Peskin QFT Solutions

Solutions to Peskin & Schroeder
Chapter 2
Zhong-Zhi Xianyu
Institute of Modern Physics and Center for High Energy Physics,

Tsinghua University, Beijing, 100084
Draft version: November 8, 2012
1 Classical electromagnetism
In this problem we do some simple calculation on classical electrodynamics. The
action without source term is given by:
S =
1
4
d
4
xF
, with F
. (1)
(a) Maxwells equations We now derive the equations of motion from the action.
Note that
F
)
=
,
F
= 0.
Then from the rst equality we get:
)
(
F
)
= 4F
.
Now substitute this into Euler-Lagrange equation, we have
0 =
(
L
(
)
)
L
A
(2)
This is sometimes called the second pair Maxwells equations. The so-called rst
pair comes directly from the denition of F
, and reads
= 0. (3)
The familiar electric and magnetic eld strengths can be written as E
i
= F
0i
and
ijk
B
k
= F
ij
, respectively. From this we deduce the Maxwells equations in terms of
E
i
and B
i
:
i
E
i
= 0,
ijk
j
B
k
0
E
i
= 0,
ijk
j
E
k
= 0,
i
B
i
= 0. (4)
E-mail: xianyuzhongzhi@gmail.com
1
Notes by Zhong-Zhi Xianyu Solution to P&S, Chapter 2 (draft version)
(b) The energy-momentum tensor The energy-momentum tensor can be dened
to be the Nother current of the space-time translational symmetry. Under space-time
translation the vector A
transforms as,
. (5)
Thus
=
L
(
L = F
+
1
4

. (6)
Obviously, this tensor is not symmetric. However, we can add an additional term
to

T
with K
being antisymmetric to its rst two indices. Its easy to see that this
term does not aect the conservation of

T
. Thus if we choose K
= F
, then:
T
=

T
= F
+
1
4

. (7)
Now this tensor is symmetric. It is called the Belinfante tensor in literature. We can
also rewrite it in terms of E
i
and B
i
:
T
00
=
1
2
(E
i
E
i
+ B
i
B
i
), T
i0
= T
0i
=
ijk
E
j
B
k
, etc. (8)
2 The complex scalar eld
The Lagrangian is given by:
L =
m
2
. (9)
(a) The conjugate momenta of and
:
=
L
, =
L
=

=
. (10)
The canonical commutation relations:
[(x), (y)] = [
(x),
(y)] = i(x y), (11)

The rest of commutators are all zero.
The Hamiltonian:
H =
d
3
x
(
L
)
=
d
3
x
(
+ m
2
)
. (12)
(b) Now we Fourier transform the eld as:
(x) =
d
3
p
(2)
3
1
2E
p
(
a
p
e
ipx
+ b
p
e
ipx
)
, (13)
thus:
(x) =
d
3
p
(2)
3
1
2E
p
(
b
p
e
ipx
+ a
p
e
ipx
)
. (14)
2
Feed all these into the Hamiltonian:
H =
d
3
x
(

+ m
2
)
=
d
3
x
d
3
p
(2)
3
2E
p
d
3
q
(2)
3
2E
q
[
E
p
E
q
(
a
p
e
ipx
b
p
e
ipx
)(
a
q
e
iqx
b
q
e
iqx
)
+p q
(
a
p
e
ipx
b
p
e
ipx
)(
a
q
e
iqx
b
q
e
iqx
)
+ m
2
(
a
p
e
ipx
+ b
p
e
ipx
)(
a
q
e
iqx
+ b
q
e
iqx
)
]
=
d
3
x
d
3
p
(2)
3
2E
p
d
3
q
(2)
3
2E
q
[
(E
p
E
q
+p q + m
2
)
(
a
p
a
q
e
i(pq)x
+ b
p
b
q
e
i(pq)x
)
(E
p
E
q
+p q m
2
)
(
b
q
a
q
e
i(p+q)x
+ a
p
b
q
e
i(p+q)x
)
]
=
d
3
p
(2)
3
2E
p
d
3
q
(2)
3
2E
q
[
(E
p
E
q
+p q + m
2
)
(
a
p
a
q
e
i(E
p
E
q
)t
+ b
p
b
q
e
i(E
p
E
q
)t
)
(2)
3
(3)
(p q)
(E
p
E
q
+p q m
2
)
(
b
q
a
q
e
i(E
p
+E
q
)t
+ a
p
b
q
e
i(E
p
+E
q
)t
)
(2)
3
(3)
(p +q)
]
=
d
3
x
E
2
p
+p
2
+ m
2
2E
p
(
a
p
a
p
+ b
p
b
p
)
=
d
3
xE
p
(
a
p
a
p
+ b
p
b
p
+ [b
p
, b
p
]
)
. (15)
Note that the last term contributes an innite constant. It is normally explained as the
vacuum energy. We simply drop it:
H =
d
3
xE
p
(
a
p
a
p
+ b
p
b
p
)
. (16)
Where we have used the mass-shell condition: E
p
=
m
2
+p
2
. Hence we at once nd
two sets of particles with the same mass m.
(c) The theory is invariant under the global transformation: e
i
,
e
i
.
The corresponding conserved charge is:
Q = i
d
3
x
(
)
. (17)
3
Rewrite this in terms of the creation and annihilation operators:
Q = i
d
3
x
(
)
= i
d
3
x
d
3
p
(2)
3
2E
p
d
3
q
(2)
3
2E
q
[
(
b
p
e
ipx
+ a
p
e
ipx
)

t
(
a
q
e
iqx
+ b
q
e
iqx
)

t
(
b
p
e
ipx
+ a
p
e
ipx
)
(
a
q
e
iqx
+ b
q
e
iqx
)
]
=
d
3
x
d
3
p
(2)
3
2E
p
d
3
q
(2)
3
2E
q
[
E
q
(
b
p
e
ipx
+ a
p
e
ipx
)(
a
q
e
iqx
b
q
e
iqx
)
E
p
(
b
p
e
ipx
a
p
e
ipx
)(
a
q
e
iqx
+ b
q
e
iqx
)
]
=
d
3
x
d
3
p
(2)
3
2E
p
d
3
q
(2)
3
2E
q
[
(E
q
E
p
)
(
b
p
a
q
e
i(p+q)x
a
p
b
q
e
i(p+q)x
)
+ (E
q
+ E
p
)
(
a
p
a
q
e
i(pq)x
b
p
b
q
e
i(pq)x
)
]
=
d
3
p
(2)
3
2E
p
d
3
q
(2)
3
2E
q
[
(E
q
E
p
)
(
b
p
a
q
e
i(E
p
+E
q
)t
a
p
b
q
e
i(E
p
+E
q
t)
)
(2)
3
(3)
(p +q)
+ (E
q
+ E
p
)
(
a
p
a
q
e
i(E
p
E
q
)t
b
p
b
q
e
i(E
p
E
q
)t
)
(2)
3
(3)
(p q)
]
=
d
3
p
(2)
3
2E
p
2E
p
(a
p
a
p
b
p
b
p
)
=
d
3
p
(2)
3
(
a
p
a
p
b
p
b
p
)
, (18)
where the last equal sign holds up to an innitely large constant term, as we did when
calculating the Hamiltonian in (b). Then the commutators follow straightforwardly:
[Q, a
] = a
, [Q, b
] = b
. (19)
We see that the particle a carries one unit of positive charge, and b carries one unit of
negative charge.
(d) Now we consider the case with two complex scalars of same mass. In this case the
Lagrangian is given by
L =
i
m
2
i
, (20)
where
i
with i = 1, 2 is a two-component complex scalar. Then it is straightforward to
see that the Lagrangian is invariant under the U(2) transformation
i
U
ij
j
with U
ij
a matrix in fundamental (self) representation of U(2) group. The U(2) group, locally
isomorphic to SU(2) U(1), is generated by 4 independent generators 1 and
1
2

a
, with
a
Pauli matrices. Then 4 independent Nother currents are associated, which are given
by
j
=
L
(
i
)
L
(
i
)
i
= (
i
)(i
i
) (
i
)(i
i
)
j
a
=
L
(
i
)
L
(
i
)
i
=
i
2
[
(
i
)
ij
j
(
i
)
ij
j
]
. (21)
4
The overall sign is chosen such that the particle carry positive charge, as will be seen in
the following. Then the corresponding Nother charges are given by
Q =i
d
3
x
(

i
)
,
Q
a
=
i
2
d
3
x
[

i
(
a
)
ij
i
(
a
)
ij

j
]
. (22)
Repeating the derivations above, we can also rewrite these charges in terms of creation
and annihilation operators, as
Q =
d
3
p
(2)
3
(
a
ip
a
ip
b
ip
b
ip
)
,
Q
a
=
1
2
d
3
p
(2)
3
(
a
ip
a
ij
a
ip
b
ip
a
ij
b
ip
)
. (23)
The generalization to n-component complex scalar is straightforward. In this case
we only need to replace the generators
a
/2 of SU(2) group to the generators t
a
in the
fundamental representation with commutation relation [t
a
, t
b
] = if
abc
t
c
.
Then we are ready to calculate the commutators among all these Nother charges and
the Hamiltonian. Firstly we show that all charges of the U(N) group commute with the
Hamiltonian. For the U(1) generator, we have
[Q, H] =
d
3
p
(2)
3
d
3
q
(2)
3
E
q
[
(
a
ip
a
ip
b
ip
b
ip
)
,
(
a
jq
a
jq
+ b
jq
b
jq
)
]
=
d
3
p
(2)
3
d
3
q
(2)
3
E
q
(
a
ip
[a
ip
, a
jq
]a
jq
+ a
jq
[a
ip
, a
jq
]a
ip
+ (a b)
)
=
d
3
p
(2)
3
d
3
q
(2)
3
E
q
(
a
ip
a
iq
a
iq
a
ip
+ (a b)
)
(2)
3
(3)
(p q)
= 0. (24)
Similar calculation gives [Q
a
, H] = 0. Then we consider the commutation among internal
U(N) charges:
[Q
a
, Q
b
] =
d
3
p
(2)
3
d
3
q
(2)
3
[
(
a
ip
t
a
ij
a
jp
b
ip
t
a
ij
b
jp
)
,
(
a
kq
t
b
k
a
q
b
kq
t
b
k
b
q
)
]
=
d
3
p
(2)
3
d
3
q
(2)
3
(
a
ip
t
a
ij
t
b
j
a
q
a
kq
t
b
k
t
a
j
a
jp
+ (a b)
)
(2)
3
(3)
(p q)
= if
abc
d
3
p
(2)
3
(
a
ip
t
c
ij
a
jp
b
ip
t
c
ij
b
jp
)
= if
abc
Q
c
, (25)
and similarly, [Q, Q] = [Q
a
, Q] = 0.
3 The spacelike correlation function
We evaluate the correlation function of a scalar eld at two points:
D(x y) = 0|(x)(y)|0, (26)
with x y being spacelike. Since any spacelike interval x y can be transformed to a
form such that x
0
y
0
= 0, thus we will simply take:
x
0
y
0
= 0, and |x y|
2
= r
2
> 0. (27)
5
Now:
D(x y) =
d
3
p
(2)
3
1
2E
p
e
ip(xy)
=
d
3
p
(2)
3
1
2
m
2
+ p
2
e
ip(xy)
=
1
(2)
3
2
0
d
1
1
d cos

0
dp
p
2
2
m
2
+ p
2
e
ipr cos
=
i
2(2)
2
r
dp
pe
ipr
m
2
+ p
2
(28)
Now we make the path deformation on p-complex plane, as is shown in Figure 2.3 in
Peskin & Schroeder. Then the integral becomes
D(x y) =
1
4
2
r

m
d
e
r
2
m
2
=
m
4
2
r
K
1
(mr). (29)
6
Chapter 3
Zhong-Zhi Xianyu

1 Lorentz group
The Lorentz group can be generated by its generators via exponential mappings.
The generators satisfy the following commutation relation:
[J
, J
] = i(g
g
nu
J
+g
). (1)
(a) Let us redene the generators as L
i
=
1
2

ijk
J
jk
(All Latin indices denote spatial
components), where L
i
generate rotations, and K
i
generate boosts. The commutators
of them can be deduced straightforwardly to be:
[L
i
, L
j
] = i
ijk
L
k
, [K
i
, K
j
] = i
ijk
L
k
. (2)
If we further dene J
i
=
1
2
(L
i
iK
i
), then the commutators become
[J
i
, J
j
] = i
ijk
J
k
, [J
i
+
, J
j
] = 0. (3)
Thus we see that the algebra of the Lorentz group is a direct sum of two identical algebra
su(2).
(b) It follows that we can classify the nite dimensional representations of the Lorentz
group by a pair (j
+
, j
), where j
= 0, 1/2, 1, 3/2, 2, are labels of irreducible repre-

sentations of SU(2).
We study two specic cases.
1. (
1
2
, 0). Following the denition, we have J
i
+
represented by
1
2

i
and J
i
represented
by 0. This implies
L
i
= (J
i
+
+J
i
) =
1
2

i
, K
i
= i(J
i
+
J
i
) =
i
2
i
. (4)
Hence a eld under this representation transforms as:
e
i
i
i
/2
i
i
/2
. (5)
1
2. (
1
2
, 0). In this case, J
i
+
0, J
i

1
2

i
. Then
L
i
= (J
i
+
+J
i
) =
1
2

i
, K
i
= i(J
i
+
J
i
) =
i
2
i
. (6)
Hence a eld under this representation transforms as:
e
i
i
i
/2+
i
i
/2
. (7)
We see that a eld under the representation (
1
2
, 0) and (0,
1
2
) are precisely the left-handed
spinor
L
and right-handed spinor
R
, respectively.
(c) Let us consider the case of (
1
2
,
1
2
). To put the eld associated with this represen-
tation into a familiar form, we note that a left-handed spinor can also be rewritten as
row, which transforms under the Lorentz transformation as:
T
L
2

T
L
2
(
1 +
i
2
i
+
1
2

i
i
)
. (8)
Then the eld under the representation (
1
2
,
1
2
) can be written as a tensor with spinor
indices:
T
L
2
V
=
(
V
0
+V
3
V
1
iV
2
V
1
+ iV
2
V
0
V
3
)
. (9)
In what follows we will prove that V
is in fact a Lorentz vector.

A quantity V
is called a Lorentz vector, if it satises the following transformation

law:
V
, (10)
where

i
2
(J
in its innitesimal form. We further note that:

(J
= i(
). (11)
and also,
ij
=
ijk
k
,
0i
=
i0
=
i
, then the combination V
= V
i
i
+ V
0
transforms according to
V
i
i
j

i
2
mn
(J
mn
)
i
j
)
V
j
i
+
(
i
2
0n
(J
0n
)
i
0
i
2
n0
(J
n0
)
0
i
)
V
0
i
=
(
i
j

i
2
mnk
k
(i)(
m
i

n
j

m
j

n
i
)
)
V
j
i
+
(
i
i
(i)(
n
i
)
)
V
0
i
=V
i
ijk
V
i
k
+V
0
i
,
V
0
V
0
+
(
i
2
0n
(J
0n
)
0i
i
2
n0
(J
n0
)
0i
)
V
i
= V
0
+
(
i
i
(i
n
i
)
)
V
i
= V
0
+
i
V
i
.
In total, we have
V
ijk
k
+
i
)
V
i
+ (1 +
i
i
)V
0
. (12)
If we can reach the same conclusion by treating the combination V
a matrix trans-
forming under the representation (
1
2
,
1
2
), then our original statement will be proved. In
fact:
V

(
1
i
2
j
+
1
2

j
j
)
V
(
1 +
i
2
j
+
1
2

j
j
)
=
(
i
+
i
2
j
[
i
,
j
] +
1
2

j
{
i
,
j
}
)
V
i
+ (1 +
i
i
)V
0
=
(
ijk
k
+
i
)
V
i
+ (1 +
i
i
)V
0
, (13)
as expected. Hence we proved that V
is a Lorentz vector.
2
2 The Gordon identity
In this problem we derive the Gordon identity,
u(p
u(p) = u(p
)
(
p
+p
2m
+
i
(p
)
2m
)
u(p). (14)
Let us start from the right hand side:
RHS. =
1
2m
u(p
)
(
(p
+p
) + i
(p
)
)
u(p)
=
1
2m
u(p
)
(
(p
+p
)
1
2
[
](p
)
)
u(p)
=
1
2m
u(p
)
(
1
2
{
}(p
+p
)
1
2
[
](p
)
)
u(p)
=
1
2m
u(p
)
(
/
p
/
p
)
u(p) = u(p
u(p) = LHS,
where we have used the commutator and anti-commutators of gamma matrices, as well
as the Dirac equation.
3 The spinor products
In this problem, together with the Problems 5.3 and 5.6, we will develop a formalism
that can be used to calculating scattering amplitudes involving massless fermions or
vector particles. This method can profoundly simplify the calculations, especially in the
calculations of QCD. Here we will derive the basic fact that the spinor products can be
treated as the square root of the inner product of lightlike Lorentz vectors. Then, in
Problem 5.3 and 5.6, this relation will be put in use in calculating the amplitudes with
external spinors and external photons, respectively.
To begin with, let k
0
and k
1
be xed four-vectors satisfying k
2
0
= 0, k
2
1
= 1 and
k
0
k
1
= 0. With these two reference momenta, we dene the following spinors:
1. Let u
L0
be left-handed spinor with momentum k
0
;
2. Let u
R0
= / k
1
u
L0
;
3. For any lightlike momentum p (p
2
= 0), dene:
u
L
(p) =
1
2p k
0
/
pu
R0
, u
R
(p) =
1
2p k
/
pu
L0
. (15)
(a) We show that / k
0
u
R0
= 0 and
/
pu
L
(p) =
/
pu
R
(p) = 0 for any lightlike p. That is,
u
R0
is a massless spinor with momentum k
0
, and u
L
(p), u
R
(p) are massless spinors with
momentum p. This is quite straightforward,
/ k
0
u
R0
= / k
0
/ k
1
u
L0
= (2g
)k
0
k
1
u
L0
= 2k
0
k
1
u
L0
/ k
1
/ k
0
u
L0
= 0, (16)
and, by denition,
/
pu
L
(p) =
1
2p k
0
/
p
/
pu
R0
=
1
2p k
0
p
2
u
R0
= 0. (17)
In the same way, we can show that
/
pu
R
(p) = 0.
3
(b) Now we choose k
0
= (E, 0, 0, E) and k
1
= (0, 1, 0, 0). Then in the Weyl
representation, we have:
/ k
0
u
L0
= 0
0 0 0 0
0 0 0 2E
2E 0 0 0
0 0 0 0
u
L0
= 0. (18)
Thus u
L0
can be chosen to be (0,
2E, 0, 0)
T
, and:
u
R0
= / k
1
u
L0
=
0 0 0 1
0 0 1 0
0 1 0 0
1 0 0 0
u
L0
=
0
0
2E
0
. (19)
Let p
= (p
0
, p
1
, p
2
, p
3
), then:
u
L
(p) =
1
2p k
0
/
pu
R0
=
1
2E(p
0
+p
3
)
0 0 p
0
+p
3
p
1
ip
2
0 0 p
1
+ip
2
p
0
p
3
p
0
p
3
p
1
+ip
2
0 0
p
1
ip
2
p
0
+p
3
0 0
u
R0
=
1
p
0
+p
3
(p
0
+p
3
)
(p
1
+ip
2
)
0
0
. (20)
In the same way, we get:
u
R
(p) =
1
p
0
+p
3
0
0
p
1
+ip
2
p
0
+p
3
. (21)
(c) We construct explicitly the spinor product s(p, q) and t(p, q).
s(p, q) = u
R
(p)u
L
(q) =
(p
1
+ip
2
)(q
0
+q
3
) (q
1
+iq
2
)(p
0
+p
3
)
(p
0
+p
3
)(q
0
+q
3
)
; (22)
t(p, q) = u
L
(p)u
R
(q) =
(q
1
iq
2
)(p
0
+p
3
) (p
1
ip
2
)(q
0
+q
3
)
(p
0
+p
3
)(q
0
+q
3
)
. (23)
It can be easily seen that s(p, q) = s(q, p) and t(p, q) = (s(q, p))
.
Now we calculate the quantity |s(p, q)|
2
:
|s(p, q)|
2
=
(
p
1
(q
0
+q
3
) q
1
(p
0
+p
3
)
)
2
+
(
p
2
(q
0
+q
3
) q
2
(p
0
+p
3
)
)
2
(p
0
+p
3
)(q
0
+q
3
)
=(p
2
1
+p
2
2
)
q
0
+q
3
p
0
+p
3
+ (q
2
1
+q
2
2
)
p
0
+p
3
q
0
+q
3
2(p
1
q
1
+p
2
q
2
)
=2(p
0
q
0
p
1
q
1
p
2
q
2
p
3
q
3
) = 2p q. (24)
Where we have used the lightlike properties p
2
= q
2
= 0. Thus we see that the spinor
product can be regarded as the square root of the 4-vector dot product for lightlike
vectors.
4
4 Majorana fermions
(a) We at rst study a two-component massive spinor lying in (
1
2
, 0) representation,
transforming according to U
L
(). It satises the following equation of motion:
i
im
2
= 0. (25)
To show this equation is indeed an admissible equation, we need to justify: 1) It is
relativistically covariant; 2) It is consistent with the mass-shell condition (namely the
Klein-Gordon equation).
To show the condition 1) is satised, we note that
is invariant under the simultane-

ous transformations of its Lorentz indices and spinor indices. That is
U()
U(
1
) =
. This implies
U
R
()
U
L
(
1
) =
,
as can be easily seen in chiral basis. Then, the combination
transforms as

U
R
()
U
L
(
1
). As a result, the rst term of the equation of motion transforms as
i
iU
R
()
U
L
(
1
)U
L
() = U
R
()
[
i
]
. (26)
To show the full equation of motion is covariant, we also need to show that the second
term i
2
transforms in the same way. To see this, we note that in the innitesimal
form,
U
L
= 1 i
i
i
/2
i
i
/2, U
R
= 1 i
i
i
/2 +
i
i
/2.
Then, under an innitesimal Lorentz transformation, transforms as:
(1 i
i
i
/2
i
i
/2),
(1 +i
i
i
/2
i
i
/2)

2
(1 +i
i
(
)
i
/2
i
(
)
i
/2)
= (1 i
i
i
/2 +
i
i
/2)
2
.
That is to say,
2
is a right-handed spinor that transforms as

2
U
R
()
2
.
Thus we see the the two terms in the equation of motion transform in the same way
under the Lorentz transformation. In other words, this equation is Lorentz covariant.
To show the condition 2) also holds, we take the complex conjugation of the equation:
i(
im
2
= 0.
Combining this and the original equation to eliminate
, we get
(
2
+m
2
) = 0, (27)
which has the same form with the Klein-Gordon equation.
(b) Now we show that the equation of motion above for the spinor can be derived
from the following action through the variation principle:
S =
d
4
x
[
i +
im
2
(
T
)
]
. (28)
Firstly, let us check that this action is real, namely S
= S. In fact,
S
d
4
x
[
T
i
im
2
(
2
)
]
5
The rst term can be rearranged as
T
i
= (
T
i
)
T
= (
) i =
i + total derivative.
Thus we see that S
= S.
Now we vary the action with respect to
, that gives
0 =
S
= i
im
2
2
2
= 0, (29)
which is exactly the Majorana equation.
(c) Let us rewrite the Dirac Lagrangian in terms of two-component spinors:
L =

(i/ m)
=
(
1
i
T
2

2
)
(
0 1
1 0
)(
m i
m
)(

1
i
2
2
)
= i
1
+ i
T
2

2
im
(
T
2

2
2
)
= i
1
+ i
2
im
(
T
2

2
2
)
, (30)
where the equality should be understood to hold up to a total derivative term.
(d) The familiar global U(1) symmetry of the Dirac Lagrangian e
i
now be-
comes
1
e
i
1
,
2
e
i
2
. The associated Nother current is
J
2
. (31)
To show its divergence
vanishes, we make use of the equations of motion:

i
1
im
2
2
= 0,
i
2
im
2
1
= 0,
i(
1
)
im
T
2

2
= 0,
i(
2
)
im
T
1

2
= 0.
Then we have
= (
1
)
1
+
1
(
2
)
2
= m
(
T
2

2
1
+
T
1

2
1
)
= 0. (32)
In a similar way, one can also show that the Nother currents associated with the global
symmetries of Majorana elds have vanishing divergence.
(e) To quantize the Majorana theory, we introduce the canonical anticommutation
relation,
{
a
(x),
b
(y)
}
=
ab
(3)
(x y),
and also expand the Majorana eld into modes. To motivate the mode expansion, we
note that the Majorana Langrangian can be obtained by replacing the spinor
2
in the
Dirac Lagrangian (30) with
1
. Then, according to our experience in Dirac theory, it
can be found that
(x) =
d
3
p
(2)
3
p
2E
p
a
[
a
a
a
(p)e
ipx
+ (i
2
)
a
a
a
(p)e
ipx
]
. (33)
6
Then with the canonical anticommutation relation above, we can nd the anticommu-
tators between annihilation and creation operators:
{a
a
(p), a
b
(q)} =
ab
(3)
(p q), {a
a
(p), a
b
(q)} = {a
a
(p), a
b
(q)} = 0. (34)
On the other hand, the Hamiltonian of the theory can be obtained by Legendre trans-
forming the Lagrangian:
H =
d
3
x
(
L

L
)
=
d
3
x
[
i
+
im
2
(
)
]
. (35)
Then we can also represent the Hamiltonian H in terms of modes:
H =
d
3
x
d
3
pd
3
q
(2)
6
2E
p
2E
q
a,b
[
(
a
a
a
(p)e
ipx
+
T
a
(i
2
)a
a
(p)e
ipx
)
(
p )
(q )
q
(
b
a
b
(q)e
iqx
(i
2
)
b
a
b
(q)e
iqx
)
+
im
2
(
a
a
a
(p)e
ipx
+
T
a
(i
2
)a
a
(p)e
ipx
)
(
p )
2
(
q )
b
a
b
(q)e
iqx
+ (i
2
)
b
a
b
(q)e
iqx
)
im
2
(
T
a
a
a
(p)e
ipx
+
a
(i
2
)a
a
(p)e
ipx
)
(
p )
T
q
(
b
a
b
(q)e
iqx
+ (i
2
)
b
a
b
(q)e
iqx
)
]
=
d
3
x
d
3
pd
3
q
(2)
6
2E
p
2E
q
a,b
{
a
a
(p)a
b
(q)
a
[
(
p )
(q )
q
+
im
2
(
p )
2
(
q )
(i
2
)
im
2
(i
2
)(
p )
T
q
]
b
e
i(pq)x
+a
a
(p)a
b
(q)
a
[
(
p )
(q )
q (i
2
) +
im
2
(
p )
2
(
q )
im
2
(i
2
)(
p )
T
q (i
2
)
]
b
e
i(p+q)x
+a
a
(p)a
b
(q)
T
a
[
(i
2
)(
p )
(q )
q +
im
2
(i
2
)(
p )
2
(
q )
(i
2
)
im
2
(
p )
T
q
]
b
e
i(p+q)x
+a
a
(p)a
b
(q)
T
a
[
(i
2
)(
p )
(q )
q (i
2
) +
im
2
(i
2
)(
p )
2
(
q )
im
2
(
p )
T
q (i
2
)
]
b
e
i(pq)x
}
=
d
3
p
(2)
3
2E
p
a,b
{
a
a
(p)a
b
(p)
a
[
(
p )
(p )
p
+
im
2
(
p )
2
(
p )
(i
2
)
im
2
(i
2
)(
p )
T
p
]
b
+a
a
(p)a
b
(p)
a
[
(
p )
(p )
p (i
2
) +
im
2
(
p )
2
(
p )
im
2
(i
2
)(
p )
T
p (i
2
)
]
b
+a
a
(p)a
b
(p)
T
a
[
(i
2
)(
p )
(p )
p +
im
2
(i
2
)(
p )
2
(
p )
(i
2
)
7
im
2
(
p )
T
p
]
b
+a
a
(p)a
b
(p)
T
a
[
(i
2
)(
p )
(p )
p (i
2
) +
im
2
(i
2
)(
p )
2
(
p )
im
2
(
p )
T
p (i
2
)
]
b
}
=
d
3
p
(2)
3
2E
p
a,b
1
2
(
E
2
p
+|p|
2
+m
2
)
[
a
a
(p)a
b
(p)
b
a
a
(p)a
b
(p)
T
a

b
]
=
d
3
p
(2)
3
E
p
2
a
[
a
a
(p)a
a
(p) a
a
(p)a
a
(p)
]
=
d
3
p
(2)
3
E
p
a
a
a
(p)a
a
(p). (36)
In the calculation above, each step goes as follows in turn: (1) Substituting the mode
expansion for into the Hamiltonian. (2) Collecting the terms into four groups, charac-
terized by a
a, a
, aa and aa
. (3) Integrating over d

3
x to produce a delta function,
with which one can further nish the integration over d
3
q. (4) Using the following
relations to simplify the spinor matrices:
(p )
2
= (p )
2
= E
2
p
+|p|
2
, (p )(p ) = p
2
= m
2
, p =
1
2
(p p ).
In this step, the a
and aa terms vanish, while the aa
and a
a terms remain. (5)

Using the normalization
b
=
ab
to eliminate spinors. (6) Using the anticommutator
{a
a
(p), a
a
(p)} =
(3)
(0) to further simplify the expression. In this step we have throw
away a constant term
1
2
E
p
(3)
(0) in the integrand. The minus sign of this term
indicates that the vacuum energy contributed by Majorana eld is negative. With these
steps done, we nd the desired result, as shown above.
5 Supersymmetry
(a) In this problem we briey study the Wess-Zumino model. Maybe it is the simplest
supersymmetric model in 4 dimensional spacetime. Firstly let us consider the massless
case, in which the Lagrangian is given by
L =
+F
F, (37)
where is a complex scalar eld, is a Weyl fermion, and F is a complex auxiliary
scalar eld. By auxiliary we mean a eld with no kinetic term in the Lagrangian and
thus it does not propagate, or equivalently, it has no particle excitation. However, in
the following, we will see that it is crucial to maintain the o-shell supersymmtry of the
theory.
The supersymmetry transformation in its innitesimal form is given by:
= i
T
2
, (38a)
= F +
)
2
, (38b)
F = i
, (38c)
where is a 2-component Grassmann variable. Now let us show that the Lagrangian
is invariant (up to a total divergence) under this supersymmetric transformation. This
8
can be checked term by term, as follows:
(
) = i
(
+ (
)
(
i
T
)
,
(
) =
(
F
+
T
)
i
F +
)
= iF
+ i
)
]
i
T
)
+ i
F + i
= iF
+ i
)
]
i
T
2
(
2
)
+ i
F + i
2
,
(F
F) = i(
F iF
,
where we have used
=
2
. Now summing the three terms above, we get:
L = i
F +
2
(
]
, (39)
which is indeed a total derivative.
(b) Now let us add the mass term in to the original massless Lagrangian:
L =
(
mF +
1
2
im
T
)
+ c.c. (40)
Let us show that this mass term is also invariant under the supersymmetry transforma-
tion, up to a total derivative:
(L) =im
T
2
F im
+
1
2
im[
T
F +
(
2
)
T
(
)
T
]
2
+
1
2
im
T
2
[F +
)
2
] + c.c.
=
1
2
imF(
T
2

T
2
) im
1
2
im(
+
1
2
im(
)
T
(
)
T
+ c.c.
=
1
2
imF(
T
2

T
2
) im
)
+
1
2
im(
)[
+
T
(
)
T
] + c.c
=im
) + c.c (41)
where we have used the following relations:
(
2
)T =
2
,
2
(
)
T
2
=
,
T
2
=
T
2
,
=
T
(
)
T
.
Now let us write down the Lagrangian with the mass term:
L =
+F
F +
(
mF +
1
2
im
T
2
+ c.c.
)
. (42)
Varying the Lagrangian with respect to F
, we get the corresponding equation of motion:

F = m
. (43)
Substitute this algebraic equation back into the Lagrangian to eliminate the eld F, we
get
L =
m
2
+
1
2
(
im
T
2
+ c.c.
)
. (44)
Thus we see that the scalar eld and the spinor eld have the same mass.
9
(c) We can also include interactions into this model. Generally, we can write a La-
grangian with nontrivial interactions containing elds
i
,
i
and F
i
(i = 1, , n), as
L =
i
+
i
i
i
+F
i
F
i
+
[
F
i
W[]
i
+
i
2
2
W[]
T
i

2
j
+ c.c.
]
, (45)
where W[] is an arbitrary function of
i
.
To see this Lagrangian is supersymmetry invariant, we only need to check the inter-
actions terms in the square bracket:
[
F
i
W[]
i
+
i
2
2
W[]
T
i

2
j
+ c.c.
]
=i
i
)
W
i
+F
i
2
W
j
(i
T
j
) +
i
2
3
W
k
(i
T
k
)
T
i

2
j
+
i
2
2
W
j
[
(
T
F
i
+
(
2
)
T
(
)
T
i
)
j
+
T
i

2
(
F
j
+
)
]
+ c.c..
The term proportional to
3
W/
3
vanishes. To see this, we note that the partial deriva-
tives with respect to
i
are commutable, hence
3
W/
i
k
is totally symmetric
on i, j, k. However, we also have the following identity:
(
T
k
)(
T
i

2
j
) + (
T
i
)(
T
j

2
k
) + (
T
j
)(
T
k
i
) = 0, (46)
which can be directly checked by brute force. Then it can be easily seen that the
3
W/
3
term vanishes indeed. On the other hand, the terms containing F also sum
to zero, which is also straightforward to justify. Hence the terms left now are
i
i
)
W
i
+ i

2
W
(
2
)
T
(
)
T
(
i
)
2
j
=i
i
W
i
)
+ i
2
W
j
i

2
W
i
)
j
=i
i
W
i
)
, (47)
which is a total derivative. Thus we conclude that the Lagrangian (45) is supersymmet-
rically invariant up to a total derivative.
Let us end up with a explicit example, in which we choose n = 1 and W[] = g
3
/3.
Then the Lagrangian (45) becomes
L =
+F
F +
(
gF
2
+i
T
2
+ c.c.
)
. (48)
We can eliminate F by solving it from its eld equation,
F +g(
)
2
= 0. (49)
Substituting this back into the Lagrangian, we get
L =
g
2
(
)
2
+ig(
T
). (50)
This is a Lagrangian of massless complex scalar and a Weyl spinor, with
4
and Yukawa
interactions. The eld equations can be easily got from by variations.
10
6 Fierz transformations
In this problem, we derive the generalized Fierz transformation, with which one can
express ( u
1
A
u
2
)( u
3
B
u
4
) as a linear combination of ( u
1
C
u
4
)( u
3
D
u
2
), where
A
is
any normalized Dirac matrices from the following set:
{
1,
=
i
2
[
],
5
,
5
= i
0
3
}
.
(a) The Dirac matrices
A
are normalized according to
tr (
A
B
) = 4
AB
. (51)
For instance, the unit element 1 is already normalized, since tr (1 1) = 4. For Dirac
matrices containing one
, we calculate the trace in Weyl representation without loss

of generality. Then the representation of
=
(
0
0
)
gives tr (
) = 2 tr (
) (no sum on ). For = 0, we have tr (

