identity and Ward identities we
want to consider the off-shell infrared problem as it appears in the CS equation. Because
∆m has infrared dimension 2, it has to be proven explicitly that the insertions
Z
Aµ Aµ , c̄A cA , H (6.44)

87
do not appear on the r.h.s. The proof is carried out best by using the Zimmermann
algebraic identities, which relate insertions of infrared dimension 2 to insertions with
infrared dimension 3. (The technique has been presented and applied for deriving the
CS equation in the spontaneously broken Higgs-Yukawa model [49] and works here in the
same way.) Then the r.h.s. contains all the 2-dimensional field polynomials explicitly and
one is able to test the CS equation with respect to these field polynomials at p2 = 0. Due
to the existence of the field mixing operators appearing e.g. in N̂V , and in the operator
∂˜θW the l.h.s. will only vanish, if the mixed 2-point functions of massless and massive
fields vanish at p2 = 0:

ΓZA =0 Γc̄Z cA = Γc̄AcZ =0 (6.45)

p2 =0 p2 =0 p2 =0

Otherwise, nonintegrable infrared divergencies appear to the next order and make it
impossible to derive the CS equation of higher orders.

7. Higher orders

7.1. The quantum numbers of higher orders breakings

We complete now the analysis of the renormalization of the standard model by proving,
that the ST identity (5.19), Ward identities of rigid symmetry (5.20) and local U(1)-
symmetry (4.69) can be established in the general form as given in section 5.2 to all
orders and lead to unique expressions for finite renormalized Green functions. The basic
ingredient of this proof is the action principle in its quantized version, as it is valid in
presence of massless particles in the framework of the BPHZL scheme [35, 39]. If the
symmetries are established to a definite order n in perturbation theory, the breakings of
the next order are restricted to be local field polynomials with definite ultraviolet and
infrared degree:
(n)
(S(Γ))(≤n−1) = 0 (S(Γ))(≤n) = ∆brs
(≤n−1) =⇒ (≤n) (7.1)
(Wα (Γ)) = 0 (Wα (Γ)) = ∆(n)
α

with
(n) (n)
dimU V ∆brs ≤ 4 dimIR ∆brs ≥ 3
(7.2)
dimU V ∆α(n) ≤ 4 dimIR ∆(n)
α ≥ 2
The ultraviolet degree of the breakings is deduced from a pure power counting analysis
of renormalizable quantum field theory, but the infrared degree is assigned due to the

88
BPHZL scheme. The BPHZL scheme implements those normalization conditions in the
scheme, which have to be fulfilled for being able to carry out infrared finite computa-
tions for off-shell Green functions in presence of massless particles. Having the Green
functions constructed in a different scheme as dimensional regularization these normal-
ization conditions have to be established finally by adjusting local counterterms. The
conditions, which ensure infrared finitiness for off-shell Green functions are read off form
the BPHZL-scheme:
ΓZ µ Aν (p2 = 0) = 0 ΓAµ Aν (p2 = 0) = 0
ΓYχ cA (p2 = 0) = 0 Γc̄Z cA (p2 = 0) = 0 (7.3)
Γc̄A cZ (p2 = 0) = 0 Γc̄A cA (p2 = 0) = 0

The breakings are furthermore restricted by the global symmetries, i.e. by electromagnetic
charge conservation (4.14)
(n) (n)
(n) Wem ∆± = ∓i∆±
Wem ∆brs = 0 (n) (7.4)
Wem ∆3 = 0

by Faddeev-Popov charge conservation

(n) (n)
Wφπ ∆brs = ∆brs Wφπ ∆(n)
α = 0

and by conservation of lepton and quark family number (4.15)

(n)
Wδi ∆brs = 0 Wδi ∆(n)
α = 0

The algebraic restrictions on the breakings derived from nilpotency of the ST operator
(4.9), algebra of Ward operators (4.10) and the consistency equation (4.13) read:
(n)
sΓcl ∆brs = 0 + O(h̄n+1 ) (7.5)
(n) (n)
Wα ∆β − Wβ ∆(n)
α = εαβγ I˜γγ ′ ∆γ ′ + O(h̄n+1 ) (7.6)
(n)
sΓcl ∆α(n) − Wα ∆brs = 0 + O(h̄ n+1
) (7.7)

The symmetry operators involved are the ones of the tree approximation, because higher
order contributions are not effective when acting in perturbation theory on a polynomial
of order h̄n .
In the higher order analysis one has to find all the breakings compatible with quan-
tum numbers and symmetries, and one has to prove, that they can be absorbed into a
redefinition of the ST operator, the Ward operators of rigid symmetry and by adjust-
ing finite counterterms, without destroying the on-shell normalization conditions on the
2-point functions. This computation is well-defined and can be carried out straightfor-
wardly. However, there are a lot of terms which have to be considered in the standard

89
model, even if CP-invariance is assumed. The analysis is simplified enormously, when
we take advantage of the fact, that the breakings are also classified under the symmetric
D
dilatational operator Wsym (6.15), and when we finally use the knowledge about local
symmetry invariants. These local invariants have been completely characterized, when
we solved the ST identiy and Ward identities in the classical approximation in generality
without using a perturbative expansion and explicit form of the operators.
We proceed now as follows: First we consider the breakings of the ST identity and
classify them in variations and non-variations, the anomalies (section 7.2):
(n)
∆brs = sΓcl Γ(n)
gen + r
(n) anom
∆brs (7.8)

