Download as pdf or txt
Download as pdf or txt
You are on page 1of 25

A little more moderation would be good.

Of course, my life hasn’t exactly been one of moderation.


Donald Trump (1946)

28
Nonabelian Gauge Theory of Strong Interactions

Elementary particles which interact strongly with each other are called hadrons (re-
call the classification in Section 24.1). According to their statistics, one distinguishes
baryons and mesons. The most prominent among these are the spin-1/2 particles
in nuclear matter, protons and neutrons. The forces between them arise, to lowest
approximation, from the exchange of a spin-0 meson called pion, and further mesons
(middle-heavy particles).
Hadrons exhibit rich mass spectra, as we have seen in Chapters 24 and 25.
These spectra were explained to a good approximation by quark models [1]. Just as
nuclei are composed of protons and neutrons, baryons are composed of three quarks,
mesons of quarks-antiquarks pairs [2].

28.1 Local Color Symmetry

In spite of its success, the initial quark model exhibited several fundamental incon-
sistencies. Most importantly, it was not compatible with the spin-statistics theorem
derived in Section 7.10. For any reasonable potential between quarks, the ground
state orbital wave function should always be without zeros implying vanishing rel-
ative angular momenta for each pair of quarks. Any higher angular momentum
would have at least one zero in the wave function which would increase the gradient
and thus the kinetic energy of the Schrödinger field via the centrifugal barrier. The
orbital wave function of the ground state of three quarks, the proton, must therefore
be symmetric under the exchange of two quarks. On the other hand, the SU(6) wave
function involving internal SU(3) and spin has the Yang tableau . But this
implies that the nucleon wave function is completely symmetric under exchange of
all positions, all SU(3), and all spin variables. This seems to contradict the fact
that quarks have spin 1/2 so that, by the spin-statistics theorem, they should be
fermions and therefore have a completely antisymmetric wave function.

1486
28.1 Local Color Symmetry 1487

To remedy this contradiction, Han and Nambu suggested that quarks should be
a triplet under a further SU(3) group of transformations now called color SU(3) [3].
Color appears as a further label to the quark fields:

u(x)
 

 d(x) 

s(x)
 
 
q(x) =  , (28.1)

 c(x) 

t(x)
 
 
b(x)

which may thus be written as

uα (x)
 

 dα (x) 

sα (x)
 
 
qα (x) =  , (28.2)

 cα (x) 

tα (x)
 
 
bα (x)

with a label α = 1, 2, 3 specifying the three colors. By postulating that all hadron
states are completely antisymmetric in the color indices, so that they are color singlet
states, the contradiction disappears and the spin-statistic relation is again valid.
After this somewhat artificial postulate the question arose how nature manages
to enforce the color antisymmetry, i.e., how it prevents color non-singlet states to
be excited. The answer suggested by Fritzsch and Gell-Mann was that color SU(3)
was a local gauge symmetry of hadronic physics. The action had to be invariant
under arbitrary nonabelian SU(3)-transformations of the quark fields
a (x)λa /2
q(x) → e−iα q(x), (28.3)

where the matrices λa now act on the three color labels of the quark field q(x) in
SU(3). The tripling of the quarks saved not only the validity of the spin-statistics
relation for quarks. It also led to the correct rate of the particle decay

π 0 → γγ

which is observed experimentally at a rate Γ = 8.4 × 10−17 s. In addition, it gave


the correct total interaction cross section observed in e+ e− collisions, which require
that each quark occurs in three color versions.
Once it was postulated that the quark action is invariant under local symmetry
transformations, it became necessary to introduce a gauge field to maintain the
invariance of the gradient term in the action. This led to the quark Lagrangian

L(x) = q̄(x)i/
D q(x) − M q̄(x)q(x) (28.4)
1488 28 Nonabelian Gauge Theory of Strong Interactions

where q(x) carries flavor indices distinguishing the quarks u, d, c, s, . . . and three
color indices. The gradient term contains the covariant derivative

/ = γ µ Dµ = γ µ (∂µ + igAµ ) ,
D (28.5)

where Aµ is a 3 × 3 matrix in color space to guarantee that Dµ transforms under


local color SU(3) transformations in the same way as the field ψ(x) itself:

Dµ q(x) → U(x)Dµ q(x) = e−iαa (x)λ


a /2
Dµ q(x). (28.6)

The covariance property is a bit harder to show than in electromagnetism. The


derivative of the field goes over into:

∂µ q(x) → ∂µ U(x)q(x) = U(x)[U −1 (x)∂µ U(x)q(x)]


= U(x)∂µ q(x) + U(x)[U −1 (x)∂µ U(x)]q(x). (28.7)

In order to remove the second term, the gauge field Aµ has to transform like
1
Aµ (x) → U(x)Aµ (x)U −1 (x) − [∂µ U(x)]U −1 (x)
ig
1
= U(x)Aµ (x)U −1 (x) + U(x)∂µ U −1 (x). (28.8)
ig
Then
h i
[∂µ + igAµ (x)] q(x) → U(x) ∂µ + iqAµ (x) + U −1 (x)∂µ U(x)
= U(x) [∂µ + igAµ (x)] q(x), (28.9)

which is the desired covariant transformation law.

28.2 Gluon Action


After the introduction of such a gauge field, an interaction has to be found describing
the dynamics of this gauge field itself. If color is never observed, the action of the
gauge field should be locally SU(3)-invariant as well. The only Lagrangian which
has this property and contains, at most, first derivatives in Aµ is given by
1
Lgluon = − tr (Fµν F µν ) (28.10)
2
where

Fµν = ∂µ Aµ − ∂ν Aµ + ig[Aµ , Aν ] (28.11)

is the field tensor, which is the nonabelian version of the covariant curl (4.807) of the
vector potential Aµ . Just as Aµ , the field tensor is a 3 × 3 matrix in color space. It
is easy to verify that the matrix Fµν transforms under SU(3) covariantly as follows

Fµν (x) → U(x)Fµν (x)U(x). (28.12)


28.3 Quantization in the Coulomb Gauge 1489

The theory is described by the Lagrangian


1
L(x) = q̄(x)i [∂µ + igAµ (x)] q(x) − tr [Fµν (x)F µν (x)] . (28.13)
2
This is a complete nonabelian analog of the gauge field Lagrangian (12.1)–(12.3)
in quantum electrodynamics. For this reason it has been given the similar name
quantum chromodynamics (QCD).
Instead of 3 × 3 matrices Aµ , Fµν , one can also use an octet of vector and tensor
fields defined by
λa a λa a
Aµ = Aµ , Fµν = F . (28.14)
2 2 µν
Thus we can write the Lagrangian of QCD as

λa
" #
µ 1 a
L = q̄(x)iγ ∂µ + ig Aaµ (x) q(x) − Fµν a
(x)Fµν (x), (28.15)
2 4

where
a
Fµν (x) = ∂µ Aaν (x) − ∂ν Aaµ (x) − gf abc Abµ (x)Acν (x). (28.16)

In terms of the eight components Aaµ (x), the Lagrangian (28.10) reads

1 a a µν
LFgluon = − Fµν F . (28.17)
4
There exists an equivalent way of expressing it in terms of two independent fields
Aaµ (x) and Fµν
a
, often used by Schwinger:

1 a a µν 1 a  
Lgluon = Fµν F − Fµν ∂µ Aaν − ∂ν Aaµ . (28.18)
4 2
The spacetime integral over this is the canonical action corresponding to the me-
chanical action of a free particle [recall (1.14)]:

p2
Z !
A= dt − + pq̇ . (28.19)
2

In the past, evidence has accumulated that the gauge field Aaµ is indeed capable
of describing the forces which bind together the quarks inside hadrons. The field
quanta carried by Aaµ are called gluons.