0
0
) = 2 tr (1
22
) =
4, and for = i = 1, 2, 3, we have tr (
i
i
) = 2 tr (
i
i
) = 2 tr (1
22
) = 4 (no sum
on i). Thus the normalized gamma matrices are
0
and i
i
.
In the same way, we can work out the rest of the normalized Dirac matrices, as:
tr (
0i
0i
) = 2 tr (
i
i
) = 4, (no sum on i)
tr (
ij
ij
) = 2 tr (
k
k
) = 4, (no sum on i, j, k)
tr (
5
5
) = 4,
tr (
5
0
) = 4, tr (
5
i
) = 4.
Thus the 16 normalized elements are:
{
1,
0
, i
i
, i
0i
,
ij
,
5
, i
5
0
,
5
i
}
. (52)
(b) Now we derive the desired Fierz identity, which can be written as:
( u
1
A
u
2
)( u
3
B
u
4
) =
C,D
C
AB
CD
( u
1
C
u
4
)( u
3
D
u
2
). (53)
Left-multiplying the equality by ( u
2
F
u
3
)( u
4
E
u
1
), we get:
( u
2
F
u
3
)( u
4
E
u
1
)( u
1
A
u
2
)( u
3
B
u
4
) =
CD
C
AB
CD
tr (
E
C
) tr (
F
D
). (54)
The left hand side:
( u
2
F
u
3
)( u
4
E
u
1
)( u
1
A
u
2
)( u
3
B
u
4
) = u
4
B
u
4
= tr (
E
B
);
the right hand side:
C,D
C
AB
CD
tr (
E
C
) tr (
F
D
) =
C,D
C
AB
CD
4
EC
4
FD
= 16C
AB
EF
,
thus we conclude:
C
AB
CD
=
1
16
tr (
C
B
). (55)
11
(c) Now we derive two Fierz identities as particular cases of the results above. The
rst one is:
( u
1
u
2
)( u
3
u
4
) =
C,D
tr (
C
D
)
16
( u
1
C
u
4
)( u
3
D
u
2
). (56)
The traces on the right hand side do not vanish only when C = D, thus we get:
( u
1
u
2
)( u
3
u
4
) =
C
1
4
( u
1
C
u
4
)( u
3
C
u
2
)
=
1
4
[
( u
1
u
4
)( u
3
u
2
) + ( u
1
u
4
)( u
3
u
2
) +
1
2
( u
1
u
4
)( u
3
u
2
)
( u
1
u
4
)( u
3
u
2
) + ( u
1
5
u
4
)( u
3
5
u
2
)
]
. (57)
The second example is:
( u
1
u
2
)( u
3
u
4
) =
C,D
tr (
C
)
16
( u
1
C
u
4
)( u
3
D
u
2
). (58)
Again, the traces vanish if
C
= C
D
with C a commuting number, which

implies that
C
=
D
. That is,
( u
1
u
2
)( u
3
u
4
) =
C
tr (
C
)
16
( u
1
C
u
4
)( u
3
C
u
2
)
=
1
4
[
4( u
1
u
4
)( u
3
u
2
) 2( u
1
u
4
)( u
3
u
2
)
2( u
1
u
4
)( u
3
u
2
) 4( u
1
5
u
4
)( u
3
5
u
2
)
]
. (59)
We note that the normalization of Dirac matrices has been properly taken into account
by raising or lowering of Lorentz indices.
7 The discrete symmetries P, C and T
(a) In this problem, we will work out the C, P and T transformations of the bilinear
, with
=
i
2
[
]. Firstly,
P

(t, x)
(t, x)P =
i
2
(t, x)
0
[
]
0
(t, x).
With the relations
0
[
0
,
i
]
0
= [
0
,
i
] and
0
[
i
,
j
]
0
= [
i
,
j
], we get:
P

(t, x)
(t, x)P =
{

(t, x)
0i
(t, x);
(t, x)
ij
(t, x).
(60)
Secondly,
T

(t, x)
(t, x)T =
i
2
(t, x)(
1
3
)[
(
1
3
)(t, x).
Note that gamma matrices keep invariant under transposition, except
2
, which changes
the sign. Thus we have:
T

(t, x)
(t, x)T =
{

(t, x)
0i
(t, x);

(t, x)
ij
(t, x).
(61)
Thirdly,
C

(t, x)
(t, x)C =
i
2
(i
0
2
)
T
(i
2
)
T
=

2
(
)
T
2
.
Note that
0
and
2
are symmetric while
1
and
3
are antisymmetric, we have
C

(t, x)
(t, x)C =
(t, x)
(t, x). (62)

12
(b) Now we work out the C, P and T transformation properties of a scalar eld .
Our starting point is
Pa
p
P = a
p
, Ta
p
T = a
p
, Ca
p
C = b
p
.
Then, for a complex scalar eld
(x) =
d
3
k
(2)
3
1
2k
0
[
a
k
e
ikx
+b
k
e
ikx
]
, (63)
we have
P(t, x)P =
d
3
k
(2)
3
1
2k
0
[
a
k
e
i(k
0
tkx)
+b
k
e
i(k
0
tkx)
]
= (t, x). (64a)
T(t, x)T =
d
3
k
(2)
3
1
2k
0
[
a
k
e
i(k
0
tkx)
+b
k
e
i(k
0
tkx)
]
= (t, x). (64b)
C(t, x)C =
d
3
k
(2)
3
1
2k
0
[
b
k
e
i(k
0
tkx)
+a
k
e
i(k
0
tkx)
]
=
(t, x). (64c)

As a consequence, we can deduce the C, P, and T transformation properties of the
current J
= i
(
)
)
, as follows:
PJ
(t, x)P = (1)

s()
i
[
(t, x)
(t, x)
(
(t, x)
)
(t, x)
]
= (1)
s()
J
(t, x), (65a)

where s() is the label for space-time indices that equals to 0 when = 0 and 1 when
= 1, 2, 3. In the similar way, we have
TJ
(t, x)T = (1)

s()
J
(t, x); (65b)

CJ
(t, x)C = J
(t, x). (65c)

One should be careful when playing with T it is antihermitian rather than hermitian,
and anticommutes, rather than commutes, with
1.
(c) Any Lorentz-scalar hermitian local operator O(x) constructed from (x) and (x)
can be decomposed into groups, each of which is a Lorentz-tensor hermitian operator and
contains either (x) or (x) only. Thus to prove that O(x) is an operator of CPT = +1,
it is enough to show that all Lorentz-tensor hermitian operators constructed from either
(x) or (x) have correct CPT value. For operators constructed from (x), this has been
done as listed in Table on Page 71 of Peskin & Schroeder; and for operators constructed
from (x), we note that all such operators can be decomposed further into a product
(including Lorentz inner product) of operators of the form
(
)(
n
) + c.c
together with the metric tensor
. But it is easy to show that any operator of this

form has the correct CPT value, namely, has the same CPT value as a Lorentz tensor
of rank (m+n). Therefore we conclude that any Lorentz-scalar hermitian local operator
constructed from and has CPT = +1.
13
8 Bound states
(a) A positronium bound state with orbital angular momentum L and total spin S
can be build by linear superposition of an electron state and a positron state, with the
spatial wave function
L
(k) as the amplitude. Symbolically we have
|L, S
L
(k)a
(k, s)b
(k, s
)|0.
Then, apply the space-inversion operator P, we get
P|L, S =
L
(k)
a
b
a
(k, s)b
(k, s
)|0 = (1)
L
L
(k)a
(k, s)b
(k, s
)|0.
(66)
Note that
b
=
a
, we conclude that P|L, S = ()
L+1
|L, S. Similarly,
C|L, S =
L
(k)b
(k, s)a
(k, s
)|0 = (1)
L+S
L
(k)b
(k, s
)a
(k, s)|0.
(67)
That is, C|L, S = (1)
L+S
|L, S. Then its easy to nd the P and C eigenvalues of
various states, listed as follows:
S
L
1
S
3
S
1
P
3
P
1
D
3
D
P + +
C + + +
(b) We know that a photon has parity eigenvalue 1 and C-eigenvalue 1. Thus we
see that the decay into 2 photons are allowed for
1
S state but forbidden for
3
S state
due to C-violation. That is,
3
S has to decay into at least 3 photons.
14
Chapter 4
Zhong-Zhi Xianyu

1 Scalar eld with a classical source
In this problem we consider the theory with the following Hamiltonian:
H = H
0
d
3
j(t, x)(x), (1)
where H
0
is the Hamiltonian for free Klein-Gordon eld , and j is a classical source.
(a) Let us calculate the probability that the source creates no particles. Obviously,
the corresponding amplitude is given by the inner product of the needed in-state and
the out-state. In our case, both in- and out-state are vacuum state. Thus:
P(0) =
out
0|0
in
2
= lim
t(1i)
0|e
i2Ht
|0
2
=
0|T exp
{
i
d
4
xH
int
}
|0
2
=
0|T exp
{
i
d
4
xj(x)
I
(x)
}
|0
2
. (2)
(b) Now we expand this probability P(0) to j
2
. The amplitude reads:
0|T exp
{
i
d
4
xj(x)
I
(x)
}
|0 =1
1
2
d
4
xd
4
y j(x)0|T
I
(x)
I
(y)|0j(y) +O(j
4
)
=1
1
2
d
4
xd
4
y j(x)j(y)
d
3
p
(2)
3
1
2E
p
+O(j
4
)
=1
1
2
d
3
p
(2)
3
1
2E
p
|
j(p)|
2
+O(j
4
). (3)
Thus:
P(0) = |1
1
2
+O(j
4
)|
2
= 1 +O(j
4
), (4)
where:

d
3
p
(2)
3
1
2E
p
|
j(p)|
2
. (5)
1
(c) We can calculate the probability P(0) exactly, to perform this, we calculate the
j
2n
term of the expansion:
i
2n
(2n)!
d
4
x
1
d
4
x
2n
j(x
1
) j(x
2n
)0|T(x
1
) (x
2n
)|0
=
i
2n
(2n 1)(2n 3) 3 1
(2n)!
d
4
x
1
d
4
x
2n
j(x
1
) j(x
2n
)
d
3
p
1
d
3
p
n
(2)
3n
1
2
n
E
p
1
E
p
n
e
ip
1
(x
1
x
2
)
e
ip
n
(x
2n1
x
2n
)
=
(1)
n
2
n
n!
(
d
3
p
(2)
3
|
j(p)|
2
2E
p
)
n
=
(/2)
2
n!
. (6)
Thus:
P(0) =
(

n=0
(/2)
n
n!
)
2
= e
. (7)
(d) Now we calculate the probability that the source creates one particle with mo-
mentum k. This time, we have:
P(k) =
k|T exp
{
i
d
4
xj(x)
I
(x)
}
|0
2
(8)
Expanding the amplitude to the rst order in j, we get:
P(k) =
k|0 +i
d
4
xj(x)
d
3
p
(2)
3
e
ipx
2E
p
k|a
p
|0 +O(j
2
)
2
=
d
3
p
(2)
3
j(p)
2E
p
2E
p
(2)
3
(p k)
2
= |
j(k)|
2
+O(j
3
). (9)
If we go on to work out all the terms, we will get:
P(k) =
n
i(2n + 1)(2n + 1)(2n 1) 3 1
(2n + 1)!
j
n+1
(k)
2
= |
j(k)|
2
e
|
j(k)|
. (10)
(e) To calculate the probability that the source creates n particles, we write down the
relevant amplitude:
d
3
k
1
d
3
k
n
(2)
3n
2
n
E
k
1
E
k
n
k
1
k
n
|T exp
{
i
d
4
xj(x)
I
(x)
}
|0. (11)
Expanding this amplitude in terms of j, we nd that the rst nonvanishing term is
the one of nth order in j. Repeat the similar calculations above, we can nd that the
amplitude is:
i
n
n!
d
3
k
1
d
3
k
n
(2)
3n
2
n
E
k
1
E
k
n
d
4
x
1
d
4
x
n
j
1
j
n
k
1
k
n
|
1

n
|0 +O(j
n+2
)
=
i
n
n!
d
3
k
1
d
3
k
n
j
n
(k)
(2)
3n
2
n
E
k
1
E
k
n
n=0
(1)
n
2
n
n!
(
d
3
p
(2)
3
|
j(p)|
2
2E
p
)
n
(12)
Then we see the probability is given by:
P(n) =

n
n!
e
, (13)
which is a Poisson distribution.
2
(f ) Its quite easy to check that the Poisson distribution P(n) satises the following
identities:
n
P(n) = 1. (14)
N =
n
nP(n) = . (15)
The rst one is almost trivial, and the second one can be obtained by acting
d
d
on
both sides of the rst identity. If we apply
d
d
again on the second identity, we get:
(N N)
2
= N
2
N
2
= . (16)
2 Decay of a scalar particle
This problem is based on the following Lagrangian:
L =
1
2
(
)
2
1
2
M
2
2
+
1
2
(
)
2
1
2
m
2
2
. (17)
When M > 2m, a particle can decay into two particles. We want to calculate the
lifetime of the particle to lowest order in .
According to the eqn (4.86) on P.107, the decay rate is given by:
d =
1
2M
d
3
p
1
d
3
p
2
(2)
6
1
4E
p
1
E
p
2
M((0) (p
1
)(p
2
))
2
(2)
4
(4)
(p
p
1
p
2
).
(18)
To lowest order in , the amplitude M is given by:
iM = i. (19)
The delta function in our case reads:
(4)
(p
p
1
p
2
) = (M E
p
1
E
p
2
)
(3)
(p
1
+p
2
), (20)
thus:
=

2
2M
d
3
p
1
d
3
p
2
(2)
6
1
4E
p
1
E
p
2
(2)
4
(M E
p
1
E
p
2
)
(3)
(p
1
+p
2
) (21)
There are two additional mass-shell constrains:
m
2
+p
2
i
= E
2
p
i
i = 1, 2. (22)
Hence:
=

2
2M
d
3
p
1
(2)
3
1
4E
2
p
1
(2)(M 2E
p
1
) =

2
4M
(
1
4m
2
M
2
)
3/2
. (23)
Then the lifetime of is:
=
1
=
4M
2
(
1
4m
2
M
2
)
3/2
. (24)
3 Linear sigma model
In this problem, we study the linear sigma model, provided by the following La-
grangian:
L =
1
2

1
2
m
2
1
4
(
i
i
)
2
. (25)
Where is a N-component scalar.
3
(a) We rstly compute the following dierential cross sections to the leading order in
:
(
1
2

1
2
), (
1
1

2
2
), (
1
1

1
1
).
Since the masses of all incoming and outgoing particles are identical, the cross section
is simply given by
(
d
d
)
COM
=
|M|
2
64
2
s
, (26)
where s is the square of COM energy, and M is the scattering amplitude. With the
help of Feynman rules, its quite easy to get
M(
1
2

1
2
) = M(
1
1

2
2
) = 2i;
M(
1
1

1
1
) = 6i. (27)
Immediately, we get
(
1
2

1
2
) = M(
1
1

2
2
) =

2
16
2
s
;
(
1
1

1
1
) =
9
2
16
2
s
. (28)
(b) Now we study the symmetry broken case, that is, m
2
=
2
< 0. Then, the scalar
multiplet can be parameterized as
= (
1
, ,
N1
, +v)
T
, (29)
where v is the VEV of ||, and equals to
2
/ at tree level.
Substitute this into the Lagrangian, we get
L =
1
2
(
k
)
2
+
1
2
(
)
2
1
2
(2
2
)
2

4

4

2

2
(
k
k
)

4
(
k
k
)
2
. (30)
Then its easy to read the Feynman rules from this expression:
k
=
i
k
2
2
2
; (31a)
k
=
i
ij
k
2
; (31b)
= 6iv; (31c)
i j
= 2iv
ij
; (31d)
= 6i; (31e)
4
i j
= 2i
ij
; (31f)
i j
k
= 2i(
ij
k
+
ik
j
+
i
jk
). (31g)
(c) With the Feynman rules derived in (b), we can compute the amplitude
M
[
i
(p
1
)
j
(p
2
)
k
(p
3
)
(p
4
)
]
,
as:
M = (2iv)
2
[
i
s 2
2
ij
k
+
i
t 2
2
ik
j
+
i
u 2
2
jk
]
2i(
ij
k
+
ik
j
+
i
jk
), (32)
where s, t, u are Mandelstam variables (See Section 5.4). Then, at the threshold p
i
= 0,
we have s = t = u = 0, and M vanishes.
On the other hand, if N = 2, then there is only one component in , thus the
amplitude reduces to
M =2i
[
2
2
s 2
2
+
2
2
t 2
2
+
2
2
u 2
2
+ 3
]
= 2i
[
s +t +u
2
2
+O(p
4
)
]
. (33)
In the second line we perform the Taylor expansion on s, t and u, which are of order
O(p
2
). Note that s + t + u = 4m
2
= 0, thus we see that O(p

2
) terms are also canceled
out.
(d) We minimize the potential with a small symmetry breaking term:
V =
2
i
+

4
(
i
i
)
2
a
N
, (34)
which yields the following equation that determines the VEV:
(
2
+
i
i
)
i
= a
iN
. (35)
Thus, up to linear order in a, the VEV
i
= (0, , 0, v) is
v =
+
a
2
2
. (36)
Now we repeat the derivation in (b) with this new VEV, and write the Lagrangian in
terms of new eld variable
i
and , as
L =
1
2
(
k
)
2
+
1
2
(
)
2
1
2
1
2
(2
2
)
2
v
3
v
k
1
4

4

2

2
(
k
k
)

4
(
k
k
)
2
. (37)
5
The
i
j

k
amplitude is still given by

M = (2iv)
2
[
i
s 2
2
ij
k
+
i
t 2
2
ik
j
+
i
u 2
2
jk
]
2i(
ij
k
+
ik
j
+
i
jk
). (38)
However this amplitude does not vanishes at the threshold. Since the vertices =
exactly even at tree level, and also s, t and u are not exactly zero in this case due to
nonzero mass of
i
. Both deviations are proportional to a, thus we conclude that the
amplitude M is also proportional to a.
4 Rutherford scattering
The Rutherford scattering is the scattering of an election by the coulomb eld of a
nucleus. In this problem, we calculate the cross section by treating the electromagnetic
eld as a given classical potential A
(x). Then the interaction Hamiltonian is:

H
I
=
d
3
xe
. (39)
(a) We rst calculate the T-matrix to lowest order. In fact:
out
p
|p
in
=p
|T exp(i
d
4
xH
I
)|p = p
|p ie
d
4
xA
(x)p
|p +O(e
2
)
=p
|p ie
d
4
xA
(x) u(p
u(p)e
i(p
p)x
+O(e
2
)
=(2)
4
(4)
(p p
) ie u(p
u(p)

A
(p
p) +O(e
2
) (40)
But on the other hand,
out
p
|p
in
= p
|S|p = p
|p +p
|iT|p. (41)
Thus to the rst order of e, we get:
p
|iT|p = ie u(p
u(p)

A
(p
p). (42)
(b) Now we calculate the cross section d in terms of the matrix elements iM.
The incident wave packet | is dened to be:
| =
d
3
k
(2)
3
e
ibk
2E
k
(k)|k, (43)
where b is the impact parameter.
The probability that a scattered electron will be found within an innitesimal element
d
3
p centered at p is:
P =
d
3
p
(2)
3
1
2E
p
out
p|
in
2
=
d
3
p
(2)
3
1
2E
p
d
3
kd
3
k
(2)
6
2E
k
2E
k
(k)
(k
)
(
out
p|k
in
)(
out
p|k
in
)
e
ib(kk
)
=
d
3
p
(2)
3
1
2E
p
d
3
kd
3
k
(2)
6
2E
k
2E
k
(k)
(k
)
(
p|iT|k
)(
p|iT|k
e
ib(kk
)
. (44)
6
In the last equality we have excluded the trivial scattering part from the S-matrix. Note
that:
p
|iT|p = iM(2)(E
p
E
p
), (45)
we have:
P =
d
3
p
(2)
3
1
2E
p
d
3
kd
3
k
(2)
6
2E
k
2E
k
(k)
(k
)|iM|
2
(2)
2
(E
p
E
k
)(E
p
E
k
)e
ib(kk
)
(46)
The cross section d is given by:
d =
d
2
b P(b), (47)
thus the integration over b gives a delta function:
d
2
b e
ib(kk
)
= (2)
2
(2)
(k
). (48)
The other two delta functions in the integrand can be modied as follows:
(E
k
E
k
) =
E
k
k
(k
) =
1
v
(k
), (49)
where we have used |v| = v = v
. Taking all these delta functions into account, we get:

d =
d
3
p
(2)
3
1
2E
p
d
3
k
(2)
3
2E
k
1
v
(k)
(k)|iM|
2
(2)(E
p
E
k
). (50)
Since the momentum of the wave packet should be localized around its central value,
we can pull out the quantities involving energy E
k
outside the integral:
d =
d
3
p
(2)
3
1
2E
p
1
2E
k
1
v
(2)|M|
2
(E
p
E
k
)
d
3
k
(2)
3
(k)
(k). (51)
Recall the normalization of the wave packet:
d
3
k
(2)
3
(k)
(k) = 1, (52)
then:
d =
d
3
p
(2)
3
1
2E
p
1
2E
k
1
v
|M(k p)|
2
(2)(E
p
E
k
). (53)
We can further integrate over |p| to get the dierential cross section d/d:
d
d
=
dp p
2
(2)
3
1
2E
p
1
2E
k
1
v
|M(k p)|
2
(2)(E
p
E
k
)
=
dp p
2
(2)
3
1
2E
p
1
2E
k
1
v
|M(k p)|
2
(2)
E
k
k
(p k)
=
1
(4)
2
|M(k, )|
2
. (54)
In the last line we work out the integral by virtue of delta function, which constrains the
outgoing momentum |p| = |k| but leave the angle between p and k arbitrary. Thus
the amplitude M(k, ) is a function of momentum |k| and angle .
7
(c) We work directly for the relativistic case. Firstly the Coulomb potential A
0
=
Ze/4r in momentum space is
A
0
(q) =
Ze
|q|
2
. (55)
This can be easily worked out by Fourier transformation, with a regulator e
mr
in-
serted:
A
0
(q, m)
d
3
xe
ipx
e
mr
Ze
4r
=
Ze
|q|
2
+m
2
. (56)
This is simply Yukawa potential, and Coulomb potential is a limiting case when m 0.
The amplitude is given by
iM(k, ) = ie u(p)
(q)u(p) with q = p k. (57)

Then we have the module square of amplitude with initial spin averaged and nal spin
summed (See 5.1 of Peskin & Schroeder for details), as:
1
2
spin
|iM(k, )|
2
=
1
2
e
2

A
(q)

A
(q)
spin
u(p)
u(k) u(k)
u(p)
=
1
2
e
2

A
(q)

A
(q) tr
[
(
/
p +m)
(/ k +m)
]
=2e
2
[
2(p

A)(k

A) +
(
m
2
(k p)
)

A
2
]
. (58)
Note that
A
0
(q) =
Ze
|p k|
2
=
Ze
4|k|
2
sin
2
(/2)
, (59)
thus
1
2
spin
|iM(k, )|
2
=
Z
2
e
4
(
1 v
2
sin
2
2
)
4|k|
4
v
2
sin
4
(/2)
, (60)
and
d
d
=
Z
2
2
(
1 v
2
sin
2
2
)
4|k|
2
v
2
sin
4
(/2)
(61)
In non-relativistic case, this formula reduces to
d
d
=
Z
2
2
4m
2
v
4
sin
4
(/2)
(62)
8
Chapter 5
Zhong-Zhi Xianyu

1 Coulomb scattering
In this problem we continue our study of the Coulomb scattering in Problem 4.4.
Here we consider the relativistic case. Lets rst recall some main points considered
before. The Coulomb potential A
0
= Ze/4r in momentum space is
A
0
(q) =
Ze
|q|
2
. (1)
Then the scattering amplitude is given by
iM(k, ) = ie u(p)
(q)u(p) with q = p k. (2)

Then we can derive the squared amplitude with initial spin averaged and nal spin
summed, as:
1
2
spin
|iM(k, )|
2
=
1
2
e
2

A
(q)

A
(q)
spin
u(p)
u(k) u(k)
u(p)
=
1
2
e
2

A
(q)

A
(q) tr
[
(
/
p + m)
(/ k + m)
]
= 2e
2
[
2(p

A)(k

A) +
(
m
2
(k p)
)
A
2
]
. (3)
Note that
A
0
(q) =
Ze
|p k|
2
=
Ze
4|k|
2
sin
2
(/2)
, (4)
thus
1
2
spin
|iM(k, )|
2
=
Z
2
e
4
(
1 v
2
sin
2
2
)
4|k|
4
v
2
sin
4
(/2)
, (5)
Now, from the result of Problem 4.4(b), we know that
d
d
=
1
(4)
2
(
1
2
spin
|M(k, )|
2
)
=
Z
2
2
(
1 v
2
sin
2
2
)
4|k|
2
v
2
sin
4
(/2)
(6)
1
k
1
k
2

p
1
p
2

e

Z
e

Z
Figure 1: The scattering of an electron by a charged heavy particle
Z
. All initial momenta
go inward and all nal momenta go outward.
This is the formula for relativistic electron scatted by Coulomb potential, and is called
Mott formula.
Now we give an alternative derivation of the Mott formula, by considering the cross
section of e
Z
e
Z
. When the mass of goes to innity and the charge of is
taken to be Ze, this cross section will reduces to Mott formula. The relevant amplitude
is shown in Figure 1, which reads
iM= Z(ie)
2
u(p
1
)
u(k
1
)
i
t
U(p
2
)
U(k
2
), (7)
where u is the spinor for electron and U is the spinor for muon, t = (k
1
p
1
)
2
is one
of three Mandelstam variables. Then the squared amplitude with initial spin averaged
and nal spin summed is
1
4
spin
|iM|
2
=
Z
2
e
4
t
2
tr
[
(/ k
1
+ m)
(
/
p
1
+ m)
]
tr
[
(/ k
2
+ M)
(
/
p
2
+ M)
]
=
Z
2
e
4
t
2
[
16m
2
M
2
8M
2
(k
1
p
1
) + 8(k
1
p
2
)(k
2
p
1
)
8m
2
(k
2
p
2
) + 8(k
1
k
2
)(p
1
p
2
)
]
. (8)
Note that the cross section is given by
(
d
d
)
CM
=
1
2E
e
2E
|v
k
1
v
k
2
|
|p
1
|
(2)
2
4E
CM
(
1
4
|M|
2
)
. (9)
When the mass of goes to innity, we have E
E
CM
M, v
k
2
0, and |p
1
| |k
1
|.
Then the expression above can be simplied to
(
d
d
)
CM
=
1
16(2)
2
M
2
(
1
4
|M|
2
)
. (10)
When M , only terms proportional to M
2
are relevant in |M|
2
. To evaluate this
squared amplitude further, we assign each momentum a specic value in CM frame:
k
1
= (E, 0, 0, k), p
1
(E, sin , 0, k cos ),
k
2
(M, 0, 0, k), p
2
(M, k sin , 0, k cos ), (11)
then t = (k
1
p
1
)
2
= 4k
2
sin
2
2
, and
1
4
|iM|
2
=
Z
2
e
4
(1 v
2
sin
2
2
)
k
2
v
2
sin
2
2
M
2
+O(M). (12)
Substituting this into the cross section, and sending M , we reach the Mott formula
again:
d
d
=
Z
2
2
(
1 v
2
sin
2
2
)
4|k|
2
v
2
sin
4
(/2)
(13)
2
k
1
p
2
k
2
p
1
e
e
+
e
e
+
k
1
p
2
p
1
k
2
e
e
+
e
e
+
Figure 2: Bhabha scattering at tree level. All initial momenta go inward and all nal momenta
go outward.
2 Bhabha scattering
The Bhabha scattering is the process e
+
e
e
+
e
. At the tree level, it consists

of two diagrams, as shown in Figure 2. The minus sign before the t-channel diagram
comes from the exchange of two fermion eld operators when contracting with in and
out states. In fact, the s- and t-channel diagrams correspond to the following two ways
of contraction, respectively:
p
1
p
2
|
/ A

/ A|k
1
k
2
, p
1
p
2
|
/ A

/ A|k
1
k
2
. (14)
In the high energy limit, we can omit the mass of electrons, then the amplitude for the
whole scattering process is:
iM= (ie)
2
[
v(k
2
)
u(k
1
)
i
s
u(p
1
)
v(p
2
) u(p
1
)
u(k
1
)
i
t
v(k
2
)
v(p
2
)
]
. (15)
Where we have used the Mandelstam variables s, t and u. They are dened as:
s = (k
1
+ k
2
)
2
, t = (p
1
k
1
)
2
, u = (p
2
k
1
)
2
. (16)
In the massless case, k
2
1
= k
2
2
= p
2
1
= p
2
2
= 0, thus we have:
s = 2k
1
k
2
= 2p
1
p
2
, t = 2p
1
k
1
= 2p
2
k
2
, u = 2p
2
k
1
= 2p
1
k
2
. (17)
We want to get the unpolarized cross section, thus we must average the ingoing spins
and sum over outgoing spins. That is:
1
4
spin
|M|
2
=
e
4
4s
2
v(k
2
)
u(k
1
) u(p
1
)
v(p
2
)
2
+
e
4
4t
2
u(p
1
)
u(k
1
) v(k
2
)
v(p
2
)
e
4
4st
[
v(p
2
)
u(p
1
) u(k
1
)
v(k
2
) u(p
1
)
u(k
1
) v(k
2
)
v(p
2
) + c.c.
]
=
e
4
4s
2
tr (/ k
1
/ k
2
) tr (
/
p
2
/
p
1
) +
e
4
4t
2
tr (/ k
1
/
p
1
) tr (
/
p
2
/ k
2
e
4
4st
[
tr (/ k
1
/ k
2
/
p
2
/
p
1
) + c.c.
]
=
2e
4
(u
2
+ t
2
)
s
2
+
2e
4
(u
2
+ s
2
)
t
2
+
4e
4
u
2
st
= 2e
4
[
t
2
s
2
+
s
2
t
2
+ u
2
(
1
s
+
1
t
)
2
]
. (18)
3
In the center-of-mass frame, we have k
0
1
= k
0
2
k
0
, and k
1
= k
2
, thus the total
energy E
2
CM
= (k
0
1
+ k
0
2
)
2
= 4k
2
= s. According to the formula for the cross section in
the four identical particles case (Eq.4.85):
(
d
d
)
CM
=
1
64
2
E
CM
(
1
4
|M|
2
)
, (19)
thus
(
d
d
)
CM
=

2
2s
[
t
2
s
2
+
s
2
t
2
+ u
2
(
1
s
+
1
t
)
2
]
, (20)
where = e
2
/4 is the ne structure constant. We integrate this over the angle to
get:
(
d
d cos
)
CM
=

2
s
[
t
2
s
2
+
s
2
t
2
+ u
2
(
1
s
+
1
t
)
2
]
. (21)
3 The spinor products (2)
In this problem we continue our study of spinor product method in last chapter. The
formulae needed in the following are:
u
L
(p) =
1
2p k
0
/
pu
R0
, u
R
(p) =
1
2p k
0
/
pu
L0
. (22)
s(p
1
, p
2
) = u
R
(p
1
)u
L
(p
2
), t(p
1
, p
2
) = u
L
(p
1
)u
R
(p
2
). (23)
For detailed explanation for these relations, see Problem 3.3
(a) Firstly, let us prove the following relation:
|s(p
1
, p
2
)|
2
= 2p
1
p
2
. (24)
We make use of the another two relations,
u
L0
u
L0
=
1
5
2
/ k
0
, u
R0
u
R0
1 +
5
2
/ k
0
. (25)
which are direct consequences of the familiar spin-sum formula
u
0
u
0
= / k
0
. We now
generalize this to:
u
L
(p) u
L
(p) =
1
5
2
/
p, u
R
(p) u
R
(p) =
1 +
5
2
/
p. (26)
We prove the rst one:
u
L
(p) u
L
(p) =
1
2p k
0
/
pu
R0
u
R0/
p =
1
2p k
0
/
p
1 +
5
2
/ k
0
/
p
=
1
2p k
0
1
5
2
/
p/ k
0
/
p =
1
2p k
0
1
5
2
(2p k / k
0
/
p)
/
p
=
1
5
2
/
p
1
2p k
0
1
5
2
/ k
0
p
2
=
1
5
2
/
p. (27)
The last equality holds because p is lightlike. Then we get:
|s(p
1
, p
2
)|
2
=| u
R
(p
1
)u
L
(p
2
)|
2
= tr
(
u
L
(p
2
) u
L
(p
2
)u
R
(p
1
) u
R
(p
1
)
)
=
1
4
tr
(
(1
5
)
/
p
2
(1
5
)
/
p
1
)
= 2p
1
p
2
. (28)
4
(b) Now we prove the relation:
tr (
n
) = tr (
1
), (29)
where
i
= 0, 1, 2, 3, 5.
To make things easier, let us perform the proof in Weyl representation, without loss
of generality. Then its easy to check that
(
)
T
=
{
, = 0, 2, 5;
, = 1, 3.
(30)
Then, we dene M =
1
3
, and it can be easily shown that M
1
M = (
)
T
, and
M
1
M = 1. Then we have:
tr (
n
) = tr (M
1
1
MM
1
2
M M
1
n
M)
= tr
[
(
1
)
T
(
2
)
T
(
n
)
T
]
= tr
[
(
1
)
T
]
= tr (
1
). (31)
With this formula in hand, we can derive the equality
u
L
(p
1
)
u
L
(p
2
) = u
R
(p
2
)
u
R
(p
1
) (32)
as follows
LHS = C u
R0/
p
1
/
p
2
u
R0
= C tr (
/
p
1
/
p
2
)
= C tr (
/
p
2
/
p
1
) = C u
L0/
p
2
/
p
1
u
L0
= RHS,
in which C
(
2
(p
1
k
0
)(p
2
k
0
)
)
1
.
(c) The way of proving the Fierz identity
u
L
(p
1
)
u
L
(p
2
)[
]
ab
= 2
[
u
L
(p
2
) u
L
(p
1
) + u
R
(p
1
) u
R
(p
2
)
]
ab
(33)
has been indicated in the book. The right hand side of this identity, as a Dirac matrix,
which we denoted by M, can be written as a linear combination of 16 matrices listed
in Problem 3.6. In addition, it is easy to check directly that
M = M
5
. Thus M
must have the form
M =
(
1
5
2
)
V

+
(
1 +
5
2
)
.
Each of the coecients V

and W
can be determined by projecting out the other one

with the aid of trace technology, that is,
V

=
1
2
tr
[
(
1
5
2
)
M
]
= u
L
(p
1
)
u
L
(p
2
), (34)
W
=
1
2
tr
[
(
1 +
5
2
)
M
]
= u
R
(p
2
)
u
R
(p
1
) = u
L
(p
1
)
u
L
(p
2
). (35)
The last equality follows from (32). Substituting V

and W
back, we nally get the

left hand side of the Fierz identity, which nishes the proof.
5
(d) The amplitude for the process at leading order in is given by
iM= (ie
2
) u
R
(k
2
)
u
R
(k
1
)
i
s
v
R
(p
1
)
v
R
(p
2
). (36)
To make use of the Fierz identity, we multiply (33), with the momenta variables changed
to p
1
k
1
and p
2
k
2
, by
[
v
R
(p
1
)
]
a
and
[
v
R
(p
2
)
]
b
, and also take account of (32),
which leads to
u
R
(k
2
)
u
R
(k
1
) v
R
(p
1
)
v
R
(p
2
)
= 2
[
v
R
(p
1
)u
L
(k
2
) u
L
(k
1
)v
R
(p
2
) + v
R
(p
1
)u
R
(k
1
) u
R
(k
2
)v
R
(p
2
)
]
= 2s(p
1
, k
2
)t(k
1
, p
2
). (37)
Then,
|iM|
2
=
4e
4
s
2
|s(p
1
, k
2
)|
2
|t(k
1
, p
2
)|
2
=
16e
4
s
2
(p
1
k
2
)(k
1
p
2
) = e
4
(1 + cos )
2
, (38)
and
d
d
(e
+
L
e
R

+
L
R
) =
|iM|
2
64
2
E
cm
=

2
4E
cm
(1 + cos )
2
. (39)
It is straightforward to work out the dierential cross section for other polarized pro-
cesses in similar ways. For instance,
d
d
(e
+
L
e
R