Here Γ(n)
gen is a general local field polynomial with quantum numbers of the classical action
and UV dimension less than four. One is not able to exclude at this stage field poly-
nomials of infrared dimension three. It is well-known, that there are anomalies in the
standard model, which are given in the next section. The anomalies have to be shown
to vanish – in the 1-loop order by inspection of diagrams, in higher orders by apply-
ing the non-renormalization theorems. These theorems state, that the anomalies of the
ST identity vanish in higher orders, if they vanish in lowest order ([46] and references
therein). Application of the non-renormalization theorems to the standard model will
be considered elsewhere and we take its validity as granted for the purpose of the pa-
per. The variations we absorb as far as possible into finite counterterms to the action.
In particular, breakings of IR dimension three cannot be absorbed into counterterms for
reasons of infrared definiteness, although they are variations. Then we establish Ward
identities of rigid symmetry and are finally able to define a unique ST identity (section
7.3). In this analysis we do not have to consider tests with respect to B-field and with
respect to antighosts, because this part has been already constructed in agreement with
the symmetries by solving the ghost equation (see section 5.4). In particular we can carry
out the variable transformation ρα → ρ′α and Ya → Ya′ as given in (5.81) and establish the
symmetries on Γnl (ρ′ , Ya′ ), as usually done (cf. e.g [3]). When we have established both,
the rigid Ward identities and the ST identity, the abelian local Ward identity is identified.
Its breakings are known to be total divergencies. The variations can be absorbed into a
redefinition of the lepton and quark couplings of the abelian subgroup and the anomalous
currents vanish, if the anomalies in the ST identity vanish. At the very end one has one
single unspecified parameter, which is not fixed on the 2-point functions. This parameter
can be finally adjusted to be the electromagnetic fine structure constant in the Thompson
limit (5.73), as it is done in the QED-like on shell schemes (see for a review [22] and
references therein).

90
7.2. The cohomolgy and the Adler-Bardeen anomaly

In the first step we concentrate completely in finding the non-variations of the break-
ings under the ST identity, the Adler-Bardeen anomalies [17, 18, 19]. In the construction
of the Callan-Symanzik equation we have shown that anomalies of the ST identity can
appear only in the 4-dimensional field polynomials (6.19). This analysis is valid to all
orders, once the CS equation is established to all orders (section 7.4).

D
Wsym ∆anom
brs =0 (7.9)

all lower dimensional polynomials have been already seen to be variations. The consistency
equation with the rigid transformation operators furthermore tells that only rigid invariant
field polynomials can contribute to the anomaly:

Wα ∆anom
brs =0 (7.10)

Therefore the algebraic problem can be indeed formulated in symmetric variables α =

+, −, 3, 4:
cα = δĝαb cb H′ = H + v
(7.11)
Vα = Oαa (θW )Vb Ĥ ′ = Ĥ + ζ̂v
and the analysis is the same as one has to carry out in the symmetric theory. Here we
see in the abstract approach that ultraviolet divergencies of the spontaneously broken
theories are not worse than the ones of the symmetric theory [7, 8].
Since we have split off the transformation of the abelian component in the ST identity,
the external field part is essentially treated as in a SU(2) gauge theory. Therefore we
remain finally with polynomials depending on vectors, scalars and fermions and arrive at
the well-known Wess-Zumino consistency condition [51]:

wα Pβ − wβ Pα = εαβγ I˜γγ ′ Pγ ′ (7.12)

where wα are the gauge transformations of the tree approximation given in (2.35) and Pα
is a 4-dimensional polynomial depending only on the propagating fields of the standard
model: Z
anom T
∆brs = ca Oaα (θW )Pα (Va , φa , fiL , fiR ) (7.13)
The solution of the consistency equation has been analyzed quite generally in [21] and
can be evaluated in the standard model without further complications. Having CP-
invariance there are even no abelian contributions, which escape the algebraic treatment
of consistency, because those terms are CP odd. We end up with the following explicit

91
expression for the anomalies (a = +, −, Z, A are physical field indices, O(θW ) is defined
in (2.28):
Z  
∆anom
brs = r1 εµνρσ O4a (θ)ca ∂ µ O4b (θW )Vbν ∂ ρ O4c (θW )Vcσ (7.14)
Z  
1
+ r2 εµνρσ O4a (θW )ca ∂ µ Vbν I˜bc ∂ ρ Vcσ − ε̂bcd (θW )Vbν Vcρ Vdσ
3
with
ε̂abc = εαβγ Oαa (θW )Oβb (θW )Oγc (θW ) (7.15)
The form of the anomaly is unique up to the addition of BRS-variations. The general
classical action contains two rigid invariant field polynomials in vectors which are odd
under parity transformations:
Z
ΓPcl (Va ) = εµνρσ O4a (θW )Vaµ (k1 Vbν I˜bc ∂ ρ Vcσ + k2 ε̂abc Vaµ Vbν Vcρ V4σ ) (7.16)

We have used the sΓcl -variations of these field polynomials to bring the anomaly in the
form given above, where it only depends on the abelian ghost combination.
In 1-loop order the coefficients of the anomaly vanish. One has to note that the purely
abelian part vanishes due to electromagnetic current conservation and therefore depends
crucially on establishment of a local Ward identity in connection with electromagnetic
current conservation.
In the following, we assume that the non-renormalization theorems on the Adler-
Bardeen is valid, if we are able to prove a local abelian Ward identity and establish the
Callan-Symanzik equation order by order in perturbation theory. These two equations are
the necessary prerequisite for proving the non-renormalization theorems in higher orders
[46].