28.3 Quantization in the Coulomb Gauge


Let us quantize the theory for the simplest nonabelian symmetry SU(2), where the
gluons Aaµ (a = 1, 2, 3) are vectors in color space. We shall follow the description of
1490 28 Nonabelian Gauge Theory of Strong Interactions

the theory in Ref. [4, 5]. In SU(2)-symmetry, the structure constants f abc reduce to
ǫabc and the covariant curl (28.16) can be written in vector notation as

Fµν (x) = ∂µ Aν (x) − ∂ν Aµ (x) − gAµ(x) × Aν (x), (28.20)

and the Lagrangian (28.18) becomes


1 1
LFgluon = Fµν · Fµν − Fµν · (∂ µ Aν − ∂ ν Aµ − gAµ × Aν ) . (28.21)
4 2
This is invariant under the nonabelian gauge transformations (28.8) whose infinitesi-
mal form is a rotation in isospace by a small angle ǫ around the direction n̂. If u ≡ ǫn̂
denotes the associated infinitesimal rotation vector, the gauge transformations take
the form
1
Aµ (x) → Auµ (x) = Aµ (x) + u(x) × Aµ (x) + ∂µ u(x)
g
Fµν → Fuµν = Fµν + u × Fµν . (28.22)

The first Euler-Lagrange equation

∂LFgluon
=0 (28.23)
∂F aµν
a
reproduces the relation (28.20) between the curl and the auxiliary tensor field Fµν .
The second Euler-Lagrange equation

∂LFgluon ∂L
∂µ = (28.24)
∂(∂µ Aaν ) ∂Aaν
yields the field equation

D µ Fµν ≡ ∂ µ Fµν − gAµ × Fµν = 0. (28.25)

The combination of Eqs. (28.20) and (28.25) coincides with the field equation that
would be obtained from the second-order formulation with the Lagrangian Lgluon of
Eq. (28.17) expressed directly in terms of ∂ µ Aν and Aµ via Eq. (28.11).
In the present first-order formulation, one is given an initial configuration of
fields Ai and F0i at some time t. From these one determines the fields at any later
time by solving the first-order equations of motion

∂0 Ai = F0i + (∇i − gAi ×) A0 , (28.26)


∂0 F0i = (∂j − gAj ×) Fji + gA0 × F0i . (28.27)

As in the abelian case of Maxwell electromagnetism [recall Eq. (7.337)], the field A0
is not a dynamical variable since the canonical momenta

∂L/∂ (∂0 Aµ ) = −F0µ = Fµ0 (28.28)


28.3 Quantization in the Coulomb Gauge 1491

vanish for µ = 0. There are only three independent field momenta. The Euler
equation for A0 is not an equation of motion but a constraint equation analogous
to the Coulomb law (7.339):

(∇k − gAk ×) Fk0 = 0. (28.29)

As in the Abelian case, it tells us that not all of the conjugate momenta F0k are
independent.
Note that in the first-order formulation, the field equation obtained from varying
Fij ,

Fij = ∂i Aj − ∂j Ai − gAi × Aj , (28.30)

is also a constraint equation which allows us to calculate Fij for given Ai at the
same time. In addition we see from Eq. (28.30) that not all field components Ak
can be treated as independent.
In order to remove the redundancy we choose the Coulomb gauge

∇k Ak = 0. (28.31)

This is always possible because of the gauge invariance of the second kind of the
Lagrangian density [recall (4.255)]. The gauge (28.31) implies that the vector field
Ai must be transverse. Therefore, the longitudinal components FL0i of the canonical
momentum F0i are not independent, but they depend on the other degrees of free-
dom through the constraint (28.29). The splitting of F0i into longitudinal FL0i and
transverse parts FT0i is defined by the equations

F0i = FT0i + FL0i , ∇i F0i = ∇i FL0i , ǫijk ∇j FL0k = 0. (28.32)

Our task is now to express A0 and FL0i in terms of the independent fields and
construct the Hamiltonian. As usual we identify the transverse components FT0i as
the electric field strengths Ei . By rewriting the longitudinal components FL0i as a
gradient of a scalar isovector f,

FL0i = −∇i f, (28.33)

we have

∇i F0i = −∇2 f. (28.34)

The independent variables are the transverse vector fields Ai and its canonically
conjugate field momentum Ei .
Inserting (28.33) into the constraint (28.29), we obtain the differential equation
for f:
 
∇2 + gAk × ∇k f = gAi × Ei . (28.35)
1492 28 Nonabelian Gauge Theory of Strong Interactions

This equation can be solved formally by introducing a Green function D ab (x, x′ ; A),
defined as a solution of the inhomogeneous differential equation
 
∇2 δ ab + gǫacb Ack ∇k D bd (x, x′ ; A) = δ ad δ (3) (x − x′ ). (28.36)

With the help of this Green function, Eq. (28.35) can be solved for the components
of f by the integral
Z
f a (x, t) = g d3 x′ D ab (x, x′ ; A)ǫbcd Ack (x′ , t)Ekd (x′ , t). (28.37)

This may be abbreviated as


f = g D[ATk , ETk ]. (28.38)
The solution D bd (x, x′ ; A) of Eq. (28.36) cannot be found explicitly, but only via
a perturbation expansion. The lowest approximation coincides with δ ab times the
Green function −1/∇2 of electrostatics. A first iteration yields additional terms up
to the order g:

ab δ ab Z
1 1

D (x, x ; A) = ′
+ g d3 x′′ ′′
ǫacb Ack ∇k + . . . . (28.39)
4π|x − x | 4π|x − x | 4π|x − x′′ |

This first-order approximation is easily verified by inserting it into Eq. (28.36).


A similar equation for A0 is obtained by taking the divergence of Eq. (28.26)
and using (28.31) and (28.32) to find
 
∇2 + gAi × ∇i A0 = ∇2 f. (28.40)

This can be solved using once more D ab (x, x′ ; A), since the operator in brackets is
the same as in Eq. (28.35):
Z
Aa0 (x, t) = d3 x′ D ab (x, x′ ; A)∇2 f b (x′ , t), (28.41)

or in short form:

A0 = D∇2 f. (28.42)

We can now construct the Hamiltonian density H by the Legendre transformation


of the Lagrangian density (28.18):

∂Ai
H = Ei · − LFgluon . (28.43)
∂t
From (28.26), (28.32), (28.33), and (28.42), we find that
h i
∂t Ai = Ei − ∇i − (∇i + gAi ×)D · ∇2 f. (28.44)
28.3 Quantization in the Coulomb Gauge 1493

Because of (28.36), the operator in brackets acting on f is explicitly transverse.