+
R
L
) =
e
4
|t(p
1
, k
1
)|
2
|s(k
2
, p
2
)|
2
64
2
E
cm
=

2
4E
cm
(1 cos )
2
. (40)
(e) Now we recalculate the Bhabha scattering studied in Problem 5.2, by evaluating
all the polarized amplitudes. For instance,
iM(e
+
L
e
R
e
+
L
e
R
)
= (ie)
2
[
u
R
(k
2
)
u
R
(k
1
)
i
s
v
R
(p
1
)
v
R
(p
2
)
u
R
(p
1
)
u
R
(k
1
)
i
t
v
R
(k
2
)
v
R
(p
2
)
]
= 2ie
2
[
s(p
1
, k
2
)t(k
1
, p
2
)
s

s(k
2
, p
1
)t(k
1
, p
2
)
t
]
. (41)
Similarly,
iM(e
+
L
e
R
e
+
R
e
L
) = 2ie
2
t(p
1
, k
1
)s(k
2
, p
2
)
s
, (42)
iM(e
+
R
e
L
e
+
L
e
R
) = 2ie
2
s(p
1
, k
1
)t(k
2
, p
2
)
s
, (43)
iM(e
+
R
e
L
e
+
R
e
L
) = 2ie
2
[
t(p
1
, k
2
)s(k
1
, p
2
)
s

t(k
2
, p
1
)s(k
1
, p
2
)
t
]
, (44)
iM(e
+
R
e
R
e
+
R
e
R
) = 2ie
2
t(k
2
, k
1
)s(p
1
, p
2
)
t
, (45)
iM(e
+
L
e
L
e
+
L
e
L
) = 2ie
2
s(k
2
, k
1
)t(p
1
, p
2
)
t
. (46)
Squaring the amplitudes and including the kinematic factors, we nd the polarized
dierential cross sections as
d
d
(e
+
L
e
R
e
+
L
e
R
) =
d
d
(e
+
R
e
L
e
+
R
e
L
) =

2
u
2
2s
(
1
s
+
1
t
)
2
, (47)
6
d
d
(e
+
L
e
R
e
+
R
e
L
) =
d
d
(e
+
R
e
L
e
+
L
e
R
) =

2
2s
t
2
s
2
, (48)
d
d
(e
+
R
e
R
e
+
R
e
R
) =
d
d
(e
+
L
e
L
e
+
L
e
L
) =

2
2s
s
2
t
2
. (49)
Therefore we recover the result obtained in Problem 5.2:
d
d
(e
+
e
e
+
e
) =

2
2s
[
t
2
s
2
+
s
2
t
2
+ u
2
(
1
s
+
1
t
)
2
]
. (50)
4 Positronium lifetimes
In this problem we study the decay of positronium (Ps) in its S and P states. To
begin with, we recall the formalism developed in the Peskin & Schroeder that treats
the problem of bound states with nonrelativistic quantum mechanics. The positronium
state |Ps, as a bound state of an electron-positron pair, can be represented in terms of
electron and positrons state vectors, as
|Ps =
2M
P
d
3
k
(2)
3
(k)C
ab
1
2m
|e
a
(k)
1
2m
|e
+
b
(k), (51)
where m is the electrons mass, M
P
is the mass of the positronium, which can be taken
to be 2m as a good approximation, a and b are spin labels, the coecient C
ab
depends
on the spin conguration of |Ps, and (k) is the momentum space wave function for
the positronium in nonrelativistic quantum mechanics. In real space, we have
100
(r) =
(m
r
)
3
exp(m
r
r), (52)
21i
(r) =
(m
r
/2)
5
x
i
exp(m
r
r/2). (53)
where m
r
= m/2 is the reduced mass. Then the amplitude of the decay process Ps 2
can be represented in terms of the amplitude for the process e
+
e
2 as
M(Ps 2) =
1
d
3
k
(2)
3
(k)C
ab
M
(
e
a
(k)e
+
b
(k) 2
)
. (54)
We put a hat on the amplitude of e
+
e
2. In the following, a hatted amplitude

always refer to this process.
(a) In this part we study the decay of the S-state positronium. As stated above, we
have to know the amplitude of the process e
+
e
2, which is illustrated in Figure 3

with the B replaced with , and is given by
i
M= (ie)
2
(p
1
)
(p
2
)
v(k
2
)
[
i(/ k
1
/
p
1
+ m)
(k
1
p
1
)
2
m
2
i(/ k
1
/
p
2
+ m)
(k
1
p
2
)
2
m
2
]
u(k
1
), (55)
where the spinors can be written in terms of two-component spinors and
in the
chiral representation as
u(k
1
) =
(
k
1

k
1

)
, v(k
2
) =
(
k
2

k
2

)
. (56)
7
We also write
as:
=
(
0
0
)
,
where
= (1,
i
) and
= (1,
i
) with
i
the three Pauli matrices. Then the
amplitude can be brought into the following form,
i
M= ie
2
(p
1
)
(p
2
)
t
(k
1
p
1
)
2
m
2
+

u
(k
1
p
2
)
2
m
2
]
, (57)
with
t
=
(
k
2

k
1

k
2

k
1

)
m
+
(
k
2

k
1

k
2

k
1

)
(k
1
p
1
)
u
=
(
k
2

k
1

k
2

k
1

)
m
+
(
k
2

k
1

k
2

k
1

)
(k
1
p
2
)
.
In the rest of the part (a), we take the nonrelativistic limit, with the momenta chosen
to be
k
1
= k
2
= (m, 0, 0, 0), p
1
= (m, 0, 0, m), p
2
= (m, 0, 0, m). (58)
Accordingly, we can assign the polarization vectors for nal photons to be
(p
1
) =
1
2
(0, 1, i, 0),
(p
2
) =
1
2
(0, 1, i, 0). (59)
Now substituting the momenta (58) into (57), noticing that
k
i
=
k
i
=
m (i =
1, 2), and (k
1
p
1
)
2
= (k
1
p
2
)
2
= m
2
, and also using the trick that one can freely
make the substitution
since the temporal component of the polarization

vectors
always vanishes, we get a much more simplied expression,

i
M= ie
2
(p
1
)
(p
2
)
)
. (60)
The positronium can lie in spin-0 (singlet) state or spin-1 (triplit) state. In the former
case, we specify the polarizations of nal photons in all possible ways, and also make
the substitution
2
(See (5.49) of Peskin & Schroeder), which leads to
i
M
s
++
= i
M
s
= i2
2e
2
, i
M
s
+
= i
M
s
+
= 0, (61)
where the subscripts denote nal photons polarizations and s means singlet. We show
the mid-step for calculating iM
s
++
as an example:
i
M
s
++
=
ie
2
2
tr
[(
(
1
+ i
2
)
3
(
1
+ i
2
) (
1
+ i
2
)
3
(
1
+ i
2
)
)
]
= i2
2e
2
.
In the same way, we can calculate the case of triplet initial state. This time, we make
the substitution
n /
2, with n = ( x i y)/
2 or n = z, corresponding to
three independent polarizations. But it is straightforward to show that the amplitudes
with these initial polarizations all vanish, which is consistent with our earlier results by
using symmetry arguments in Problem 3.8.
8
Therefore it is enough to consider the singlet state only. The amplitude for the decay
of a positronium in its
1
S
0
state into 2 then follows directly from (54), as
M
(
1
S
0
2) =
(x = 0)
M
s
, (62)
where (x = 0) =
(m/2)
3
/ according to (52). Then the squared amplitude with
nal photons polarizations summed is
spin
M(
1
S
0
2)
2
=
|(0)|
2
2m
(
|M
s
++
|
2
+|M
s
|
2
)
= 16
5
m
2
. (63)
Finally we nd the decay width of the process Ps(
1
S
0
) 2, to be
(
1
S
0
2) =
1
2
1
4m
d
3
p
1
d
3
p
2
(2)
6
2E
1
2E
2
M(
1
S
0
2)
2
(2)
4
(4)
(p
Ps
p
1
p
2
)
=
1
2
1
4m
d
3
p
1
(2)
3
4m
2
M(
1
S
0
2)
2
(2)(mE
1
)
=
1
2

5
m, (64)
where an additional factor of 1/2 follows from the fact that the two photons in the nal
state are identical particles.
(b) To study the decay of P state (l = 1) positronium, we should keep one power of
3-momenta of initial electron and positron. Thus we set the momenta of initial and nal
particles, and also the polarization vectors of the latter, in e
e
+
2, to be
k
1
= (E, 0, 0, k), k
2
= (E, 0, 0, k),
p
1
= (E, E sin , 0, E cos ), p
2
= (E, E sin , 0, E cos ),
(p
1
) =
1
2
(0, cos , i, sin ),
(p
2
) =
1
2
(0, cos , i, sin ). (65)
Here we have the approximate expression up to linear order in k:
k
1
=
k
2
=
m
k
2
3
+O(k
2
),
k
2
=
k
1
=
m +
k
2
3
+O(k
2
),
1
(k
1
p
1
)
2
m
2
=
1
2m
2

k cos
2m
3
+O(k
2
),
1
(k
1
p
2
)
2
m
2
=
1
2m
2
+
k cos
2m
3
+O(k
2
).
Consequently,
t
= 2m
2
(
1
s
+
3
c
mk
(
3
+ 2
)
+O(k
2
),
u
= 2m
2
(
1
s
+
3
c
mk
(
3
+ 2
)
+O(k
2
),
where we use the shorthand notation s
= sin and c
= cos . We can use these

expansion to nd the terms in the amplitude i
M of linear order in k, to be
i
O(k)
=ie
2
(p
1
)
(p
2
)
k
2m
[
2c
(
1
s
+
3
c
2c
(
1
s
+
3
c
9
+
(
3
+ 2
)
+
(
3
+ 2
)
]
,
(66)
Feeding in the polarization vectors of photons, and also make the substitution

n /
2 or 1/
2 for triplet and singlet positronium, respectively, as done in last part,

we get
i
|
O(k)
= 0, i
|
O(k)
= i2s
(1 + c
)e
2
k/m,
i
M
+
|
O(k)
= i2
2e
2
k/m, i
M
+
|
O(k)
= i2
2s
2
e
2
k/m,
i
|
O(k)
= 0, i
|
O(k)
= i2s
(1 + c
)e
2
k/m,
i
|
O(k)
= 0, i
|
O(k)
= 0. (67)
The vanishing results in the last line indicate that S = 0 state of P-wave positronium
cannot decay to two photons.
(c) Now we prove that the state
|B(k) =
2M
P
d
3
p
(2)
3
i
(p)a
p+k/2
i
b
p+k/2
|0 (68)
is a properly normalized state for the P-wave positronium. In fact,
B(k)|B(k) = 2M
P
d
3
p
(2)
3
d
3
p
(2)
3

j
(p
)
i
(p)
0|b
p
+k/2
j
a
p
+k/2
a
p+k/2
i
b
+k/2
|0
= 2M
P
d
3
p
(2)
3
d
3
p
(2)
3

j
(p
)
i
(p)
0|b
p
+k/2
i
b
+k/2
|0(2)
3
(3)
(p
p)
= 2M
P
d
3
p
(2)
3

j
(p)
i
(p)0|b
p+k/2
i
b
p+k/2
|0
= 2M
P
d
3
p
(2)
3

j
(p)
i
(p)0| tr (
j
i
)|0(2)
3
(3)
(0)
= 2M
P
(2)
3
(3)
(0), (69)
which is precisely the needed normalization of a state. In this calculation we have used
the anticommutation relations of creation and annihilation operators, as well as the
normalization of the wave function and the matrices.
(d) Now we evaluate the partial decay rate of the S = 1 P-wave positronium of
denite J into two photons. The states for the positronium is presented in (c), with the
matrices chosen as
=
i
, J = 0,
1
2

ijk
n
j
k
, J = 1,
1
3
h
ij
j
, J = 2,
(70)
and the wave function given by (53).
10
Firstly, consider the J = 0 state, in which case we have
iM(
3
P
0

) =
1
d
3
k
(2)
3
i
(k)
(
1
i
)
ab
i
M
(
e
a
(k)e
+
b
(k)
)
, (71)
where , = + or are labels of photons polarizations and a, b = or are spinor
indices. For amplitude i
M, we only need the terms linear in k, as listed in (67). Let us

rewrite this as
i
M
(
e
a
(k)e
+
b
(k)
)
= F
ab
,i
k
i
.
In the same way, the wave function can also be put into the form of
i
(x) = x
i
f(r),
with r = |x|. Then the integration above can be carried out to be
iM(
3
P
0

) =
i
6m
i
ab
F
ab
,j
[
i

x
j
i
(x)
]
x=0
=
i
6m
i
ab
F
ab
,i
f(0). (72)
On the other hand, we have chose the direction of k to be in the x
3
-axis, then F
ab
,1
=
F
ab
,2
= 0 as a consequence. Therefore,
iM(
3
P
0

) =
i
6m
f(0)
(
F
,3
F
,3
)
=
7
24
msin . (73)
Square these amplitudes, sum over the photons polarizations, and nish the phase space
integration in the same way as what we did in (a), we nally get the partial decay rate
of the J = 0 P-wave positronium into two photons to be
(
3
P
0
) =
1
576
7
m. (74)
The positronium in
3
P
1
state, namely the case J = 1, cannot decay into two photons
by the conservation of the angular momentum, since the total angular momentum of
two physical photons cannot be 1. Therefore let us turn to the case of J = 2. In this
case we should average over the initial polarizations of the positronium, which can be
represented by the symmetric and traceless polarization tensors h
ij
n
, with n = 1, 2, , 5
the labeled of 5 independent polarizations. Let us choose these tensors to be
h
ij
1
=
1
2
(
i2
j3
+
i3
j2
), h
ij
2
=
1
2
(
i1
j3
+
i3
j1
),
h
ij
3
=
1
2
(
i1
j2
+
i2
j1
), h
ij
4
=
1
2
(
i1
j1
i2
j2
),
h
ij
5
=
1
2
(
i1
j1
i3
j3
). (75)
Then the decay amplitude for a specic polarization of J = 2 Ps can be represented as
iM
n
(
3
P
2

) =
1
d
3
k
(2)
3
i
(k)
(
1
3
h
ij
n
j
)
ab
i
M
(
e
a
(k)e
+
b
(k)
)
=
1
3m
h
ij
n
j
ab
F
ab
,i
f(0). (76)
Now substituting all stus in, we nd the nonvanishing components of the decay ampli-
tude to be
iM
2
(
3
P
2

) =
7
48
im,
11
iM
2
(
3
P
2

) =
7
48
imsin
2
,
iM
5
(
3
P
2

) = 2
7
48
imsin
2
. (77)
Squaring these amplitudes, summing over photons polarizations and averaging the ini-
tial polarization of the positronium (by dividing the squared and summed amplitude by
5), we get
1
5
spin
M
n
(
3
P
2
2)
2
=

7
m
2
120
(1 + sin
2
+ 4 sin
4
). (78)
Finally, we nish the phase space integration and get the partial decay rate of
3
P
2
positronium into 2 photons to be
(
3
P
2
) =
19
19200
7
m. (79)
5 Physics of a massive vector boson
In this problem, the mass of electron is always set to zero.
(a) We rstly compute the cross section (e
+
e
B) and the decay rate (B

e
+
e
). For the cross section, the squared amplitude can be easily found to be
1
4
spin
|iM|
2
=
1
4
spin
ig
(i)
v(p
u(p)
2
= 2g
2
(p p
). (80)
Note that we have set the mass of electrons to be zero. Then the cross section can be
deduced from (4.79). Lets take the initial momenta to be
p =
1
2
(E, 0, 0, E), p
=
1
2
(E, 0, 0, E), (81)
with E being the center-of-mass energy. Then its easy to get
=
g
2
4E
(2)(M
B
E) =
g
2
4E
(2)2M
B
(M
2
B
s) = g
2
(M
2
B
s), (82)
where s = E
2
.
To deduce the decay rate, we should average polarizations of massive vector B instead
of two electrons. Thus the squared amplitude in this case reads
1
3
spin
|iM|
2
=
8
3
g
2
(p p
). (83)
The decay rate can be found from (4.86):
=
1
2M
B
d
3
p
(2)
3
d
3
p
2
3
1
2E
p
1
2E
p
(
1
3
|M|
2
)
(2)
4
(4)
(p
B
p p
)
=
1
2M
B
d
3
p
(2)
3
1
4E
2
p
(
16
3
g
2
E
2
p
)
(2)(M
B
2E
p
)
=
4
(2)
2
2M
B
dp
4
3
g
2
E
2
p
1
2
(
1
2
M
B
E
p
) =
g
2
M
B
12
. (84)
We see the cross section and the decay rate satisfy the following relation, as expected:
(e
+
e
B) =
12
2
M
B
(B e
+
e
)(s M
2
). (85)
12
k
1
p
2
p
1
k
2
e
e
+
B
+
k
1
p
1
k
2
p
2
e
e
+
B
Figure 3: The tree diagrams of the process e
e
+
+B. All initial momenta go inward and
all nal momenta go outward.
(b) Now we calculate the cross section (e
e
+
+B) in COM frame. The related
diagrams are shown in Figure 3. The amplitude reads:
iM= (ie)(ig)
(p
1
)e
(p
2
) v(k
2
)
[
i
/ k
1
/
p
1
i
/ k
1
/
p
2
]
u(k
1
), (86)
where
is the polarization of photon while e
is the polarization for B. Now we square

this amplitude,
1
4
spin
|iM|
2
=
1
4
e
2
g
2
g
tr
[
(
(/ k
1
/
p
1
)
t
+
(
/
p
1
/ k
2
)
u
)
/ k
1
(/ k
1
/
p
1
)
t
+
(
/
p
1
/ k
2
)
u
)
/ k
2
]
= 8e
2
g
2
[
(k
1
p
1
)(k
2
p
1
)
t
2
+
(k
1
p
1
)(k
2
p
1
)
u
2
+
2(k
1
k
2
)(k
1
k
2
k
1
p
1
k
2
p
1
)
tu
]
= 2e
2
g
2
[
u
t
+
t
u
+
2s(s + t + u)
tu
]
= 2e
2
g
2
[
u
t
+
t
u
+
2sM
2
B
tu
]
(87)
Then the cross section can be evaluated as
(
d
d
)
CM
=
1
2E
k
1
2E
k
2
|v
k
1
v
k
2
|
|p
1
|
(2)
2
4E
CM
(
1
4
|M|
2
)
=
e
2
g
2
32
2
s
(
1
M
2
B
s
)
[
u
t
+
t
u
+
2sM
2
B
tu
]
. (88)
We can also write this dierential cross section in terms of squared COM energy s and
scattering angle . To do this, we note that
s = E
2
CM
, t = (M
2
B
E
2
CM
) sin
2
2
, u = (M
2
B
E
2
CM
) cos
2
2
. (89)
Then we have
(
d
d
)
CM
=
e
2
g
2
(1 M
2
B
/s)
16
2
s sin
2
[
1 + cos
2
+
4sM
2
B
(s M
2
B
)
2
]
, (90)
and
(
d
d cos
)
CM
=
g
2
(1 M
2
B
/s)
2s sin
2
[
1 + cos
2
+
4sM
2
B
(s M
2
B
)
2
]
. (91)
13
(c) The dierential cross obtained in (b) diverges when 0 or . Now let us
study the former case, namely 0.
If we cut of the integral from
2
c
m
2
e
/s, then we have:
c
(
d
d cos
)
CM
sin d
g
2
(1 M
2
B
/s)
2s
[
2 +
4sM
2
B
(s M
2
B
)
2
]
1m
2
e
/s
dt
1 t
2
g
2
(1 M
2
B
/s)
4s
[
2 +
4sM
2
B
(s M
2
B
)
2
]
log
(
s
m
2
e
)
=
g
2
2
1 + M
4
B
/s
2
s M
2
B
log
(
s
m
2
e
)
(92)
Now we calculate the following expression:
1
0
dxf(x)(e
+
e
B)
E
CM
=(1x)s
=
1
0
dx
[

2
1 + (1 x)
2
x
log
(
s
m
2
e
)
]
g
2
(
M
2
B
(1 x)s
)
=
g
2
2
1 + M
4
B
/s
2
s M
2
B
log
(
s
m
2
e
)
(93)
6 The spinor products (3)
This problem generalize the spinor product formalism to the processes involving
external photons.
(a) Firstly we can represent photons polarization vectors in terms of spinors of denite
helicity. Let the momentum of the photon be k, and p be a lightlike momentum such
that p k = 0. Then, the polarization vector
(k) of the photon can be taken to be
+
(k) =
1
4p k
u
R
(k)
u
R
(p),
(k) =
1
4p k
u
L
(k)
u
L
(p), (94)
where the spinors u
L,R
(k) have been introduced in Problems 3.3 and 5.3. Now we use
this choice to calculate the polarization sum:
+
(k)
+
(k) +
(k)
(k)
=
1
4p k
[
u
R
(k)
u
R
(p) u
R
(p)
u
R
(k) + u
L
(k)
u
L
(p) u
L
(p)
u
L
(k)
]
=
1
4p k
tr
[
/
p
/ k
]
= g
+
p
+ p
p k
. (95)
When dotted into an amplitude with external photon, the second term of the result
vanishes. This justies the denitions above for photons polarization vectors.
(b) Now we apply the formalism to the process e
+
e
2 in the massless limit. The

relevant diagrams are similar to those in Figure 3, except that one should replace the
label B by . To simplify expressions, we introduce the standard shorthand notations
as follows:
p = u
R
(p), p] = u
L
(p), p = u
L
(p), [p = u
R
(p). (96)
14
Then the spin products become s(p
1
, p
2
) = [p
1
p
2
] and t(p
1
, p
2
) = p
1
p
2
. Various
expressions get simplied with this notation. For example, the Fierz identity (37) now
reads [k
2
k
1
[p
1
p
2
= 2[p
1
k
2
]k
1
p
2
. Similarly, we also have k
1
k
2
]p
1
p
2
] =
2k
1
p
1
[p
2
k
2
].
Now we write down the expression for tree amplitude of e
+
R
e
L

R
L
. For illustra-
tion, we still keep the original expression as well as all explicit mid-steps. The auxiliary
lightlike momenta used in the polarization vectors are arbitrarily chosen such that the
calculation can be mostly simplied.
iM(e
+
R
e
L

L
R
)
= (ie)
2
(p
1
)
+
(p
2
) u
L
(k
2
)
[
i
/ k
1
/
p
1
i
/ k
1
/
p
1
]
u
L
(k
1
)
=ie
2
k
2
p
1
][k
1
p
2
(k
2
p
1
)(k
1
p
2
)
[
k
2
(/ k
1
/
p
1
)
k
1
]
t
+
k
2
(/ k
1
/
p
2
)
k
1
]
u
]
=ie
2
k
2
p
1
][k
1
p
2
2u
[
k
2
k
1
]k
1
k
1
] k
2
p
1
]p
1
k
1
]
t
+
k
2
k
1
]k
1
k
1
] k
2
p
2
]p
2
k
1
]
u
]
=
2ie
2
u
[
k
1
k
2
[p
1
k
1
]k
2
p
2
[k
1
k
1
] k
2
p
1
[k
1
p
1
]k
2
p
2
[k
1
p
1
]
t
+
k
2
k
2
[k
1
p
1
]k
1
p
2
[k
1
k
1
] k
2
k
2
[p
2
p
1
]p
2
p
2
[k
1
k
1
]
u
]
= 2ie
2
k
2
p
1
[k
1
p
1
]k
2
p
2
[k
1
p
1
]
tu
, (97)
where we have used the spin sum identity
/
p = p]p +p[p in the third equality, and also
the Fierz transformations. Note that all spinor products like pp and [pp], or p
k
and [p
k] vanish. Square this amplitude, we get
iM(e
+
R
e
L

L
R
)
2
= 4e
4
t
u
. (98)
In the same way, we calculate other polarized amplitudes:
iM(e
+
R
e
L

R
L
)
=ie
2
[k
1
p
1
k
2
p
2
]
4
(k
1
p
1
)(k
2
p
2
)
[
[k
2
(/ k
1
/
p
1
)
k
1
t
+
[k
2
(/ k
1
/
p
2
)
k
1
u
]
= 2ie
2
k
2
p
1
[k
1
p
2
]k
2
p
2
[k
1
p
2
]
tu
(99)
Note that we have used a dierent set of auxiliary momenta in photons polarizations.
After evaluating the rest two nonvanishing amplitudes, we get the squared polarized
amplitudes, as follows:
M(e
+
R
e
L

L
R
)
2
=
M(e
+
L
e
R

R
L
)
2
= 4e
4
t
u
, (100)
M(e
+
L
e
R

L
R
)
2
=
M(e
+
L
e
R

L
R
)
2
= 4e
4
u
t
. (101)
Then the dierential cross section follows straightforwardly,
d
d cos
=
1
16s
(
1
4
spin
|iM|
2
)
=
2
2
s
(
t
u
+
u
t
)
, (102)
which is in accordance with (5.107) of Peskin & Schroeder.
15
Chapter 6
Zhong-Zhi Xianyu

1 Rosenbluth formula
In this problem we derive the dierential cross section for the electron-proton scatter-
ing in the lab frame, assuming that the scattering energy is much higher than electrons
mass, and taking account of the form factors of the proton. The result is known as
Rosenbluth formula. The relevant diagram is shown in Figure 1. Let us rstly work out
k
1
k
2
p
1
p
2
e
p
e
p
Figure 1: The electron-proton scattering. The blob denotes form factors that includes the
eect of strong interaction. All initial momenta go inward and all nal momenta go outward.
the kinematics. In the lab frame, the momenta can be parameterized as
k
1
= (E, 0, 0, E), p
1
= (E
, E
sin , 0, E
cos ), k
2
= (M, 0, 0, 0), (1)
and p
2
can be found by momentum conservation, k
1
+ k
2
= p
1
+ p
2
. With the on-shell
condition p
2
2
= M
2
, we nd that
E
=
ME
M + 2E sin
2
2
. (2)
We also use q = k
1
p
1
to denote the momentum transfer and t = q
2
its square. Note
that we have set the electron mass to zero.
Now we write down the amplitude M.
iM= (ie)
2

U(p
2
)
[
F
1
(q
2
) +
i
2M
F
2
(q
2
)
]
U(k
2
)
i
t
u(p
1
)
u(k
1
), (3)
1
where U is the spinor for the proton and u is for the electron, M is the mass of the
proton. At this stage, we convert this expression into a more convenient form by means
of the Gordon identity (see Problem 3.2):
iM= (ie)
2

U(p
2
)
[
(F
1
+F
2
)
(p
2
+k
2
)
2M
F
2
]
U(k
2
)
i
t
u(p
1
)
u(k
1
), (4)
Now, the modular-squared amplitude with initial spins averaged and nal spins
summed, is
1
4
|M|
2
=
e
4
4q
4
tr
[(
(F
1
+F
2
)
(p
2
+k
2
)
2M
F
2
)
(/ k
2
+M)
(F
1
+F
2
)
(p
2
+k
2
)
2M
F
2
)
(
/
p
2
+M)
]
tr
[
/ k
1
/
p
1
]
=
4e
4
M
2
q
4
[
(
2E
2
+ 2E
2
+q
2
)
(F
1
+F
2
)
2
(
2F
1
F
2
+F
2
2
(
1 +
q
2
4M
2
)
)(
(E +E
)
2
+q
2
(
1
q
2
4M
2
)
)
]
. (5)
There are two terms in the square bracket in the last expression. We rewrite the rst
factor in the second term as
2F
1
F
2
+F
2
2
(
1 +
q
2
4M
2
)
= (F
1
+F
2
)
2
F
2
1
+
q
2
4M
2
F
2
2
,
and combine the (F
1
+F
2
)
2
part into the rst term, which leads to
1
4
|M|
2
=
4e
4
M
2
q
4
[
q
4
2M
2
(F
1
+F
2
)
2
+ 4
(
F
2
1

q
2
4M
2
F
2
2
)
EE
cos
2
2
]
,
where we have used the following two relations which can be easily justied:
E
E =
q
2
2m
(6)
q
2
= 4E
E sin
2
2
. (7)
Now we can put the squared amplitude into its nal form:
1
4
|M|
2
=
16e
4
E
2
M
3
q
4
(
M + 2E sin
2
2
)
[
(
F
2
1

q
2
4M
2
F
2
2
)
cos
2
2

q
2
2M
2
(F
1
+F
2
)
2
sin
2
2
]
, (8)
On the other hand, we can derive the A+B 1 +2 dierential cross section in the
lab frame as
d
L
=
1
2E
A
2E
B
|v
A
v
B
|
d
3
p
1
d
3
p
2
(2)
6
2E
1
2E
2
|M|
2
(2)
4
(4)
(p
1
+p
2
p
A
p
B
). (9)
In our case, E
A
= E, E
B
= M, E
1
= E
, and |v
A
v
B
| 1, thus
d
L
=
1
4EM
d
3
p
1
d
3
p
2
(2)
6
2E
1
2E
2
|M|
2
(2)
4
(4)
(p
1
+p
2
p
A
p
B
)
=
1
4EM
E
2
dE
d cos d
(2)
3
2E
2E
2
|M|
2
(2)
(
E
+E
2
(E
) E M
)
=
1
4EM
E
2
dE
d cos d
(2)
2
2E
2E
2
|M|
2
[
1 +
E
E cos
E
2
(E
)
]
1
(
E
ME
M + 2E sin
2
2
)
2
=
1
4EM
d cos
8
|M|
2
E
M + 2E sin
2
2
where we use the notation E
2
= E
2
(E
) to emphasize that E
2
is a function of E
. That
is,
E
2
=
M
2
+E
2
+E
2
2E
E cos .
Then,
(
d
d cos
)
L
=
1
32
(
M + 2E sin
2
2
)
2
|M|
2
(10)
So nally we get the dierential cross section, the Rosenbluth formula, as
(
d
d cos
)
L
=

2
2E
2
(
1 +
2E
M
sin
2
2
)
sin
4
2
[
(
F
2
1

q
2
4M
2
F
2
2
)
cos
2
2

q
2
2M
2
(F
1
+F
2
)
2
sin
2
2
]
. (11)
2 Equivalent photon approximation
In this problem we study the scattering of a very high energy electron from a target
in the forward scattering limit. The relevant matrix element is
M= (ie) u(p
u(p)
ig
q
2
(q). (12)
(a) First, the spinor product in the expression above can be expanded as
u(p
u(p) = A q
+B q
+C
1
+D
2
. (13)
Now, using the fact that q
u(p
u(p) = 0, we have
0 = Aq
2
+Bq q 4AEE
sin
2
2
+Bq q B
2
.
(b) It is easy to nd that
1
= N(0, p
cos p, 0, p
sin ),
2
= (0, 0, 1, 0),
where N = (E
2
+ E
2
2EE
cos )
1/2
is the normalization constant. Then, for the
right-handed electron with spinor u
+
(p) =
2E(0, 0, 1, 0)
T
and left-handed electron
with u
(p) =
2E(0, 1, 0, 0)
T
, it is straightforward to show that
u
+
(p
) =
2E
(0, 0, cos

2
, sin

2
)
T
, u
(p
) =
2E
(sin

2
, cos

2
, 0, 0), (14)
and,
u
(p
)
1
u
(p)
EE
E +E
|E E
|
, (15)
u
(p
)
2
u
(p) i
EE
, (16)
u
(p
)
1
u
(p) = u
(p
)
2
u
(p) = 0. (17)
That is to say, we have
C
EE
E +E
|E E
|
, D
= i
EE
. (18)
3
(c) The squared amplitude is given by
|M
|
2
=
e
2
(q
2
)
2
(q)

M
(q)
(
C
1
+D
2
)(
C
1
+D
2
)
(19)
Averaging and summing over the initial and nal spins of the electron respectively, we
get
1
2
|M|
2
=
e
2
2(q
2
)
2
(q)

M
(q)
[
(
|C
+
|
2
+|C
|
2
)
1
+
(
|D
+
|
2
+|D
|
2
)
2
+
(
C
+
D
+
+C
2
+
(
C
+
D
+
+C
1
]
=
e
2
(q
2
)
2
(q)

M
(q)EE
2
[
(
E +E
E E
)
2
1
+
2
]
(20)
Then the cross section reads
d =
1
2E2M
t
d
3
p
(2)
3
2E
d
3
p
t
(2
3
)2E
t
(
1
2
|M|
2
)
(2)
4
(4)
(
p
i
)
=
e
2
2E2M
t
d
3
p
(2)
3
2E
EE
2
3(q
2
)
2
[
(
E +E
E E
)
2
+ 1
]
d
3
p
t
(2
3
)2E
t
|
(q)|
2
(2)
4
(4)
(
p
i
)
=
1
2E2M
t
dx
[
1 +
(
2 x
x
)
2
]

0
d

2
sin
4(1 cos )
2
d
3
p
t
(2
3
)2E
t
|
(q)|
2
(2)
4
(4)
(
p
i
)
. (21)
where we have used the trick described in the nal project of Part I (radiation of gluon
jets) to separate the contractions of Lorentz indices, and x (E E
)/E. Now let

us focus on the integral over the scattering angle in the last expression, which is
contributed from the following factor:

0
d

2
sin
4(1 cos )
4

0
d
which is logarithmically divergent as 0.