7.3. The establishment of symmetries

We start the consideration in 1-loop order and take for the lowest order the usual
standard model Lagrangian as given in section 2. If one calculates the finite Green
functions Γren with the Feynman rules of the tree approximation in a specific scheme
to 1-loop order, the ST identity is in general broken by the local field polynomial ∆brs :

(SΓren )(≤1) = ∆brs (7.17)

Because the coefficient of the anomaly vanishes in 1-loop order, the breaking can be
rewritten as a variation of integrated field polynomials:

∆brs = sΓcl Γ(1)

gen (7.18)

92
and
dimU V Γ(1)
gen ≤ 4 dimIR Γ(1)
gen ≥ 3 (7.19)
In the BPHZL scheme it is obvious that we do not have to introduce counterterms with
respect to a photon mass term, because the variation of the photon mass term has infrared
dimension 2. Z Z
sΓcl A Aµ = 2∂ µ cA Aµ + ...
µ
(7.20)

But all further field polynomials appear in principle in Γ(1) gen . In order to be able to
establish the on-shell conditions and conditions on the residua of all propagating fields,
we have to show that we have not to dispose of those field polynomials which are fixed by
the normalization conditions. These field polynomials are listed in (5.13). Due to the fact
that we have eliminated the antighost contributions by using the ghost equations (5.81),
the normalization conditions specified on the ghost 2-point functions, are now translated
into normalization conditions on the external field part. Explicitly we are not able to
dispose of the terms Yχ cZ and Yχ cA for establishing on-shell conditions without introducing
infrared divergencies (cf. eqs. (5.115), (5.116) and also (7.4)). The terms ρα agαb ∂cb are
kept arbitrary and are finally adjusted on the residua of ghost propagators. Therefore
we find the following list of field polynomials, which are not available for adjusting finite
counterterm contributions
Z 
Γgen
bil = − 41 (∂ µ Vaν − ∂ ν Vaµ )Zab
V
(∂µ Vνb − ∂ν Vµb )+ 21 Vaµ MVab Vµb

+ 12 ∂ µ φa Zab
S
∂µ φb − 21 MH2 H 2 (x)

+ iZfRi f¯iR ∂/fiR + iZfLi f¯iL ∂/fiL − Mfi (f¯iR fiL + f¯iR fiL )

+ ρµα agαb ∂µ cb + Yχ mgχb cb (7.21)

These terms can be eliminated if they are in one to one correspondence with sΓcl -
invariants. From the detailed considerations of the classical approximation it is seen,
R R R
that there are left three polynomials, namely Z µ Aµ , Yχ cZ and Yχ cA which do not
correspond to sΓcl -invariants. Therefore we are able to write
Z Z Z
∆brs = sΓcl Γ′break + u1 sΓcl MZ2 Z µ Aµ + u′2 MZ sΓcl Yχ cZ + u′3 MZ sΓcl Y χ cA
Z  
δ µ δ
= sΓcl Γbreak + u1 sΓcl Aµ − Z Γcl
δZ µ δAµ
Z  δ δ 
+u2 sΓcl (sin θW cZ + cos θW cA ) sin θW + cos θW Γcl
δcZ δcA
Z  δ δ 
+u3 sΓcl (cos θW cZ − sin θW cA ) sin θW + cos θW Γcl (7.22)
δcZ δcA

93
There we have rewritten the field polynomials with infrared dimension 3 into field op-
erators acting on the classical action, and have the remaining terms shifted into Γbreak .
Γbreak consists of all integrated CP-even field polynomials except the ones listed in (7.21),
i.e. it has especially infrared dimension 4. Applying the consistency equation between the
Ward operators of rigid symmetry and the ST identity (7.5) we find that the breakings
of the Ward operators take the following form:
 (≤1) Z  
δ µ δ
Wα Γren = ∆inv
α + Wα Γbreak + u1 Wα Aµ
− Z Γcl (7.23)
δZ µ δAµ
Z  δ δ 
+u3 Wα (cos θW cZ − sin θW cA ) sin θW + cos θW Γcl
δcZ δcA
∆inv
α comprises all field polynomials, which are sΓcl -invariants.

sΓcl ∆inv
α = 0 (7.24)

Considering the list of all possible breakings compatible with the algebra of rigid symmetry
and discrete and global symmetries it is seen, that ∆inv
α itself can be written as a sΓcl and
Wα -variation. Explicitly we find the following list of contributions:
X Z 
(1) δ
∆α = Wα uk Nk Γcl + v Γcl (7.25)
k δH
and Nk comprises the following field operators
Z 
δ δ δ δ δ δ 
NZA = Z +A + cZ + cA − ρ3 − σ3 (7.26)
δZ δA δcZ δcA δρ3 δσ3
Z   δ δ 
NZA
f = sin θW Z + cos θW A cos θW − sin θW − ρ3 ∂(sin θW cZ + cos θW cA )
δZ δA
  
δ δ 
+ sin θW cZ + cos θW cA cos θW − sin θW
δcZ δcA
Z   δ δ  δ 
Nc Z c Z = cos θW cZ − sin θW cA cos θW − sin θW −σ3
δcZ δcA δσ3
Z   δ δ 
Ncg
Z cA
= sin θW cZ + cos θW cA cos θW − sin θW
δcZ δcA
Z
δ δ δ δ
N φ+ = φ+ + φ− − Y+ − Y−
δφ+ δφ− δY+ δY−
Z  
δ δ
Nχ = χ − Yχ
δχ δYχ
Z  
δ δ δ δ
Nei = eLi L + L eLi − ψeRi R − R ψeRi
δei δei δψei δψei
Z  
δ δ δ δ
Ndi = dLi L + L dLi − ψdRi R − R ψdRi
δdi δdi δψdi δψdi
For absorbing these polynomials into the Ward operators (5.20) we have finally to note
that the expansion of the coefficients to 1-loop order can be also rewritten into a field

