Professional Documents
Culture Documents
where d(g) is the Haar measure on the Lie group G and we integrate over all group elements g G.
To show that M 1 [A] is gauge-invariant, first note that gauge transformation is transitive, i.e., for g, h G,
(Ah )g = (hAh1 dhh1 )g = ghAh1 g 1 gdhh1 g 1 dgg 1 = ghA(gh)1 d(gh)(gh)1 = Agh .
Gauge-invariance of M [A] follows from gauge-invariance of the Haar measure:
Z
Z
Z
Z
1
h
h g
gh
1
g
M [A ] =
d(g) (G(A ) ) =
d(g) (G(A )) =
d(gh ) (G(A )) =
gG
gh1 G
gG
g0
gG
M [A] = det
G(Ag )
.
g
g=e
4. Degree of divergence
(D2 )r
1
F . After gauge-fixing and
1+
Consider a Lagrangian density of the form L(A) = F
4
2
introducing ghost fields, we find the following quantum Lagrangian
( 2 )r
1
( 2 )r
Lq (A, c, c) = L(A) A 1 +
A
+
D c.
c
1
+
2
2
2
The gauge propagator is determined by the quadratic terms in A:
( 2 )r
1
( 2 )r
( 2 )r
1
1
+
1
+
A 1 +
A
A .
2
2
2
2
2
Integrating by parts and throwing away the surface integrals, we find
( 2 )r
1
( 2 )r
( 2 )r
1
+
1
+
A 1 +
A
1
+
A
A
2
2
2
2
2
1
( 2 )r
( 2 )r
1 1
1
= A 1 +
1
+
A
+
A = A 1
1
A
A ,
2
2
2
2
1
( 2 )r
( 2 )r
2
1
+
is the Greens function of the propagator.
1
1+
where = 1 +
2
2
( 2 )r
Similarly, the ghost propagator is determined by c 1 +
c. We solve for the propagator by
2
Fourier transform. Each derivative will become a momentum in the Fourier space. Thus the gauge and
ghost propagators are proportional to p22r . The vertices are determined by the higher-order terms in A
and c. They are
( 2 )r
(D2 )r
1
(D2 )r
A 1+
[A , A ] + [A , A ] 1 +
A + c 1 +
[A , c].
2
2
2
2
Each covariant derivative D = d + A will either add a momentum factor or an extra gauge leg. Therefore,
there are vertices with n = 3, . . . , 2r + 4 gauge legs and 2r + 4 n momentum factors. The ghost-gluon
interaction is determined by the third term. We see that there is a 3-vertex with 2r + 1 momentum factors.
When r = 0, this reduces to the Feynman rules for
Z Yang-Mills theory.
2r+4
X
(1)
n=3
2r+4
X
n=3
2r+4
X
n=3
find
2r+4
X
(2r + 4 n)Vn = 2r + 4 (2r + 4)L + (2r + 2)I E, which we substitute into (1) to find
n=3
D = 2r + 4 + (d 2r 4)L E.
In d = 4, D = 4 E 2r(L 1).
5. One-loop renormalization
The one-loop diagram for a ghost-ghost interaction is
k
p
c, a
, f
, g
c, d
c, e
kp
2
p
c, b
i
de . The gluon propagator con(p k)2
i
f g . The ghost-gluon vertices contribute gp f af d and g(p k) f beg . Thus the total conk2
tribution is
Z
2 Z
dd k p (p k)
p (p k)
2 af d beg
2
af d bdf g
(i) f f de f g g
= f f
.