(d) We reintroduce the mass of the electron into the denominator to cut o the diver-
gence, namely, let q
2
= 2(EE
pp
cos ) + 2m
2
. Then we can expand q
2
, treating
m
2
and as small quantities, as
q
2
(1 x)E
2
x
2
1 x
m
2
.
Then the polar angle integration near = 0 becomes
0
d
3
[
2
+
x
2
(1 x)
2
m
2
E
2
]
2

1
2
log
E
2
m
2
. (22)
(e) Combining the results above, the cross section can be expressed as
d =
1
2E2M
t
dx
[
1 +
(
2 x
x
)
2
]
0
d
3
[
2
+
x
2
(1 x)
2
m
2
E
2
]
2
4
d
3
p
t
(2
3
)2E
t
|
(q)|
2
(2)
4
(4)
(
p
i
)
=
1
2E2M
t
dx
1 + (1 x)
2
x
2
log
E
2
m
2
d
3
p
t
(2
3
)2E
t
|
(q)|
2
(2)
4
(4)
(
p
i
)
. (23)
3 Exotic contributions to g 2
(a) The 1-loop vertex correction from Higgs boson is
u(p
u(p) =
(
i
2
)
2
d
d
k
(2)
d
i
(k p)
2
m
2
h
u(p
)
i
/ k +
/
q m
i
/ k m
u(p)
=
i
2
2
1
0
dx
1x
0
dy
d
d
k
(2)
d
2 u(p
)N
u(p)
(k
2
)
3
, (24)
with
N
= (/ k +
/
q +m)
(/ k +m),
k
= k xp +yq,
= (1 x)m
2
+xm
2
h
x(1 x)p
2
y(1 y)q
2
+ 2xyp q.
To put this correction into the following form:
F
1
(q) +
i
2m
F
2
(q), (25)
we rst rewrite N
as
N
= A
+B(p
+p)
+C(p
p)
,
where term proportional to (p
p) can be thrown away by Ward identity q
(q) = 0.
This can be done by gamma matrix calculations, leading to the following result:
N
=
[
(
2
d
1
)
k
2
+ (3 + 2x x
2
)m
2
+ (y xy y
2
)q
2
]
+ (x
2
1)m(p
+p)
.
Then, use Gordon identity, we nd
N
=
[
(
2
d
1
)
k
2
+ (x + 1)
2
m
2
+ (y y
2
xy)q
2
]
+
i
2m
2m
2
(1 x
2
).
Comparing this with (25), we see that
F
2
(q = 0) = 2i
2
m
2
1
0
dx
1x
0
dy
d
4
k
(2)
4
1 x
2
(k
2
)
3
=

2
(4)
2
1
0
dx
(1 x)
2
(1 +x)
(1 x)
2
+x(m
h
/m)
2
. (26)
To carry out the integration over x, we use the approximation that m
h
m. Then
F
2
(q = 0)

2
(4)
2
1
0
dx
[
1
1 +x(m
h
/m)
2

1 +x x
2
(m
h
/m)
2
]

2
(4)
2
(m
h
/m)
2
[
log(m
2
h
/m
2
)
7
6
]
. (27)
5
(b) According to (a), the limits on and m
h
is given by
F
2
(q = 0) =

2
(4)
2
(m
h
/m)
2
[
log(m
2
h
/m
2
)
7
6
]
< 1 10
10
. (28)
For = 3 10
6
and m
h
> 60 GeV, we have F
2
(q = 0) 10
18
10
10
. From the
bound from muons anomalous magnetic moment, in which case = 6 10
4
,
6
Chapter 7
Zhong-Zhi Xianyu

1 Optical theorem in
4
theory
In this problem we check the optical theorem in phi-4 theory to order
2
. Firstly,
the total cross section
tot
at this order receives contributions from tree level only.
The squared amplitude is simply
2
. Then its easy to get the total cross section by
complementing kinematic factors. That is,
tot
=

2
16s
, (1)
where s = E
2
CM
and E
CM
is the COM energy. Then, consider the imaginary part of the
scattering amplitude. The contribution comes from 1-loop diagram in s-channel this
time. We compute this amplitude directly,
iM=
1
2
(i)
2
d
d
k
(2)
d
i
k
2
m
2
i
(k p)
2
m
2
=

2
2
d
d
k
(2)
d
1
0
dx
1
(k
2
)
2
=
i
2
2(4)
2
[
2
+ log 4
1
0
dx log
(
m
2
x(1 x)s
)
]
. (2)
Therefore,
ImM=

2
2(4)
2
1
0
dxIm
[
log
(
m
2
x(1 x)s
)
]
. (3)
The argument in the logarithm is real, thus the imaginary part of the logarithm equals
to 0 or depending on the argument is positive or negative. (More precisely, the
imaginary part is but not due to our i prescription.) Then we see this logarithm
contributes an constant imaginary part , only when
1
1 4m
2
/s
2
< x <
1 +
1 4m
2
/s
2
.
Thus we have
ImM=

2
32
1 4m
2
/s =

2
p
CM
16E
CM
. (4)
1
2 Alternative regulators in QED
In this problem we compute the rst order corrections to Z
1
and Z
2
in QED, using
cut-o regularization and dimensional regularization respectively. The renormalization
condition is chosen to be,
(q = 0) = Z
1
1

, (5)
Z
1
2
= 1
d
d
/
p
/ p=m
. (6)
We begin with dimensional regularization instead of momentum cut-o.
(b) Dimensional regularization. We rstly calculate F
1
(0):
u(p
(p, p
)u(p)
=(ie)
2
d
d
k
(2)
d
ig
(k p)
2
2
u(p
i
/ k +
/
q m
i
/ k m
u(p)
=ie
2
d
d
k
(2)
d
u(p
(/ k +
/
q + m)
(/ k + m)
u(p)
(
(k p)
2
2
)(
(k + q)
2
m
2
)(
k
2
m
2
)
=ie
2
d
d
k
(2)
d
1
0
dx
1x
0
dy
2 u(p
)N
u(p)
(k
2
)
3
, (7)
where
k
= k xp + yq,
= (1 x)m
2
+ x
2
x(1 x)p
2
y(1 y)q
2
+ 2xyp q,
N
(/ k +
/
q + m)
(/ k + m)
.
The next step is to put N
into a suitable form. The calculation is basically in parallel

with Section 6.3 of Peskin & Schroeder. Here we show some details. The rst step is to
nish the summation over dummy Lorentz indices. Note that we are using dimensional
regularization, thus we should use (A.55) in Peskin & Schroeder. The result is:
N
=2/ k
(/ k +
/
q) + 4m(2k + q)
(d 2)m
2
+ (4 d)
[
(/ k +
/
q)
/ k m(/ k +
/
q)
/ k
]
.
Note that d will be sent to 4 at the end of the calculation. Thus in the square bracket
in this expression, only the combination / k
/ k contributes to the nal result. Thus we

simply have
N
= 2/ k
(/ k +
/
q) + 4m(2k + q)
(d 2)m
2
+ (4 d)/ k
/ k. (8)
Here and following, we are free to drop o terms in N
which contribute nothing to nal

results. The equal sign should be understood in this way. The next step is to rewrite
N
in terms of k
instead of k:
N
= (2 d)/ k
/ k
2
[
x
/
p y
/
q
]
[
x
/
p + (1 y)
/
q
]
+ 4m
[
2xp + (1 2y)q
]
2m
2
.
Terms linear in k
has been dropped since they integrate to zero. The third step is to
put N
into a linear combination of
, (p+p
. Terms proportional to (pp
will be
2
dropped due to Ward identity. The basic strategy is using substitution q
= (p
p)
,
on shell condition u(p
)
/
p
= u(p
)m and
/
pu(p) = mu(p). Here we show the detailed steps
for the second term above:
2
[
x
/
p y
/
q
]
[
x
/
p + (1 y)
/
q
]
=2
[
x
2
/
p
/
p y(1 y)
/
q
/
q xy
/
q
/
p + x(1 y)
/
p
/
q
]
=2
[
x
2
(2p
/
p)
/
p y(1 y)(2q
/
q)
/
q xy
/
q
/
p + x(1 y)(
/
p
/
q)
/
q
]
=2
[
2x
2
mp
x
2
m
2
+ y(1 y)q
2
2xymq
+ xm
/
q x(1 y)
/
q
/
q
]
=2
[
x(x + 2)m
2
+ (x + y)(1 y)q
2
+ 2x
2
mp
+ 2xmp
]
.
Combining this with other terms, and also make the momentum symmetrization, which
amounts to the substitution / k
/ k

(
2
d
1
)
k
2
, we get,
N
=
(2d)
2
d
/ k
/ k
+
[
2(x
2
+ 2x 1)m
2
2(x + y)(1 y)q
2
]
+ 2x(1 x)m(p
+ p)
. (9)
Now we employ Gordons identity
u(p
u(p) = u(p
)
[
p
+ p
2m
+
i
2m
]
u(p),
to put N
into a linear combination of
and
,
N
=
[
(2d)
2
d
k
2
2(x
2
4x + 1)m
2
2(x + y)(1 y)q
2
]
2x(1 x)mi
.
Now we have put the vertex
into the following form:
F
1
(q) +
i
2m
F
2
(q). (10)
We are interest in F
1
(q), which is related to Z
1
by Z
1
= F
1
(q = 0). Finishing
momentum integral:
F
1
(0)
=2ie
2
1
0
dx
1x
0
dy
d
d
k
(2)
d
1
(k
2
)
3
[
(2d)
2
d
k
2
2(x
2
4x + 1)m
2
]
=
2e
2
(4)
d/2
1
0
dx
1x
0
dy
[
(2 d)
2
(2
d
2
)
4
2d/2
+ (x
2
4x + 1)m
2
(3
d
2
)
3d/2
]
, (11)
and sending d = 4 4:
F
1
(0) =
2e
2
(4)
2
1
0
dx(1 x)
[
2
+ log 4
log
(
(1 x)
2
m
2
+ x
2
)
2 +
(x
2
4x + 1)m
2
(1 x)
2
m
2
+ x
2
]
, (12)
we reach the needed result Z
1
= F
1
(0). Now we turn to Z
2
. The correction of rst
order is given by Z
2
= (d/d
/
p)
/ p=m
. We compute (
/
p) using dimensional regulariza-
tion,
i(
/
p) = (ie)
2
d
d
k
(2)
d
i
/ k m
i
(p k)
2
2
3
=e
2
d
d
k
(2)
d
(2 d)/ k + dm
(k
2
m
2
)
(
(p k)
2
2
)
=e
2
d
d
k
(2)
d
1
0
dx
(2 d)x
/
p + dm
(k
2
)
2
=
ie
2
(4)
d/2
1
0
dx
(2
d
2
)
2d/2
[
(2 d)x
/
p + dm
]
, (13)
where k
= k xp and = (1 x)m
2
+ x
2
x(1 x)p
2
. Then we can compute
d(
/
p)
d
/
p
=
e
2
(4)
d/2
1
0
dx
[
(2
d
2
)
2d/2
(2 d)x
(2
d
2
)(2
d
2
)
3d/2
d
d
/
p
(
(2 d)x
/
p + dm
)
]
=
e
2
(4)
d/2
1
0
dx
[
(2
d
2
)
2d/2
(2 d)x
+
(3
d
2
)
3d/2
2x(1 x)
/
p
(
(2 d)x
/
p + dm
)
]
. (14)
Then, setting
/
p = m and d = 4 with 0, we get
d(
/
p)
d
/
p
/ p=m
=
2e
2
(4)
2
1
0
dxx
[
2
+ + log 4
log
(
(1 x)
2
m
2
+ x
2
)
1
2(1 x)(2 x)m
2
(1 x)
2
m
2
+ x
2
]
(15)
Now we have found both Z
1
and Z
2
, but the relation Z
1
= Z
2
is still not manifest.
To make it more transparent, we rewrite the logarithm term as follows,
1
0
dx(1 x) log
(
(1 x)
2
m
2
+ x
2
)
=
1
0
dx(1 2x + x) log
(
(1 x)
2
m
2
+ x
2
)
=
1
0
dx
[
(1 x)
(1 x)(1 x
2
)m
2
(1 x
2
)m
2
+ x
2
xlog
(
(1 x)
2
m
2
+ x
2
)
]
. (16)
Combining this with other terms, and also using the fact
xdx =
(1 x)dx, we get
F
1
(0) =
2e
2
(4)
2
1
0
dxx
[
2
+ log 4
log
(
(1 x)
2
m
2
+ x
2
)
1
2(1 x)(2 x)m
2
(1 x)
2
m
2
+ x
2
]
. (17)
Now it is clear that Z
1
= Z
2
. Thus Z
1
= Z
2
keeps unaected at this order when
dimensional regularization is used.
(a) Momentum cut-o. Now we repeat the calculation above but use momentum
cut-o instead. We can directly borrow some results above. All we need to do is setting
d 4 and adding a UV momentum cut-o , as well as the following integral formulae:

d
4
k
(2)
4
1
(k
2
)
2
=
i
16
2
[
log
(
1 +

2
2
+
]
,
4

d
4
k
(2)
4
k
2
(k
2
)
3
=
i
16
2
[
log
(
1 +

2
)
+
(4
2
+3)
2(
2
+)
2

3
2
]
,

d
4
k
(2)
4
1
(k
2
)
3
=
i
32
2
4
(
2
+ )
2
.
We begin with (11):
F
1
(0) =2ie
2
1
0
dx
1x
0
dy

d
4
k
(2)
4
1
(k
2
)
3
[
k
2
2(x
2
4x + 1)m
2
]
=
e
2
8
2
1
0
dx(1 x)
[
log
(
1 +

2
)
+
(x
2
4x + 1)m
2

3
2
]
, (18)
In the same way, we get
i(
/
p) = 2e
2

d
4
k
(2)
4
1
0
dx
x
/
p 2m
(k
2
)
2
, (19)
and
d(
/
p)
d
/
p
/ p=m
=
e
2
8
2
1
0
dx
[
xlog
(
1 +

2
)
x +
2x(1 x)(x 2)m
2
]
. (20)
This shows that Z
1
= Z
2
with cut-o regularization.
3 Radiative corrections in QED with Yukawa inter-
action
(a) We calculate the rst order corrections to Z
1
and Z
2
, as was done in Problem 7.2.
Firstly, we compute
, which is similar to the corresponding QED correction:

u(p
(p, p
)u(p)
=(i
2
/
2)
2
d
d
k
(2)
d
i
(k p)
2
m
2
u(p
)
i
/ k +
/
q m
i
/ k m
u(p)
=
i
2
2
1
0
dx
1x
0
dy
d
d
k
(2)
d
2 u(p
)N
u(p)
(k
2
)
3
, (21)
where
k
= k xp + yq,
= (1 x)m
2
+ xm
2
x(1 x)p
2
y(1 y)q
2
+ 2xyp q,
N
= (/ k +
/
q + m)
(/ k + m).
Then we put this correction into the following form, in parallel with steps of Problem
7.2. That is: (1) replace k by k
in N
:
N
= / k
/ k
+
(
x
/
p + (1 y)
/
q + m
)
(x
/
p y
/
q + m);
(2) rewrite the numerator N
by gamma matrix relations and equations of motion, as

N
=
[
(
2
d
1
)
k
2
+ (3 + 2x x
2
)m
2
+ (y xy y
2
)q
2
]
+ (x
2
1)m(p
+ p)
;
(3) use Gordon identity to further transform N
into:
N
=
[
(
2
d
1
)
k
2
+ (x + 1)
2
m
2
+ (y y
2
xy)q
2
]
+
i
2m
2m
2
(1 x
2
).
5
Then, we can read o F
1
from the coecient of
, as:
F
1
(0) = i
2
1
0
dx
1x
0
dy
d
d
k
(2)
d
1
(k
2
)
3
[
(
2
d
1
)
k
2
+ (x + 1)
2
m
2
]
=

2
2(4)
2
1
0
dx(1 x)
[
2
+ log 4
log
(
(1 x)
2
m
2
+ xm
2
)
1 +
(x + 1)
2
m
2
(1 x)
2
m
2
+ xm
2
]
. (22)
Using the trick identity (16) again, we nally get
F
1
(0) =

2
2(4)
2
1
0
dxx
[
2
+ log 4
log
(
(1 x)
2
m
2
+ xm
2
)
+
2(1 x
2
)m
2
(1 x)
2
m
2
+ xm
2
]
. (23)
Then we compute (
/
p).
i(
/
p) = (i/
2)
2
d
d
k
(2)
d
i
/ k m
i
(p k)
2
2
=

2
2
d
d
k
(2)
d
1
0
dx
x
/
p + m
(k
2
)
2
=
i
2
2(4)
2
1
0
dx
[
2
+ log 4 log
]
(x
/
p + m), (24)
where k
= k xp and = (1 x)m
2
+ x
2
x(1 x)p
2
. Then we have
d(
/
p)
d
/
p
/ p=m
=

2
2(4)
2
1
0
dxx
[
2
+ log 4
log
(
(1 x)
2
m
2
+ xm
2
)
+
2(1 x
2
)m
2
(1 x)
2
m
2
+ xm
2
]
. (25)
Thus we have proved that Z
1
= Z
2
holds for 1-loop scalar corrections.
(b) Now consider the 1-loop corrections to Yukawa vertex. We focus on the divergent
part only. The equalities below should be understood to be hold up to a nite part.
Then, for vertex correction from photon, we have
(p, p
photon
=(ie)
2
d
d
k
(2)
d
i
(k p)
2
i
/ k +
/
q m
i
/ k m
=ie
2
1
0
dx
1x
0
dy
d
d
k
(2)
d
2dk
2
(k
2
)
3
=
d
2
e
2
2(4)
d/2
1
0
dx
1x
0
dy
(2
d
2
)
2d/2
=
4e
2
(4)
2
2
(26)
In the same way,
(p, p
scalar
=
(
i
2
)
2

d
d
k
(2)
d
i
(k p)
2
m
2
i
/ k +
/
q m
i
/ k m
=
i
2
2
1
0
dx
1x
0
dy
d
d
k
(2)
d
2k
2
(k
2
)
3
=

2
2(4)
2
2
. (27)
6
On the other hand, the 1-loop corrections for electrons self-energy also come from
two parts: one is the photon correction, which has been evaluated in Problem 7.1,
i(
/
p)
photon
=
ie
2
(4)
d/2
1
0
dx
(2
d
2
)
2d/2
[
(2 d)x
/
p + dm
]
=
ie
2
(
/
p 4m)
(4)
2
2
, (28)
and the other is the scalar correction:
i(
/
p)
scalar
=
(
i
2
)
2

d
d
k
(2)
d
i
/ k m
i
(p k)
2
m
2
=
i
2
(
/
p + 2m)
4(4)
2
2
, (29)
To sum up, we have got the total vertex correction:
(p, p
) = (p, p
photon
+ (p, p
scalar
=
4e
2
2
/2
(4)
2
2
, (30)
and also:
d(
/
p)
d
/
p
/ p=m
=
d
[
(
/
p)
photon
+ (
/
p)
scalar
]
d
/
p
/ p=m
=
e
2
+
2
/4
(4)
2
2
. (31)
7
Final Project I. Radiation of Gluon Jets
Zhong-Zhi Xianyu

In this nal project we study some basics of the process e
+
e
q q. We will focus
mainly on the radiative corrections from virtual and soft gluons.
(a) First we calculate the 1-loop vertex correction to M(e
+
e
q q) from virtual
gluon. The amplitude is given by
i
1
M= Q
f
(ie)
2
(ig)
2
v(k
2
)
u(k
1
)
i
q
2
u(p
1
)
[
d
d
k
(2)
d
i
/ k
i
/ k
/
q
i
(k p
1
)
2
2
]
v(p
2
). (1)
Now we simplify the loop integral in the standard way, as was done in Problem 7.2. The
result is
i
1
M= ig
2
[
d
d
k
(2)
d
1
0
dx
1x
0
dy
2
(
(2d)
2
d
k
2
2(1 x)(x + y)q
2
)
(k
2
)
3
]
iM
0
=
2g
2
(4)
d/2
1
0
dx
1x
0
dy
[
(2 d)
2
4
2d/2
(2
d
2
) +
(1 x)(x + y)q
2
3d/2
(3
d
2
)
]
iM
0
=
2g
2
(4)
2
1
0
dx
1x
0
dy
[
2
+ log 4 log 2 +
(1 x)(x + y)q
2
]
iM
0
,
(2)
where
iM
0
= Q
f
(ie)
2
u(p
1
)
v(p
2
)
1
q
2
v(k
2
)
u(k
1
) (3)
is the tree amplitude, and
k
= k xq yp
1
, = x(1 x y)q
2
y(1 y)p
2
1
+ y
2
. (4)
With the external legs amputated, the result is,
i
1
M=
2g
2
(4)
2
1
0
dx
1x
0
dy
[
log
(
y
2
y
2
x(1 x y)q
2
)
+
(1 x)(x + y)q
2
y
2
x(1 x y)q
2
]
iM
0
, (5)
1
Notes by Zhong-Zhi Xianyu Solution to P&S, Final Project I (draft version)
Then the cross section is given by
(e
+
e
q q) =
4
2
3s
3|F
1
(q
2
= s)|
2
, (6)
with
F
1
(q
2
= s) = Q
2
f
+
Q
2
f
g
2
1
0
dx
1x
0
dy
[
log
(
y
2
y
2
x(1 x y)s
)
+
(1 x)(x + y)s
y
2
x(1 x y)s
]
(7)
We will carry out the Feynman integration in (e).
(b) Now we simplify the 3-body phase space integral
d
3
=
d
3
k
1
d
3
k
2
d
3
k
3
(2)
9
2E
1
2E
2
2E
3
(4)
(q k
1
k
2
k
3
) (8)
in the center of mass frame. It is convenient to introduce a new set of variables x
i
=
2k
i
q/q
2
, (i = 1, 2, 3). In the center-of-mass frame, x
i
= 2E
i
/E
q
Then one can show
that all Lorentz scalars involving nal states only can be represented in terms of x
i
and
particles masses. In fact, we only need to check (k
1
+ k
2
)
2
, (k
2
+ k
3
)
2
and (k
3
+ k
1
)
2
.
From instance,
(k
1
+ k
2
)
2
= (q k
3
)
2
= q
2
+ m
2
3
2q k
3
= s(1 x
3
) + m
2
3
. (9)
Similarly,
(k
2
+ k
3
)
2
= s(1 x
1
) + m
2
1
, (k
3
+ k
1
)
2
= s(1 x
2
) + m
2
2
. (10)
To simply the phase integral, we rst integrate out k
3
with spatial delta function that
restricts k
3
= k
1
+ k
2
:
d
3
=
d
3
k
1
d
3
k
2
(2)
6
2E
1
2E
2
2E
3
(2)(E
q
E
1
E
2
E
3
). (11)
The integral measure can be rewritten as
d
3
k
1
d
3
k
2
= k
2
1
k
2
2
dk
1
dk
2
d
1
d
12
, (12)
where d
1
is the spherical integral measure associated with d
3
k
1
, and d
12
is the
spherical integral of relative angles between k
1
and k
2
. The former spherical inte-
gral can be directly carried out and results in a factor 4. To nish the integral with
d
12
= d cos
12
d
12
, we make use of the remaining delta function, which can be rewrit-
ten as
(E
q
E
1
E
2
E
3
) =
E
3
k
1
k
2
(
cos
12
E
2
3
k
2
1
k
2
2
2
2k
1
k
2
)
, (13)
by means of E
3
=
k
2
1
+ k
2
2
+ 2k
1
k
2
cos +
2
. Thus
d
1
d
12
(E
q
E
1
E
2
E
3
) =
8
2
E
3
k
1
k
2
.
Now using k
1
dk
1
= E
1
dE
1
and k
2
dk
2
= E
2
dE
2
, we have
d
3
=
dk
1
dk
2
k
2
1
k
2
2
8(2)
5
E
1
E
2
E
3
8
2
E
3
k
1
k
2
=
1
32
3
dE
1
dE
2
=
s
128
3
dx
1
dx
2
. (14)
2
To determine the integral region for m
1
= m
2
= 0 and m
3
= , we note that there are
two extremal cases: k
1
and k
2
are parallel or antiparallel. In the former case, we have
E
q
= E
1
+ E
2
+ E
3
= E
1
+ E
2
+
(E
1
+ E
2
)
2
+
2
, (15)
which yields
2E
q
(E
1
+ E
2
) = E
2
q

2
, (16)
while in the latter case,
E
q
= E
1
+ E
2
+
(E
1
E
2
)
2
+
2
, (17)
which gives
(E
q
2E
1
)(E
q
2E
2
) =
2
. (18)
These two boundary cases can be represented by x
i
variables as
x
1
+ x
2
= 1

2
s
; (19)
(1 x
1
)(1 x
2
) =

2
s
. (20)
The integral thus goes over the region bounded by these two curves.
(c) Now we calculate the dierential cross section for the process e
+
e
q gg to lowest
order in and
g
. The amplitude is
iM= Q
f
(ie)
2
(ig)
(k
3
)
u(k
1
)
[
i
/ k
1
+ / k
3
i
/ k
2
+ / k
3
]
v(k
2
)
i
q
2
v(p
2
)
u(p
1
). (21)
Then, the squared amplitude is
1
4
|iM|
2
=
Q
2
f
g
2
e
4
4s
2
(g
) tr (
/
p
1
/
p
2
)
tr
[
(
1
/ k
1
+ / k
3
1
/ k
2
+ / k
3
)
/ k
2
1
/ k
1
+ / k
3
1
/ k
2
+ / k
3
)
/ k
1
]
=
4Q
2
f
g
2
e
4
3s
2
(
8p
1
p
2
)
[
4(k
1
k
2
)(k
1
k
2
+ q k
3
)
(k
1
+ k
3
)
2
(k
2
+ k
3
)
2
+
(
1
(k
1
+ k
3
)
4
+
1
(k
2
+ k
3
)
4
)
(
2(k
1
k
3
)(k
2
k
3
)
2
(k
1
k
2
)
)
]
. (22)
We have used the trick described on Page 261 of Peskin & Schroeder when getting
through the last equal sign. Now rewrite the quantities of nal-state kinematics in
terms of x
i
, and set 0, we obtain
1
4
|iM|
2
=
2Q
2
f
g
2
e
4
3s
2
(
8p
1
p
2
)
[
2(1 x
3
)
(1 x
1
)(1 x
2
)
+
1 x
1
1 x
2
+
1 x
2
1 x
1
]
=
8Q
2
f
g
2
e
4
3s
x
2
1
+ x
2
2
(1 x
1
)(1 x
2
)
. (23)
3
Thus the dierential cross section, with 3 colors counted, reads
d
dx
1
dx
2
COM
=
1
2E
p
1
2E
p
2
|v
p
1
v
p
2
|
s
128
3
(
1
4
|M|
2
)
=
4
2
3s
3Q
2
f

g
2
x
2
1
+ x
2
2
(1 x
1
)(1 x
2
)
, (24)
where we have used the fact that the initial electron and positron are massless, which
implies that 2E
p
1
= 2E
p
2
=
s and |v
p
1
v
p
2
| = 2 in the center-of-mass frame.
(d) Now we reevaluate the averaged squared amplitude, with kept nonzero in (22).
The result is
1
4
|iM|
2
=
8Q
2
f
g
2
e
4
3s
F(x
1
, x
2
,
2
/s), (25)
where
F
(
x
1
, x
2
,

2
s
)
=
2(x
1
+ x
2
1 +

2
s
)(1 +

2
s
)
(1 x
1
)(1 x
2
)
+
[
1
(1 x
1
)
2
+
1
(1 x
2
)
2
](
(1 x
1
)(1 x
2
)

2
s
)
. (26)
The cross section, then, can be got by integrating over dx
1
dx
2
:
(e
+
e
q qg) =
1
2E
p
1
2E
p
2
|v
p
1
v
p
2
|
s
128
3
dx
1
dx
2
(
1
4
|M|
2
)
=
4
2
3s
3Q
2
f

g
2
2
s
0
dx
1
1
t
s(1x
1
)
1x
1
2
s
dx
2
F
(
x
1
, x
2
,

2
s
)
=
4
2
3s
3Q
2
f

g
2
[
log
2

2
s
+ 3 log

2
s
+ 5
1
3

2
+O(
2
)
]
. (27)
(e) It is straightforward to nish the integration over Feynman parameters in (a),
yielding
F
1
(q
2
= s) = Q
2
f

Q
2
f
g
4
[
log
2

2
s
+ 3 log

2
s
+
7
2

1
3

2
i
(
2 log

2
s
+ 7
)
+O(
2
)
]
. (28)
Then the cross section, to the order of
g
, is given by
(e
+
e
q q) =
4
2
3s
3Q
2
f
{
1

g
2
[
log
2

2
s
+ 3 log

2
s
+
7
2

1
3

2
]
+O(
2
)
}
. (29)
(f ) Combining the results in (d) and (e), we reach the nal result:
(e
+
e
q q + q qg) =
4
2
3s
3Q
2
f
[
1 +
3
g
4
]
. (30)
Note that all divergent terms as 0 cancel out in this expression. We also note that
the correct QCD correction should include an additional factor of 4/3 in the second term
in the square bracket, which comes from color indices summation.
4
Chapter 9
Zhong-Zhi Xianyu

1 Scalar QED
The Lagrangian for scalar QED reads
L =
1
4
F
+ (D
(D
) m
2
, (1)
with
F
,
D
= (
+ ieA
).
(a) Expanding the covariant derivative, its easy to nd the corresponding Feynmans
rules:
= 2ie
2
g
= ie(p
1
p
2
)
with all momenta pointing inwards.

The propagators are standard. We will work in the Feynman gauge and set = 1,
then the propagator for photon is simply
i
p
2
+ i
,
and the propagator for scalar is
i
p
2
m
2
+ i
.
1
(b) Now we calculate the spin-averaged dierential cross section for the process e
+
e
. The scattering amplitude is given by

iM= (ie)
2
v(k
2
)
u(k
1
)
i
s
(p
1
p
2
)
. (2)
Then the spin-averaged and squared amplitude is
1
4
spins
|M|
2
=
e
4
4s
2
tr
[
(
/
p
1
/
p
2
)/ k
1
(
/
p
1
/
p
2
)/ k
2
]
=
e
4
4s
2
[
8(k
1
p
1
k
1
p
2
)(k
2
p
1
k
2
p
2
) 4(k
1
k
2
)(p
1
p
2
)
2
]
. (3)
We may parameterize the momenta as
k
1
= (E, 0, 0, E), p
1
= (E, p sin , 0, p cos ),
k
2
= (E, 0, 0, E), p
2
= (E, p sin , 0, p cos ),
with p =
E
2
m
2
. Then we have
1
4
spins
|M|
2
=
e
4
p
2
2E
2
sin
2
. (4)
Thus the dierential cross section is:
(
d
d
)
CM
=
1
2(2E)
2
p
8(2)
2
E
(
1
4
|M|
2
)
=

2
8s
(
1
m
2
E
2
)
3/2
sin
2
. (5)
(c)
= 2ie
2
d
d
k
(2)
d
i
k
2
m
2
(ie)
2
d
d
k
(2)
d
(p 2k)
(p 2k)
(k
2
m
2
)((p k)
2
m
2
)
=e
2
d
d
k
(2)
d
2
(
(p k)
2
m
2
)
(p 2k)
(p 2k)
(k
2
m
2
)
(
(p k)
2
m
2
)
=e
2
d
d
k
(2)
d
1
0
dx
2
(
k
2
+ (1 x)
2
p
2
m
2
)
+ (1 2x)
2
p
+ 4k
(k
2
)
2
. =e
2
d
d
k
(2)
d
1
0
dx
2
k
2
(1
2
d
) + 2
(
(1 x)
2
p
2
m
2
)
(1 2x)
2
p
(k
2
)
2
=
ie
2
(4)
d/2
1
0
dx
[
(1
d
2
)(1
d
2
)2
2d/2
+
(2
d
2
)
2d/2
(
2
(
(1 x)
2
p
2
m
2
)
(1 2x)
2
p
)
]
=
ie
2
(4)
d/2
1
0
dx
(2
d
2
)
2d/2
[
2
(
(1 x)
2
x(1 x)
)
p
2
(1 2x)
2
p
]
. (6)
We can symmetrize the integrand as (1 x)
2
1
2
(
(1 x)
2
+ x
2
)
, then we get
=
ie
2
(4)
d/2
1
0
dx
(2
d
2
)
2d/2
(1 2x)
2
(p
2
)
. (7)
2
2 Quantum statistical mechanics
In this problem we study the path integral formulation in statistical mechanics. The
theory can be described by the partition function:
Z = tr e
H
, (8)
where H is the Hamiltonian of the system. It is a function of the generalized coordinates
q and the corresponding conjugate momentum p. In this problem, we simply assume
the Hamiltonian has the following form:
H =
p
2
2m
+ V (q). (9)
We assume the dimension of the conguration space is d, then both q and p have d com-
ponents. Then we assume the eigenstates of both q and p form a complete orthonormal
basis of the Hilbert space:
1 =
d
d
q |qq|; 1 =
d
d
p
(2)
d
|pp|. (10)
Then the partition function can be written as
Z = tr e
H
=
d
d
q q|e
H
|q. (11)
(a) Now we derive a path integral expression for the partition function. Following the
same way of deriving path integral in a quantum eld theory, we separate the quantity
e
H
into N factors:
e
H
= e
H
e
H
, (N factors),
then inserting a complete basis between each pair of adjacent factors, as
e
H
=
d
d
q
1
d
d
q
N1
q|e
H
|q
N1
q
N1
|e
H
|q
N2
q
1
|e
H
|q.
Now we focus on one factors:
q
i+1
|e
H
|q
i
= q
i+1
|e
(
1
2m
p
2
+V (q)
)
|q
i
= e
V (q
i
)
q
i+1
|e

2m
p
2
|q
i
,
and
q
i+1
|e

2m
p
2
|q
i
=
d
d
p
i+1
d
d
p
i
(2)
d
(2)
d
q
i+1
|p
i+1
p
i+1
|e
p
2
/2m
|p
i
p
i
|q
i
d
d
p
(2)
d
e
ip(q
i+1
q
i
)
e
p
2
/2m
=
[
m
2
]
d/2
e
m(q
i+1
q
i
)
2
/2
.
Inserting all this into the partition function, we get:
Z =
[
m
2
]
Nd/2
N
i=0
d
d
q
i
exp
[
m(q
i+1
q
i
)
2
2
V (q
i
)
]
, (12)
with q
N+1
= q
0
.
Now let N , then we have
Z =
Dq exp
[
dL
E
()
]
, (13)
3
where the integral measure is dened by
Dq = lim
N
[
m
2(N)
]
Nd/2
N
i=0
d
d
q
i
, (14)
and L
E
() is a Lagrangian in Euclidean form:
L
E
() =
m
2
(
dq
d
)
2
+ V (q()). (15)
Note that the periodic integral on comes from the trace in the partition function.
(b) Now we study an explicit example, a simple harmonic oscillator, which can by
dened by the Lagrangian
L
E
=
1
2
q
2
+
1
2

2
q
2
. (16)
Our task is to complete the path integral to nd a expression for the partition function
of harmonic oscillator. This can be easily done by a Fourier transformation of the
coordinates q() with respect to . Since the time direction is periodic, the Fourier
spectrum of q is discrete. That is,
q() =
1/2
n
e
2in/
q
n
, (17)
Then we have:
d L
E
() =
d
1
2
m,n
[(
2i
)
2
mn +
2
]
q
m
q
n
e
2i(m+n)/
=
1
2
m,n
[(
2i
)
2
mn +
2
]
q
m
q
n
m,n
=
1
2
nZ
[(
2
)
2
n
2
+
2
]
q
n
q
n
=
1
2
nZ
[(
2
)
2
n
2
+
2
]
|q
n
|
2
(18)
Then the path integral can be written as
Z = C
dq
0
e
2
q
2
0

n>0
dReq
n
dImq
n
exp
[

2
(
4
2
n
2
2
+
2
)
|q
n
|
2
]
=
C
n>0
[
4
2
n
2
2
+
2
]
1
=
C
n>0
[
1 +
1
(n)
2
(

2
)
2
]
1
= C sinh
1
(/2) = C
n0
exp
[
(n +
1
2
)
]
. (19)
(c) From now on we will consider the statistics of elds. We study the statistical
properties of boson system, fermion system, and photon system.
For a scalar eld, the Lagrangian is given by
L
E
() =
d
3
x
1
2
[

2
(, x) +
(
(, x)
)
2
+ m
2
2
(, x)
]
. (20)
Following the method we used to deal with the simple harmonic oscillator, here we
decompose the scalar eld (, x) into eigenmodes in momentum space:
(, x) =
1/2
n
e
2in/
d
3
k
(2)
3
e
ikx
n,k
. (21)
4
Then the Lagrangian can also be rewritten in terms of modes, as
d L
E
() =
dd
3
x
n,n
d
3
kd
3
k
(2)
6
1
2
[
(
2i
)
2
n
n k
k + m
2
]
n,k
,k
e
i2(n
+n)/+i(k+k
)x
=
1
2
d
3
k
(2)
3
[
(
2
)
2
n
2
+k
2
+ m
2
]
|
n,k
|
2
=
d
3
k
(2)
3
[
1
2

2
k
|
0,k
|
2
+
n>0
(
(
2
)
2
n
2
+
2
k
)
|
n,k
|
2
]
, (22)
where
2
k
= k
2
+ m
2
. Then the partition function, as a path integral over the eld
congurations can be represented by
Z = C

n>0,k
Re
n,k
Im
n,k
exp
[
(
(
2
)
2
n
2
+
2
k
)
|
n,k
|
2
]
(23)
By the calculation similar to that in (b), we get
Z = C
k
[
n>0
(
4
2
n
2
2
+
2
k
)]
1
= C
k
exp
[
k
(
n +
1
2
)
]
. (24)
This product gives the meaning to the formal expression
[
det(
2
+ m
2
)
]
1/2
with
proper regularization.
(d) Then consider the fermionic oscillator. The action is given by
S =
d L
E
() =
d
(
()

() +

()()
)
. (25)
The antiperiodic boundary condition ( + ) = () is crucial to expanding the
fermion into modes:
() =
1/2
nZ+1/2
e
2i/
n
. (26)
Then the partition function can be evaluated to be
Z =

n
d
n
d
n
[
nZ+1/2
n
(
2in
+
)
n
]
= C()
nZ+1/2
(
2in
+
)
= C()
n=0
(
4
2
(n +
1
2
)
2
2
+
2
)
= C() cosh
(
1
2

)
= C()
(
e
/2
+ e
/2
)
, (27)
with the form of a two-level system, as expected.
(e) Finally we consider the system of photons. The partition function is given by
Z =
DA
DbDc exp
[
dd
3
x
(
1
2
A
2
A
b
2
c
)
]
= C()
[
det()
]
4(1/2)
det(
2
), (28)
5
where the rst determinant comes from the integral over the vector eld A
while the
second one comes from the integral over the ghost elds. Therefore
Z = C()
[
det(
2
)
]
2(1/2)
, (29)
which shows the contributions from the two physical polarizations of a photon. Here we
see the eect of the ghost elds of eliminating the additional two unphysical polarizations
of a vector eld.
6
Chapter 10
Zhong-Zhi Xianyu