94
differentiation acting on Γcl . Denoting with Wα(0) the Ward operator of the tree approxi-
mation (2.112) then we write

Wα = Wα (0) + δWα(1) + O(h̄2 ) (7.27)

with
Z 
δ δ δ δ
δWα(1) Γcl = Wα(0) δrZ Z + δrA A + δθV (Z −A ) (7.28)
δZ δA δA δZ
  δ δ 
g
+ δr33 cos θW cZ − sin θW cA cos θW − sin θW
δcZ δcA
  δ δ 
g
+ δr34 cos θW cZ − sin θW cA sin θW + cos θW
δcZ δcA
  δ δ 
g
+ δr43 sin θW cZ + cos θW cA cos θW − sin θW
δcZ δcA
N
X 
S
F
δ
+ δr+ Nφ+ + δrχS Nχ + (δrli Nei + δrqi Ndi ) + δv Γcl
i=1 δH
Therefrom it is seen that the scalar and fermion contributions are immediately absorbed
into a redefinition of the tree Ward operators compatible with the algebra. For the ghosts
and vectors only parts of the invariants are absorbed, but a straigthforward calculation
shows that all the remaining contributions can be shifted into a Γ̂break , which again in-
cludes only interaction terms. Therefore we remain with
 (≤1)
WΓren = −δWα(1) Γcl + Wα (Γbreak + Γ̂break ) (7.29)

If one goes back with these expressions to the ST identity we have now for reasons of
consistency to split off therein the corresponding contributions δS (1) , because otherwise
the consistency relations are not valid to the next order.

S(Γren )(≤1) = −δS (1) Γcl + sΓcl (Γbreak + Γ̂break ) (7.30)

where
Z 
(1) δ δ δ δ
δS Γcl = sΓcl δrZ Z + δrA A + δθV (Z −A ) (7.31)
δZ δA δA δZ
  δ δ 
g
+ δr33 cos θW cZ − sin θW cA cos θW − sin θW
δcZ δcA
  δ δ 
g
+ δr34 cos θW cZ − sin θW cA sin θW + cos θW
δcZ δcA
  
g δ δ 
+ δr44 sin θW cZ + cos θW cA sin θW + cos θW Γcl
δcZ δcA
Defining the generating functional of Green functions of the standard model by

Γ = Γcl + Γren − Γbreak − Γ̂break + O(h̄2 ) (7.32)

95
and the Ward operators and ST operator to 1-loop order by

Wα = Wα(o) + δWα(1) S(Γ) = S (0) (Γ) + δS (1) (Γ) (7.33)

we have proceeded to absorb all breakings into counterterms of the action and a redefini-
tion of the symmetry operators compatible with the algebra.

S(Γ) = 0 + O(h̄2 ) Wα Γ = 0 + O(h̄2 ) (7.34)

The field polynomials, on which the normalization conditions on the 2-point functions
are established, are not touched in the construction. It is worth to note that at higher
orders as it was in the classical approximation the ST identity is only completely spec-
ified, if we construct simultanouesly the Ward identities of rigid symmetry. By now we
have suppressed those contributions which depend on the external field φ̂a . They do not
contribute to anomalies and the absorption of their breakings proceeds as in [50], where
we have carried out the same analysis in the abelian Higgs model.
Since we have only determined the normalization conditions on the 2-point functions in
the above construction, the finite Green functions are not unique by now. First we have
to fix the remaining coupling constant by a normalization condition on an interaction
vertex as given e.g in (5.73). From the construction of the general classical invariant
action it is seen, that furthermore the abelian couplings of fermions are not specified.
The contributions which remain arbitrary can be read off from the fermionic part Γgen matter
gen
(5.60) and the external field part Γext.f. (5.64) of the general invariant classical action.
They are obtained in their explicit form by expanding Gδi to 1-loop order and setting all
other coefficients to their tree value. We denote with Γδi the corresponding sΓcl -invariant
field polynomials.
NF X
X (1)
Γ′ = Γ + Gδi Γδi (7.35)
i=1 δ=l,q

and Γ and Γ′ satisfy both the ST identity and rigid Ward identities in the same form.
For fixing these undetermined parameters one has to use the local U(1) Ward identity
(4.69). Having constructed Γ in accordance with ST identity and rigid symmetry, it is
obvious that the abelian Ward identity is only broken by total divergencies in 1-loop
order. Consistency (4.68) furthermore restricts the breakings to be again sΓcl -invariants
and rigid invariants. One has
  1 
e δ 1 δ 
wQ − ∂ sin Θ + cos Θ Γ (7.36)
cos θW rZ δZ rA δA
 1 1 
=✷ sin ΘBZ + cos ΘBA
rZ rA
matter
+δg1 ∂µ jµ + δgδi ∂µ j δi + ri ∂µ Jianom + O(h̄2 )