Combining further (28.44) and (28.35) or (28.37), we obtain that
Z Z h i
3
d x Ei · ∂t Ai = d3 x E2i + g(Ei × Ai ) · D · ∇2 f
Z h i Z h i
= d3 x E2i − f · ∇2 f = d3 x E2i + (∇i f)2 . (28.45)

If we now rewrite the Lagrangian density as


1 1
LFgluon = Fµν · Fµν − Fµν · (∂ µ Aµ + gAµ × Aν )
4 2
1 2 1 1
= (F0k − B2i ) = (Ek − ∇k f)2 − B2k , (28.46)
2 2 2
and identify the magnetic field strength as
1
Bi = ǫijk Fjk , (28.47)
2
we find the Hamiltonian
1
Z h i
H= d3 x E2i + B2i + (∇i f)2 . (28.48)
2
The last term is like the familiar instantaneous Coulomb interaction discussed in
QED in Section 12.3 [see the last term in Eq. (12.81)].
We now express the generating functional WC [j] in the Coulomb gauge in terms
of the independent coordinates and momenta, Ai and Ei , as the functional integral

1 1
Z  Z  
WC [j] = DETi DATi exp i 4
d x Ek · Ȧk − E2k − B2k − Ak · jk (28.49)
2 2
where the superscript T indicates the transverse spatial components of the field, and
f is the functional (28.38) depending on ETi and ATi as specified in Eq. (28.37). Note
that the spatial source term at the end has a negative sign, so that the covariant
functional to be derived below in Eq. (28.63) will contain the four-dimensional scalar
coupling of the vector field to the source with a positive sign: Aµ · jµ .
The transverse field ETi is somewhat awkward to handle. Therefore we introduce
an initially dummy variable EL by
Z Z
DETi = DETi DEL δ (3) [EL ], (28.50)

where δ (3) [EL ] ≡ x δ (3) (EL (x)) is the δ-functional in four-dimensional spacetime.
Q

Then we define three independent transverse and longitudinal components Ei by

1 1
 
ETi = δij − ∇i 2 ∇j Ei , E L ≡ ∇i ∇j E j , (28.51)
∇ ∇2
1494 28 Nonabelian Gauge Theory of Strong Interactions

in terms of which we can rewrite the measure (28.50) as


Z Z
DETi = DEi Jδ[∇j Ej ], (28.52)

where J is a field-independent, and thus irrelevant, Jacobian of the transformation


from the three Ei to ETi , EL , and
3 Y
3
dEia (x).
YY
DEi ≡
x i=1 a=1

The same decomposition is applied to the gauge fields ATi , so that the generating
functional has the functional integral representation
Z
WC [j] = const × DEi DAi δ[∇k Ek ]δ[∇k Ak ]
1 1 1
 Z  
× exp i d4 x Ek · Ȧk − E2k − B2k − (∇k f)2 − Ak · jk . (28.53)
2 2 2
From this expression we can derive the Feynman diagram rules in the Coulomb
gauge. As in the abelian case, these are not covariant, and the Lorentz covariance of
the emerging S-matrix is not obvious. The Coulomb gauge is only useful to derive
the generating functional from the canonical formalism.
In order to find the Feynman rules for the covariant and gauge invariant S-matrix,
one should start out with a covariant-looking form of the generating functional
instead of (28.53). There, f is a function of E and A given by Eq. (28.37). It is
possible to introduce f as a variable of integration and fix its value to satisfy (28.37)
by a δ-functional
Z
Df δ [f − gD · Ak × Ek ] = 1. (28.54)

If the generating functional (28.53) is multiplied by this expression, it obviously


remains unchanged. Let Det M be the Jacobian of the transformation from f to
(∇2 + gAi × ∇i )f. The factor in front of f is a functional matrix in isospin space:
 
M ab (x, x′ ) = ∇2 δ ab + gǫabc Aci (y)∇i δ (4) (x − x′ )
h i
= ∇2 δ ab δ (3) (x − x′ ) + gǫabc G(x, x′ )Aci (x′ )∇i δ(x0 − x′0 ), (28.55)

where G(x, y) is the Green function satifying ∇2 G(x, y) = δ (3) (x − x′ ). With the
help of Eq. (28.35), we can now rewrite (28.54) as
Z Z
Df δ [f − gD · Ak × Ek ] = Det M Df δ[(∇2 + gAi × ∇i )f − gAi × Ei ], (28.56)

and the generating functional (28.53) becomes


Z
WC [j] = Det M DAi DEi Df δ[∇i Ai ] δ[∇i Ei ]δ[(∇2 + gAi × ∇i )f − gAi × Ei ]
1h
 Z  i2  
× exp i Ek · Ak − fk2 + B2k + (∇k f) − ji · Ai d4 x . (28.57)
2
28.3 Quantization in the Coulomb Gauge 1495

Next we change variables from Ei to

F0i = Ei − ∇i f, (28.58)

using (28.32) and (28.33). Then we rewrite the measure of integration in (28.57) as

DEi Df δ[∇i Ei ] δ[(∇2 + gAi × ∇i )f − gAi × Ei ]


= DF0i Df δ[∇i F0i + ∇2 f] δ[∇2 f − gAi × F0i ]
= DF0i Df δ[∇i F0i + gAi × F0i ] δ[∇2 f − gAi × F0i ]. (28.59)

Now we perform the integration over Df using the last δ-function in (28.59). The
Jacobian is just Det∇2 , i.e., an irrelevant infinite constant which shall be absorbed
into the definition of M. Thus we obtain
Z
WC [j] = Det M DAi DF0i δ[∇i Ai ] δ[∇i F0i + gAi × F0i ] (28.60)
1 1
 Z  
× exp i d4 x F0i · ∂0 Ai − F20i − (∇i Aj −∇j Aj +gAi × Aj )2 − ji · Ai .
2 4
The exponent has been found by setting in the exponent of (28.57)

E2k + (∇k f)2 = (Ek − ∇k f)2 = F20i ,

and omitting the mixed term since it vanishes upon integration over x, due to the
transversality of Ek .
Next we express the last factor in the measure of (28.60) as a functional integral
over a dummy field variable A0 :
YZ dA0
δ[∇i F0i + gAi × F0i ] = exp {iA0 · (∇0 F0i − gAi × F0i )}
x 2π
Z  Z 
= const.× DA0 exp i d4 x F0i (gA0 × Ai − ∇i A0 ) .(28.61)