Draft version: March 27, 2013
1 One-Loop structure of QED
(a) In this problem we show that any photon n-point amplitude with n an odd number
vanishes.
Now we evaluate explicitly the one-point photon amplitude and three-point photon
amplitude at 1-loop level to check Furrys theorem. The one-point amplitude at 1-loop
level is simply given by
i
(1)
= (ie)
_
d
d
k
(2)
d
i tr [
(/ k + m)]
k
2
m
2
= 0, (1)
and the three-point amplitude consists of two diagrams:
i
(3)
= (ie)
3
_
d
d
k
(2)
d
(1)
_
tr
_
i
/ k m
i
/ k +
/
p
1
m
i
/ k +
/
p
1
+
/
p
2
m
_
+ tr
_
i
/ k +
/
p
1
+
/
p
2
m
i
/ k +
/
p
1
m
i
/ k m
_
_
. (2)
(b) Next we will show that the potential logarithmic divergences in photon four-point
diagrams cancel with each other. Since the divergence in this case does not depend
on external momenta, we will set all external momenta to be zero for simplicity. For
the same reason we will also set the fermions mass to be zero. Then the six diagrams
contributing the four-point amplitude can be evaluated as:
(Divergent part of i
)
=
_
d
d
k
(2)
d
1
(k
2
)
4
_
tr [
/ k
/ k
/ k
/ k] + tr [
/ k
/ k
/ k
/ k] + tr [
/ k
/ k
/ k
/ k]
+ tr [
/ k
/ k
/ k
/ k] + tr [
/ k
/ k
/ k
/ k] + tr [
/ k
/ k
/ k
/ k]
_
. (3)
Now lets focus on the rst trace, which can be worked out explicitly, to be
tr [
/ k
/ k
/ k
/ k] = 32k
8k
2
_
k
+ k
+ k
+ k
1
+ 4(k
2
)
2
(g
+ g
+ g
). (4)
Then, we symmetrize the momentum factors according to k
k
2
g
/4 and k
(k
2
)
2
(g
+g
+g
)/24. (Since the divergence can be at most logarithmic,

so it is safe to set spacetime dimension d = 4 at this stage.) Then the rst trace term
reduces to,
tr [
/ k
/ k
/ k
/ k]
4
3
(k
2
)
2
(g
2g
+ g
). (5)
The other ve terms can be easily got by permuting indices. Then it is straightforward
to see that the six terms sum to zero.
2 Renormalization of Yukawa theory
In this problem we study the pseudoscalar Yukawa Lagrangian,
L =
1
2
(
)
2
1
2
m
2
2
+

(i/ M) ig

5
, (6)
where is a real scalar and is a Dirac Fermion.
(a) Lets gure out how the supercial degree of divergence D depends on the number
of external lines. From power counting, its easy to see that D can be represented by
D = 4L P
f
2P
s
, (7)
where L is the no. of loops, P
f
is the no. of internal fermion lines, and P
s
is the no. of
internal scalar lines. We also note the following simple relations:
L = P
f
+ P
s
V + 1 ,
2V = 2P
f
+ N
f
,
V = 2P
s
+ N
s
.
Then we can deduce
D = 4L P
f
2P
s
= 4(P
f
+ P
s
V + 1) P
f
2P
s
= 4
3
2
N
f
N
s
. (8)
Guided by this result, we can nd all divergent amplitudes as follows.
D = 2 D = 1
D = 0 D = 0
We note that we have ignored the vacuum diagram, which simply contributes an in-
nitely large constant, the potentially divergent diagrams with odd number of external
scalars are also ignored, since they actually vanish. This result shows that the original
theory cannot be renormalized unless we including a new
4
interaction, as
L =
4!
4
. (9)
2
(b) Now let us evaluate the divergent parts of all 1-loop diagrams of Yukawa theory.
First we consider the two point function of scalar. The one-loop contribution to this
amplitude is shown as follows.
+ +
The d = 4 pole of rst two loop diagrams can be determined as
=
i
2
_
d
d
k
(2)
d
i
k
2
m
2

im
2
(4)
2
1
. (10)
= (ig)
2
_
d
d
k
(2)
d
tr
_
i
/ k M

5
i
(/ k
/
p) M

5
_
4ig
2
(p
2
2M
2
)
(4)
2
1
.
(11)
From this we nd the divergent part of the counterterm to be
m

(m
2
8g
2
M
2
)
(4)
2
1
=
4g
2
(4)
2
1
. (12)
Then we come to the two point function of fermion, the 1-loop correction of which is
given by the following two diagrams.
+
From the pole of the loop diagram
= g
2
_
d
d
k
(2)
d
5
i
/ k M

5
i
(k p)
2
m
2

ig
2
(
/
p 2M)
(4)
2
1
, (13)
we nd the following counterterms:
M

2g
2
M
(4)
2
1

g
2
(4)
2
1
. (14)
The following two diagrams contribute to 1-loop corrections to Yukawa coupling and
4
coupling, respectively.
Since the divergent part of diagram is independent of external momenta, we can set all
these momenta to be zero. Then the loop diagram is
= g
3
_
d
d
k
(2)
d
5
i
/ k M

5
i
/ k M

5
i
k
2
m
2

g
3
5
(4)
2
2
(15)
3
=
(i)
2
2
_
d
d
k
(2)
d
_
i
k
2
m
2
_
2
i
2
(4)
2
1
. (16)
= (1)g
4
_
d
d
k
(2)
d
tr
__
5
i
/ k M
_
4
_

8ig
4
(4)
2
1
. (17)
Note that there are 3 permutations for the rst diagram and 6 permutations for the
second diagram. Then we can determine the divergent part of counterterm to be
g

2g
3
(4)
2
1

3
2
48g
4
(4)
2
1
. (18)
3 Field-strength renormalization in
4
theory
In this problem we evaluate the two-loop corrections to scalars two-point function
in
4
theory in the massless limit. There are three diagrams contribute in total.
The rst diagram reads
=
(i)
2
6
_
d
d
k
(2)
d
d
d
q
(2)
d
i
k
2
m
2
i
q
2
m
2
i
(p k q)
2
m
2
=
i
2
6
_
d
d
k
E
(2)
d
d
d
q
E
(2)
d
i
k
2
E
+ m
2
i
q
2
E
+ m
2
i
(p
E
k
E
q
E
)
2
+ m
2
=
i
2
6
_
d
d
k
E
(2)
d
1
k
2
E
+ m
2
1
(4)
d/2
_
1
0
dx
(2
d
2
)
[m
2
+ x(1 x)(p
E
k
E
)
2
]
2d/2
=
i
2
(2
d
2
)
6(4)
d/2
_
1
0
dxdy
_
d
d
k
E
(2)
d
[x(1 x)]
d/22
(1 y)
1d/2
(3
d
2
)/(2
d
2
)
_
(k
E
yp
E
)
2
+ y(1 y)p
2
E
+
_
1 y +
y
x(1x)
_
m
2
3d/2
=
i
2
6(4)
d
_
1
0
dxdy
(3 d)[x(1 x)]
d/22
(1 y)
1d/2
_
y(1 y)p
2
E
+
_
1 y +
y
x(1x)
_
m
2
3d
. (19)
Now we take m
2
= 0 and d = 4 4. Then we have
=
i
2
12(4)
4
(1 + )(p
2
E
)
1
+ =
i
2
12(4)
4
p
2
E
_
1
log(p
2
E
) +
_
=
i
2
12(4)
4
p
2
_
1
log(p
2
) +
_
. (20)
The second diagram actually vanishes in m 0 limit. In fact,
=
i
2
_
d
d
k
(2)
d
i
k
2
m
2
=
i
2(4)
d/2
(1
d
2
)
m
1d/2
m 0. (21)
4
The third diagram reads ip
2
Z
. Therefore we can choose the counterterm
Z
, under the
MS scheme, to be
Z
=

2
12(4)
4
_
1
log M
2
_
. (22)
Thus the eld strength counterterm receives a nonzero contribution at this order. In
the massless limit, it is
(2)
=
i
2
12(4)
4
p
2
log
M
2
p
2
. (23)
4 Asymptotic behavior of diagrams in
4
theory
In this problem we calculate the four point amplitude in
4
theory to 2-loop order
in s , t xed, limit. The tree level result is simply i, and the 1-loop result can
be easily evaluated to be
i
1
M=
(i)
2
2
_
d
d
k
(2)
d
i
k
2
m
2
_
i
(p
s
k)
2
m
2
+
i
(p
t
k)
2
m
2
+
i
(p
u
k)
2
m
2
_
i
i
2
2(4)
2
_
3
_
2
+ log 4
_
log s log t log u
_
i
=
i
2
2(4)
2
_
log s + log t + log u
_

i
2
(4)
2
log s (24)
In the last step we take the limit s . In this limit t can be ignored and u s.
We see the divergent part of the counterterm coecient
at 1-loop order is

3
2
(4)
2
1
. (25)
Now we consider the two-loop correction.
=
(i)
3
4
_ _
d
d
k
(2)
d
i
k
2
i
(p
s
k)
2
_
2
=
i
3
4(4)
d
_ _
1
0
dx
(2
d
2
)
[x(1 x)s]
2d/2
_
2
i
3
(4)
4
_
1
2

1
log s +
1
2
log
2
s
_
. (26)
In the last line we only keep the divergent terms as 0 and s .
=
(i)
3
2
_
d
d
kd
d
q
(2)
2d
i
k
2
i
(p
s
k)
2
i
q
2
i
(k p
3
q)
2
=
i
3
2
_
d
d
k
(2)
d
1
k
2
(p
s
k)
2
_
i
(4)
d/2
_
1
0
dx
(2
d
2
)
[x(1 x)(k
E
p
3E
)
2
]
2d/2
_
5
=
i
3
2(4)
d/2
_
1
0
dx
(2
d
2
)
[x(1 x)]
2d/2
_
d
d
k
E
(2)
d
1
k
2
E
(p
sE
k
E
)
2
[(k
E
p
3E
)
2
]
2d/2
=
i
3
2(4)
d/2
_
1
0
dx
(2
d
2
)
[x(1 x)]
2d/2
_
d
d
k
E
(2)
d
_
1
0
dy
_
1y
0
dz
z
1d/2
(k
2
E
+ )
4d/2
(4
d
2
)
(2
d
2
)
=
i
3
2(4)
d
_
dxdydz
z
1d/2
[x(1 x)]
2d/2
(4 d)
4d
, (27)
where = ys + zp
2
3E
(yp
sE
+ zp
3E
)
2
.
Then we nd

i
3
(4)
4
_
1
2

1
log s +
1
2
log
2
s
_
. (28)
The same result for the third diagram. Then we have
+ +
i
3
(4)
4
_
3
2

3
log s +
3
2
log
2
s
_
. (29)
Now we come to the counterterm. The fourth diagram reads
=
(i)(i
)
2
_
d
d
k
(2)
d
i
k
2
i
(p
s
k)
2
3
3
2(4)
2
1
i
(4)
2
2
_
1

2
log s +

2
8
log
2
s +
_
3i
3
(4)
4
_
1
2

1
2
log s +
1
8
log
2
s
_
(30)
The same result for the fth diagram. Then we have
+ + + +
i
3
(4)
4
_
3
2

3
4
log
2
s
_
. (31)
So much for the s-channel. The t and u-channel results can be obtained by replacing s
with t and u respectively. In the limit s and t-xed, we can simply ignore t and
treating u s, then the total 2-loop correction in this limit is
i
2
M
3i
3
2(4)
4
log
2
s. (32)
The double pole 1/
2
has been absorbed by
.
In summary, we have the following asymptotic expression for the 4-point amplitude
to 2-loop order in the s and t-xed limit:
iM= i
i
2
(4)
2
log s
3i
3
2(4)
4
log
2
s + . (33)
6
Chapter 11
Zhong-Zhi Xianyu

1 Spin-wave theory
(a) Firstly we prove the following formula:
Te
i(x)
e
i(0)
= e
[D(x)D(0)]
. (1)
Where D(x) = T(x)(0) is the time-ordered correlation of two scalars. The left hand
side of this equation can be represented by path integral, as
1
Z[0]
De
i(x)
e
i(0)
exp
[
i
d
d
xd
d
y
1
2
(x)D
1
(x y)(y)
]
. (2)
This expression precisely has the form Z[J]/Z[0], with J(y) = (y x) (0). Thus we
have
Z[J]/Z[0] =
1
2
d
d
xd
d
y J(x)D(x y)J(y) = exp
[
D(x) D(0)
]
, (3)
which is just the right hand side of the formula.
(b) The operator being translational invariant O[(x)] = O[(x) ] can depend
on only through
. And the only relevant/marginal Lorentz-invariant operator

satisfying this condition is
1
2
()
2
.
(c) From now on we use bold x to denote coordinate and italic x to denote its length,
x |x|. We can use the result in (a) to evaluate s(x)s
(0), as
s(x)s
(0) = A
2
e
i(x)
e
i(0)
= A
2
e
D(x)D(0)
. (4)
Note that the correlation function
D(x) =
1
d
d
k
E
(2)
d
1
k
2
E
e
ik
E
x
(5)
is the solution to the following equation:
2
D(x y) =
(d)
(x y). (6)
1
Since D(x) is a function of the length only, namely D(x) = D(x), thus we have

x
d1
x
(
x
d1

x
D(x)
)
=
(1 +
d
2
)
d
d/2
(x)
x
d1
. (7)
Then its easy to nd
D(x) =
(1 +
d
2
)
d(d 2)
d/2
1
x
d2
, for d = 2,
1
2
log x, for d = 2.
(8)
Then we have
Dimension d d = 1 d = 2 d = 3 d = 4
D(x)
1
2
x
1
2
log x
1
4x
1
4
2
x
2
ss
e
x
1/x
2
e
1/x
e
1/x
2
Since 0 when d 2, the correlation function ss
in this case is independent of

length x.
2 A zeroth-order natural relation
We study N = 2 linear sigma model coupled to fermions:
L =
1
2

i
+
1
2

2
1
4
(
i
i
)
2
+

(i/ ) g

(
1
+ i
5
2
), (9)
with
i
a two-component eld, i = 1, 2.
(a) Now, under the following transformation:
1
cos
2
sin ;
2
1
sin +
2
cos ; e
i
2
/2
, (10)
the rst three terms involving
i
only keep invariant. The fourth term, as the kinetic
term of a chiral fermion, is also unaected by this transformation. Thus, to show the
whole Lagrangian is invariant, we only need to check the last term, and this is really
the case:
g

(
1
+ i
5
2
)
g

e
i
5
/2
[
(
1
cos
2
sin ) + i
5
(
1
sin +
2
cos )
]
e
i
5
/2
=g

e
i
5
/2
e
i
5
(
1
+ i
5
2
)e
i
5
/2
= g

(
1
+ i
5
2
). (11)
(b) Now let acquire a vacuum expectation value v, which equals to
2
/ classically.
Then, in terms of new variables = (v + (x), (x)), the Lagrangian reads
L =
1
2
(
)
2
+
1
2
(
)
2
1
4
(
4
+
4
)
1
2

2
2
v
3
v
2
+

(i/ gv) g

( + i
5
). (12)
That is, the fermion acquire a mass m
f
= gv.
2
(c) Now we calculate the radiative corrections to the mass relation m
f
= gv. The
renormalization conditions we need are as follows.
q
p
p
= g
5
at q
2
= 0, p
2
= p
2
= m
2
f
. (13)
= 0. (14)
These two conditions xed g and v so that they receive no radiative corrections. Then
we want to show that the mass of the fermion m
f
receives nite radiative correction at
1-loop. Since the tadpole diagrams of sum to zero by the renormalization condition
above, the fermions self-energy receive nonzero contributions from the following three
diagrams:
The rst two 1-loop diagrams can be evaluated as
(e) = (ig)
2
d
d
k
(2)
d
i
/ k m
f
i
(k p)
2
2
2
= g
2
d
d
k
(2)
d
1
0
dx
x
/
p + m
f
(k
2
1
)
2
=
ig
2
(4)
d/2
1
0
dx
(2
d
2
)
2d/2
1
(x
/
p + m
f
)
=
ig
2
(4)
2
1
0
dx(x
/
p + m
f
)
[
2
+ log 4 log
1
]
(15)
(f) = g
2
d
d
k
(2)
d
5
i
/ k m
f
5
i
(k p)
2
= g
2
d
d
k
(2)
d
1
0
dx
x
/
p m
f
(k
2
2
)
2
=
ig
2
(4)
d/2
1
0
dx
(2
d
2
)
2d/2
2
(x
/
p m
f
)
=
ig
2
(4)
2
1
0
dx(x
/
p m
f
)
[
2
+ log 4 log
2
]
. (16)
This leads to
(e) + (f) =
ig
2
(4)
2
1
0
dx
{
2x
/
p
[
2
+ log 4
1
2
log(
1
2
)
]
+ m
f
log

2
1
}
(17)
We see that the correction to the fermions mass m
f
from these two diagrams is nite.
Besides, the third diagram, namely the counterterm, contributes the mass correction
through
g
v. The the total correction to m
f
is nite only when
g
is nite. Let us
check this by means of the rst renormalization condition (13) stated above. The 1-loop
contributions to (13) are as follows.
3
(a) = (ig)
2
g
d
d
k
(2)
d
i
/ k m
f
5
i
/ k m
f
i
(k p)
2
2
2
= ig
3
d
d
k
(2)
d
(/ k + m
f
)
5
(/ k + m
f
)
(k
2
m
2
f
)
2
(
(k p)
2
2
2
) = ig
3
d
d
k
(2)
d
1
(k
2
m
2
f
)
(
(k p)
2
2
2
)
=ig
3
d
d
k
(2)
d
1
0
dx
1
(k
2
1
)
2
=
g
3
5
(4)
d/2
1
0
dx
(2
d
2
)
2d/2
1
=
g
3
5
(4)
2
1
0
dx
[
2
+ log 4 log
1
]
(18)
(b) = g
3
d
d
k
(2)
d

5
i
/ k m
f
5
i
/ k m
f
5
i
(k p)
2
2
2
=ig
3
d
d
k
(2)
d
5
(/ k + m
f
)
5
(/ k + m
f
)
5
(k
2
m
2
f
)
2
(
(k p)
2
2
2
) = ig
3
d
d
k
(2)
d
1
(k
2
m
2
f
)
(
(k p)
2
2
2
)
= ig
3
d
d
k
(2)
d
1
0
dx
1
(k
2
2
)
2
=
g
3
5
(4)
d/2
1
0
dx
(2
d
2
)
2d/2
2
=
g
3
5
(4)
2
1
0
dx
[
2
+ log 4 log
2
]
(19)
(c) = (ig)g(2iv)
d
d
k
(2)
d
5
i
/ k m
f
i
(k p)
2
2
2
i
(k p)
2
= 4ig
2
v
5
d
d
k
(2)
d
1
0
dx
1x
0
dy
(x + y)
/
p + m
f
(k
2
3
)
3
=
2g
2
v
5
(4)
2
1
0
dx
1x
0
dy
(x + y)
/
p + m
f
3
(20)
(d) = (ig)g(2iv)
d
d
k
(2)
d
i
/ k m
f
5
i
(k p)
2
2
2
i
(k p)
2
= 4ig
2
v
5
d
d
k
(2)
d
1
0
dx
1x
0
dy
(x + y)
/
p + m
f
(k
2
3
)
3
=
2g
2
v
5
(4)
2
1
0
dx
1x
0
dy
(x + y)
/
p + m
f
3
(21)
Thus,
(a) + (b) + (c) + (d) =
g
5
(4)
2
1
0
dx
[
g
2
log

2
1
+ 4
1x
0
dy
m
2
f
3
]
. (22)
3 The Gross-Neveu model
The Gross-Neveu Model is a theory of fermions in 1 + 1 dimensional spacetime:
L =

i
i/
i
+
1
2
g
2
(
i
)
2
, (23)
4
with i = 1, , N. The gamma matrices are taken as
0
=
2
,
1
= i
1
, where
i
is
the familiar Pauli matrices. We also dene
5
=
0
1
=
3
.
(a) The theory is invariant under the transformation
i

5
i
. It is straightforward
to check this. We note that:
i
=
0
=
5
, (24)
thus:
L

5
i/
5
i
+
1
2
g
2
(
i
)
2
=

i
i/
i
+
1
2
g
2
(
i
)
2
. (25)
However, a mass term will transform as m
i

i
m
i

i
, thus a theory respecting
this chiral symmetry does not allow such a mass term.
(b) The supercial renormalizability of the theory (by power counting) is obvious since
[g] = 0.
(c) The model can be phrased in another equivalent way:
Z =
DD exp
[
i
d
2
x
(
i
i/
i
1
2g
2
i
)
]
. (26)
This can be justied by integrating out ,
D exp
[
i
d
2
x(
1
2g
2
i
)
]
= N exp
[
i
d
2
x
g
2
2
(
i
)
2
]
. (27)
which recovers the following path integral:
Z =
D exp
[
i
d
2
x
(
i
i/
i
+
1
2
g
2
(
i
)
2
)
]
. (28)
(d) We can also integrate out the fermions
i
to get the eective potential for the
auxiliary eld :
D exp
[
i
d
2
x
(
i
i/
i
i
)
]
=
[
det(i/ )
]
N
=
[
det(
2
+
2
)
]
N
= exp
[
d
2
k
(2)
2
N log(k
2
+
2
)
]
. (29)
The integral is divergent, which should be regularized. We use the dimensional regular-
ization:
d
d
k
E
(2)
d
N log(k
2
E
+
2
) = N
d
d
k
E
(2)
d
[

1
k
2
E
+
2
]
=0
=iN
(d/2)(
2
)
d/2
(4)
d/2
. (30)
Now we set d = 2 and send 0,
d
d
k
E
(2)
d
N log(k
2
E
+
2
) =
iN
2
4
(
2
+ log 4 log
2
+ 1
)
. (31)
Thus the eective potential is
V
e
() =
1
2g
2
2
+
N
4

2
(
log

2
2
1
)
(32)
by modied minimal subtraction.
5
(e) Now we minimize the eective potential:
0 =
V
e
=
1
g
2
+
N
2
log

2
2
, (33)
and nd nonzero vacuum expectation values = e
/g
2
N
. The dependence of this
result on the renormalization condition is totally in the dependence on the subtraction
point .
(f ) There is a well-known result that the loop expansion is equivalent to the expansion
in powers of . This is true because the integrand of the partition function can be put
into the form of e
iS/
. That is, appears as an overall coecient of the action. In our
case, we see that the overall factor N plays the same role. Thus by the same argument,
we conclude that the loop expansion is equivalent to the expansion in powers of 1/N.
6
Chapter 12
Zhong-Zhi Xianyu

1 Beta Function in Yukawa Theory
In this problem we calculate the 1-loop beta functions in Yukawa theory. All needed
ingredients have been given in Problem 10.2 Here we list the needed counterterms:
=
g
2
2(4
2
)
(
2
log M
2
)
; (1)
=
2g
2
(4
2
)
(
2
log M
2
)
; (2)
g
=
g
3
(4)
2
(
2
log M
2
)
; (3)
=
3
2
48g
4
2(4)
2
(
2
log M
2
)
. (4)
Here is the UV cuto and M is the renormalization scale. Then, the beta functions
to lowest order are given by
g
= M

M
(
g
+
1
2
g
0
+ g
0
)
=
5g
3
(4)
2
; (5)
= M

M
(
+ 2
0
)
=
3
2
+ 8g
2
48g
4
(4)
2
. (6)
2 Beta Function of the Gross-Neveu Model
We evaluate the function of the 2-dimensional Gross-Neveu model with the La-
grangian
L =

i
(i/ )
i
+
1
2
g
2
(
i
)
2
, (i = 1, , N) (7)
to 1-loop order. The Feynman rules can be easily worked out to be
k
i j =
(
i
/ k
)
ij
1
i j
k l
= ig
2
(
ij
+
i
jk
)
Now consider the two-point function
(2)
ij
(p). The one-loop correction to
(2)
ij
(p) comes
from the following two diagrams:
i j + i j
It is easy to see the loop diagram contains a factor of
d
2
k tr [/ k
1
], which is zero under
dimensional regularization. Thus the wave function renormalization factor receives no
contribution at 1-loop level, namely
= 0.
Then we turn to the 4-point function
(4)
ijk
. There are three diagrams in total,
namely,
(a)
n m
i j
k
+
(b)
n
m
i
k
j
+
(c)
m
n
i
j
k
We calculate them in turn. The rst one:
(a) = (ig
2
)
2
d
d
k
(2)
d
(
mn
+
n
mk
)
(
i
/ k
)
(
ij
mn
+
in
jm
)
(
i
/ k
)
= g
4
(
(2N + 2)
ij
+
1
2

i
jn
(
d
d
k
(2)
d
1
k
2
(8)
The second diagram reads:
(b) =
1
2
(ig
2
)
2
d
d
k
(2)
d
(
mj
+
m
nj
)
(
i
/ k
)
(
im
kn
+
in
km
)
(
i
/ k
)
=
g
4
2
(
ij
k
(
+
i
jn
(
d
d
k
(2)
d
1
k
2
(9)
The third diagram:
(c) = (ig
2
)
2
d
d
k
(2)
d
(
mj
nk
+
mn
jk
)
(
i
/ k
)
(
im
+
i
mn
)
(
i
/ k
)
=g
4
(
1
2

ij
k
(
+ (2 2N)
i
jk
d
d
k
(2)
d
1
k
2
(10)
2
Summing up the three diagrams and using dimensional regularization with d = 2 ,
we get
2g
4
(N 1)(
ij
+
i
jk
d
d
k
(2)
d
1
k
2
2(N 1)ig
4
4
2
(
ij
+
i
jk
). (11)
Only the divergent terms are kept in the last expression, from which we can read the
counterterm
g
=
(N 1)g
4
2
(
2
log M
2
)
. (12)
Thus the function is
(g
2
) = M

M
(
g
) =
(N 1)(g
2
)
2
, (13)
and
(g) =
(N 1)g
3
2
. (14)
It is interesting to see that the 1-loop function vanishes for N = 1. This is because
we have the Fierz identity 2(
)(
) = (
)(
), and the Gross-Neveu model

in this case is equivalent to massless Thirring model, which is known to have vanishing
function.
3 Asymptotic Symmetry
In this problem we study a bi-scalar model, given by the following Lagrangian:
L =
1
2
(
(
1
)
2
+ (
2
)
2
)

4!
(
4
1
+
4
2
)

12
2
1
2
2
. (15)
(a) First, we calculate the 1-loop beta functions
and
. The relevant 1-loop

diagrams for calculating
are:
The relevant diagrams for calculating
are:
Here the single line represents
1
and double line represents
2
. Since the divergent
parts of these diagrams are all independent of external momenta, we can therefore
simply ignore them. Then its easy to evaluate them, as follows.
=
(i)
2
2
d
4
k
(2)
4
i
k
2
i
k
2

i
2
2(4)
2
2
, (16)
3
=
(i/3)
2
2
d
4
k
(2)
4
i
k
2
i
k
2

i
2
18(4)
2
2
. (17)
The t-channel and u-channel give the same result. Thus we can determine
to be

9
2
+
2
6(4)
2
2
. (18)
On the other hand,
= =
(i)(i/3)
2
d
4
k
(2)
4
i
k
2
i
k
2

i
6(4)
2
2
, (19)
= = (i/3)
2
d
4
k
(2)
4
i
k
2
i
k
2

i
2
9(4)
2
2
. (20)
Then we have

3 + 2
2
3(4
2
)
2
. (21)
Its easy to see that eld strengths for both
1
and
2
receives no contributions from
1-loop diagrams. Thus the 1-loop beta functions can be evaluated as
=
d
d
=
3
2
+
2
/3
(4)
2
; (22)
=
d
d
=
2 + 4
2
/3
(4)
2
. (23)
(b) Now we derive the renormalization equation for /:
d
d
(

)
=
1
=

3(4)
2
[
(/)
2
+ 4(/) 3
]
. (24)
Then it is easy to see that / = 1 is an IR xed point.
(c) In 4 dimensions, the functions for and are shifted as
= +
3
2
+
2
/3
(4)
2
; (25)
= +
2 + 4
2
/3
(4)
2
. (26)
But it is easy to show that the terms containing cancel out in the function for /,
and the result is the same as (24). This is true because / still remains dimensionless
in 4 dimensions. Therefore we conclude that there are three xed points of the RG
ow for / at 0, 1, and 3. We illustrate this in the diagram of RG ow in the - plane,
with the deviation of dimension = 0.01, in Figure 1.
4
0.0 0.2 0.4 0.6 0.8 1.0
0.0
0.2
0.4
0.6
0.8
1.0
Figure 1: The RG ow of the theory (15) in 4 dimensions with = 0.01. Three

nontrivial xed points are shown by blue dots.
5
Chapter 13
Zhong-Zhi Xianyu

1 Correlation-to-scaling exponent
In this problem we consider the eect of the deviation of the coupling from its xed
point to the two-point correlation function G(M, t) in d = 4 dimensions. Symbolically,
we can always write
G(M, t) = G
(M, t) +
G(M, t)
, (1)
where

is the running coupling, dened to be the solution of the following renormal-
ization group equation:
d
d log
=
2
)
d 2 + 2(
)
. (2)
As the rst step, let us expand the function of

around the xed point, as
(
) = (
) +
d(
)
d
) +O((
)
2
)
= (
) +O((
)
2
). (3)
Then the renormalization group equation reads
d
d log

2(
)
d 2 + 2(
)
=

), (4)
where and on the right hand side are critical exponents, which in our case are dened
to be
=
d 2 + 2(
)
d(2
))
, =
1
2
2(
)
.
Dont confuse the critical exponent with the function. Now, from this equation we
can solve the running coupling

to be
+
(
(
0
)
)
(

0
)
/
. (5)
Now let
0
be the scale at which the bare coupling is dened. Then we get
)
/
. (6)
1
2 The exponent
We have found the counterterm
Z
to O(
2
) with MS scheme in Problem 10.3, to
be
Z
=

2
12(4)
4
[
1
log M
2
]
. (7)
Then the anomalous dimension to O(
2
) is given by
=
1
2
M

M

Z
=

2
12(4)
4
. (8)
This result can be easily generalized to the O(N)-symmetric
4
theory, by replacing the
Feynman rule of the
4
coupling i with
2i(
ij
k
+
ik
j
+
i
jk
),
which is equivalent to multiplying the diagram (??) by the following factor:
4
(
ik
m
+
i
km
+
im
k
)(
jk
m
+
j
km
+
jm
k
)
= 12(N + 2)
ij
, (9)
and the anomalous dimension (8) obtained above should be multiplied by 12(N + 2),
which leads to
= (N + 2)

2
(4)
4
, (10)
which is the same as (13.47) of Peskin&Schroeder.
3 The CP
N
model
(a) The Lagrangian of the CP
N
model can be written as
L =
1
g
2
(
j
|
z
j
|
2
j
z
z
j
2
)
, (11)
with z
j
(j = 1, N + 1) the components of a vector in (N + 1) dimensional complex
space, subject to the constraint
j
|z
j
|
2
= 1 (12)
and the identication
(e
i
z
1
, , e
i
z
N+1
) (z
1
, , z
N+1
). (13)
Now we prove that the Lagrangian given above is invariant under the following local
transformation:
z
j
(x) e
i(x)
z
j
(x), (14)
as,
g
2
L
(e
i
z
j
)
2
+
e
i
z
(e
i
z
j
)
2
=
(
|
z
j
|
2
+|
|
2
+ 2Re
(
i(
)z
z
j
)
)
(
|z
z
j
|
2
+|
|
2
+ 2Re
(
i(
)z
i
z
i
z
z
j
)
)
= g
2
L. (15)
2
Then we show that the nonlinear model with n = 3 is equivalent to the CP
N
model with N = 1. To see this, we substitute n
i
= z
i
z into the Lagrangian of the
nonlinear sigma model, L =
1
2g
2
|
n
i
|
2
, to get
L =
1
2g
2
)
i
z +z
2
=
1
2g
2
i
[
2(
)(
z)z
z + (
z)
2
z
2
+ (
)
2
z
2
]
=
1
2g
2
i
[
2(
)(
z) +
(
z
z +z
)
2
2(z
z)(z
)
]
=
1
2g
2
i
[
2(
)(
z) +
[
(z
z)]
2
2(z
z)(z
)
]
. (16)
Then after a proper normalization of the eld z, it is straightforward to see that the
Lagrangian above reduces to
L =
1
g
2
(
|
z|
2
2|z
z|
2
)
, (17)
which is indeed the CP
1
model.
(b) The Lagrangian (11) can be obtained by the following Lagrangian with a gauge
eld A
and a Lagrange multiplier which expresses the local gauge symmetry and the
constraint explicitly:
L =
1
g
2
(
|D
z
j
|
2
(
|z
j
|
2
1
)
)
, (18)
with D
+ iA
. Now let us verify this by functionally integrating out the gauge

eld A
as well as the Lagrange multiplier to get

Z =
D
2
z
i
DA
D exp
[
i
g
2
d
2
x
(
|D
z
j
|
2
(
|z
j
|
2
1
)
)
]
=
D
2
z
i
DA
(
|z
j
|
2
1
)
exp
[
i
g
2
d
2
x|D
z
j
|
2
]
=
D
2
z
i
DA
(
|z
j
|
2
1
)
exp
[
i
g
2
d
2
x
(
A
+ 2iA
j
)z
j
+|
z
j
|
2
)
]
=N
D
2
z
i
(
|z
j
|
2
1
)
exp
[
i
g
2
d
2
x
(
|
z
j
|
2
|z
z
j
|
)
2
]
. (19)
(c) On the other hand one can also integrate out z
i
eld in the Lagrangian (18), as
Z =
Dz
i
DA
D exp
[
i
g
2
d
2
x
(
|D
z
j
|
2
(
|z
j
|
2
1
)
)
]
=
DA
D exp
[
N tr log(D
2
) +
i
g
2
d
2
x
]
(20)
We assume that the expectation values for A
and are constants. Then the exponent

can be evaluated by means of dimensional regularization, as
iS = N tr log(D
2
) +
i
g
2
d
2
x
=
[
N
d
d
k
(2)
d
log
(
k
2
+A
)
+
i
g
2
]
V
(2)
3
i
[
N
4
(
log
M
2
A
2
+ 1
)
( A
2
) +
1
g
2
]
V
(2)
, (21)
where V
(2)
=
d
2
x, and we have used the MS scheme to subtract the divergence. Now
we can minimize the quantity in the square bracket in the last line to get
A
= 0, = M
2
exp
(
4
gN
2
)
. (22)
(d) The meaning of the eective action S is most easily seen from its diagrammatic
representations. For instance, at the 1-loop level, we know that the logarithmic terms in
the eective action is simply the sum of a series of 1-loop diagrams with n 0 external
legs, where the number of external legs n is simply the power of corresponding elds in
the expansion of S. Therefore, to the second order in A and in , the eective action is
represented precisely by the following set of diagrams,
where the dashed lines represent , curved lines represent A
, and the internal loop are

z eld. Then it is straightforward to see that the correct kinetic terms for and A
are generated from these diagrams. That is, the gauge eld A
becomes dynamical due

to quantum corrections. The gauge invariance of the resulted kinetic term F
can
also be justied by explicit calculation as was done in Problem 9.1.
4
Final Project II. The Coleman-Weinberg Potential
Zhong-Zhi Xianyu

In this nal project, we work out some properties of Coleman-Weinberg model,
illustrating some basic techniques of the renormalization group. The Coleman-Weinberg
model is simply a theory of scalar electrodynamics, described by the Lagrangian
L =
1
2
F
+ (D
(D
) m
2

6
(
)
2
, (1)
with a complex scalar and D
= (
+ ieA
).
(a) Consider the case of spontaneous breaking of the U(1) gauge symmetry (x)
e
i(x)
(x), caused by a negative squared mass, namely m
2
=
2
< 0. The scalar then
acquires a nonzero vacuum expectation value (VEV)
0
=
||
2
. We split this VEV
out of the scalar eld, namely
=
0
+
1
2
[
(x) + i(x)
]
, (2)
with the new eld (x) and (x) being real. At the tree level, it is easy to nd
0
=
3
2
/ by minimize the scalar potential V () =
2
+

6
(
)
2
. We also introduce
v =
2
0
for convenience. Then, rewrite the Lagrangian in terms of these new eld
variables, we get
L =
1
4
(F
)
2
+
1
2
(
)
2
+
1
2
(
)
2
+
1
2
e
2
v
2
A
1
2
(2
2
)
2

24
(
4
+
4
+ 2
2
2
+ 4v
2
+ 4v
3
) +evA
+eA
) +
1
2
e
2
A
(
2
+
2
+ 2v). (3)
Then we see that the vector eld A
acquires a mass, equal to m

A
= ev at the classical
level.
(b) Now we calculate the 1-loop eective potential of the model. We know that 1-loop
correction of the eective Lagrangian is given by
L =
i
2
log det
[