96
The breakings are given by

∂ µ jµmatter = wQ Γcl ∂ µ jµδi = wδi Γcl (7.37)

wδi denotes the non-integrated version of the operators of lepton and quark familiy con-
servation (4.15). The anomalous contribution is determined to
Z  
µ
∂ Jµanom = r1 εµνρσ ∂ µ O4b (θW )Vbν ∂ ρ O4c (θW )Vcσ (7.38)
Z  
1
+ r2 εµνρσ ∂ µ Vbν I˜bc ∂ ρ Vcσ − ε̂bcd (θW )Vbν Vcρ Vdσ
3
The coefficients of the anomalous currents in the Ward identity are related to the ones
of the ST identity. (This can be seen by establishing the abelian Ward identity as a
sΓcl -variation of a ghost equation.) In particular, they vanish in 1-loop order, and they
vanish to all orders, if the non-renormalization theorems are valid in the standard model.
Vanishing of the purely abelian current anomaly to all orders can be proved only by means
of the local abelian Ward identity [46].
The absorption of non-anomalous currents proceeds as in the classical approximation
(cf. (5.70) – (5.73)): The lepton and quark family currents are absorbed by fixing the by
(1)
now undetermined constants Gδi in (7.35) and the matter current ∂j matter is absorbed
into the overall normalization of the Ward identity.
The finite renormalized Green functions are constructed in 1-loop order uniquely:
They satisfy the ST identity, the Ward identities of rigid symmetry and the local abelian
Ward identity. The 2-point functions of physical fields and of Faddeev-Popov ghosts have
one particle properties, and especially the mass matrices of massiv massless particles are
diagonalized at p2 = 0. This property ensures that the Green functions of the next order
exist in renormalized perturbation theory.

7.4. Induction to all orders

Having constructed the Green functions of 1-loop order in accordance with the sym-
metries and in accordance with off-shell infrared existence (7.3), the action principle can
be applied to the renormalized Green functions of the next order in the same way, as it
applied, when we proceeded from lowest order to 1-loop (7.17) and (7.18). For this reason
we are able to carry out the proof to all orders by induction (7.1). Assuming that the
ST identity (5.19), the Ward identities of rigid symmetry (5.20) and the local U(1) Ward
identity (4.69) are established for the Green functions to order n − 1, then one can make
the induction step to order n. The important point is the fact that the UV dimension

97
and IR dimension of the breakings is not changed, because the counterterms, we had to
add for establishing the symmetries, are compatible with UV and IR dimension 4. In
particular, the Γef f , which governs the perturbative expansion of Green functions in the
BPHZL scheme, is a 4-4 insertion (see [3] for details).
Because the one-loop breakings have been absorbed in accordance with the algebraic
properties and the consistency equation

sΓ S(Γ) = 0 for any Γ (7.39)

sΓ sΓ = 0 if S(Γ) = 0
Wα S(Γ) − sΓ Wα Γ = 0 for any Γ
h i
Wα , Wβ = εαβγ I˜γγ ′ Wγ ′

the breakings of order n are algebraically characterized by (7.5). Therefore we are able
to proceed from order n − 1 to order n in the same way as from lowest order to 1-loop
order, since we did not use explicit expressions of 1-loop order, but only algebraic and
power counting properties.
The only ingredient of 1-loop order has been the characterization of the anomaly can-
didates by the CS equation (7.9). In order to close the arguments we have finally to
derive the Callan-Symanzik to order n − 1. The CS equation of 1-loop is given in (6.38).
Since the symmetries are established to order n the the unsymmetric soft field polynomial
∆≤3
m vanishes. The construction of the higher order CS equation proceeds for the hard
breakings as given in section 6.2, especially there are the same number of independent
parameters and sΓcl -invariants. All differentiations with respect to couplings and mass
parameters act on the parameters, which appear in higher orders as corrections in the
ST identity and the Ward identities of rigid symmetry. As it is for the differentiation
with respect to MW (6.35) in 1-loop order, they have all supplemented by field opera-
tors in order to commute with the ST operator and Ward operators of rigid symmetry.
The explicit expressions can be read off from eqs. (7.28) and (7.30) and will be given in
detail elsewhere. We denote with ∂˜λ , λ = e, mH , mfi , θW , θG the rigid and sΓcl symmet-
ric operators of higher orders. (The weak mixing angle and the ghost angle are given
by the on-shell definition (5.107)). Also the higher order field differentiation operators
(6.24), (6.25) and (6.26), which correspond to the anomalous dimensions, are modified
in an obvious way. The soft breakings of the CS equation are constructed as in the tree
approximation, because they are completely characterized by their algebraic properties.
The CS equation is then finally given by
 NF X
X
m∂m + βe e∂˜e − βMW ∂˜θW + βmH mH ∂˜mH + βmfi mfi ∂˜mfi (7.40)
i=1 f

98
 − γV (NV − NB + 2ξ∂ξ + 2ξ∂ ˆ ) − βθ ∂˜θ − γc Nc
ξ̂ G G

ˆ ) − γS NS − γ N − γ̃S ÑS
− γ̂V (N̂V − γ̂B N̂B + 2(ξ + ξ)∂ ξ̂ Ŝ Ŝ
NF
X 
− (γFli NFLl + γFqi NFLqi + γei NeRi + γui NuRi + γdi NdRi ) Γ
i ϕ̂o =0
i=1
Z  X  δΓ  X  δΓ
= (1 + βλ ∂λ )v + (1 + βλ ∂λ )ζ̂v
λ
δH λ δ Ĥ
 Z
δΓ δΓ δΓ m2H
+v(γS + γ̃S ) + ζ̂vγŜ + + γ̃S q̂a I˜ab Yb
δH δ Ĥ 2 δ ϕ̂o
Infrared existence of the CS equation can be proved as in 1-loop order, since the conditions
for off-shell infrared existence (7.3) have been maintained in the construction of symmetry
operators.
With the establishment of the CS equation the construction of standard model Green
functions to all orders is completed. The ST identity and Ward identities of rigid sym-
metry
S(Γ) = 0 Wα Γ = 0 (7.41)
the local abelian Ward identity
  
1 δ 1 δ   1 1 
g1 w Q − ∂ sin Θ + cos Θ Γ=✷ sin ΘBZ + cos ΘBA (7.42)
rZ δZ rA δA rZ rA
with
wQ ≡ wem − w3 (7.43)
define uniquely the Green functions of the standard model of electroweak interactions to
all orders of perturbation theory in the on-shell scheme.