Finally, we write the term 41 (∇i Aj − ∇j Ai + gAi × Aj )2 in the exponent of (28.60)


as
1 1
Z   
DFij exp i Fij · Fij − Fij · (∇i Aj − ∇j Ai + gAi × Aj ) , (28.62)
4 2
which is a standard Gaussian integral. Inserting (28.62) and (28.61) into (28.60),
we obtain
Z
WC [j] = Det M DAµ DFµν δ[∇i Ai ]
1 1
 Z 
4
× exp i d x − F0i · F0i + Fij · Fij
2 4
1

− Fij · (∇i Aj − ∇j Ai + gAi × Aj ) + F0i (∂0 Ai − ∇i A0 + gA0 × Ai )
2 Z  Z 
= Det M DAµ DFµν δ[∇i Ai ] exp i d4 x[LFgluon + jµ · Aµ ] . (28.63)
1496 28 Nonabelian Gauge Theory of Strong Interactions

Were it not for the factor Det M, this would directly define covariant Feynman
rules. In order to calculate the effect of thei factor, it is useful to reexpress it in
terms of an effective Lagrangian density. Recalling the explicit form of the functional
matrix (28.55), we factorize it as

Det M = Det ∇2 · Det [ 1̂ + M̂ ], (28.64)

where

M̂ = gǫabc G(x, y)Aci (y) · ∇i δ(x0 − y0 ), 1̂ = δ ab δ 4 (x − y). (28.65)

The determinant Det ∇2 is again an irrelevant infinite constant. The second factor
is expanded as

Det ( 1̂+ M̂) = exp Tr log( 1̂ + M̂ ) (28.66)



X (−1) n−1 Z h i
= exp d4 x1 . . . d4 xn tr M̂ (x1 , x2 )M̂ (x2 , x3 ) . . . M̂ (xn , x1 ) .
n=0 n

The trace symbol tr runs only over isospin indices. Inserting (28.65), this becomes

gn Z 3
" ( Z
d x1 . . . d3 xn dt tr [T · Ai1 (x1 , t)∇i1 G(x1 , x2 )
X
Det ( 1̂+ M̂) = exp δ(0) −
n=0 n

× T · Ai2 (x2 , t)∇i2 G(x2 , x3 ) · · · T · Ain (xn , t)∇in G(xn , x1 ) ,(28.67)

where (T a )bc = ǫabc and Tr includes the trace over isospin indices.
Since (28.67) is a power series in the exponent, it is an effective correction in
each order to the Feynman rules obtained from LFgluon alone.

28.4 General Functional Quantization of Gauge Fields


Equation (28.63) can be further simplified. We can perform the functional integra-
a
tion over Fµν and obtain
Z  Z 
W [j] = Det M DAµ δ [∇i Ai (x)] exp i d4 x[Lgluon (x) + jµ (x) · Aµ (x)] , (28.68)

where L(x) is the second-order Lagrangian density (28.17). Except for the fac-
tor Det M δ [∇i Ai (x)], this expression looks the same as for a standard scalar field
theory:
Z  Z 
4
W [j] ∼ Dφ exp i d x[L(x) + j(x)φ(x)] . (28.69)

For the abelian gauge theory QED we have shown in Section 14.16 how to derive
such a factor following an intuitive argument due to Faddeev and Popov. Here we
may do the same for the nonabelian case. Recall how the argument went in the
28.4 General Functional Quantization of Gauge Fields 1497

abelian case. There we expressed the quadratic part of the action in the bilocal
form
1 1
Z Z
A0 = − d4 x (∂µ Aν − ∂ν Aµ )2 = d4 xd4 x′ Aµ (x)D µν (x, x′ )Aν (y), (28.70)
4 2
with a functional matrix
 
D µν (x, x′ ) = − ∂ 2 g µν − ∂ µ ∂ ν δ 4 (x − x′ ).

This cannot be inverted, since it involves only the transverse components of Aµ ,


while ignoring the longitudinal components. Hence the Euclidean version of the
functional integral of Eq. (28.68) contains no Gaussian exponential involving the
longitudinal components of Aµ , and therefore the functional integral over these
diverges. In the abelian case this was a consequence of the invariance of the exponent
in (14.347) under the gauge transformation (14.345). Here it is a consequence of the
invariance of the exponent of (28.68) under the gauge transformation

Aµ → Agµ . (28.71)

The right-hand side emerges as a result of applying the element g of the gauge group
G to the field Aµ :
" #
1
Agµ · L = U(g) Aµ · L + U −1 (g)∂µ U(g) U −1 (g). (28.72)
ig

The exponent in the functional integral (28.72) is constant on the orbits of the gauge
group, which are formed by all Agµ for fixed Aµ , whiled dg runs over the entire group
G. This causes a divergence of the functional integral for the generating functional
W [j]. The amplitude W [j = 0] is therefore proportional to the “volume” of orbits
x dg(x), and this factor should be extracted before defining W [j = 0]. Thus the
Q

functional integral should not be performed over all fluctuations of the gauge fields,
but only over the different orbits of Aµ defined by the symmetry transformations of
the gauge group.
To implement this idea, we choose a “hypersurface” in the manifold of all fields
which intersects each orbit only once. Let

fa (Aµ ) = 0, a = 1, 2, . . . N (28.73)

describe such hypersurface, where N is the dimension of the group. We shall assume
that the equation
fa (Agµ ) = 0
has a unique solution g for any given field Aµ . We are going to integrate over all
different hypersurfaces of this kind, instead of integrating over the manifold of all
fields. The conditions fa (Aµ ) = 0 define a particular gauge, the Coulomb gauge
fa (Aµ ) = ∇i Aai being just a particular example.
1498 28 Nonabelian Gauge Theory of Strong Interactions

Before proceeding further, let us recall briefly some simple facts about group
representations. For any two group elements g, g ′ ∈ G, the product gg ′ is also ∈ G,
and their representations satisfy the same multiplication law

U(g)U(g ′ ) = U(gg ′).

The invariant Hurwitz measure over the group G is invariant under this operation,
so that

dg ′ = d(gg ′). (28.74)

If we parametrize U(g) in the neighborhood of the identity as

U(g) = 1 + iu · L + O(u2 ),

then in the neighborhood of the identity we may choose


Y
dg = dua , g ∼ 1. (28.75)
a

Let us define the functional ∆f [Aµ ] by


Z
δ[fa (Agµ (x))] = 1,
Y
∆f [Aµ ] Dg (28.76)
a

where Z Y Z 
Dg ≡ dg(x) . (28.77)
x

Without the gauge invariance, the vacuum-to-vacuum amplitude would be given by


Z  Z 
4
DAµ exp i d x Lgluon (x) . (28.78)

We now insert the left-hand side of Eq. (28.76) into the functional integral (28.78)
without changing the result:
Z  Z 
δ[fa (Agµ (x))] exp i d4 x Lgluon (x) .
Y
DgDAµ∆f [Aµ ] (28.79)
a