2
L
]
=0
+L, (4)
where is the uctuating elds and L denotes counterterms.
1
Notes by Zhong-Zhi Xianyu Solution to P&S, Final Project II (draft version)
Let the background value of the complex scalar be
cl
. By the assumption of Poincare
symmetry,
cl
must be a constant. For the same reason, the background value of the
vector eld A
must vanish. In addition, we can set

cl
to be real without loss of
generality. Then we have
(x) =
cl
+
1
(x) + i
2
(x),
where
1
(x),
2
(x), together with A
(x), now serve as uctuating elds. Expanding

the Lagrangian around the background elds and keeping terms quadratic in uctuating
elds only, we get
L =
1
2
F
+ ieA
)(
cl
+
1
+ i
2
)
2
m
2
cl
+
1
+ i
2
cl
+
1
+ i
2
4
=
1
2
A
[
g
(
2
+ 2e
2
2
cl
)
]
A
+
1
2

1
(
2
m
2
2
cl
)
1
+
1
2

2
(
2
m
2

3

2
cl
)
2
2e
cl
A
2
+ , (5)
where denotes terms other than being quadratic in uctuating elds. Now we
impose the Landau gauge condition
= 0 to the Lagrangian, which removes the

o-diagonal term2e
cl
A
2
. Then, according to (4), the 1-loop eective Lagrangian
can be evaluated as
i
2
log det
[

2
L
]
=0
=
i
2
[
log det
(
(
2
+ 2e
2
2
cl
) +
)
+ log det
(
2
+m
2
+
2
cl
)
+ log det
(
2
+m
2
+

3

2
cl
)
]
=
i
2
d
d
k
(2)
d
[
tr log(k
2
+ 2e
2
2
cl
)
3
+ tr log(k
2
+m
2
+
2
cl
) + tr log(k
2
+m
2
+

3

2
cl
)
]
=
(
d
2
)
2(4)
d/2
[
3(2e
2
2
cl
)
d/2
+ (m
2
+
2
cl
)
d/2
+ (m
2
+

3

2
cl
)
d/2
]
. (6)
In the second equality we use the following identity to evaluate the determinant: trick
to evaluate the determinant. Let A be a n 1 matrix and B a 1 n matrix, then
det(I +AB) =
n1
( +BA),
where A and B are matrices of n 1 and 1 n, respectively, is an arbitrary complex
number and I is the n n identity matrix. In our case, this gives
det
(
(
2
+ 2e
2
2
cl
) +
)
= 2e
2
2
cl
(
2
+ 2e
2
2
cl
)
3
.
Then the second equality follows up to an irrelevant constant term. The third equality
makes use of the trick of (11.72) in Peskin & Schroeders book. Then, for d = 4 and
0, we have
i
2
log det
[

2
L
]
=0
=
1
4(4)
2
[
3(2e
2
2
cl
)
2
(
log(2e
2
2
cl
)
)
+ (m
2
+
2
cl
)
2
(
log(m
2
+
2
cl
)
)
+ (m
2
+

3

2
cl
)
2
(
log(m
2
+

3

2
cl
)
)
]
, (7)
2
where we dene
2
+ log 4 +
3
2
for brevity.
Now, with MS scheme, we can determine the counterterms in (4) to be
L =
1
4(4)
2
[
2
+ log 4 log M
2
]
(
3(2e
2
2
cl
)
2
+ (m
2
+
2
cl
)
2
+ (m
2
+

3

2
cl
)
2
)
.
(8)
where M is the renormalization scale. Now the eective potential follows directly from
(4), (7) and (8):
V
e
[
cl
] = m
2
2
cl
+

6

4
cl
1
4(4)
2
[
3(2e
2
2
cl
)
2
(
log
M
2
2e
2
2
cl
+
3
2
)
+ (m
2
+
2
cl
)
2
(
log
M
2
m
2
+
2
cl
+
3
2
)
+ (m
2
+

3

2
cl
)
2
(
log
M
2
m
2
+

3

2
cl
+
3
2
)
]
. (9)
(c) Now taking the mass parameter
2
= m
2
= 0, then the eective potential (9)
becomes
V
e
[
cl
] =

6

4
cl
+
1
4(4)
2
[
12e
4
4
cl
(
log
2e
2
2
cl
M
2

3
2
)
+
10
9

2
4
cl
(
log

2
cl
M
2

3
2
)
]

6

4
cl
+
3e
4
4
cl
(4)
2
(
log
2e
2
2
cl
M
2

3
2
)
. (10)
In the second line we use the fact that is of the order e
4
to drop the
2
term. Then
the minimal point of this eective potential can be easily worked out to be
2
cl
=
M
2
2e
2
exp
(
1
8
2
9e
4
)
. (11)
As e
4
, we see that
cl
is of the same order with M. Thus the eective potential
remains valid at this level of perturbation theory.
(d) We plot the eective potential as a function of
cl
in Figure (1). The purple curve
with m
2
= 5 10
7
M
2
corresponds the case with no spontaneous symmetry breaking.
The blue curve shows that as m
2
goes to 0 from above, new local minima is formed.
Finally, the orange and red curves correspond to broken symmetry, and in the case of
the orange curve with m
2
= 0, the symmetry is dynamically broken.
(e) Now we calculate functions of the Coleman-Weinberg model to 1-loop level at
high energies, where we can send the mass parameter m
2
to zero. It is convenient to
work in the Feynman gauge = 1. Then the relevant Feynman rules can be read from
the Lagrangian (3) to be:
=
i
k
2
, =
i
k
2
, =
i
k
2
,
= i, = i, =
i
3
,
3
0 1 2 3 4 5 6 7
0.06
0.04
0.02
0.00
0.02
cl
10
2
M
V
e
f
f
c
l
1
0
2
M
Figure 1: The eective potential V

e
as a function of
cl
, with dierent values of
m
2
/M
2
= 5 10
7
, 2.4 10
7
, 0 and 1 10
7
from top to bottom, respectively.
= 2ie
2
, = 2ie
2
= e(k
1
k
2
)
We rst nd the 1-loop wave function renormalization. For eld, there is only one
diagram with nonzero contribution,
p
p k
which reads,
e
2
d
d
k
(2)
d
i
k
2
i
(p k)
2
(p +k)
(p k)

2ie
2
p
2
(4)
2
2
. (12)
Then we have
=
2e
2
(4)
2
(
2
M
2
)
, (13)
and it is straightforward to see that
. For photons wave function renormalization

(vacuum polarization), we need to evaluate the following three diagrams,
The sum of the three diagrams is
e
2
d
d
k
(2)
d
i
k
2
i
(p k)
2
(2k p)
(2k p)
+ 2
1
2
2ie
2
d
d
k
(2)
d
i
k
2

(p k)
2
(p k)
2
4

ie
2
3(4)
2
2
(p
2
), (14)
which gives
A
=
e
2
3(4)
2
(
2
log M
2
)
. (15)
Then we turn to the 1-loop corrections to couplings. For scalar self-coupling , we
consider the 1-loop corrections to
4
term in the Lagrangian. There are six types of
diagrams contributing, listed as follows:
For each type there are several dierent permutations of internal lines giving identical
result, or more concretely, 3 permutations for each of the rst three types, and 6 per-
mutations for each of the last three types. Now we evaluate them in turn. We set all
external momenta to zero to simplify the calculation. Then,
(a) =
(i)
2
2
d
d
k
(2)
d
(
i
k
2
)
2
i
2
2(4)
2
2
, (16)
(b) =
(i/3)
2
2
d
d
k
(2)
d
(
i
k
2
)
2
i
2
18(4)
2
2
, (17)
(c) =
(2ie
2
)
2
2
d
d
k
(2)
d
(
i
k
2
)
2
8ie
4
(4)
2
2
, (18)
(d) =
ie
2
3
d
d
k
(2)
d
i
k
2
(
i
k
2
)
2
(k
)
ie
2
3(4)
2
2
, (19)
(e) = (2ie
2
)e
2
d
d
k
(2)
d
i
k
2
(
i
k
2
)
2
(k
)
2ie
4
(4)
2
2
, (20)
(f) = e
4
d
d
k
(2)
d
(
i
k
2
)
2
(
i
k
2
)
2
(k
)
2
ie
4
(4)
2
2
, (21)
Then multiplying (a)(c) by 3 and (d)(f) by 6, we nd
=
5
2
/3 2e
2
+ 18e
4
(4)
2
2
. (22)
Finally we consider the 1-loop corrections to e. For this purpose we calculate 1-loop
diagrams with three external lines with 1 A
, 1 and 1 respectively, shown as follows.

Now we calculate them in turn.
(a) = e
3
d
d
k
(2)
d
i
k
2
(
i
(p k)
2
)
2
(2p k)
2
(2k 2p)
5
0.0 0.1 0.2 0.3 0.4 0.5
0.0
0.1
0.2
0.3
0.4
0.5
e
2
Figure 2: The renormalization group ow of Coleman-Weinberg model.
= ie
3
d
d
k
(2)
d
1
0
dx
4x[(2 x)p k
]
2
[k
(1 x)p]
[k
2
+x(1 x)p
2
]
3
ie
3
d
d
k
(2)
d
1
0
dx
k
2
(1 x)p
2(2 x)(p k
)k
[k
2
+x(1 x)p
2
]
3
= ie
3
d
d
k
(2)
d
1
0
dx
[(1 x)
2
d
(2 x)]k
2
p
[k
2
+x(1 x)p
2
]
3
2e
3
(4)
2
2
, (23)
(b) = e
(
i
3
)
d
d
k
(2)
d
(
i
k
2
)
2
2k
= 0, (24)
(c) = (d) = e(2ie
2
)
d
d
k
(2)
d
i
k
2
i
(p k)
2
(k +p)

3e
3
(4)
2
2
. (25)
Summing the four diagrams, we nd that
e
=
2e
3
(4)
2
(
2
log M
2
)
. (26)
Now we are ready to calculate functions:
e
= M

M
(
e
+
1
2
(
A
+
)
)
=
e
3
48
2
, (27)
= M

M
(
+ 2
)
=
5
2
18e
2
+ 54e
4
24
2
. (28)
The trajectory of renormalization group ows generated from these functions are
shown in Figure 2.
(f ) The eective potential obtained in (c) is not a solution to the renormalization
group equation, since it is only a rst order result in perturbation expansion. However,
it is possible to nd an eective potential as a solution to the RG equation, with the
result in (c) serving as a sort of initial condition. The eective potential obtained in
this way is said to be RG improved.
6
The Callan-Symansik equation for the eective potential reads
(
M

M
+
+
e
cl
cl
)
V
e
(
cl
, , e; M) = 0. (29)
The solution to this equation is well known, that is, the dependence of the sliding energy
scale M is described totally by running parameters,
V
e
(
cl
, , e; M) = V
e
(
cl
(M
),
(M
), e(M
); M
)
, (30)
where barred quantities satisfy
M

M
=
, e), M
e
M
=
e
(
, e), M

cl
M
=
, e)
cl
(31)
The RG-improved eective potential should be such that when expanded in terms of
coupling constants and e, it will recover the result in (c) at the given order. For
simplicity here we work under the assumption that e
4
, so that all terms of higher
orders of coupling constants than and e
4
can be ignored. In this case, the perturbative
calculation in (c) gives
V
e
=

6

4
cl
+
3e
4
4
cl
(4)
2
(
log
2e
2
2
cl
M
2

3
2
)
. (32)
Now we claim that the RG-improved edition of this result reads
V
e
=
4
cl
+
3 e
4
4
cl
(4)
2
(
log 2 e
2
3
2
)
. (33)
To see this, we rstly solve the renormalization group equations (31),
(M
) = e
4
(

e
4
+
9
4
2
log
M
M
)
, (34)
e
2
(M
) =
e
2
1 (e
2
/24
2
) log(M
/M)
, (35)
cl
(M
) =
cl
(
M
M
)
2e
2
/(4)
2
, (36)
where the unbarred quantities , e and
cl
are evaluated at scale M. Now we substitute
these results back into the RG-improved eective potential (33) and expand in terms of
coupling constants. Then it is straightforward to see that the result recovers (32). To
see the spontaneous symmetry breaking still occurs, we note that the running coupling
(M
) ows to negative value rapidly for small M
=
cl
, while e(M
) changes mildly
along the
cl
scale, as can be seen directly from Figure 2. Therefore the the coecient
before
4
cl
is negative for small
cl
and positive for large
cl
. As a consequence, the
minimum of this eective potential should be away from
cl
= 0, namely the U(1)
symmetry is spontaneously broken.
To nd the scalar mass m
in this case (with = 0), we calculate the second deriva-

tive of the eective potential V
e
with respect to
cl
. Since the renormalization scale
M can be arbitrarily chosen, we set it to be M
2
= 2e
2
2
cl
to simplify the calculation.
Then the vanishing of the rst derivative of V
e
at
cl
=
cl
implies that = 9e
4
/8
2
.
Insert this back to V
e
in (32), we nd that
V
e
=
3e
4
4
cl
16
2
(
log

2
cl
2
cl

1
2
)
. (37)
7
0.0 0.1 0.2 0.3 0.4 0.5
0.0
0.1
0.2
0.3
0.4
0.5
e
2
0.0 0.1 0.2 0.3 0.4 0.5
0.0
0.1
0.2
0.3
0.4
0.5
e
2
0.0 0.1 0.2 0.3 0.4 0.5
0.0
0.1
0.2
0.3
0.4
0.5
e
2
0.0 0.1 0.2 0.3 0.4 0.5
0.0
0.1
0.2
0.3
0.4
0.5
e
2
Figure 3: The renormalization group ows of Coleman-Weinberg model in d = 4
spacetime dimensions, with = 0.005, 0.01, 0.1 and 1 in the upper-left, upper-right,
lower-left and lower-right diagram, respectively.
Then, taking the second derivative of this expression with respect to
cl
, we get the
scalar mass m
2
= 3e
4
2
cl
/2
2
= 3e
4
v
2
/4
2
. Recall that the gauge bosons mass m
A
is
given by m
A
= e
2
v
2
at the leading order, thus we conclude that m
2
/m
2
A
= 3e
2
/4
2
at
the leading order in e
2
.
(h) When the spacetime dimension is shifted from 4 as d = 4 , the functions
e
and
are also shifted to be
e
= e +
e
3
48
2
,
= +
5
2
18e
2
+ 54e
4
24
2
. (38)
We plot the corresponding RG ow diagrams for several choice of in Figure 3, where
we also extrapolate the result to = 1.
8
Chapter 15
Zhong-Zhi Xianyu

1 Brute-force computations in SU(3)
(a) The dimension of SU(N) group is d = N
2
1, when N = 3 we get d = 8.
(b) Its easy to see that t
1
, t
2
, t
3
generate a SU(2) subgroup of SU(3). Thus we have
f
ijk
=
ijk
for i, j, k = 1, 2, 3. Just take another example, lets check [t
6
, t
7
]:
[t
6
, t
7
] = i(
1
2
t
3
+
3
2
t
8
),
thus we get
f
678
=
3
2
, f
673
=
1
2
.
Then what about f
376
?
[t
3
, t
7
] =
i
2
t
6
f
376
=
1
2
= f
673
.
(c) C(F) =
1
2
. Here F represents fundamental representation.
(d) C
2
(F) =
4
3
, d(F) = 3, d(G) = 8, thus we see that d(F)C
2
(F) = d(G)C(F).
2 Adjoint representation of SU(2)
The structure constants for SU(2) is f
abc
=
abc
, thus we can write down the repre-
sentation matrices for its generators directly from
(t
b
G
)
ac
= if
abc
= i
abc
.
More explicitly,
t
1
G
=
_
_
_
0 0 0
0 0 i
0 i 0
_
_
_, t
2
G
=
_
_
_
0 0 i
0 0 0
i 0 0
_
_
_, t
3
G
=
_
_
_
0 i 0
i 0 0
0 0 0
_
_
_, (1)
1
Then,
C(G) = tr (t
1
G
t
1
G
) = tr (t
2
G
t
2
G
) = tr (t
3
G
t
3
G
) = 2,
C
2
(G)I
3
= t
1
G
t
1
G
+ t
2
G
t
2
G
+ t
3
G
t
3
G
= 2I
3
C
2
(G) = 2.
Here I
3
is the 3 3 unit matrix.
3 Coulomb potential
(a) We calculate vacuum expectation value for Wilson loop U
P
(z, z), dened by
U
P
(z, z) = exp
_
ie
_
P
dx
(x)
_
. (2)
By denition, we have
U
P
(z, z) =
_
DA
exp
_
iS[A
] ie
_
P
dx
(x)
_
, (3)
where
S[A
] =
_
d
4
x
_
1
4
F
1
2
(
)
2
_
. (4)
Thus U
P
(z, z) is simply a Gaussian integral, and can be worked out directly, as
U
P
(z, z) = exp
_
1
2
_
ie
_
P
dx
__
ie
_
P
dy
_
_
d
4
k
(2)
4
ig
k
2
+ i
e
ik(xy)
_
(5)
Here we have set 0 to simplify the calculation. Working out the momentum integral,
we get
U
P
(z, z) = exp
_
e
2
8
2
_
P
dx
_
P
dy
(x y)
2
_
. (6)
The momentum integration goes as follows
_
d
4
k
(2)
4
e
ik(xy)
k
2
+ i
= i
_
d
4
k
E
(2)
4
e
ikE(xy)
k
2
E
=
i
(2)
4
_
2
0
d
_

0
dsin
_

0
d sin
2
_

0
dk
E
k
3
E
e
ikE|xy| cos
k
2
E
=
i
4
3
_

0
dk
E
k
E
_

0
d sin
2
e
ikE|xy| cos
=
i
4
2
_

0
dk
E
k
E
J
1
_
k
E
|x y|
_
k
E
|x y|
=
i
4
2
(x y)
2
. (7)
Where J
1
(x) is Bessel function and we use the fact that
_
0
dxJ
1
(x) = 1.
(b) Now taking a narrow rectangular Wilson loop P with width R in x
1
direction
(0 < x
1
< 1) and length T in x
0
direction (0 < x
0
< T) and evaluate U
P
. When
the integral over dx and dy go independently over the loop, divergence will occur as
|x y|
2
0. But what we want to show is the dependence of U
P
on the geometry of
the loop, namely the width R and length T, which should be divergence free. Therefore,
when T R, the integral in Wilson loop is mainly contributed by time direction and
can be expressed as
U
P
(z, z) exp
_
2e
2
8
2
_
T
0
dx
0
_
0
T
dy
0
1
(x
0
y
0
)
2
R
2
i
_
, (8)
2
and we have add a small imaginary part to the denominator for the reason that will be
clear. Carry out the integration, we nd
_
T
0
dx
0
_
0
T
dy
0
1
(x
0
y
0
)
2
R
2
i
TR
2T
R
arctanh
_
T
R + i
_
=
i
T
R
.
Therefore,
U
P
= exp
_
ie
2
4R
T
_
= e
iV (R)T
, (9)
which gives the familiar result V (R) = e
2
/4R.
(c) For the Wilson loop of a non-Abelian gauge group, we have
U
P
(z, z) = tr
_
Pexp
_
ig
_
P
dx
A
a
(x)t
a
r
_
_
, (10)
where t
a
r
is the matrices of the group generators in representation r. We expand this
expression to the order of g
2
,
U
P
(z, z) = tr (1) g
2
_
P
dx
_
P
dy
A
a
(x)A
b
(y) tr (t
a
r
t
b
r
) +O(g
3
)
= tr (1)
_
1 g
2
C
2
(r)
_
P
dx
_
P
dy
A
a
(x)A
b
(y)
_
+O(g
3
). (11)
Compared with the Abelian case, we see that to order g
2
, the non-Abelian result is
given by making the replacement e
2
g
2
C
2
(r). Therefore we conclude that V (R) =
g
2
C
2
(r)/4R in non-Abelian case.
4 Scalar propagator in a gauge theory
In this problem we study very briey the heat kernel representation of Green func-
tions/propagator of a scalar eld living within a gauge eld background.
(a) To begin with, we consider the simplest case, in which the background gauge
eld vanishes. Then we can represent the Green function D
F
(x, y) of the Klein-Gordon
equation, dened to be
(
2
+ m
2
)D
F
(x, y) = i
(4)
(x y) (12)
with proper boundary conditions, by the following integral over the heat kernel function
D(x, y, T):
D
F
(x, y) =
_

0
dT D(x, y, T). (13)
The heat kernel satises the following Schrodinger equation:
_
i

T
(
2
+ m
2
)
_
D(x, y, T) = i(T)
(4)
(x y). (14)
The solution to this equation can be represented by
D(x, y, T) = x|e
iHT
|y =
_
d
4
k
(2)
4
d
4
k
(2)
4
x|kk|e
iHT
|k
|y
=
_
d
4
k
(2)
4
d
4
k
(2)
4
e
i(k
2
+m
2
)T
e
ikx+ik
y
(2)
4
(4)
(k k
)
3
=
_
d
4
k
(2)
4
e
i(k
2
m
2
)T
e
ik(xy)
, (15)
with H =
2
+m
2
. Integrating this result over T, with the +i prescription, we recover
the Feynman propagator for a scalar eld:
_

0
dT D(x, y, T) =
_
d
4
k
(2)
4
e
ik(xy)
_

0
dT e
i(k
2
m
2
+i)T
=
_
d
4
k
(2)
4
ie
ik(xy)
k
2
m
2
+ i
. (16)
(b) Now let us turn on a background Abelian gauge eld A
(x). The corresponding

Schrodinger equation then becomes
_
i

T

_
_
ieA
(x)
_
2
+ m
2
_
_
D(x, y, T) = i(T)
(4)
(x y), (17)
the solution of which, x|e
iHT
|y, can also be expressed as a path integral,
x|e
iHT
|y = lim
N
_
N
i=1
_
dx
i
x
i
exp
_
it
__
ieA
(x)
_
2
+ m
2
x
i1
_
_
,
(18)
where we have identify x = x
N
, y = x
0
, and t = T/N. Then,
x
i
|e
it[(ieA(x))
2
+m
2
]
|x
i1
=
_
d
4
k
i
(2)
4
x
i
|e
it[
2
ieA(x)
+m
2
]
|k
i
k
i
|e
it[ie
A(x)e
2
A
2
(x)]
|x
i1
=
_
d
4
k
i
(2)
4
x
i
|e
it[k
2
i
+eA(xi)k
i
+m
2
]
|k
i
k
i
|e
it[ek
i
A(xi1)e
2
A
2
(xi1)]
|x
i1
=
_
d
4
k
i
(2)
4
e
it[k
2
i
+eki(A(xi)+A(xi1))e
2
A
2
(xi1)+m
2
i]
e
iki(xixi1)
= C exp
_
it
4
_
x
i
x
i1
t
+ eA(x
i
) + eA(x
i1
)
_
2
it(m
2
e
2
A
2
(x
i1
))
_
C exp
_
it
4
_
dx
dt
_
2
iteA(x)
dx
dt
itm
2
_
. (19)
In the last line we take the continuum limit, and C is an irrelevant normalization con-
stant. Then we get
D(x, y, T) =
_
Dx exp
_
i
_
T
0
dt
_
_
dx
dt
_
2
+ m
2
_
ie
_
T
0
dx(t) A(x(t))
_
(20)
5 Casimir operator computations
(a) In the language of angular momentum theory, we can take common eigenfunctions
of J
2
=
a
T
a
T
a
and J
z
= T
3
to be the representation basis. Then the representation
matrix for T
3
is diagonal:
t
3
j
= diag (j, j + 1, , j 1, j).
Thus
tr (t
3
j
t
3
j
) =
j
m=j
m
2
=
1
3
j(j + 1)(2j + 1).
4
Then we have
tr (t
3
r
t
3
r
) =
i
tr (t
3
ji
t
3
ji
) =
1
3
i
j
i
(j
i
+ 1)(2j
i
+ 1) = C(r),
which implies that
3C(r) =
i
j
i
(j
i
+ 1)(2j
i
+ 1). (21)
(b) Let the SU(2) subgroup be spanned by T
1
, T
1
and T
3
. Then in fundamental
representation, the representation matrices for SU(2) subgroup of SU(N) can be taken
as
t
i
N
=
_

i
/2 0
2(N2)
0
(N2)2
0
(N2)(N2)
_
. (22)
Where
i
(i = 1, 2, 3) are Pauli matrices. We see that the representation matrices for
SU(2) decomposes into a doublet and (N 2) singlet. Then its easy to nd that
C(N) =
1
3
_
1
2
(
1
2
+ 1)(2
1
2
+ 1)
_
=
1
2
, (23)
by formula in (a).
In adjoint representation, the representation matrices (t
i
)
ab
= if
aib
(a, b = 1, , N
2
1, i = 1, 2, 3). Thus we need to know some information about structure constants. Here
we give a handwaving illustration by analyzing the structure of fundamental representa-
tion matrices a little bit more. Note that therere three types of representation matrices,
listed as follows. For convenience, lets call them t
A
, t
B
and t
C
:
t
A
=
_
A
22
0
2(N2)
0
(N2)2
0
(N2)(N2)
_
. (24)
t
B
=
_
A
22
B
2(N2)
B
(N2)2
0
(N2)(N2)
_
. (25)
t
C
=
_
1
2
tr (C)I
22
0
2(N2)
0
(N2)2
C
(N2)(N2)
_
. (26)
In which, t
A
is just the representation matrices for SU(2) subgroup. Thus we see that
there are 3 t
A
, 2(N 2) t
B
and (N 2)
2
t
C
in total. Its also obvious that there is no
way to generate a t
A
from commutators between two t
C
or between a t
B
and t
C
, the
only way to generate t
A
are commutators between two t
A
or between to t
B
. Then, t
A
commutators correspond to the triplet representation os SU(2) subgroup, and 2(N2)-
t
B
commutators correspond to the doublet representation of SU(2). In this way we
see that adjoint representation matrices for SU(2) subgroup decompose into 1 triplet,
2(N 1) doublets and (N 2)
2
singlets.
Then we can calculate C(G), again, by using formula in (a), as:
C(G) =
1
3
_
1(1 + 1)(2 1 + 1) + 2(N 2)
1
2
(
1
2
+ 1)(2
1
2
+ 1)
= N. (27)
(c) Let U SU(N) be N N unitary matrix, S be a symmetric N N matrix,
and A be an antisymmetric N N matrix. Then we can use S and A to build two
representations for SU(N) respectively, as
S USU
T
, A UAU
T
.
5
Its easy to verify that they are indeed representations. Lets denote these two rep-
resentation by s and a. Its also obvious to see that the dimensions of s and a are
d(s) = N(N + 1)/2 and d(a) = N(N 1)/2 respectively.
Accordingly, the generator T
a
acts on S and A as:
S T
a
S + S(T
a
)
T
, A T
a
A + A(T
a
)
T
. (28)
To get C
2
(s) and C
2
(a), we can make use of the formula
d(r)C
2
(r) = d(G)C(r). (29)
Thus we need to calculate C(r) and C(a). By formula in (a), we can take an generator
in SU(2) subgroup to simplify the calculation. Lets take
t
3
N
=
1
2
diag(1, 1, 0, , 0),
Then we have:
S =
_
_
_
_
S
11
S
1n
.
.
.
.
.
.
.
.
.
S
n1
S
nn
_
_
_
_
t
3
N
S + S(t
3
N
)
T
=
1
2
_
_
_
_
_
_
_
_
2S
11
0 S
13
S
1n
0 2S
22
S
23
S
2n
S
31
S
32
0 0
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
S
n1
S
n2
0 0
_
_
_
_
_
_
_
_
A =
_
_
_
_
_
_
_
0 A
12
A
1n
A
21
0
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. A
n1,n
A
n1
A
n,n1
0
_
_
_
_
_
_
_
t
3
N
A+A(t
3
N
)
T
=
1
2
_
_
_
_
_
_
_
_
0 0 A
13
A
1n
0 0 A
23
A
2n
A
31
A
32
0 0
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
A
n1
A
n2
0 0
_
_
_
_
_
_
_
_
Thus we see that the representation matrices for T
3
, in both s representation and a
representation, are diagonal. They are:
t
3
s
= diag(1, 0,
1
2
,
1
2
. .
N2
, 1,
1
2
, ,
1
2
. .
N2
, 0, , 0
. .
(N2)(N1)/2
); (30)
t
3
a
= diag(0,
1
2
, ,
1
2
. .
2(N2)
, 0, , 0
. .
(N2)(N3)/2
). (31)
Here we have rearrange the upper triangular elements of S and A by line.
Then we get
C(s) = tr (t
3
s
)
2
=
1
2
(N + 2); (32)
C(a) = tr (t
3
a
)
2
=
1
2
(N 2). (33)
Then,
C
2
(s) =
d(G)C(s)
d(s)
=
(N
2
1)(N + 2)/2
N(N + 1)/2
=
(N 1)(N + 2)
N
; (34)
C
2
(a) =
d(G)C(a)
d(a)
=
(N
2
1)(N 2)/2
N(N 1)/2
=
(N + 1)(N 2)
N
. (35)
6
At last lets check the formula implied by (15.100) and (15.101):
_
C
2
(r
1
) + C
2
(r
2
)
_
d(r
1
)d(r
2
) =
C
2
(r
i
)d(r
i
), (36)
in which the tensor product representation r
1
r
2
decomposes into a direct sum of
irreducible representations r
i
. In our case, the direct sum of representation s and a is
equivalent to the tensor product representation of two copies of N. That is,
N N

= s + a.
Thus, we have,
_
C
2
(N) + C
2
(N)
_
d(N)d(N) =
_
N
2
1
2N
+
N
2
1
2N
_
N
2
= N(N
2
1);
and
C
2
(s)d(s) + C
2
(a)d(a) =
_
C(s) + C(a)
d(G) = N(N
2
1).
Thus formula (36) indeed holds in our case.
7
Chapter 16
Zhong-Zhi Xianyu

1 Arnowitt-Fickler gauge
In this problem we perform the Faddeev-Popov quantization of Yang-Mills theory in
Arnowitt-Fickler gauge (also called axial gauge), namely A
3a
= 0. More generally, we
may write the gauge condition as n
A
a
= 0 with n
an arbitrary space-like vector of

unit norm (n
2
= 1). The condition A
3a
= 0 corresponds simply to the choice n
= g
3
.
This gauge has the advantage that the Faddeev-Popov ghosts do not propagate and do
not couple to gauge elds, as we will show below.
Our starting point, the partition function, reads
Z =
_
DA
(n A
a
)e
iS[A]
det
_

(n A
a
)
b
_
, (1)
with S =
1
4
_
d
4
x(F
a
)
2
the classical action for the gauge eld, and the Faddeev-Popov
determinant is given by
det
_

(n A
a
)
b
_
= det
_
1
g
n
ab
f
abc
n
A
c
_
=
_
DbDc exp
_
i
_
d
4
xb
a
_
n
ab
f
abc
n
A
c
_
c
b
_
. (2)
When multiplied by the delta function (n A
a
), the second term in the exponent
above vanishes, which implies that the ghost and antighost do not interact with gauge
eld. Meanwhile, they do not propagator either, since there does not exist a canonical
kinetic term for them. Therefore we can safely treat the Faddeev-Popov determinant
as an overall normalization of the partition function and ignore it. Then, the partition
function reduces to
Z = lim
0
_
DA
exp
_
i
_
d
4
x
_
1
4
(F
a
)
2
1
2
(n A
a
)
2
_
_
= lim
0
_
DA
exp
_
i
_
d
4
x
_
1
2
A
a
_
g
_
A
a
gf
abc
(
A
a
)A
b
A
c
1
4
g
2
f
eab
f
ecd
A
a
A
b
A
c
A
d
_
_
. (3)
1
where we have convert the delta function (n A
a
) into a limit of Gaussian function.
Then we see that the three-point or four-point gauge boson vertices share the same
Feynman rules with the ones in covariant gauge. The only dierence arises from the
propagator. Let us parameterize the propagator in momentum space as
D
(k) = Ag
+ Bk
+ C(k
+ n
) + Dn
. (4)
Then, the equation of motion satised by the propagator,
_
g
k
2
k
+
1
_
D
(k) = ig
(5)
gives
A =
i
k
2
, B =
1 k
2
k n
C, C =
1
k n
A, D = 0. (6)
Note that the gauge xing parameter should be sent to 0. Therefore the propagator
reads
D(k)
=
i
k
2
_
g
(k n)
2

k
+ k
k n
_
. (7)
2 Scalar eld with non-Abelian charge
(a) Firstly we write down the Lagrangian for the Yang-Mills theory with charged scalar
eld, as
L =
1
4
(F
a
)
2
+ (D
(D
), (8)
where the covariant derivative D
=
_
+ igA
a
t
a
r
_
with t
a
r
the matrices of gauge
group generators in representation r. For simplicity we ignore the possible mass term
for the scalar. Then, it is straightforward to derive the Feynman rules for this theory by
expanding this Lagrangian. The rules for the propagator and self-interactions of gauge
boson are independent of matter content and are the same with the ones given in Figure
16.1 in Peskin&Schroeder. The only new ingredients here are the gauge boson-scalar
eld interactions, which generate the following Feynman rules:
, a , b
= ig
2
(t
a
r
t
b
r
+ t
b
r
t
a
r
)g
,
p
1
p
2
, a
= igt
a
r
(p
1
p
2
)
,
(b) To compute the function of coupling g, we introduce some additional Feynman
rules involving counterterms:
p
1
p
2
a
= ig
1
t
a
r
(p
1
p
2
)
,
2
p
= ip
2
2
i
m
,
p
a
b
= i
ab
3
(p
2
g
).
Then the function is given by
= gM

M
_
1
+
2
+
1
2

3
_
. (9)
To determine the counterterms, we evaluate the following relevant 1-loop diagrams.
But the calculations can be simplied a lot if we observe that the combination
1

2
is determined by pure gauge sector, and is independent of matter content. This may be
most easily seen from the counterterm relation
1

2
=
c
1
c
2
, where the right hand
side comes from ghost contribution which is a pure gauge quantity. We will demonstrate
this counterterm relation explicitly in the next problem for fermionic matter. Therefore,
we can borrow directly the result of Peskin & Schroeder, or from the result of Problem
16.3(a),
2
=
g
2
(4)
2
C
2
(G)
_
2
log M
2
_
. (10)
On the other hand,
3
can be found by evaluating the loop corrections to the gauge
bosons self-energy. The contributions from the gauge boson loop and ghost loop have
already been given in eq.(16.71) in Peskin&Schroeder, while the rest of the contribu-
tions is from the scalar-loop, and is simply the result we have found in Problem 9.1(c),
multiplied by the gauge factor tr (t
a
r
t
b
r
) = C(r) and the number of scalar n
s
. Combining
these two parts gives the divergent part of
3
:
3
=
g
2
(4)
2
_
5
3
C
2
(G)
1
3
n
s
C(r)
_
2
log M
2
_
. (11)
Then it is straightforward to see that
=
g
3
(4)
2
_
11
3
C
2
(G)
1
3
n
s
C(r)
_
. (12)
3 Counterterm relations
In this problem we calculate the divergent parts of counterterms in Yang-Mills theory
with Dirac spinors at 1-loop level, to verify the counterterm relations, which is a set of
constraints set by gauge invariance. To begin with, let us rewrite the Lagrangian in its
renormalized form, with counterterms separated, as
L =
1
4
(
A
a
A
a
)
2
+

(i/ m) c
a
2
c
a
+ gA
a
t
a
gf
abc
(
A
a
)A
b
A
c
1
4
g
2
(f
eab
A
a
A
b
)(f
ecd
A
c
A
d
) g c
a
f
abc
(A
b
c
c
)
1
4

3
(
A
a
A
a
)
2
+

(i
2
/
m
)
c
2
c
a
2
c
a
+ g
1
A
a
g
3g
1
f
abc
(
A
a
)A
b
A
c
1
4

4g
1
(f
eab
A
a
A
b
)(f
ecd
A
c
A
d
) g
c
1
c
a
f
abc
(A
b
c
c
). (13)
3
Then the counterterm relations we will verify are
2
=
3g
1

3
=
1
2
(
4g
1

3
) =
c
1
c
2
. (14)
Note that
1
and
2
have been given in (16.84) and (16.77) in Peskin&Schroeder. Here
we simply quote the results:
1
=
g
2
(4)
2
_
C
2
(r) + C
2
(G)
_
2
log M
2
_
, (15)
2
=
g
2
(4)
2
C
2
(r)
_
2
log M
2
_
. (16)
Therefore,
2
=
g
2
(4)
2
C
2
(G)
_
2
log M
2
_
. (17)
(a) Firstly let us check the equality
1