99
8. Conclusions and outlook

In this article we have constructed the finite renormalized Green functions of the
standard model of electroweak interactions to all orders of perturbation theory. Special
attention has been paid to the construction of 2-point functions in the on-shell scheme.
Only if the Green functions have one-particle properties in the LSZ-limit (apart from the
problem of unstable particles), can one proceed to construct the S-matrix and finally prove
unitarity of the physical S-matrix. These properties are the main requirements for being
able to interpret a quantum field theory as a physical theory of fundamental interactions.
Since the standard model contains massless particles, mass diagonalization of massless and
massive fields is connected with off-shell infrared existence of finite renormalized Green
functions.
The analysis has been carried out using the method of algebraic renormalization, which
until now was applied mainly to theories with semisimple gauge groups. In order to apply
the method to the standard model with the non-semisimple SU(2) × U(1) group we had
to generalize the method of algebraic renormalization at some points. In particular, we
had to obtain the symmetry operators by means of their algebraic properties instead of
postulating them in an explicit form a priori. The parameters, which appear in the gen-
eral solution of the algebra, correspond to field redefinitions of individual fields and, in
particular, to non-diagonal field redefinitions of neutral massive/massless fields. The ad-
justment of the latter parameters is essential for diagonalizing the mass matrix of neutral
vectors at p2 = 0. Due to the non-semisimple group structure, the abelian component of
the action contains additional free parameters, which are not specified by the Slavnov-
Taylor identity. These are interpreted as the couplings of the currents of lepton and quark
family conservation. Classically, these currents are conserved in the standard model, if
one neglects mixing of quarks due to the CKM matrix. In the general case there are
classically the conserved currents of fermion family number conservation and of baryon
number conservation. These currents are not gauged, but are not distinguished from the
electromagnetic current in the theoretical prescription, since they have the same quan-
tum numbers. In order to characterize the interactions prescribed by the standard model
as the ones of weak and electromagnetic interaction, the electromagnetic and the lepton
and quark number currents have to be identified and fixed by a Ward identity. Because
the electromagnetic Ward identity of current conservation cannot be derived for off-shell
Green functions, we have to use a specific form of the abelian local Ward identity. This
identity is the functional generalization of the Gell-Mann Nishijima relation. We want
to point out that in the general case, with quark family mixing, the local Ward identity
becomes even more important for correct adjustment of the electromagnetic current.

100
 An abelian Ward identity cannot be derived a priori, but has to be characterized in
the group structure as being abelian. For this reason, we have to require invariance under
the nonabelian rigid symmetries. The construction of Green functions in agreement with
rigid symmetry restricts the gauge fixing sector and the number of independent parameters
appearing therein. In order to be able to diagonalize the mass matrix of neutral ghosts
at p2 = 0, one has to introduce an additional ghost angle into the BRS-transformations
of antighosts. In the on-shell scheme, this angle is related to the ghost mass ratio in a
way similar to that in which the vector mass ratio is related to the weak mixing angle.
From the Callan-Symanzik equation, one can see that the ghost mass ratio indeed has to
be introduced as an independent parameter of the theory, since it has independent higher
order corrections.
The most remarkable consequence of the higher order construction is the observation,
that the standard model provides exactly the right number of parameters to bring the
propagators to a form in which they have one-particle properties in the LSZ-limit. As
we have pointed out, one has to adjust all of these parameters and one has to take into
account all deformations allowed by the algebra. If we had not succeeded with the analysis
as prescribed in the paper, then we would have had to prove that one-particle properties
are the consequence of a symmetry. Such a procedure has to be carried out finally in
the unphysical sector proving mass degeneracy for all unphysical fields by means of the
Slavnov-Taylor identity [14, 20]. From this point of view the vector and the unphysical
sector will be analysed carefully, when the renormalization is extended to CP-violating
interactions.
Having constructed the symmetry operators, we can apply them immediately to ex-
plicit one-loop and higher loop calculations. The parameters appearing therein are mainly
determined on 2-point functions, which are listed for one-loop order in the literature. One
is then able to prove, if finite Green functions satisfy the Slavnov-Taylor identity. In di-
mensional regularization one has to pay most attention to such breakings, which are
absorbed into parity violating counterterms of the effective action. Due to parity non-
conservation, it is not evident that the finite Green functions satisfy the Slavnov-Taylor
identity, if poles are subtracted by means of d-dimensional symmetric counterterms.
A first insight into higher order non-local contributions can be gained by considering
the Callan-Symanzik equation. The Callan-Symanzik equation of the standard model of
electroweak interaction has a completely different form from that of the corresponding
symmetric theory. It contains mixed field operators between massless/massive neutral
fields and, in particular, β-functions with respect to the independent mass parameters
of the theory. It is, however, not a matter of taste, whether one wants to derive the

101
Callan-Symanzik equation in terms of physical fields or symmetric fields, since it does not
even exist, if one does not include the mass diagonalization conditions of massless/massive
particles at p2 = 0 in the construction. For this reason, the considerations which concern
the renormalization group analysis of the unbroken SU(2)×U(1)-theory are not applicable
to the standard model. It was one of the main intentions of the present work, to make the
differences between unbroken and broken theories apparent. In particular, what one has
to consider in the spontaneously broken case, are the large mass logarithms, which are
induced from the lowest order β-functions to higher orders. These large-mass logarithms
are specific for the model in its spontaneously broken form and have been analysed in a
much simpler broken theory in [45]. The corresponding systematic investigation is now
also feasible in the electroweak standard model.