Now we perform a gauge transformation Aµ (x) → Agµ (x) defined by (28.72) on


Aµ (x). Under this, the action in (28.78) is invariant. We may also convince ourselves
that the functional ∆f [Aµ ] defined by (28.76) is gauge invariant:
Z Y
g ′
∆−1 dg ′(x) δ[fa (Agg
Y
f [Aµ ] = µ (x))]
x x,a
Z Y

d(g(x)g ′(x)) δ[fa (Agg
Y
= µ (x))]
x x,a
Z Y
′′
dg ′′(x) δ[fa (Agµ (x))],
Y
=
x x,a
28.4 General Functional Quantization of Gauge Fields 1499

so that indeed
∆f [Agµ ] = ∆f [Aµ ]. (28.80)
Hence we may write Eq. (28.79) also as
Z Z  Z 
d4 x Lgluon (x) ,
Y
Dg DAµ ∆f [Aµ ] δ[fa (Aµ )] exp i
x,a

and we find that the integrand of the group integration is independent of g(x). RThis
was the observation of Faddeev and Popov, who saw that the functional integral Dg
is simply an infinite factor independent of the fields. It can therefore be dropped
from the amplitude, so that the generating functional W [j] may be defined as
Z  Z 
a 4 µ
Y
Wf [j] = DAµ ∆f [Aµ ] δ[f (A)] exp i d x[L(x) + j (x) · Aµ (x)] . (28.81)
a

Faddeev and Popov also gave the canonical derivation of Eq. (28.63) as discussed in
the preceeding section.
Before demonstrating the equivalence of Eqs. (28.68) and (28.81), we shall
compute ∆f [Aµ ]. Since the factor ∆f [Aµ ] is multiplied by a δ[f a (Aµ (x))] in
Q

Eq. (28.81), it suffices to compute ∆f [Aµ ] only for vector fields Aµ which satisfy
Eq. (28.73). Let us define the functional matrix Mf by
Z
a
(Agµ (x)) a
d4 y [Mf (x, y)]ab ub (y) + O(u2 ).
X
f = f (Aµ (x)) + (28.82)
b

Then we find from Eq. (28.76) that


Z Yn o Z Y
a
∆−1
f [Af ] = dua (x)δ[f (Agµ (x))] = {dua (x)δ[Mf u]} .
x,a x,a

The integral receives a contribution only from A-fields satisfying fa (Aµ ) = 0, so


that
∆f [Aµ ] = Det Mf = exp {Tr log Mf } . (28.83)

The hypersurface equation fa = 0 is just the gauge condition, and for the
Coulomb gauge adopted in the preceeding section, we had
f a (Aµ ) = ∇i Aai = 0,

which becomes, after an infinitesimal gauge transformation,


1  2 ab 
f a (Agµ ) = ∇i Aai + ∇ δ − gǫabc Aci ∇i ub (x) + O(u2 ). (28.84)
g
From this we identify
1 1
 
[Mf (x, x )]ab ∼ ∇2 δ ab − g abc 2 Aci ∇i δ 4 (x − x′ ) ∼ [M(x, x′ )]ab ,

(28.85)
g ∇
1500 28 Nonabelian Gauge Theory of Strong Interactions

which shows that Eq. (28.68) is indeed a special case of Eq. (28.81) for f a = ∇i Aai .
Being in possession of Eq. (28.81), we are free to use many different gauges other
than the Coulomb gauge. If we choose, for example, the manifestly covariant Landau
gauge condition ∂ µ Aµ = 0, then Eq. (28.82) takes the form

1
∂ µ Agµ (x) = ∂ µ Aµ (x) + [∂ 2 u + g∂ µ (Aµ × u)] + O(u2 ), (28.86)
g
so that Mf is given by

1
[ML (x, x′ )]ab = (∂ 2 δab − gǫabc Acµ ∂ µ )δ 4 (x − x′ ) (28.87)
g

when Aµ is restricted to ∂ µ Aµ = 0. Removing the trivial factor (1/g)∂ 2 from


ML (x, x′ ), we have
n o
∆L ≡ Det ML ∼ exp Tr ln(1̂ + M̂L ) , (28.88)

where
∂ 4
Z
M̂Lab (x, x′ ) = gǫabc DF (x − z)Acµ (z) δ (z − y)d4z, (28.89)
∂zµ

and DF (x−z) is the usual Feynman propagator satisfying −∂ 2 DF (x−y) = δ 4 (x−y).


More explicitly we can write

(−g)n
( Z h
d4 x1 . . . d4 xn tr ∂ λ DF (x1 − x2 )T · Aµ (x2 )∂ µ DF (x2 − x3 )
X
∆L = exp −
n=0 n
× . . . DF (xn − x1 )T · Aλ (x1 )]} . (28.90)

The need to have the extra factor ∆f [Aµ ] δ[f a (Aµ (x)] was first noted by Feynman.
Q
a
We can write Eq. (28.81) as
Z   Z 
a 4 µ
Y
Wf [j] = DAµ δ[f (Aµ (x))] exp i Seff + d x j (x) · Aµ (x) , (28.91)
a

where Z
Aeff = d4 x Lgluon (x) − iTr ln ML . (28.92)

At this point it is useful to observe that the additional term −iTr ln ML in the
effective action may be thought of as arising from loops generated by a fictitious
isotriplet of complex scalar fields c obeying Fermi statistics, whose presence and
interactions can be described by the following action
Z h i Z
d4 x ∂ µ c† (x) · Aµ (x) × c(x) ∼ d4 xd4 x′ c†a (x) [ML (x.x′ )]ab cb (x′ ).
X
Ac = −
a,b
(28.93)
28.4 General Functional Quantization of Gauge Fields 1501

With this, Eq. (28.91) may be written as


Z Z   Z 
a † 4 µ
Y
WL [j] = DAµ δ [f (Aµ (x))] Dc Dc exp i A + Ac + d x j · Aµ (x) .
a
(28.94)
Since c and c† appear quadratically in the action (28.93), the path integral Dc† Dc
R

can trivially be done with the result


Z
Dc† Dc exp(iAc ) ∼ Det ML = exp {Tr ln ML } , (28.95)
and we can expand
h i
exp {Tr ln ML } ∼ exp Tr ln(1 + M̂L )
1 1
 