2
=
c
1
c
2
. The 1-loop contributions to
c
1
come from the following three diagrams:
p
1
k
p
2
c a
b
The rst diagram reads
(g)
3
f
ade
f
ebf
f
fdc
_
d
d
k
(2)
d
i
k
2
i
(p
1
k)
2
i
(p
2
k)
2
p
2
(p
2
k)
(p
1
k)
g
3
f
ade
f
ebf
f
fdc
_
d
d
k
(2)
d
k
(p
2
k)
k
6
=
1
4
g
3
f
ade
f
ebf
f
fdc
p
2
_
d
d
k
(2)
d
1
k
4
i
4
g
3
f
ade
f
ebf
f
fdc
p
2

i
(4)
2
2
. (18)
The second diagram reads
(g)
2
gf
ade
f
ebf
f
fdc
_
d
d
k
(2)
d
i
k
2
i
(p
1
k)
2
i
(p
2
k)
2
p
2
k
_
g
(k p
2
q)
+ g
(q p
1
+ k)
+ g
(p
1
+ p
2
2k)
ig
3
f
ade
f
ebf
f
fdc
_
d
d
k
(2)
d
1
k
6
_
p
2
k
2
+ k
(k p
2
) 2k
(k p
2
)
3
4
ig
3
f
ade
f
ebf
f
fdc
p
2
_
d
d
k
(2)
d
1
k
4

3
4
ig
3
f
ade
f
ebf
f
fdc
p
2

i
(4)
2
2
. (19)
To simplify the structure constant product, we make use of the Jacobi identity,
0 = f
ebf
(f
abd
f
dce
+ f
bcd
f
dae
+ f
cad
f
dbe
) = 2f
abd
f
dce
f
ebf
f
caf
C
2
(G),
then we have
f
ade
f
ebf
f
fdc
=
1
2
f
abc
C
2
(G). (20)
4
Note that the third diagram reads g
c
1
f
abc
p
2
, thus we see that to make the sum of
these three diagrams nite, the counterterm coecient
c
1
should be
c
1

g
2
C
2
(G)
2(4)
2
_
2
log M
2
_
. (21)
Then consider
c
2
. This coecient should absorb the divergence from the following
diagram:
p
k
a
b
This diagram reads
(g)
2
f
bcd
f
dca
_
d
d
k
(2)
d
i
k
2
i
(p k)
2
(p k)
=g
2
C
2
(G)
ab
_
d
d
k
(2)
d
_
1
0
dx
p (k
+ xp)
(k
2
)
2
g
2
2
C
2
(G)
ab
p
2
i
(4)
2
2
+ terms indep. of p
2
. (22)
The corresponding counterterm contributes i
c
2
p
2
, therefore we have
c
2

g
2
C
2
(G)
2(4)
2
_
2
log M
2
_
. (23)
Combining (21), (23) and (17), we see that the equality
1
2
=
c
1
c
2
is satised.
(b) Now lets verify the equality
1
2
=
3g
1

3
. In this case the calculation turns
out to be more cumbersome, though. The coecient
3
has been given by (16.74) in
Peskin&Schroeder. The result is
3
=
g
2
(4)
2
_
5
3
C
2
(G)
4
3
n
f
C(r)
__
2
log M
2
_
. (24)
Thus we only need to calculate
3g
1
. The relevant loop diagrams are listed as follows.
p
p
b
c
a
For simplicity, we have set the external momenta to be p, p and 0 for the three
external gauge boson lines labeled with (a), (b) and (c). Then the contribution of
the counterterm to this vertex is given by
g
3g
1
f
abc
(2g
). (25)
5
To extract the divergent part from
3g
1
, we have to evaluate the loop diagrams shown
above. Let us calculate them now in turn. The rst diagram reads
=
1
2
g(ig
2
)f
ade
_
f
def
f
bcf
(g
)
+ f
dbf
f
ecf
(g
) + f
dcf
f
ebf
(g
_
d
d
k
(2)
d
i
k
2
i
(k p)
2
_
g
(p + k)
+ g
(2k + p)
+ g
(k 2p)
=
1
2
ig
3
f
abc
C
2
(G)
3
2
(g
_
d
d
k
(2)
d
1
k
2
1
(k p)
2
_
g
(p + k)
+ g
(2k + p)
+ g
(k 2p)
1
2
ig
3
f
abc
C
2
(G)
9
2
(g
)
_
d
d
k
(2)
d
1
k
4
9
4
ig
3
f
abc
C
2
(G)(g
)
i
(4)
2
2
. (26)
There are two additional diagrams associated with this diagram by the two cyclic per-
mutations of the three external momenta. One gives
9
4
ig
3
f
abc
C
2
(G)(g
)
i
(4)
2
2
,
while the other yields zero. Therefore the sum of these three diagrams gives:
9
4
ig
3
f
abc
C
2
(G)(2g
)
i
(4)
2
2
, (27)
Then we come to the second diagram, which reads
= g
3
f
adf
f
bed
f
cfe
_
d
d
k
(2)
d
i
k
2
i
k
2
i
(p + k)
2
_
g
(2p + k)
+ g
(p 2k)
+ g
(k p)
_
g
(p + k)
+ g
(2k p)
+ g
(2p + k)
_
g
2g
+ g
ig
3
_
1
2
f
abc
C
2
(G)
_
d
d
k
(2)
d
_
1
0
dx
2(1 x)k
2
(k
2
)
3
1
4
_
2(8 + 15x)g
+ (30x 23)
_
g
+ g
_
_
ig
3
_
1
2
f
abc
C
2
(G)
13
4
_
2g
_
_
d
d
k
(2)
d
1
k
4
13
8
ig
3
f
abc
C
2
(G)
_
2g
i
(4)
2
2
. (28)
The third diagram reads
= (g)
3
f
daf
f
ebd
f
fce
_
d
d
k
(2)
d
(1)
_
i
k
2
_
2
i
(k + p)
2
(p + k)
=ig
3
1
2
f
abc
C
2
(G)
_
d
d
k
(2)
d
_
1
0
dx
2(1 x)k
2
(k
2
)
3
6
1
d
_
xg
xg
+ (1 x)g
1
24
ig
3
f
abc
C
2
(G)(g
+ g
2g
)
i
(4)
2
2
. (29)
There is again a similar diagram with ghost loop running reversely, which gives
1
24
ig
3
f
abc
C
2
(G)(g
2g
+ g
)
i
(4)
2
2
.
Then these two diagrams with ghost loops sum to
1
24
ig
3
f
abc
C
2
(G)(2g
)
i
(4)
2
2
. (30)
Finally we consider the fourth diagram with fermion loop. There are also two copies
with fermions running in opposite directions. One (shown in the gure) gives
= n
f
(ig)
3
tr (t
a
t
c
t
b
)
_
d
d
k
(2)
d
(1) tr
_
i
/ k
i
/ k
i
/ k +
/
p
_
4
3
n
f
g
3
tr (t
a
t
c
t
b
)(2g
)
i
(4)
2
2
, (31)
while the other gives
4
3
n
f
g
3
tr (t
a
t
b
t
c
)(2g
)
i
(4)
2
2
.
Thus they sum to
4
3
n
f
g
3
tr
_
t
a
[t
c
, t
b
]
_
(2g
)
i
(4)
2
2
=
4
3
in
f
g
3
C(r)f
abc
(2g
)
i
(4)
2
2
. (32)
Now, sum up the four groups of diagrams, we get
g
3
(4)
2
2
f
abc
_
_
9
4
+
13
8

1
24
_
C
2
(G) +
4
3
n
f
C(r)
_
(2g
), (33)
and consequently,
3g
1
=
g
2
(4)
2
_
2
3
C
2
(G)
4
3
n
f
C(r)
_
_
2
log M
2
_
. (34)
Thus,
3g
1

3
=
g
2
(4)
2
C
2
(G)
_
2
log M
2
_
, (35)
which equals to
1
2
(17), as expected.
(c) Now lets move to the relation
1
2
=
1
2
(
4g
1

3
). This time we have to evaluate
4g
1
, which is determined by the divergent part of the following ve types of diagrams:
c
d
a
b
7
Firstly the counterterm itself contributes to the 1-loop corrections with
4g
1
_
f
abe
f
cde
(g
) + f
ace
f
bde
(g
)
+ f
ade
f
bce
(g
. (36)
To evaluate the loop diagrams, we set all external momenta to zero for simplicity. The
rst diagram then reads
=
1
2
(ig
2
)
2
_
f
abg
f
efg
(g
) + f
aeg
f
bfg
(g
)
+ f
afg
f
beg
(g
)
__
f
efh
f
cdh
(g
)
+ f
ech
f
fdh
(g
) + f
edh
f
fch
(g
)
_
_
d
d
k
(2)
d
_
i
k
2
_
2
ig
4
2(4)
2
2
_
f
abg
f
efg
f
efh
f
cdh
(2g
2g
)
+ f
abg
f
efg
f
ech
f
fdh
(g
) + f
abg
f
efg
f
edh
f
fch
(g
)
+ f
aeg
f
bfg
f
efh
f
cdh
(g
) + f
afg
f
beg
f
efh
f
cdh
(g
)
+ tr (t
a
t
b
t
d
t
c
)(2g
+ g
) + tr (t
a
t
b
t
c
t
d
)(2g
+ g
)
+ tr (t
a
t
b
t
c
t
d
)(2g
+ g
) + tr (t
a
t
b
t
d
t
c
)(2g
+ g
)
_
=
ig
4
2(4)
2
2
_
tr (t
a
t
b
t
c
t
d
)(4g
8g
+ 10g
)
+ tr (t
a
t
b
t
d
t
c
)(4g
+ 10g
8g
)
_
, (37)
where we have used (20) and f
abc
= i(t
a
)
bc
with t
a
the generators in adjoint representa-
tion. There are two additional diagrams similar to this one, which can be obtained by
exchange of labels as (b c) and (b d). Therefore the total contribution from
these three diagrams is
ig
4
(4)
2
2
_
tr (t
a
t
b
t
c
t
d
)(7g
8g
+ 7g
)
+ tr (t
a
t
b
t
d
t
c
)(7g
+ 7g
8g
)
+ tr (t
a
t
c
t
b
t
d
)(8g
+ 7g
+ 7g
)
_
. (38)
The second diagram has ve additional counterparts. The one displayed in the gure
reads
= (ig
2
)g
2
f
aeg
f
bgf
_
f
efh
f
cdh
(g
)
+ f
ech
f
fdh
(g
) + f
edh
f
fch
(g
_
d
d
k
(2)
d
_
i
k
2
_
3
(g
2k
+ g
)(g
2g
+ g
)
= g
4
f
aeg
f
bgf
_
f
efh
f
cdh
(g
)
+ f
ech
f
fdh
(g
) + f
edh
f
fch
(g
_
d
d
k
(2)
d
_
1
k
2
_
2_
2g
+ 5g
4g
g
4
4
i
(4)
2
2
_
f
aeg
f
bgf
f
efh
f
cdh
(6g
)
8
+ f
aeg
f
bgf
f
ech
f
fdh
(13g
+ 4g
2g
)
+ f
aeg
f
bgf
f
edh
f
fch
(13g
2g
+ 4g
)
_
=
g
4
4
i
(4)
2
2
_
if
cdh
tr (t
a
t
b
t
h
)(6g
)
tr (t
a
t
b
t
d
t
c
)(13g
+ 4g
2g
)
tr (t
a
t
b
t
c
t
d
)(13g
2g
+ 4g
)
_
=
g
4
4
i
(4)
2
2
_
tr (t
a
t
b
t
c
t
d
)(13g
8g
+ 10g
)
+ tr (t
a
t
b
t
d
t
c
)(13g
+ 10g
8g
)
_
, (39)
where we have used the fact that f
aeg
f
bgf
f
efh
f
cdh
= if
cdh
tr (t
a
t
b
t
h
) = tr
_
t
a
t
b
[t
c
, t
d
]
_
,
and f
aeg
f
bgf
f
ech
f
fdh
= tr (t
a
t
b
t
d
t
c
), etc. There are additional ve diagrams associ-
ated with this one, namely,
c d
a b
(a)
c b
a d
(b)
a d
c b
(c)
d b
c a
(d)
a c
b d
(e)
In the diagrams above, (a) gives the identical result to the one have just evaluated, while
(b) and (c) give identical expression, so do (d) and (e). We can nd (b) from the result
above by the exchange (a c), and (d) by the exchange (a d). Then we sum
up all six diagrams, which is equivalent to summing the original one with (b) and (d)
and multiplying the result by 2:
2
ig
4
4(4)
2
2
_
tr (t
a
t
b
t
c
t
d
)(13g
8g
+ 10g
)
+ tr (t
a
t
b
t
d
t
c
)(13g
+ 10g
8g
)
+ tr (t
c
t
b
t
a
t
d
)(13g
8g
+ 10g
)
+ tr (t
c
t
b
t
d
t
a
)(13g
8g
+ 10g
)
+ tr (t
d
t
b
t
c
t
a
)(13g
8g
+ 10g
)
+ tr (t
d
t
b
t
a
t
c
)(13g
8g
+ 10g
)
_
=
ig
4
2(4)
2
2
_
tr (t
a
t
b
t
c
t
d
)(23g
16g
+ 23g
)
+ tr (t
a
t
b
t
d
t
c
)(23g
+ 23g
16g
)
+ tr (t
a
t
c
t
b
t
d
)(16g
+ 23g
+ 23g
)
_
, (40)
where we use the cyclic symmetry of trace and also the relation tr (t
a
t
b
t
c
t
d
) = tr (t
d
t
b
t
c
t
a
).
The third diagram reads
= g
4
f
aeh
f
bhg
f
cfe
f
dgf
_
d
d
k
(2)
d
_
i
k
2
_
4
(g
2g
+ g
)
(g
2g
+ g
)(g
2g
+ g
)
9
(g
2g
+ g
)
= g
4
tr (t
a
t
b
t
d
t
c
)
_
d
d
k
(2)
d
1
(k
2
)
4
_
34k
+ k
4
(g
+ g
)
+ 3k
2
(g
+ g
+ g
+ g
= g
4
tr (t
a
t
b
t
d
t
c
)
_
d
d
k
(2)
d
1
(k
2
)
2
_
34
24
(g
+ g
+ g
)
+ (g
+ g
) +
3
2
(g
+ g
)
_
=
ig
4
12(4)
2
2
tr (t
a
t
b
t
d
t
c
)(47g
+ 47g
+ 17g
). (41)
Combined with the other two similar diagrams, we get
ig
4
12(4)
2
2
_
tr (t
a
t
b
t
c
t
d
)(47g
+ 17g
+ 47g
)
+ tr (t
a
t
b
t
d
t
c
)(47g
+ 47g
+ 17g
)
+ tr (t
a
t
c
t
b
t
d
)(17g
+ 47g
+ 47g
(42)
The fourth diagram with ghost loop is given by
= (g)
4
f
eah
f
hbg
f
gdf
f
fce
_
d
d
k
(2)
d
(1)
_
i
k
2
_
4
k
ig
4
24(4)
2
2
tr (t
a
t
b
t
d
t
c
)(g
+ g
+ g
). (43)
There are six distinct diagrams with ghost loops, with dierent permutations of external
labels (Lorentz and gauge). They sum to
ig
4
12(4)
2
2
_
tr (t
a
t
b
t
c
t
d
) + tr (t
a
t
b
t
d
t
c
) + tr (t
a
t
c
t
b
t
d
)
(g
+ g
+ g
). (44)
Finally the diagram with fermion loop reads
= (ig)
4
n
f
tr (t
a
r
t
b
r
t
d
r
t
c
r
)
_
d
d
k
(2)
d
() tr
_
i
/ k
i
/ k
i
/ k
i
/ k
_
=g
4
n
f
tr (t
a
r
t
b
r
t
d
r
t
c
r
)
_
d
d
k
(2)
d
1
(k
2
)
4
_
4(g
+ g
)(k
2
)
2
8(g
+ g
+ g
+ g
)k
2
+ 32k
=g
4
n
f
tr (t
a
r
t
b
r
t
d
r
t
c
r
)
_
d
d
k
(2)
d
1
(k
2
)
2
_
4(g
+ g
)
4(g
+ g
) +
4
3
(g
+ g
+ g
=
4ig
4
n
f
3(4)
2
2
tr (t
a
r
t
b
r
t
d
r
t
c
r
)(g
+ g
2g
). (45)
Combined with the similar diagrams with dierent permutations, we get
8ig
4
n
f
3(4)
2
2
_
tr (t
a
r
t
b
r
t
c
r
t
d
r
)(g
2g
+ g
)
+ tr (t
a
r
t
b
r
t
d
r
t
c
r
)(g
+ g
2g
)
10
+ tr (t
a
r
t
c
r
t
b
r
t
d
r
)(2g
+ g
+ g
). (46)
Now, we sum up the rst four types of diagrams, namely, (38), (40), (42), and (44), and
nd the result to be
2ig
4
3(4)
2
2
_
tr (t
a
t
b
t
c
t
d
)(g
+ 2g
)
+ tr (t
a
t
b
t
d
t
c
)(g
+ 2g
)
+ tr (t
a
t
c
t
b
t
d
)(2g
)
_
=
2ig
4
3(4)
2
2
_
g
_
2 tr (t
a
t
c
t
b
t
d
) tr (t
a
t
b
t
c
t
d
) tr (t
a
t
b
t
d
t
c
)
_
+ g
_
2 tr (t
a
t
b
t
c
t
d
) tr (t
a
t
b
t
d
t
c
) tr (t
a
t
c
t
b
t
d
)
_
+ g
_
2 tr (t
a
t
b
t
d
t
c
) tr (t
a
t
b
t
c
t
d
) tr (t
a
t
c
t
b
t
d
)
_
_
=
ig
4
3(4)
2
2
C
2
(G)
_
g
(f
ade
f
bce
f
ace
f
bde
)
+ g
(f
ade
f
bce
f
abe
f
cde
) + g
(f
ace
f
bde
+ f
abe
f
cde
)
=
ig
4
3(4)
2
2
C
2
(G)
_
f
abe
f
cde
(g
)
+ f
ace
f
bde
(g
) + f
ade
f
bce
(g
. (47)
Similar manipulations on (46) gives
4ig
4
3(4)
2
2
n
f
C
2
(r)
_
f
abe
f
cde
(g
)
+ f
ace
f
bde
(g
) + f
ade
f
bce
(g
. (48)
Therefore, we nally nd
4g
1
to be
g
2
3(4)
2
_
C
2
(G) + 4n
f
C
2
(r)
_
2
log M
2
_
, (49)
and it is straightforward to see that
4g
1

3
= 2(
1
2
).
11
Chapter 17
Zhong-Zhi Xianyu

1 Two-Loop renormalization group relations
(a) In this problem we study the higher orders of QCD function. Formally, we have
(g) =
b
0
(4)
2
g
3
b
1
(4)
4
g
5
b
2
(4)
6
g
7
+ . (1)
The we can deduce the corresponding function for
s
g
2
/(4), namely,
=
2b
0
4

2
s
2b
1
(4)
2
3
s
2b
2
(4)
3
4
s
+ . (2)
Integrate this equation, we get
_
Q
d
2
=
_
s(Q
2
)
d
s
_
b
0
4
2
s
+
b
1
(4)
2
3
s
+
b
2
(4)
3
4
s
+
_
1
. (3)
The integral can be carried out approximately, as
log(Q/)
2
=
4
b
0
_
1
s
(Q
2
)
+
b
1
4b
0
log

s
(Q
2
)
1 +
b1
4b0
s
(Q
2
)
+
_
. (4)
Then the running coupling
s
(Q
2
) can be solved iteratively, to be,
s
(Q
2
) =
4
b
0
_
1
log(Q/)
2

b
1
b
2
0
log log(Q/)
2
[log(Q/)
2
]
2
+
_
. (5)
(b) Now we substitute (5) into the e
+
e
annihilation cross section, we get

(e
+
e
hadrons)
=
0
_
3
f
Q
2
f
_
_
1 +

s
+ a
2
_

s
_
2
+O(
3
s
)
_
=
0
_
3
f
Q
2
f
_
_
1 +
4
b
0
1
log(Q/)
2

4b
1
b
3
0
log log(Q/)
2
[log(Q/)
2
]
2
+
_
. (6)
Since the expression for the cross section is independent of renormalization scheme to
the order showed above, we conclude that the function coecients b
0
and b
1
are also
independent of the renormalization scheme.
1
2 A Direct test of the spin of the gluon
(a) We repeat the calculations in Part (c) of the Final Project I, with the gluon-quark
vertex replaced by a Yukawa vertex.
iM= Q
q
(ie)
2
(ig) u(k
1
)
_
i
/ k
1
+ / k
3
i
/ k
2
+ / k
3
_
v(k
2
)
i
q
2
v(p
2
)
u(p
1
) (7)
Then, use the trick described in Final Project I, we have
1
4
|iM|
2
=
Q
2
q
g
2
e
4
4s
2
tr (
/
p
1
/
p
2
)
tr
_
_
1
/ k
1
+ / k
3
1
/ k
2
+ / k
3
_
/ k
2
_
1
/ k
1
+ / k
3
1
/ k
2
+ / k
3
_
/ k
1
_
=
32Q
2
q
g
2
e
4
3s
2
(p
1
p
2
)(k
1
k
3
)(k
2
k
3
)
_
1
(k
1
+ k
3
)
2
+
1
(k
2
+ k
3
)
2
_
2
. (8)
Rewrite this in terms of x
q
, x
q
and x
3
, we get
1
4
|iM|
2
=
4Q
2
q
g
2
e
4
3s
2
(1 x
q
)(1 x
q
)
_
1
1 x
q
+
1
1 x
q
_
2
=
4Q
2
q
g
2
e
4
3s
2
x
2
3
(1 x
q
)(1 x
q
)
. (9)
Note the phase space integral for 3-body nal state is deduced in Final Project 1 to be
_
d
3
=
s
128
3
_
dx
q
dx
q
,
thus the dierential cross section is given by
d
2
dx
1
dx
2
(e
+
e
q qS) =
s
128
3

1
4
|M|
2
=
4
2
Q
2
q
3s

g
4
x
2
3
(1 x
q
)(1 x
q
)
. (10)
(b) Now let x
a
> x
b
> x
c
. Then there are six ways to associated the original three
variables x
q
, x
q
and x
3
to these three ordered ones. Note that the integral measure
dx
a
dx
b
does not change for dierent possibilities since the change of integral variables
(x
q
, x
q
) (x
q
, x
3
) or (x
q
, x
3
) generate an Jacobian whose absolute value is 1, due to
the constraint x
q
+ x
q
+ x
3
= 2. Therefore, summing up all 6 possibilities, we get
d
2
dx
a
dx
b
(e
+
e
q qS)
x
2
c
(1 x
a
)(1 x
b
)
+
x
2
b
(1 x
c
)(1 x
a
)
+
x
2
a
(1 x
b
)(1 x
c
)
, (11)
for q qS nal state, and
d
2
dx
a
dx
b
(e
+
e
q qS)
x
2
a
+ x
2
b
(1 x
a
)(1 x
b
)
+
x
2
b
+ x
2
c
(1 x
b
)(1 x
c
)
+
x
2
c
+ x
2
a
(1 x
c
)(1 x
a
)
, (12)
We plot these two distributions on the x
a
x
b
plain with the range x
a
> x
b
> x
c
, as
shown in Figure
2
0.5 0.6 0.7 0.8 0.9 1.0
0.5
0.6
0.7
0.8
0.9
1.0
x
a
x
b
0 1
0.5 0.6 0.7 0.8 0.9 1.0
0.5
0.6
0.7
0.8
0.9
1.0
x
a
x
b
0 1
Figure 1: The dierential cross sections of e
+
e
q qg as a function of x
1
and x
2
,
assuming gluon is a vector/scalar particle in left/right diagram.
3 Quark-gluon and gluon-gluon scattering
In this problem we evaluate the cross sections for two processes: (a) q q gg, (b)
gg gg.
(a) There are three diagrams contributing the process q(k
1
) q(k
2
) g(p
1
)g(p
2
) at the
tree level, as shown in Fig. 17.11 of Peskin&Schroeder. The amplitudes associated with
these diagrams are listed as follows:
iM
1
= (ig)
2
v(k
2
)/
(p
2
)
i(/ k
1
/
p
1
)
(k
1
p
1
)
2
/
(p
1
)u(k
1
)t
b
t
a
, (13a)
iM
2
= (ig)
2
v(k
2
)/
(p
1
)
i(/ k
1
/
p
2
)
(k
1
p
2
)
2
/
(p
2
)u(k
1
)t
a
t
b
, (13b)
iM
3
= (ig)gf
abc
_
g
(p
2
p
1
)
(2p
2
+ p
1
)
+ g
(p
2
+ 2p
1
)
i
(k
1
+ k
2
)
2
v(k
2
)
u(k
1
)
(p
1
)
(p
2
)t
c
. (13c)
It is convenient to evaluate these diagrams with initial and nal states of denite he-
licities. By P and CP symmetry of QCD, there are only two independent processes,
namely q
L
q
R
g
R
g
R
and q
L
q
R
g
R
g
L
, that could be nonzero. Lets evaluate them in
turn for the three diagrams. To begin with, we set up the kinematics:
k
1
= (E, 0, 0, E), p
1
= (E, E sin , 0, E cos ),
k
2
= (E, 0, 0, E), p
2
= (E, E sin, 0, E cos ). (14)
Then,
u
L
(k
1
) =
2E(0, 1, 0, 0), v
L
(k
2
) =
2E(1, 0, 0, 0).
L
(p
1
) =
1
2
(0, cos , i, sin ),
R
(p
1
) =
1
2
(0, cos , i, sin ),
L
(p
2
) =
1
2
(0, cos , i, sin),
R
(p
2
) =
1
2
(0, cos , i, sin). (15)
3
Now we begin the calculation. (In the following we use s
sin and c
= cos .)
iM
1
(q
L
q
R
g
R
g
R
) =
ig
2
E
2
t
b
t
a
t
(0, 0, 1, 0)
_
_
_
_
_
s
1 + c
1 + c
1 c
1 c
_
_
_
_
_
_
_
_
_
_
1 c
1 + c
1 + c
1 c
_
_
_
_
_
_
_
_
_
_
s
1 c
1 c
1 + c
1 + c
_
_
_
_
_
_
_
_
_
_
0
1
0
0
_
_
_
_
_
= ig
2
t
b
t
a
2E
2
t
(1 cos ) sin = ig
2
t
b
t
a
sin. (16)
iM
2
(q
L
q
R
g
R
g
R
) =
ig
2
E
2
t
a
t
b
u
(0, 0, 1, 0)
_
_
_
_
_
s
1 c
1 c
1 + c
1 + c
_
_
_
_
_
_
_
_
_
_
1 + c
1 c
1 c
1 + c
_
_
_
_
_
_
_
_
_
_
s
1 + c
1 + c
1 c
1 c
_
_
_
_
_
_
_
_
_
_
0
1
0
0
_
_
_
_
_
= ig
2
t
b
t
a
2E
2
u
(1 + cos ) sin = ig
2
t
a
t
b
sin. (17)
iM
3
(q
L
q
R
g
R
g
R
) =
g
2
f
abc
t
c
E
2
s
(0, 0, 1, 0)
_
_
4
_
_
_
_
_
c
_
_
_
_
_
_
_
_
_
_
_
_
0
1
0
0
_
_
_
_
_
= g
2
f
abc
t
c
sin = ig
2
[t
a
, t
b
] sin. (18)
Thus we nd that
iM(q
L
q
R
g
R
g
R
) =
_
iM
1
+ iM
2
+ iM
3
_
(q
L
q
R
g
R
g
R
) = 0. (19)
In the same manner, we calculate the amplitude for q
L
q
R
g
R
g
L
. This time, we
nd:
iM
1
(q
L
q
R
g
R
g
L
) =ig
2
t
b
t
a
sin , (20a)
iM
2
(q
L
q
R
g
R
g
L
) =ig
2
t
a
t
b
t
u
sin , (20b)
iM
3
(q
L
q
R
g
R
g
L
) = 0, (20c)
Therefore,
iM(q
L
q
R
g
R
g
L
) = ig
2
_
t
b
t
a
+ t
a
t
b
t
u
_
sin , (21)
and by crossing symmetry,
iM(q
L
q
R
g
L
g
R
) = ig
2
_
t
b
t
a
+ t
a
t
b
u
t
_
sin. (22)
4
There are two more nonzero amplitudes with q
R
q
L
initial states, which are identical
to the amplitudes above. Then we nd the spin- and color-summed/averaged squared
amplitude to be
1
3
2

1
2
2
spin,color
|M|
2
=
1
36
2 g
4
sin
2
_
_
tr (t
b
t
a
t
a
t
b
) + 2 tr (t
b
t
a
t
b
t
a
)
t
u
+ tr (t
a
t
b
t
b
t
a
)
t
2
u
2
_
+ (t u)
_
=
8
2
2
s
9
(1 cos
2
)
__
16
3
_
1 +
t
2
u
2
_
4t
3u
_
+ (t u)
_
=
512
2
2
s
27
_
t
u
+
u
t

9(t
2
+ u
2
)
4s
2
_
. (23)
Therefore the dierential cross section is given by
d
dt
=
32
2
s
27s
2
_
t
u
+
u
t

9(t
2
+ u
2
)
4s
2
_
. (24)
(b) Now consider the process g(k
1
)g(k
2
) g(p
1
)g(p
2
). The four tree level diagrams
are shown in Fig. 17.12 of Peskin&Schroeder. Their amplitudes are given by:
iM
1
= g
2
f
abc
f
cde
i
s
_
g
(k
1
k
2
)
+ g
(k
1
+ 2k
2
)
(2k
1
+ k
2
)
_
g
(p
2
p
1
)
(p
1
+ 2p
2
)
+ g
(p
1
+ 2p
2
)
(k
1
)
(k
2
)
(p
1
)
(p
2
),
(25a)
iM
2
= g
2
f
ace
f
bde
i
t
_
g
(k
1
+ p
1
)
(2p
1
k
1
)
(2k
1
p
1
)
_
g
(k
2
+ p
2
)
(2p
2
k
2
)
+ g
(p
2
2k
2
)
(k
1
)
(k
2
)
(p
1
)
(p
2
),
(25b)
iM
3
= g
2
f
ade
f
bce
i
u
_
g
(k
1
+ p
2
)
(2p
2
k
1
)
(2k
1
p
2
)
_
g
(k
2
+ p
1
)
(2p
1
k
2
)
+ g
(p
1
2k
2
)
(k
1
)
(k
2
)
(p
1
)
(p
2
),
(25c)
iM
4
=ig
2
_
f
abe
f
cde
_
(k
1
)
(p
1
)(k
2
)
(p
2
) (k
1
)
(p
2
)(k
2
)
(p
1
)
_
+ f
ace
f
bde
_
(k
1
) (k
2
)
(k
1
)
(p
2
) (k
1
)
(p
2
)(k
2
)
(p
1
)
_
+ f
ade
f
bce
_
(k
1
) (k
2
)
(p
1
)
(p
2
) (k
1
)
(p
1
)(k
2
)
(p
2
)
_
. (25d)
The choice for all external momenta and nal states polarizations are the same with that
in (a). Now to evaluate the amplitude g
R
g
R
g
R
g
R
, we also need the initial states
polarization vectors for right-handed gluons with momenta k
1
and k
2
, which are given
by
R
(k
1
) =
1
2
(0, 1, i, 0),
R
(k
2
) =
1
2
(0, 1, , 0). (26)
Then after some calculations, we nd,
iM
1
=ig
2
f
abe
f
cde
cos , (27a)
iM
2
= ig
2
f
ace
f
bde
19 + 7 cos 11 cos
2
+ cos
3
4(1 cos )
; (27b)
iM
3
= ig
2
f
ade
f
bce
19 7 cos 11 cos
2
cos
3
4(1 + cos )
; (27c)
5
iM
4
=ig
2
_
f
abe
f
cde
cos +
1
4
f
ace
f
bde
(3 + 2 cos cos
2
)
+
1
4
f
ade
f
bce
(3 2 cos cos
2
)
_
. (27d)
The sum of these four amplitudes is
iM(g
R
g
R
g
R
g
R
) =2ig
2
_
f
abe
f
cde
cos f
ace
f
bde
_
2
1 cos
+ cos
_
f
ade
f
bce
_
2
1 + cos
cos
__
= 4ig
2
_
f
ace
f
bde
1
1 cos
+ f
ade
f
bce
1
1 + cos
_
=2ig
2
_
f
ace
f
bde
s
t
+ f
ade
f
bce
s
u
_
. (28)
We can also obtain the amplitudes for g
L
g
R
g
L
g
R
and g
L
g
R
g
R
g
L
from the
result above by crossing symmetry, namely the change of variables (s, b) (u, d) and
(s, b) (t, c), which gives
iM(g
L
g
R
g
L
g
R
) = 2ig
2
_
f
ace
f
bde
u
t
+ f
abe
f
cde
u
s
_
, (29)
iM(g
L
g
R
g
R
g
L
) =2ig
2
_
f
abe
f
cde
t
s
f
ade
f
bce
t
u
_
. (30)
The amplitudes for g
L
g
L
g
L
g
L
, g
R
g
L
g
R
g
L
and g
R
g
L
g
L
g
R
are identical to
the amplitudes for g
R
g
R
g
R
g
R
, g
L
g
R
g
L
g
R
and g
L
g
R
g
R
g
L
, respectively, due
to parity conservation of QCD. It can be shown by the conservation of angular mo-
mentum that other helicity amplitudes all vanish. Therefore we have found all required
amplitude. To get the cross section, we take the square of these results.
|M(g
R
g
R
g
R
g
R
)|
2
= 4g
4
_
f
ace
f
bde
f
acf
f
bdf
s
2
t
2
+ f
ade
f
bce
f
adf
f
bcf
s
2
u
2
+ 2f
ace
f
bde
f
adf
f
bcf
s
2
tu
_
= 4g
4
_
tr (t
a
t
a
t
b
t
b
)
_
s
2
t
2
+
s
2
u
2
_
+ 2 tr (t
a
t
b
t
a
t
b
)
s
2
tu
_
= 288g
4
_
s
2
t
2
+
s
2
u
2
+
s
2
tu
_
, (31)
where t
a
is the generator of SU(3) group in adjoint representation which is related to
the structure constants by f
abc
= i(t
a
)
bc
. Thus tr (t
a
t
a
t
b
t
b
) = (C
2
(G))
2
d(G) = 72, and
tr (t
a
t
b
t
a
t
b
) = tr (t
a
t
b
[t
a
, t
b
]) + tr (t
a
t
a
t
b
t
b
) =
1
2
if
abc
tr ([t
a
, t
b
]t
c
) + (C
2
(G))
2
d(G)
=
1
2
f
abc
f
abd
tr (t
c
t
d
) + (C
2
(G))
2
d(G) =
1
2
(C
2
(G))
2
d(G),
which is 36 for SU(3). Similarly, we can work out the square of other amplitudes, to be
|M(g
L
g
R
g
L
g
R
)|
2
= 288g
4
_
u
2
t
2
+
u
2
s
2
+
u
2
st
_
, (32)
|M(g
L
g
R
g
R
g
L
)|
2
= 288g
4
_
t
2
s
2
+
t
2
u
2
+
t
2
su
_
. (33)
Therefore, the spin-averaged and squared amplitudes is
1
8
2