Acknowledgements I want to thank K. Sibold for initial common work, many help-
ful discussions and a critical reading of the manuscript. I am grateful to G. Weiglein,
B.A. Kniehl and A. Denner for comments on explicit one-loop renormalization.

102
A The quantum numbers of fields

In this appendix we list the quantum numbers of fields. We give the electromagnetic
charge Qem , the Faddeev-Popov charge Qφπ and the properties under charge conjugation
C and parity transformation P. Parity transformation to massive fermions is assigned in
accordance with parity conservation in electromagnetic interactions. The infrared (dimIR )
and ultraviolet (dimU V ) dimensions of fields is adjusted in agreement with the BPHZL-
scheme [34, 35].

dimU V dimIR Qem Qφπ C P (xµ → xµ )

3
eL 2
2 -1 0 −iγ 2 eR∗ γ 0 eR
3
eR 2
2 -1 0 −iγ 2 eL∗ γ 0 eL
3
uL 2
2 + 23 0 −iγ 2 uL∗ γ 0 uR
3
uR 2
2 + 23 0 −iγ 2 uR∗ γ 0 uL
3
dL 2
2 − 13 0 −iγ 2 dL∗ γ 0 dR
3
dR 2
2 − 13 0 −iγ 2 dR∗ γ 0 dL
3 3
νL 2 2
0 0 CP : −iγ 2 γ 0 ν L∗
5
ψeR 2
2 -1 1 −iγ 2 ψeL∗ γ 0 ψeL
5
ψeL 2
2 -1 1 −iγ 2 ψeR∗ γ 0 ψeR
5
ψuR 2
2 + 23 1 −iγ 2 ψuL∗ γ 0 ψuL
5
ψuL 2
2 + 23 1 −iγ 2 ψuR∗ γ 0 ψuR
5
ψdR 2
2 − 13 1 −iγ 2 ψdL∗ γ 0 ψdL
5
ψdL 2
2 − 13 1 −iγ 2 ψdR∗ γ 0 ψdR
5 5
ψνR 2 2
0 1 CP : −iγ 2 γ 0 ψνR∗

Table 1: Quantum numbers of the fermion fields

103
 dimU V dimIR Qem Qφπ C P (xµ → xµ )
W±µ 1 2 ±1 0 −W∓µ Wµ±
Zµ 1 2 0 0 −Z µ Zµ
Aµ 1 1 0 0 −Aµ Aµ
ρµ± 3 3 ±1 -1 −ρµ∓ ρµ±
ρµ3 3 3 0 -1 −ρµ3 ρµ3
c± 0 1 ±1 1 −c∓ c±
cZ 0 1 0 1 −cZ cZ
cA 0 0 0 1 −cA cA
σ± 4 4 ±1 -2 −σ∓ σ±
σ3 4 4 0 -2 −σ3 σ3
c̄± 2 3 ±1 -1 −c̄∓ c̄±
c̄Z 2 3 0 -1 −c̄Z c̄Z
c̄A 2 1 0 -1 −c̄A c̄A
B± 2 3 ±1 0 −B∓ B±
BZ 2 3 0 0 −BZ BZ
BA 2 2 0 0 −BA BA
φ± 1 1 ±1 0 φ∓ φ±
H 1 2 0 0 H H
χ 1 2 0 0 −χ χ
Y± 3 3 ±1 -1 Y∓ Y±
YH 3 3 0 -1 YH YH
Yχ 3 3 0 -1 −Yχ Yχ
φ̂± 1 1 ±1 0 φ̂∓ φ̂±
Ĥ 1 2 0 0 Ĥ Ĥ
χ̂ 1 2 0 0 −χ̂ χ̂
q̂± 1 1 ±1 1 +q̂∓ q̂±
q̂H 1 2 0 1 q̂H q̂H
q̂χ 1 2 0 1 −q̂χ q̂χ
ϕ̂0 2 2 0 0 ϕ̂0 ϕ̂0

Table 2: Quantum numbers of boson fields

104
References
[1] LEP Collaboration ALEPH, DELPHI, L3, OPAL, the LEP Electroweak Working
Group and the SLD Heavy Flavour Group, CERN-PPE/96-183.

[2] Reports of the Working Group on Precision Calculations for the Z-resonance, CERN
Yellow Report, CERN 95-03, eds. D. Bardin, W. Hollik and G. Passarino.

[3] O. Piguet, S. Sorella, Algebraic Renormalization, Lecture Notes in Physics 28,

Springer Verlag, 1995.

[4] S.L. Glashow, Nucl. Phys. B22 (1961) 579.

[5] S. Weinberg, Phys. Rev. Lett. 19 (1967) 1264.

[6] A. Salam, in: Proceedings of the 8th Nobel Symposium, p. 367, ed. N. Svartholm,
Almqvist and Wiksell, Stockholm 1968.

[7] G. ’t Hooft, Nucl. Phys. B33 (1971) 167.

[8] K. Symanzik, in: Cargese Lectures in Physics vol. 5, ed. D. Bessis, New York, 1972.

[9] B.W. Lee and J. Zinn-Justin, Phys. Rev. D7 (1973) 1049.

[10] L.D. Faddeev, Phys. Lett. B25 (1967) 29.