= exp Tr M̂L − Tr M̂L2 + . . . − (−)n Tr M̂Ln + . . . . (28.96)
2 n
The terms in the exponent may arise from loop diagrams of the Fermi fields c.
Let us now specify the Feynman rules resulting from WL [j] of Eq. (28.94), fol-
lowing from the general rules desribed in Chapter 14. The gauge boson propagator
is determined from theZ
free-field functional
R 4
DA δ[∂ µ A ]ei d x{− 4 (∂µ Aν −∂ν Aµ ) + j (x)·Aµ (x)} .
1 2 µ
W 0 [j] =
L µ µ (28.97)
A convenient way of evaluating Eq. (28.97) is to express the δ-functional as
i
 Z 
δ[∂ µ Aµ (x)] ∼ lim exp d4 x[∂ µ Aµ (x)]2 , (28.98)
α→0 2α
Q √
where we have discarded an irrelevant infinite constant x 2πα. Then we arrive
at the generating functional of free vector boson Green functions:
i 1
Z  Z   
0 4 2 µν µ ν
WL [j] = lim DAµ exp − d xAµ (x) · −∂ g + ∂ ∂ 1 − Aν (x)
α→0 2 Z
α 
+i d4 x jµ (x) · Aµ (x)
i
 Z 
= lim exp − d4 xd4 x′ jµ (x) DFµν (x − x′ ; α) jν (x′ ) , (28.99)
α→0 2
µν ′
where DF (x − x ; α) is the free vector boson propagator
d4 k
" #
Z
1 kµ kν
DFµν (x ′
− x ; α) = 4
exp {ik · (x − x′ )} 2 −gµν + 2 (1 − α) . (28.100)
(2π) k + iǫ k
In the limit α → 0, this becomes
d4 k
!
1 kµ kν
Z
DFµν (x ′
−x) = 4
exp {ik · (x − x′ )} 2 −gµν + 2 , (28.101)
(2π) k + iǫ k
which is transverse in spacetime. The rest of the Feynman rules can be derived as
usual. They are recorded in Figs. 28.1 and 28.2. In addition, the following rule must
be kept in mind: the ghost-ghost-vector vertex may carry a “dot” which indicates
that a ghost line is differentiated. Note that a ghost line cannot be dotted at both
ends and that a ghost loop carries an extra minus sign.
1502 28 Nonabelian Gauge Theory of Strong Interactions

Figure 28.1 Propagators in the Yang-Mills theory. Wavy lines are vector mesons. Dashed
lines are scalar ghosts.

Figure 28.2 Vertices in the Yang-Mills theory. Note that the ghost with index c in the
last diagram is pictured by a dotted line.

28.5 Equivalence of Landau and Coulomb Gauges

Formally, the S-matrix computed in the Landau gauge is the same as that com-
puted in the Coulomb gauge [5]. An element of the unrenormalized S-matrix is
obtained from the corresponding Green functions by removing single particle propa-
gators corresponding to external lines, taking the Fourier transform of the resulting
“amputated” Green function, and placing external momenta on the mass shell. The
demonstration to be presented is basically correct, except that the S-matrix of a
gauge theory is plagued by infrared divergences and may not even be defined. In
fact, this may be the reason why massless Yang-Mills particles are not seen in na-
ture. The point of presenting this demonstration is at this point purely pedagogical.
The technique will be useful in the discussion of spontaneously broken versions of
gauge theories.
Let us first establish a relation between WC [j] and WL [j]. From Eq. (28.68) we
obtain for the first:
Z  Z 
4 µ
WC [j] = DAµ ∆C [Aµ ]δ[∇i Ai ] exp iA[Aµ ] + i d x j · Aµ , (28.102)
28.5 Equivalence of Landau and Coulomb Gauges 1503

where ∆C = Det M satisfies


Z
∆C [Aµ ] Dgδ[∂ µ Agµ (x)] = 1. (28.103)

Inserting the left-hand side of Eq. (28.103) into the integrand of the functional
integration in Eq. (28.102), we write
Z Z
WC [j] = Dg DAµ ∆C [Aµ ]∆C [Aµ ]δ[∇i Ai ]
 Z 
µ
× δ[∂ Agµ ] exp iA[Aµ ] + i 4
d xj · Aµ . µ

Now we perform a gauge transformation of the integration variables Aµ (x):


Aµ (x) → Agµ (x).
Recalling the gauge invariance of the action A, the functional ∆f , and the metric
DAµ (x), we find that
Z
WC [j] = DAµ ∆C [Aµ ]δ[∂ µ Aµ ] exp(iA[Aµ ]) (28.104)
Z Y  Z 
−1
δ(∇i Aig ) exp i
−1
d4 xjµ · Aµg
Y
×∆C [Aµ ] dg(x)
x x
Z  Z 
µ 4 µ
Agµ0
Y
= DAµ ∆L [Aµ ] δ(∂ Aµ (x)) exp iA[Aµ ] + i d xj · .
x

As before, Agµ0 is the gauge transform of Aµ , which satisfies ∂ µ Aµ = 0, such that


( " # )
1
L· ∇i Agi 0 = ∇i U(g0 ) L · Ai + U −1 (g0 )∇i U(g0 ) U −1 (g0 ) = 0. (28.105)
ig
When deriving Eq. (28.104), we have used the fact that
Z Y Z (Y Y )
−1
δ(∇i Aig
Y
∆C [Aµ ] dg(x) = ∆C [Aµ ] dua(x)
x x x a
!
1
∇i Agi 0 − M[Agµ0 ]u ∼ ∆C [Aµ ]∆−1 g0
Y
× δ C [Aµ ] = 1.
x g
It is possible to solve Eq. (28.105) for Agµ0 as a power series in Aµ . This can be done
beginning with
1
 
g0
Ai = δij − ∇i 2 ∇j Aj + O(A2µ ).

The source jµ in the Coulomb gauge will be restricted to
j0 = 0, ∇i ji = 0, (28.106)
so that we may write
Z Z
d4 x jµ · Agµ0 = d4 x jµ · Fµ (x; Aλ ),
1504 28 Nonabelian Gauge Theory of Strong Interactions

where

Fµ (x; Aλ ) = Aµ (x) + O(A2λ ). (28.107)

Carrying this construction to high orders we can finally express WC in terms of WL


as follows:
" ( Z !)#
4 µ 1 δ
WC [j] = exp i d xj (x) · Fµ x; WL [j]|j=0 . (28.108)
i δjλ

It is helpful to visualize the generating functionals (28.104) or (28.108) with the


help of Feynman diagrams. The two expressions imply that the Green functions
in the Coulomb gauge are the same as those in the Landau gauge, if the source
is suitably restricted by equations like those in (28.106). We only must take into
account extra vertices between source and field, represented by the term
Z
d4 x jµ · (Fµ − Aµ ). (28.109)

Then one may construct Green functions in the Coulomb gauge from the Feynman
rules of the Landau gauge. This relationship becomes much simpler if we go to
the mass shell. In this case, we ought to compare only the terms having a pole
in each of the external momentum square p2i . Of all the diagrams generated by
the extra couplings of (28.109), only those survive in this limit in which the whole
effect of the extra vertices can be reduced to a type of self-energy insertion into the
corresponding external line. The other corrections introduced by (28.109) do not
contribute to poles of the Green functions at p2i = 0, and therefore do not contribute
to the S-matrix. Hence in the limit p2i → 0, the unrenormalized S-matrix elements
in Coulomb gauge C and Landau gauge L with propagators differ by coming from
the different propagators

′ ZL ′ ZC
lim Dµν (p; L) = (gµν + . . .), lim Dµν (p; C) = (gµν + . . .).
2
p →0 p2
+ iǫ 2
p →0 p2+ iǫ
In particular, the ratio
σ 2 = ZC /ZL (28.110)
is different from unity. In general, the unrenormalized S-matrix elements in the two
gauges C and L are related to each order by

SC = σ n SL = (ZC /ZL )n/2 SL ,

so that the renormalized S-matrix element


−n/2 −n/2
Sren ≡ ZC SC = Z L SL

is independent of the gauge in which the calculation was done.