dd k 2
(2)d k 2 (p k)2
(2)d
k (p k)2
tributes
Recall that we can define an inner product on a Lie algebra g by the Killing form K : g g 7 R defined
by K(X, Y ) = tr(ad(X)ad(Y )). In components, K(T a , T b ) = tr(ad(T a )ad(T b )) = f acd f bdc . For compact Lie
algebras, we can take f acd f bcd = ab C. Note that
Z
Z
Z
Z
Z
p (p k)
p2
pk
p2
p (p k )
d
d
d
dd k 2
=
d
k
d
k
=
d
k
dd k
,
2
2
2
2
2
2
2
k (p k)
k (p k)
k (p k)
k (p k)
(p k )2 k 2
Z
Z
1
p (p k)
p2
d
d
1
d k 2
=
k (p k)2
da1
da2
da1 da2 dd k e
d ke
a1 k2 a2 (pk)2
(a1 +a2 ) k a
a2
1 +a2
da1 da2 dd k e
2
a a
p
a 1+a2 p2
2a2
(a1 +a2 ) k2 a +a
kp+ a
da1 da2
(a1 + a2 )
d/2
a2
1 +a2
a1 a2
p2
e a1 +a2 p .
Now let a1 = s, a2 = (1 )s such that da1 da2 = sdsd. We can rewrite the integral as
Z 1
Z 1
Z
Z
d
1
d/2
d/2 s(1)p2
d/2
=
+
2
d
ds
s
e
=
g2C
. There are two one-loop diagrams for a ghost-gluon interaction:
cc (p, p)ba p2 ab 1
16 2
c, a
c, a
p
, d
pk
c, i
, e
c, h
, f
, g
c, c
c, d
, b
pk
, h
c, e
, i
c, f
c, g
k q
c, c
p q
q
, b
k q
p q
i
i
i
k p (k + q)
de
hi
f g = ig 3 f adh f bdf f chf 2
.
k2
(p k)2
(k + q)2
k (p k)2 (k + q)2
The first diagram gives more complicated terms due to the 3-gluon vertex
i
i
i
de
hi
f g
2
2
k
(p k)
(k + q)2
(p (p k) (q k) + (p k) p (k + 2q) + p (p k)(2k + q) )
= ig 3 f adh f chf f bf d
.
k 2 (p k)2 (k + q)2
1
(2)d
dd k
X
,
k 2 (p k)2 (k + q)2
1 abc
f C. Power-counting shows that only the kk terms in X give rise to a pole in .
2
Z
dd k
p k 2
1
cAc (p, q, q, p),cba igf abcp + if abc g 3 C
d
2
2
(2) k (p k)2 (k + q)2
Z
dd k
1
abc
= igf p 1 +
,
(2)d (p + q k )2 k 2
where we have used the change of variable k = k q. We have already shown that
1
2
dd k
.
(2)d (p k)2 k 2
Hence
g2C
cAc (p, q, q, p),cba igf p 1 +
.
16 2
g2C a 2 a
g 3 C abc a b c
To cancel the divergences, we add the counterterms Lc.t. =
c
c
+
f c A c .
16 2
16 2
6. (g)
dg
The beta function, defined as (g) = , expresses the dependence of the coupling g on mass scale .
d
For (g) = b1 g 3 b2 g 5 + O(g 7 ), we can write the differential equation as
Z
Z
Z
dg
dg
1
d
=
=
b
3
5
7
3
2
2
b1 g + b2 g + O(g )
b1 g 1 + b g + O(g 4 )
1
Z
1
b2
dg
b2 2
2 log g + O(g 2 ).
=
1 g + O(g 4 ) =
b1 g 3
b1
2b1 g 2
b1
abc
The limits of integration is from to a cut-off scale , which we define later. The integral can be evaluated
as
g()
1
1
b2
log
+
+ O(g 2 ).
log =
b2
log g 2 () = 0 and apply (2) to itself to get
b1
b2
2
2
2
1
log log 2 + O 1/ log 2 .
= b1 log 2 +
g 2 ()
b1
g 2 ()
1
1
1
1
= (1 vg 2 + O(g 4 )) and 2 = 2 2v + O(g 2 ).
g
g
g
g
2v
Define by
= log 2 . Then (3) can be written as
b1
2
b2
2
=
b
log
+ log 2
2v
+
log
log
1
2
2
2
b1
g ()
= b1 log
= b1 log
2
2
2
2
+ O 1/ log 2
2 !
2
log
b2
2
2
b2
2
b1
b1
log 2
2
2
b2
(3)