[11] C. Becchi, A. Rouet, R. Stora, Phys. Lett. B52 (1974) 344.

[12] I.V. Tyutin, “Gauge invariance in field theory and statistical mechanics”, Lebedev
preprint FIAN, no 39 (1975), unpublished.

[13] C. Becchi, A. Rouet, R. Stora, Comm. Math. Phys. 42 (1975) 127.

[14] C. Becchi, A. Rouet, R. Stora, Ann. Phys. (NY) 98 (1976) 287.

[15] J.H. Lowenstein, Phys. Rev. D4 (1971) 2281.

J.H. Lowenstein, Comm. Math. Phys. 24 (1971) 1.

[16] Y.M.P. Lam, Phys. Rev. D6 (1972) 2145,

Phys. Rev. D7 (1973) 2943.

[17] S.L. Adler, Phys. Rev. 117 (1969) 2426.

[18] J.S. Bell and R. Jackiw, Nuovo Cimento 60 (1969) 47.

[19] W.A. Bardeen, Phys. Rev. 184 (1969) 48.

105
[20] T. Kugo and I. Ojima, Phys. Lett. B73 (1978) 458;
Prog. theor. Phys. suppl. 66 (1979) 1.

[21] G. Bandelloni, C. Becchi, A. Blasi and R. Collina, Ann. Inst. Henri Poincaré 28
(1978) 225, 255.

[22] W. Hollik, in: Precision tests of the standard elektroweak model, ed. P.Langacker,
World Scientific, 1995.

[23] M. Böhm, W. Hollik, M. Igarashi, F.Jegerlehner, in: Radiative Corrections: Results

and Perspectives, eds. N. Dombey, F. Boudjema, New York, 1990.

[24] G. Passarino, M. Veltman, Nucl. Phys. B160 (1979) 151.

[25] M. Consoli, Nucl. Phys. B160 (1979) 208.

[26] A. Sirlin, Phys. Rev. D22 (1980) 971. W.J. Marciano, A. Sirlin, Phys. Rev. 22 (1980)
2695. A. Sirlin, W.J. Marciano, Nucl. Phys. 189 (1981) 442.

[27] J. Fleischer, F. Jegerlehner, Phys. Rev. D23 (1981) 2001.

[28] K.I. Aoki, Z. Hioki, R. Kawabe, M. Konuma, T. Muta, Suppl. Prog. Theor. Phys.
73 (1982) 1; Z. Hoki, Phys. Rev. Lett. 65 (1990) 683, E:1692; Z. Phys. C49 (1991)
287.

[29] D.Yu. Bardin, P.Ch. Christova, O.M. Fedorenko, Nucl. Phys. B175 (1980) 435; Nucl.
Phys. B197 (1982) 1.

[30] M. Consoli, S. LoPresti, L. Maiani, Nucl. Phys. B223 (1983) 474.

[31] M. Böhm, W. Hollik, H. Spiesberger, Fortschr. Phys. 34 (1986) 687.

[32] W. Hollik, Fortschr. Phys. 38 (1990) 165.

[33] F. Jegerlehner, Proceedings of the 1990 Theoretical Advanced Study Institute in Ele-
mentary Particle Physics, Boulder, Colorado, eds. M. Cvetic and P. Langacker, 1991,
Singapore.

[34] W. Zimmermann, Comm. Math. Phys. 15 (1969) 208.

[35] J.H. Lowenstein, Comm. Math. Phys. 47 (1976) 53.

[36] A. Denner, S. Dittmaier and G. Weiglein, Phys. Lett. B333 (1994) 420; Nucl. Phys.
B440 (1995) 95.

106
[37] A. Denner, Fortschr. Phys. 41 (1993) 307.

[38] P. Breitenlohner, D. Maison, Comm. Math. Phys. 52 (1977) 11,39,35.

[39] T.E. Clark, J.H. Lowenstein, Nucl. Phys. B113 (1976) 109.

[40] R. Stuart, in Perspectives for electroweak interactions in e+ e− collisions, Ringberg

castle, Germany, ed. B.A. Kniehl, World Scientific, Singapore, 1995.

[41] C.G. Callan, Phys. Rev. D2 (1970) 1541;

K. Symanzik, Comm. Math. Phys. 18 (1970) 227.

[42] A. Connes, J. Lott, Nucl. Phys. B (Proc. Supp.) 18 (1991) 29;

A. H. Chamseddine, G. Felder, J. Fröhlich, Nucl. Phys. B353 (1991) 689.

[43] R. Oehme and W. Zimmermann, Comm. Math. Phys. 97 (1985) 569;

R. Oehme, K. Sibold and W. Zimmermann, Phys. Lett. B147 (1984) 115;
W. Zimmermann, Comm. Math. Phys. 97 (1985) 211;
R. Oehme, Prog. theor. Phys. suppl. 86 (1986) 215.

[44] J.Kubo, K. Sibold, W. Zimmermann, Nucl. Phys. B259 (1994) 291.

[45] E. Kraus, Ann. Phys. (NY) 240 (1995) 367.

[46] G. Bandelloni, C. Becchi, A. Blasi, R. Collina, Comm. Math. Phys. 72 (1980) 239.

[47] E. Kraus, G. Weiglein, Preprint in preparation.

[48] W. Zimmermann, Ann. Phys. (NY) 77 (1973) 536.

[49] E. Kraus, Z. Phys. C 60 (1993) 741.

[50] E. Kraus, K. Sibold, Z. Phys. C 68 (1995) 331.

[51] J. Wess, B. Zumino, Phys. Lett. B49 (1974) 52.

107