As a consequence, WC [j] is equal to the expression (28.104), and thus it is be
equal to WL [j] except that the coefficient of jµ is the gauge-transformed vector field
28.6 Perturbative QCD 1505

Agµ0 , instead of Aµ itself. For the S-matrix, the only consequence of this difference
is that the renormalization constants attached to each external line depend on the
gauge. Thus we have shown that ultimately the S-matrix can be calculated from
WL [j].
As pointed out earlier, the only flaw in the above argument is that the singularity
at p2i = 0 is not in general a simple pole.

28.6 Perturbative QCD


For scattering processes at high energy, the strong interaction becomes effectively
so weak that quarks behave approximately like free pointlike particles. This phe-
nomenon of strong interactions is called asymptotic freedom. It was discovered ex-
perimentally in deep-inelastic scattering at the Stanford Linear Accelerator and led
Feynman to his famous parton hypothesis [11, 12, 13]. By combining this hypothesis
with Gell-Mann’s field theory of quarks held together by a color octet of non-abelian
gauge fields, the present-day quantum field theory of strong interactions was born.
It was called Quantum Chromo Dynamics (QCD) [3]. Theoretically the asymptotic
freedom was first observed for charged vector bosons by V.S. Vanyashin and M.V.
Terantjev as early as 1965 [8], and for the nonabelian gauge theory with local SU(2)-
symmetry by I.B. Khriplovich [9]. But it was not until 1972 that it was noted in
the western physics community, first by ’t Hooft who failed to publish it and told
it only to colleagues, and later by Politzer, Gross, and Wilczek [14, 15] who earned
the Nobel prize of 1973 for it.
For a sufficiently small number of quark flavors Nf < 33/2, the behavior of
the beta function in QCD is opposite to that in Eq. (20.31) of φ4 -theory and in
Eq. (27.30) for QED. It starts out with negative slope and remains negative at
least for a certain range. This is what causes the asymptotic freedom when the
energy-momenta of the particles in all Feynman diagrams grow large. To study this
behavior analytically, we introduce a small number ǫ and imagine living in 4 + ǫ
dimensions with less than Nf = 17 quark flavors. Then we define an analogue of
the fine-structure constant α = e2 /4π for strong interactions, namely the constant
“alpha-strong” αs ≡ g 2 /(4π)2 . By a perturbative calculation of the scale dependence
of the coupling constant, one finds the β-function of the theory as β(g) ≡ µ∂µ αs
[recall (20.59)]. To lowest order, this is
β(αs ) = ǫαs − bαs2 , with b = 11 − 2Nf /3. (28.111)
Then the solution of the differential equation for the scale-dependent coupling con-
stant becomes, by analogy with (20.126):
Z αs (µ) dαs
log µ = . (28.112)
αs ǫαs − bαs2
Because of the sign change with respect to (20.126), the movement of αs for increas-
ing mass scale µ is opposite to that in the φ4 -theory and in QED. The coupling
constant αs = g 2 /(4π)2 goes towards a fixed point g ∗ = ǫ/b in the ultraviolet limit.
1506 28 Nonabelian Gauge Theory of Strong Interactions

By comparison with (20.289), we find the solution after changing the signs of ǫ
and b to satisfy
µ−ǫ 1 b
= − . (28.113)
αs (µ0 ) αs (µ) ǫ
This shows that αs tends to the fixed point αs∗ = ǫ/b in the limit of large µ, i.e., in
the UV-limit. In four spacetime dimensions, we let the auxiliary parameter ǫ go to
zero and see that the solution (28.113) tends to
1 1 b µ
= − ln . (28.114)
αs (µ0 ) αs (µ) ǫ µ0
This is solved for αs (µ) by
1
αs (µ) = µ. (28.115)
αs−1 (µ0 ) + b log
µ0
The first term in the denominator can be absorbed in the logarithm, and we can
rewrite αs (µ) as
1 4π
αs (µ) = 2 = . (28.116)
µ 2 µ2

1
2
b log 11 − Nf log 2
Λ2QCD 3 ΛQCD

The mass ΛQCD is the dimensionally transmuted coupling constant of QCD. For
µ → ∞, the coupling constant αs (µ) goes to zero. This is why the nonabelian gauge
theory with less than Nf < 33/2 flavors is ultraviolet-free.
If higher loops are included in the calculations of QCD, the β-function has the
expansion [18]

β(g) = ǫαs−2(β0 αs2 + β1 αs3 + β2 αs4 + β3 αs5 ) + . . . , (28.117)

where
2 b
β0 = 11 − Nf = ,
3 2
38Nf
β1 = 102 − ,
3
325Nf 2 5033Nf 2857
β2 = − + , (28.118)
54 18 2
1093Nf 3
! !
2 6472ζ(3) 50065 6508ζ(3) 1078361
β3 = + Nf + − Nf +
729 81 162 27 162
149753
+ 3564ζ(3) + .
6
The various partial sums are plotted in Fig. 28.3.
28.7 Approximate Chiral Symmetry 1507

β(αs ) αs
0.1 0.2 0.3 0.4 0.5 0.6
-1

-2

-3
4 3 5 2
-4

-5

Figure 28.3 Flow of the coupling constant αs towards the origin as the scale parameter
µ approaches infinity (ultraviolet limit). For the opposite flow direction (infrared limit),
the arrows are reversed and the coupling strength becomes large leading to the formation
of a fat attractive flux bundle between quarks.

If only the term αs3 is included in the differential equation for the scale-
dependence of the coupling constant, the result changes from (28.116) to

µ2
ln ln
αs (µ) 1 b1 Λ2QCD
= − 3 . (28.119)
4π µ2 b0 2 µ2
b0 ln 2 ln 2
ΛQCD ΛQCD

The presently best fit to the experimental coupling constants yields the value

ΛQCD ≈ 217 ± 25MeV. (28.120)

Since the theory is asymptotically free, many properties of it can be studied in


perturbation theory. The results can be compared quite well with the experimental
data observed in high-energy collisions. Details are amply available in the literature
[16, 17].

28.7 Approximate Chiral Symmetry


At first, the masses of the quarks are approximated to be equal to zero. Then
the Lagrangian density has an additional invariance. If we restrict our attention
only to the three quarks u, d, s, the flavor group is SU(3). For massless quarks, the
symmetry is extended to SU(3) × SU(3), where the extension involves SU(3) flavor.
Its Noether current densities are
λa
jaµ (x) = q̄(x)γ µ q(x),
2
µ λa
j5a (x) = q̄(x)γ µ γ5 q(x), (28.121)
2
1508 28 Nonabelian Gauge Theory of Strong Interactions

where the λa matrices act on the flavor-SU(3) indices. The current densities are
color singlets. The corresponding color octet currents are not observable, due to
local SU(3) color invariance.
After field quantization the field components satisfy the local SU(3) × SU(3)
commutation rules derived in (8.281) and discussed in Section 25.3. The charges Qa
and axial charges Q5a , defined by
Z
Qa = d3 xj 0 ,
Z
Q5a = d3 xj5a
0
, (28.122)

form the Lie algebra

[Qa , Qb ] = ifabc Qc , (28.123)


[Qa , Q5b ] = ifabc Q5c , (28.124)
[Q5a , Q5b ] = ifabc Qc . (28.125)

From these we may form the chiral charges

QLa = (Qa − Q5a )/2, QR


a = (Qa + Q5a )/2, (28.126)

which generate the two commuting groups SU(3)L and SU(3)R , respectively. Due
to the chiral invariance of the massless quark gluon Lagrangian density (28.15), the
currents are conserved:

∂µ j µ (x) = 0,
∂µ j5µ (x) = 0. (28.127)

The ground state of the theory breaks the axial part of this symmetry spontaneously.
This gives rise to non-zero quark masses. The spontaneous breakdown is accompa-
nied by massless pseudoscalar Nambu-Goldstone bosons, which are identified with
the pion and its flavor octet partners.
In nature, quarks are not massless and the axial charges are not conserved. The
masses of non-strange quarks are, however, very small so that the axial charges
with the SU(2)-indices a = 1, 2 are approximately conserved. This is the basis of
the PCAC hypothesis (Partial Conservation of Axial vector Current). The nonzero
masses of mu , md , ms . . . raise the mass of the pion and the other pseudoscalar mesons
to the experimental nonzero values. The quark masses which give a consistent
picture of experimental data are [7]

mu 4.5 ± 1.4
   

 md 


 7.9 ± 2.4 

ms 155 ± 50
   
   
  =   MeV. (28.128)

 mc 


 1270 ± 50 

mt 40000 ± 10000
   
   
mb 4250 ± 100
Notes and References 1509

In recent years, much insight into the theory has been gained from computer
simulations of lattice models of the theory. They have confirmed that the theory
has the desired properties to explain the many strongly interacting particles observed
in the laboratory.

Notes and References


[1] If the confining forces are derived from Monte Carlo simulations of gluons in an SU(3)-lattice
gauge theory, the hadronic mass spectrum of the quark model was calculated by
R.D. Loft and T.A. DeGrand, Phys. Rev. D 39, 2678 (1989).
If the b-quarks are included, the corresponding hyperons are discussed by
C. Alexandrou, A. Borelli, S. Güsken, F. Jegerlehner, K. Schilling, G. Siegert, R. Sommer,
Phys. Lett. B 337, 340 (1994) (hep-lat/9407027).
[2] H. Kleinert, Phys. Lett. B 59, 163 (1975) (http://klnrt.de/50); Phys. Lett. B 62, 77
(1976) (http://klnrt.de/51).
[3] H. Fritzsch, M. Gell-Mann, Intern. Conf. on Duality and Symmetry in Hadron Physics,
Weizmann Science Press, 1971;
H. Fritzsch, M. Gell-Mann, H. Leutwyler, Phys. Lett. B 47, 365 (1973).
[4] V.N. Popov and L.D. Faddeev, Phys. Letters B 25 (1967) 29;
B.S. DeWitt, Phys. Rev. 162, 1195 (1967);
V.N. Gribov, Nucl. Phys. B 139, 1 (1978).
[5] The presentation in this section follows closely the excellent review article by
E.S. Abers and T.D. Lee, Phys. Rep. C 9, 1 (1973).
[6] In addition to the references in [4] see
N.P. Konopleva and V.N. Popov, Kalibrovochnye Polya (Atomizdat, Moscow, 1972), in
Russian.
R.P. Feynman, Acta Phys. Polonica 26, 697 (1963);
B.S. DeWitt, Phys. Rev. 162, 1195, 1239 (1967);
S. Mandelstam, Phys. Rev. 175, 1580 (1968);
E.S. Fradkin and I.V. Tuytin, Phys. Letters B 30, 562 (1969); Phys. Rev. D 2, 2841 (1970);
M.T. Veltman, Nucl. Phys. B 21, 288 (1970);
G. ’t Hooft, Nucl. Phys. B 33, 173 (1971).
[7] H. Kleinert, Collective Quantum Fields, Lectures presented at the First Erice Sum-
mer School on Low-Temperature Physics, 1977, in Fortschr. Physik 26, 565-671 (1978)
(http://klnrt.de/55); J. Gasser and H. Leutwyler, Phys. Rep. 87, 77 (1982);
G. Harrison et al. Phys. Lett. B 47, 493 (1984).
[8] V.S. Vanyashin and M.V. Terantjev, Sov. Phys. (JETP) 48, 565 (1965).
[9] I.B. Khriplovich, Sov. J. Nucl. Phys. 10, 235 (1969) [Yad. Fiz. 10. 409 (1969).
[10] H.D. Politzer, Phys. Rev. Lett. 30, 134 (1973);
D.J. Gross and F. Wilczek, Phys. Rev. D 8, 3633 (1973).
[11] R.P. Feynman, Phys. Rev. Lett. 23, 1415 (1969); See also his lecture on The behavior of
hadron collisions at extreme energies in Proceedings of the 1969 Stony Brook conference
edited by C.N. Yang et al., Gordon and Breach, New York, p. 237. The name “parton” first
appeared here. See also the textbook:
R.P. Feynman, Photon-Hadron Interactions, Benjamin, Reading, 1972.
[12] J.D. Bjorken, Phys. Rev. 179, 1527 (1969).
1510 28 Nonabelian Gauge Theory of Strong Interactions

[13] F.E. Close, An Introduction to Quarks and Partons, Academic Press, London, 1980.
[14] H.D. Politzer, Phys. Rev. Lett. 30, 1346 (1973).
[15] D.J. Gross and F. Wilczek, Phys. Rev. D 8, 3633 (1973); D 9, 980 (1974).
[16] M.E. Peskin and D.V. Schroeder, An Introduction to Quantum Field Theory, Westview
Press, N.Y., 1995. See especially Chapter 16.
[17] J.C. Collins, Foundations of Perturbative QCD, Cambridge Monographs on Particle Physics,
Nuclear Physics and Cosmology, 2013;
T. Muta, Foundations of Quantum Chromodynamics, World Scientific Lecture Notes in
Physics-Vol. 78, Third edition, 2010;
Y.L. Dokshitzer, V.A. Khoze, A.H. Mueller, S.I. Troyan, Basics of Perturbative QCD, Edi-
tion Frontières, Paris, 1991;
G. Sterman, J. Smith, J.C. Collins, J. Whitmore, R. Brock, J. Huston, J. Pumplin, Wu-
Ki Tung, H. Weerts, Chien-Peng Yuan, S. Kuhlmann, S. Mishra, J.G. Morfin, F. Olness,
J. Owens, Jianwei Qui, D.E. Soper, Handbook of perturbative QCD, Reviews of Modern
Physics, 67, 157 (1995).
[18] T. van Ritbergen, J.A.M. Vermaseren, S.A. Larin, Phys. Lett. B 400, 379 (1997).

You might also like