1
2
2
|M
2
| =
1
8
2
2
2
2 288g
4
_
6
2tu
s
2

2us
t
2

2st
u
2
_
= 72
2
2
s
_
3
tu
s
2

us
t
2

st
u
2
_
. (34)
6
Thus the dierential cross section is
d
dt
(gg gg) =
9
2
s
2s
2
_
3
tu
s
2

us
t
2

st
u
2
_
. (35)
4 The gluon splitting function
In this problem we calculate the gluon splitting function P
gg
(z) by evaluating the
amplitude of the virtual process g gg, as shown in Fig. 2.
Figure 2: The Gluon splitting process.
The momenta of initial and nal states are taken to be the same with that of Fig.
17.16 of Peskin&Schroeder. That is, we have
p = (p, 0, 0, p), q = (zp, p
, 0, zp), k = ((1 z)p, p
, 0, (1 z)p), (36)
and the polarization vectors associated with gluons are,
i
L
(p) =
1
2
(1, i, 0),
i
R
(p) =
1
2
(1, i, 0),
i
L
(q) =
1
2
(1, i,
p
zp
),
i
R
(q) =
1
2
(1, i,
p
zp
),
i
L
(k) =
1
2
(1, i,
p
(1 z)p
),
i
R
(k) =
1
2
(1, i,
p
(1 z)p
). (37)
Then we can evaluate the amplitude for the process g gg directly, which is given by
iM
abc
= gf
abc
__
(q) (p)
__
(p + q)
(k)
_
+
_
(q)
(k)
__
(k q) (p)
_
(k) (p)
__
(p + k)
(q)
_
. (38)
We evaluate the amplitudes with denite initial and nal polarizations in turn:
iM
abc
(g
L
(p) g
L
(q)g
L
(k)) =
2
_
1
1 z
+
1
z
_
gf
abc
p
, (39a)
iM
abc
(g
L
(p) g
L
(q)g
R
(k)) =
2z
1 z
gf
abc
p
, (39b)
iM
abc
(g
L
(p) g
R
(q)g
L
(k)) =
2(1 z)
z
gf
abc
p
, (39c)
iM
abc
(g
L
(p) g
R
(q)g
R
(k)) = 0. (39d)
By parity invariance, the amplitudes with right-handed initial gluon are dictated by the
results above. Note further that f
abc
f
abc
= 24, thus we have
1
2

1
8
spin,color
|M|
2
=
1
2

1
8
2 24 2g
2
p
2
__
1
1 z
+
1
z
_
2
+
z
2
(1 z)
2
+
(1 z)
2
z
2
_
=
12g
2
p
2
z(1 z)
_
1 z
z
+
z
1 z
+ z(1 z)
_
7
=
2e
2
p
2
z(1 z)
P
(0)
gg
(z), (40)
where the superscript (1) represents the part of the splitting function contributed from
the diagram calculated above, in parallel with the notation of Peskin&Schroeder. (See
17.100. for instance.) Therefore we get
P
(1)
gg
= 6
_
1 z
z
+
z
1 z
+ z(1 z)
_
. (41)
Besides, there should be a term proportional to (1 z) in P
gg
, which comes from the
zeroth order, as well as the corrections from P
qg
and P
gg
, where P
qg
(z) =
1
2
(z
2
+
(1 z)
2
). Now lets take it to be A(1 z), then the coecient A can be determined
by the following normalization condition (namely the momentum conservation):
1 =
_
1
0
dz z
_
2n
f
P
qg
(z) + P
(1)
gg
(z) + A(1 z)
, (42)
where n
f
is the number of fermion types, and the coecient 2 is from contributions
of both quarks and anti-quarks. To carry out the integral, we use the prescription
1
1z

1
(1z)+
, then it is straightforward to nd that A =
11
2

1
3
n
f
. Therefore,
P
gg
= 6
_
1 z
z
+
z
(1 z)
+
+ z(1 z)
_
+
_
11
2

n
f
3
_
(1 z). (43)
5 Photoproduction of heavy quarks
In this problem we study the production of a pair of heavy quark-antiquark by the
scattering of a photon o a proton. At the leading order at the parton level, the process
is contributed from the photon-gluon scattering, as shown in Figure 3.
Figure 3: Tree diagrams for the photoproduction of heavy quarks at the parton level.
The corresponding amplitude can be read from a similar process e
+
e
in QED.
From (5.105) of Peskin & Schroeder, we have the amplitude for e
+
e
2, which reads
(adapted to our notation for external momenta)
1
4
|M(e
+
e
2)|
2
= 2e
4
_
k
1
p
2
k
1
p
1
+
k
1
p
1
k
1
p
2
+ 2m
2
_
1
k
1
p
1
+
1
k
1
p
2
_
m
4
_
1
k
1
p
1
+
1
k
1
p
2
_
2
_
. (44)
Then the amplitude M(g Q

Q) can be obtained by making the exchange (k
1
, k
2
)
(p
1
, p2), replacing e
4
by e
2
g
2
, and also including the factor
1
8
Q
2
q
tr (t
a
t
a
) =
1
2
Q
2
q
taking
account of the color average, the electric charge of quarks, and the summation of color
indices, respectively. Then the amplitude in the present case is
1
4 8
|M(g Q

Q)|
2
= e
2
g
2
Q
2
q
_
p
1
k
2
p
1
k
1
+
p
1
k
1
p
1
k
2
+ 2m
2
_
1
p
1
k
1
+
1
p
1
k
2
_
8
m
4
_
1
p
1
k
1
+
1
p
1
k
2
_
2
_
. (45)
In partons center-of-mass frame, we have k
1
= (E, 0, 0, E), k
2
= (E, 0, 0, E), p
1
=
(E, p sin , 0, p cos ) and p
2
= (E, p sin , 0, p cos ), with p
2
= E
2
m
2
. Then p
1
k
1
=
E(E p cos ) and p
1
k
2
= E(E + p cos ). Then the dierential cross section is
d
d cos
=
s
Q
2
q
16
p
E
3
_
E
2
+ p
2
cos
2
2m
2
E
2
p
2
cos
2

2m
4
(E
2
p
2
cos
2
)
2
_
. (46)
Then the cross section for photon and proton initial state is given by
((k
1
) + p(k
2
) Q

Q) =
_
dxf
g
(x) ((k
1
) + g(xk
2
) Q

Q). (47)
6 Behavior of parton distribution functions at small
x
(a) In this problem we study the solution of A-P equations at small x with certain
approximations. Firstly, we show that the A-P equations,
d
d log Q
f
g
(x, Q) =

s
(Q
2
)
_
1
x
dz
z
_
P
gq
(z)
f
_
f
f
_
x
z
, Q
_
+ f
f
_
x
z
, Q
_
_
+ P
gg
(z)f
g
_
x
z
, Q
_
_
, (48)
d
d log Q
f
f
(x, Q) =

s
(Q
2
)
_
1
x
dz
z
_
P
qq
(z)f
f
_
x
z
, Q
_
+ P
qg
(z)f
g
_
x
z
, Q
_
_
, (49)
d
d log Q
f
f
(x, Q) =

s
(Q
2
)
_
1
x
dz
z
_
P
qq
(z)f
f
_
x
z
, Q
_
+ P
qg
(z)f
g
_
x
z
, Q
_
_
, (50)
can be rewritten as a dierential equation with variable = log log(Q
2
/
2
). To see this,
we note that d/d log Q = 2e
d/d, and to 1-loop order,

s
(Q) = 2/
_
b
0
log(Q/)
_
=
(4/b
0
)e
, so we have
d
d
f
g
(x, ) =
2
b
0
_
1
x
dz
z
_
P
gq
(z)
f
_
f
f
_
x
z
,
_
+ f
f
_
x
z
,
_
_
+ P
gg
(z)f
g
_
x
z
,
_
_
, (51)
d
d
f
f
(x, ) =
2
b
0
_
1
x
dz
z
_
P
qq
(z)f
f
_
x
z
,
_
+ P
qg
(z)f
g
_
x
z
,
_
_
, (52)
d
d
f
f
(x, ) =
2
b
0
_
1
x
dz
z
_
P
qq
(z)f
f
_
x
z
,
_
+ P
qg
(z)f
g
_
x
z
,
_
_
. (53)
(b) Now we apply the approximation that 1) gluon PDF dominates the integrand in
the A-P equations and 2) the function g(x, Q) = xf
g
(x, Q) is slowly varying with x when
x is small. Then, dene w = log(1/x), which gives d/dw = xd/dx, we can calculate
2
w
g(x, ) =x
d
dx
_
x

f
g
(x, Q)
_
x
d
dx
_
2x
b
0
_
1
x
dz
z
P
gg
(z)f
g
_
x
z
, Q
_
_
9
=
2
b
0
xP
gg
(x)f
g
(x, Q)
2x
b
0
_
1
x
dzP
gg
(z)
d
dx
_
x
z
f
g
_
x
z
, Q
__
2
b
0
xP
gg
(x)f
g
(x, Q). (54)
From the result of Problem 17.4 we know that xP
gg
(x) = 6 as x 0. Therefore the
A-P equation for f
g
becomes
2
w
g(x, ) =
12
b
0
g(x, ). (55)
Then we verify that
g = K(Q
2
) exp
_
_
48
b
0
w(
0
)
_
1/2
_
(56)
is an approximation solution to the dierential equation above when w 1, where
K(Q
2
) is an initial condition. We apply
2
/w on this expression, to get
2
w
g(w, ) =
1
4
48
b
0
w(
0
)
exp
_
_
48
b
0
w(
0
)
_
1/2
_
_
2(
0
)
K(Q
2
)
+
_
1 +
_
48
b
0
w(
0
)
_
K(Q
2
)
_
. (57)
In the limit w 1, the square root term in the last line dominates, thus
2
w
g(w, )
b
0
12
K(Q
2
) exp
_
_
48
b
0
w(
0
)
_
1/2
_
=
12
b
0
g(w, ). (58)
(c) Then we consider the A-P equation for quarks. If we adopt the approximation in
(b) again, namely, the gluon PDF dominates and the function q(x, ) = xf
f
(x, Q) is
slowly varying, then we have
q(x, ) = x

f
f
(x, ) =
2x
b
0
_
1
x
dz
z
P
qg
(z)f
g
_
x
z
,
_
=
2
b
0
_
1
x
dzP
qg
(z) g
_
x
z
,
_
=
1
b
0
_
1
x
dz
_
z
2
+ (1 z)
2
_
g
_
x
z
,
_
2
b
0
_
1
6
(2z
3
2z
2
+ 3z) g
_
x
z
,
_
_
1
x
2
3b
0
g(x, ), (59)
where we have used x 1 and q(x, )/x 0. Then, we verify that
q =
_

0
27b
o
w
K(Q
2
) exp
_
_
48
b
0
w(
0
)
_
1/2
_
(60)
is again an approximate solution to the equation derived above, in the limit w 1. In
fact,
q(x, ) = exp
_
_
48
b
0
w(
0
)
_
1/2
__
2
3b
0
K(Q
2
)
+
1
18
3
b
0
w(
0
)
_
K(Q
2
) + 2(
0
)
K(Q
2
)
_
_
2
3b
0
K(Q
2
) exp
_
_
48
b
0
w(
0
)
_
1/2
_
=
2
3b
0
g(x, ). (61)
(d) We use the tted formula of K(Q
2
) to plot the PDFs of gluon and quarks in
Figure.
10
0.00 0.05 0.10 0.15 0.20
0.0
0.5
1.0
1.5
2.0
2.5
3.0
x
xf
g
x,Q
2
xf
f
x,Q
2
Figure 4: Approximate parton distribution functions at small x with Q = 500GeV.

11
Chapter 18
Zhong-Zhi Xianyu

1 Matrix element for proton decay
(a) We estimate the order of magnitude of the proton lifetime, through the decay
p e
+
0
, based on the following operator,
O
X
=
2
m
2
X
ijk
e
R
u
Ri
u
Lj
d
Lk
, (1)
where m
X
is the scale of this higher dimensional operator, whose typical value is around
10
16
GeV, and i, j, k are color indices for quarks, , , are spinor indices. Then,
the amplitude M of this decay process should be proportional to m
2
X
. Note that the
amplitude M has mass dimension 1, thus we should have M m
3
p
m
2
X
with m
p
the
proton mass. Now take m
X
10
16
GeV and m
p
1GeV, we have the decay width

1
8
1
2m
p
|M|
2
1
16
m
5
p
m
4
X
10
65
GeV 10
33
yr
1
. (2)
(b) Now we consider the rst order QCD correction to the estimation above. The
correction comes from virtual gluon exchange among three quarks in the operator. To
evaluate these 1-loop diagrams, we rstly xed the renormalization condition of O
X
to
be
u
Lj
d
Lk
u
Ri
= i
ijk
. (3)
The 1-loop diagrams are shown in Figure 1. The Feynman rules can be written in two-
component spinor notations. The left-handed spinors propagator reads i(p )/p
2
, the
right-handed spionrs propagator is i(p )/p
2
, the QCD interaction between quark and
gluon is i[
Li

(t
a
)
ij
Lj
+
Ri
(t
a
)
ij
Rj
], and the vertex corresponding O
X
reads
i
ijk
. Then the rst diagram reads

(a) = i(ig)
2
imn
(t
a
)
mj
(t
a
)
nk
_
d
d
q
(2)
d
i
q
2
_
iq
q
2

_
iq
q
2

1
Figure 1: 1-loop QCD correction to the eective operator of proton decay.
=g
2
2
3
_
ijk
_
d
d
q
(2)
d
q
q
6
=g
2
2
3
_
ijk
16
i
4(4)
2
2
=
8g
2
3(4)
2
2
i
ijk
, (4)
where the Pauli matrices is simplied as follows,
= (
T
= (
T
= 4(
T
= 4(
= 16
, (5)
in which we used the fact that
= i
2
and
T
2
=
2
. In the computation of this

diagram, we also used
imn
(t
a
)
mj
(t
a
)
nk
= (2/3)
ijk
. The coecient of this equality
can be easily justied by contracting both sides with
ijk
. Similarly, we compute the
second diagram, as follows,
(b) = i(ig)
2
imn
(t
a
)
mj
(t
a
)
nk
_
d
d
q
(2)
d
i
q
2
_
iq
q
2

_
iq
q
2

=g
2
2
3
_
ijk
4
i
4(4)
2
2
=
2g
2
3(4)
2
2
i
ijk
, (6)
where we used the identity (
= 2
. The third diagram gives the same

result as the second one. Therefore, we get the counterterm for the operator O
X
in MS
scheme to be
OX
=
4g
2
(4)
2
_
2
log M
2
_
, (7)
where M
2
is the renormalization scale. We further recall that the eld strength renor-
malization counterterm for quarks in QCD is given by
2
=
4g
2
3(4)
2
_
2
log M
2
_
, (8)
then the anomalous dimension of operator O
X
is given by
= M

M
_
OX
+
3
2

2
_
=
4g
2
(4)
2
. (9)
Therefore this QCD correction will enhance the operator strength by a factor of
_
log(m
2
X
/
2
)
log(m
2
p
/
2
)
_
a0/2b0
, (10)
2
where 200GeV, a
0
= 4 is the coecient from anomalous dimension, and b
0
=
11 (2/3)n
f
= 7 is the 1-loop coecient of QCD function. Taking m
X
= 10
16
GeV
and m
p
= 1GeV, this factor is about 2.5. Then the decay rate of proton is enhanced by
a factor of 2.5
2
6.3.
2 Parity-violating deep inelastic form factor
(a) We rstly compute the amplitude of the neutrino deep inelastic scattering through
charged current interaction, which reads
iM(p
X) =
ig
2
2m
2
W
u(k
_
1
5
2
_
u(k)
_
d
4
xe
iqx
X|
_
J
+
(x) + J
(x)
_
|P. (11)
Then the squared amplitude with initial protons spins averaged and nal state X
summed is
1
2
|M|
2
=
1
2
g
4
4m
4
W
spin
_
u(k
_
1
5
2
_
u(k) u(k)
_
1 +
5
2
_
u(k
)
_
X
_
d
X
P|
_
J
+
(x) + J
(x)
_
|XX|
_
J
+
(0) + J
(0)
_
|P. (12)
The trace factor can be straightforwardly worked out to be
L
spin
_
u(k
_
1
5
2
_
u(k) u(k)
_
1 +
5
2
_
u(k
)
_
= tr
_
_
1
5
2
_
/ k
_
1 +
5
2
_
/ k
_
= 2
_
k
+ k
k k
+ i
_
. (13)
Then, use the optical theorem, we have
L
X
_
d
X
P|
_
J
+
(x) + J
(x)
_
|XX|
_
J
+
(0) + J
(0)
_
|P
= 2 Im
_
L
W
()
_
, (14)
with
W
()
= 2i
_
d
4
xe
iqx
P|T
_
J
(x)J
+
(0)
_
|P, (15)
Therefore, the cross section is
(p
X) =
1
2s
_
d
3
k
(2)
3
1
2k

1
2
|M|
2
=
1
2s
_
dxdy
ys
(4)
2

g
4
4m
4
W
Im
_
L
W
()
_
, (16)
and the dierential cross section is thus given by
d
2
dxdy
(p
X) =
yG
2
F
2
2
Im
__
k
+ k
k k
+ i
_
W
()
. (17)
3
(b) The lepton momentum tensor obtained in (a) is
L
= 2(k
+ k
k k
+ i
). (18)
Then it is straightforward to see that q
= (k k
= 0 and q
= 0. As a
consequence, any term in W
()
proportional to q
or q
is irrelevant. Therefore we
can rewrite the tensor W
()
in terms of three form factors W
()
i
(i = 1, 2, 3). That is,
W
()
= g
W
()
1
+ P
W
()
2
+ i
W
()
3
+ . (19)
Then the deep inelastic scattering cross section becomes
d
2
dxdy
(p
X) =
yG
2
F
2
2
_
2(k k
) ImW
()
1
+ 2(P k)(P k
) ImW
()
2
4
_
(P k)(q k
) (q k)(P k
)
_
ImW
()
3
_
. (20)
(c) Now we evaluate ImW
()
1,2,3
in the parton model. Firstly, W
()
can be written as
W
()
= 2i
_
d
4
xe
iqx
_
1
0
d
f
f
f
()
1
q
f
(P)
T{J
(x)J
+
(0)}
q
f
(P)
_
, (21)
and be evaluated in terms of Feynman diagrams displayed in Fig. 18.10 of Peskin &
Schroeder. For the rst diagram, we have
2i
_
1
0
d
_
f
d
()
1
u(p)
_
1
5
2
_
i
/
p +
/
q + i
_
1
5
2
_
u(p)
+f
u
()
1
u(p)
_
1
5
2
_
i
/
p +
/
q + i
_
1
5
2
_
u(p)
_
, (22)
where p = P. Then, averaging/summing over initial/nal spin states gives
2
_
1
0
d
_
f
d
()
1

1
2
tr
_
/
p
1
5
2
(
/
p +
/
q)
1
5
2
_
+ f
u
()
1

1
2
tr
_
/
p
1
5
2
(
/
p +
/
q)
1
5
2
_
_
1
2p q + q
2
+ i
_
1
0
d
_
_
f
d
() + f
u
()
__
4
2
P
2P qg
_
+
_
f
d
() f
u
()
_
2i
_
1
2p q + q
2
+ i
, (23)
where we have dropped terms containing q
or q
in the last line. Then it is easy to

read from this expression that
ImW
()
1
= 2P q
_
1
0
d
_
f
d
() + f
u
()
Im
_
1
2p q + q
2
+ i
_
, (24)
ImW
()
2
=
_
1
0
d 4
_
f
d
() + f
u
()
Im
_
1
2p q + q
2
+ i
_
, (25)
ImW
()
3
= 2
_
1
0
d
_
f
d
() f
u
()
Im
_
1
2p q + q
2
+ i
_
, (26)
where 2P q = ys, and
Im
_
1
2p q + q
2
+ i
_
=

ys
( x). (27)
4
Note that the second diagram in Fig. 18.10 of Peskin & Schroeder does not contributes,
as explained in the book. Therefore we conclude that
ImW
()
1
=
_
f
d
(x) + f
u
(x)
, (28)
ImW
()
2
=
4x
ys
_
f
d
(x) + f
u
(x)
, (29)
ImW
()
3
=
2
ys
_
f
d
(x) f
u
(x)
. (30)
(d) The analysis above can be easily repeated for the left-handed current J
fL
of single
avor f, dened by J
fL
=

f
P
L
f where P
L
(1
5
)/2. Then, dene
W
fL
= 2i
_
d
4
xe
iqx
T{J
fL
(x)J
fL
(0)}
P
_
, (31)
and its decomposition,
W
fL
= g
W
1fL
+ P
W
2fL
+ i
W
3fL
+ . (32)
We see that it amounts to the replacement in the nal result that d f and u

f.
Therefore,
ImW
()
1fL
=
_
f
f
(x) + f
f
(x)
, (33)
ImW
()
2fL
=
4x
ys
_
f
f
(x) + f
f
(x)
, (34)
ImW
()
3fL
=
2
ys
_
f
f
(x) f
f
(x)
. (35)
(e) Now we perform OPE on W
fL
. Firstly,
_
d
4
xe
iqx
J
fL
(x)J
fL
(0)
_
d
4
xe
iqx
_
q
P
L
q(x) q
P
L
q(0) + q
P
L
q(x) q
P
L
q(0)
_
= q
P
L
i(i/ +
/
q)
(i + q)
2

P
L
q +
_
, q q
_
. (36)
Then, the rst term in the last line can be written as
q
P
L
i(i/ +
/
q)
(i + q)
2

P
L
q =
1
2
_
q
i(i/ +
/
q)
(i + q)
2

q q
i(i/ +
/
q)
(i + q)
2

5
q
_
=
i
2
q
_
2
(
(i
)
) g
/
q i
(i + q)
1
Q
2
n=0
_
2iq
Q
2
_
n
q, (37)
where we have symmetrize the indices for the rst two terms in the square bracket
and antisymmetrize the indices for the third term, by using the equalities
1
2
(
) = g
+ g
and
1
2
(
5
) = i
, and
terms proportional to q
or q
have also been dropped. The (anti)symmetrization can

be understood by looking at (32), where the terms with no
5
are symmetric on while
the term involving
5
is antisymmetric on . Therefore, when including the second
term in (36), we should keep terms of even powers in q for symmetric indices and of
odd powers in q for antisymmetric .
5
Now, with these understood, and using the denition of twist-2, spin-n operator,
O
(n)1n
f
= q
f
(1
(iD
2
) (iD
n)
)q
f
traces, (38)
we have,
i
_
d
4
xe
iqx
J
fL
(x)J
fL
(0) =
n>0, even
2
(2q
1
) (2q
n2
)
(Q
2
)
n1
O
(n)1n2
f
1
2
g
n>0, even
(2q
1
) (2q
n
)
(Q
2
)
n
O
(n)1n
f
i
n>0 odd
(2q
1
) (2q
n1
)
(Q
2
)
n
O
(n)1n1
f
. (39)
Then, using P|O
(n)1n
f
|P = 2A
n
f
P
1
P
n
, we can get W
fL
to be
W
fL
= 8P
n>0, even
(2q P)
n2
(Q
2
)
n1
A
n
f
2g
n>0, even
(2q P)
n
(Q
2
)
n
A
n
f
+ 4i
n>0, odd
(2q P)
n1
(Q
2
)
n
A
n
f
. (40)
So we can read out
W
1fL
= 2
n>0, even
(2q P)
n
(Q
2
)
n
A
n
f
, (41)
W
2fL
= 8
n>0, even
(2q P)
n2
(Q
2
)
n1
A
n
f
, (42)
W
3fL
= 4
n>0, odd
(2q P)
n1
(Q
2
)
n
A
n
f
. (43)
(f ) Now we use W
3fL
obtained above to derive a sum rule for parton distribution f
f
,
dened by
f
f
(x, Q
2
) =
ys
2
ImW
3fL
(x, Q
2
), (44)
where x = Q
2
/ and = 2P q = ys. The the analytic behavior of W
3fL
on the
v-complex plane is shown in Fig. 18.11 of Peskin & Schroeder. Thus we can dene the
contour integral
I
3n
=
_
d
2i
1
n
W
3fL
(, Q
2
), (45)
where the contour is a small circle around the origin = 0. This integral picks up the
coecient of
(n1)
term, namely, I
3n
= 4A
n
f
/(Q
2
)
n
. On the other hand, the contour
can be deformed as shown in Fig. 18.12 of Peskin & Schroeder. Then the integral can
be evaluated as
I
3n
= 2
_

Q
2
f
d
2i
1
n
(2i) ImW
3fL
(, Q
2
) =
4
(Q
2
)
n
_
1
0
dxx
n1
f
f
(x, Q
2
). (46)
Therefore we get the sum rule,
_
1
0
dxx
n1
f
f
(x, Q
2
) = A
n
f
. (47)
6
3 Anomalous dimensions of gluon twist-2 operators
In this problem we nish evaluating anomalous dimension matrix
n
in (18.180) of
Peskin & Schroeder, given by
n
=
g
2
(4)
2
_
a
n
ff
a
n
fg
a
n
gf
a
n
gg
_
(48)
where a
n
ff
has already been evaluated explicitly in the book. Here we evaluate the
remaining three elements. The needed Feynman rules involving operators O
(n)
f
and
O
(n)
g
are listed as follows:
k
= / ( k)
n1
,
k
a
b
= 2
_
g
( k)
n
+ k
2
( k)
n2
2k
(
)
( k)
n1
_
ab
,
k
1

k
2
k
3
a
b
c
=2igf
abc
g
j=1
( k
1
)
j1
( k
2
)
nj
+ (cyclic permutations on ak
1
, bk
2
, ck
3
) + .
In the last expression, we list only terms containing a metric tensor g
, and the ignored

terms (marked by ) are irrelevant in the following calculations. To be clear, we have
introduced a source J
(n)
to these operators, namely, we write L = J
(n)
1n
O
(n)1n
f,g
,
with J
(n)
1n
=
1

n
, and
2
= 0. As can be easily seen, this source automatically
projects the operator O
(n)
f,g
to its symmetric and traceless part.
(a) Firstly, we consider a
n
fg
, which can be got by evaluating the following two diagrams.
p
k
With the Feynman rules listed above, the rst diagram reads,
(ig)
2
_
d
4
k
(2)
4
(1) tr
_
t
b
i
/ k
/
i
/ k
t
a
i
/ k
/
p
_
( k)
n1
=ig
2
tr [t
a
t
b
]
_
d
4
k
(2)
4
_
1
0
dx
2(1 x)
(k
2
)
3
( k)
n1
tr [
/ k/ / k
(/ k
/
p)]. (49)
We need to extract terms of proportional to g
( p)
n
and of logarithmical divergence.
This needs some manipulations on the numerator of the integrand. We rstly evaluate
the gamma trace, keep terms containing at least two powers of k, and shift the variable
k
= k
xp
. Then we pick up terms containing two k
, which contributes
7
to logarithmical divergence. At last we symmetrize the indices according to k
k
2
g
/4. The detailed steps are given as follows.

( k)
n1
tr [
/ k/ / k
(/ k
/
p)]
_
16( k)
n
k
_
4( k)
n
(k 2p) kg
_
4( k
)
n1
( p)k
2
g
_
16x
n
( p)
n
k
_
4nx
n
(k
p)( k
)( p)
n1
g
+ 4n(x 2)x
n1
(k
p)( k
)( p)
n1
g
+ 4x
n
( p)
n
k
2
g
_
8(n 1)x
n1
(k
p)( k
)( p)
n1
g
+ 4x
n1
( p)
n
k
2
g
_
4x
n
_
( p)
n
k
2
g
_
nx
n
+ n(x 2)x
n1
+ 4x
n
_
( p)
n
k
2
g
_
2(n 1)x
n1
+ 4x
n1
_
( p)
n
k
2
g
=(2nx
n
+ 2x
n1
)( p)
n
k
2
g
. (50)
Then it is straightforward to nish the loop integral,
ig
2
tr [t
a
t
b
]( p)
n
g
_
1
0
dx2(1 x)(2nx
n
+ 2x
n1
)
_
d
4
k
(2)
4
k
2
(k
2
)
3
=
g
2
(4)
2
2(n
2
+ n + 2)
n(n + 1)(n + 2)
(2
d
2
)
2d/2
( p)
n
ab
g
. (51)
The second diagram contributes an identical term for n even. The two diagrams sum to
g
2
(4)
2
2(n
2
+ n + 2)
n(n + 1)(n + 2)
(2
d
2
)
2d/2

_
2( p)
n
ab
g
_
. (52)
Therefore the corresponding counterterm reads
fg
=
g
2
(4)
2
2(n
2
+ n + 2)
n(n + 1)(n + 2)
(2
d
2
)
(M
2
)
2d/2
, (53)
and the anomalous dimension element reads
n
fg
= M

M

fg
=
g
2
(4)
2
4(n
2
+ n + 2)
n(n + 1)(n + 2)
, (54)
and thus,
a
n
fg
=
4(n
2
+ n + 2)
n(n + 1)(n + 2)
. (55)
(b) Then we consider a
n
gf
and a
n
gg
. This time we need to evaluate the following four
diagrams.
The rst diagram contributes to a
n
gf
, which reads
2(ig)
2
_
d
4
k
(2)
4
t
a
i
/
p / k
t
a
_
i
k
2
_
2
8
_
g
( k)
n
+ k
2
( k)
n2
2k
(
)
( k)
n1
_
=2ig
2
C
2
(N)
_
d
4
k
(2)
4
_
1
0
dx
2(1 x)
(k
2
)
3
_
(
/
p / k)
( k)
n
+ / (
/
p / k)/ ( k)
n2
k
2
_
/ (
/
p / k)/ k + / k(
/
p / k)/
_
( k)
n1
_
(56)
To nd the pieces proportional to / ( p)
n1
and of logarithmical divergence, we manip-
ulate on the expression in the square bracket, shifting the variable k
= k
xp
,
extracting terms with two factors of k
, symmetrizing the integrand with k
k
2
g
/4, and throwing away terms proportional to

2
(= 0). This gives
_
(
/
p / k)
( k)
n
_
+
_
/ (
/
p / k)/ ( k)
n2
k
2
_
__
/ (
/
p / k)/ k + / k(
/
p / k)/
_
( k)
n1
_
=
_
2(
/
p / k)( k)
n
_
+
_
2/
_
(p k)
_
( k)
n2
k
2
_
_
2
_
_
(p k)
_
/ k +
_
(p k) k
_
/ (
/
p / k)( k)
_
( k)
n1
_
_
2nx
n1
/ k
( k
)( p)
n1
_
+
_
2x
n1
(k
p)( k
)( p)
n2
+ 2(n 2)(1 x)x
n2
(k
p)( k
)( p)
n2
+ (1 x)x
n2
( p)
n1
k
2
_
+
_
2x
n1
/ k
( k
)( p)
n1
2(n 1)(1 x)x
n2
( k
)( p)
n1
2/
_
x
n1
k
2
( p)
n1
(n 1)x
n1
(k
p)( k
)( p)
n2
+ (n 1)(1 x)x
n2
(k
p)( k
)( p)
n2
_
2nx
n1
/ k
( k
)( p)
n1
_
_
n
2
x
n1
_
/ k
2
( p)
n1
+
_
x
n1
+ n(1 x)x
n2
_
/ k
2
( p)
n1
+
_
(n 1)(1 x)x
n2
+ 2x
n1
_
/ k
2
( p)
n1
=
_
n
2
x
n1
+ x
n2
_
/ k
2
( p)
n1
. (57)
Then we have
2ig
2
C
2
(N)/ ( p)
n1
_
1
0
dx2(1 x)
_
n
2
x
n1
+ x
n2
_
_
d
4
k
(2)
4
k
2
(k
2
)
3
=
g
2
C
2
(N)
(4)
2
2(n
2
+ n + 2)
n(n
2
1)
(2
d
2
)
2d/2
/ ( p)
n1
, (58)
which gives the counterterm coecient,
n
gf
=
g
2
C
2
(N)
(4)
2
2(n
2
+ n + 2)
n(n
2
1)
(2
d
2
)
(M
2
)
2d/2
(59)
Then, in a similar way as in (a), we get
n
gf
=
g
2
C
2
(N)
(4)
2
4(n
2
+ n + 2)
n(n
2
1)
, (60)
and for N = 3, C
2
(N) = 4/3, we get
a
n
gf
=
16
3
(n
2
+ n + 2)
n(n
2
1)
. (61)
9
The second to fourth diagrams contribute to a
n
gg
. Now we evaluate them in turn.
The second one reads,
2g
2
f
ace
f
bde
cd
_
d
4
k
(2)
4
_
i
k
2
_
2
i
(p k)
2
_
g
(p + k)
+ g
(p 2k)
+ g
(k 2p)
_
g
(p + k)
+ g
(2k p)
+ g
(2p k)
_
g
( k)
n
+ k
2
( k)
n2
2k
(
)
( k)
n1
_
2ig
2
C
2
(G)
ab
_
d
4
k
(2)
4
_
1
0
dx
2(1 x)
(k
2
)
3
_
8( k)
n
k
_
( k)
n
k
2
+ 2( k)
n
(k p) 8( k)
n1
( p)(k p) + 4( k)
n2
( p)
2
k
2
_
g
_
2ig
2
C
2
(G)g
ab
( p)
n
_
1
0
dx2(1 x)
_
_
3 +
n
2
_
x
n
1
2
nx
n1
2x
n2
_
_
d
4
k
(2)
4
k
2
(k
2
)
3
=
g
2
C
2
(G)
(4)
2
_
4
n + 2

6
n + 1
+
4
n

4
n 1
_
(2
d
2
)
2d/2
(2)g
ab
( p)
n
. (62)
The third diagram reads (where an additional 1/2 is the symmetry factor),
1
2
2ig
2
f
acd
f
bcd
_
d
4
k
(2)
4
i
k
2
i
(p k)
2
_
g
(p + k)
+ g
(p 2k)
+ g
(k 2p)
j=1
_
g
( (p k))
j1
( p)
nj
g
( p)
j1
( k)
nj
_
ig
2
C
2
(G)g
ab
( p)
n
n
j=1
_
1
0
dx
_
(1 + x)x
nj
(x 2)(1 x)
j1
_
_
d
4
k
(2)
4
1
(k
2
)
2
=
g
2
C
2
(G)
(4)
2
g
ab
(2
2
d
)
2d/2
( p)
n
j=1
_
1
j
+
1
j + 1
+
1
n j + 1
+
1
n j + 2
_
g
2
C
2
(G)
(4)
2
_
2
n
j=2
1
j
+
1
n + 1
+ 1
_
(2
2
d
)
2d/2
(2)g
( p)
n
. (63)
The contribution from fourth diagram is identical to the one from the third diagram.
Summing the last three diagram together, we get
g
2
C
2
(G)
(4)
2
_
4
(n + 1)(n + 2)
+
4
n(n + 1)
4
n
j=2
1
j
2
_
(2
d
2
)
2d/2
(2)g
ab
(p)
n
. (64)
Thus the corresponding counterterm is
g
=
g
2
C
2
(G)
(4)
2
_
4
(n + 1)(n + 2)
+
4
n(n + 1)
4
n
j=2
1
j
2
_
(2
d
2
)
(M
2
)
2d/2
. (65)
10
As a result,
n
gg
= M

M
(
g
+
3
)
=
2g
2
(4)
2
__
4
(n + 1)(n + 2)
+
4
n(n + 1)
4
n
j=2
1
j

1
3
_
C
2
(G)
4
3
n
f
C(N)
_
, (66)
therefore, for N = 3, C
2
(N) = 4/3 and C(N) = 1/2, we have,
a
n
gg
= 6
_
4
(n + 1)(n + 2)
+
4
n(n + 1)
4
n
j=2
1
j

1
3

2
9
n
f
_
. (67)
4 Deep inelastic scattering from a photon
(a) The A-P equation for parton distributions in the photon can be easily written
down by using the QED splitting functions listed in (17.121) of Peskin & Schroeder.
Taking account of quarks electric charge properly, we have,
d
d log Q
f
q
(x, Q) =
3Q
2
q
_
1
x
dz
z
_
P
ee
(z)f
q
_
x
z
, Q
_
+ P
e
(z)f
_
x
z
, Q
_
_
, (68)
d
d log Q
f
q
(x, Q) =
3Q
2
q
_
1
x
dz
z
_
P
ee
(z)f
q
_
x
z
, Q
_
+ P
e
(z)f
_
x
z
, Q
_
_
, (69)
d
d log Q
f
(x, Q) =
q
3Q
2
q
_
1
x
dz
z
_
P
e
(z)
_
f
q
_
x
z
, Q
_
+ f
q
_
x
z
, Q
_
_
+ P
(z)f
_
x
z
, Q
_
_
, (70)
where the splitting functions are
P
ee
(z) =
1 + z
2
(1 z)
+
+
3
2
(1 z), (71)
P
e
(z) =
1 + (1 z)
2
z
, (72)
P
e
(z) = z
2
+ (1 z)
2
, (73)
P
(z) =
2
3
(1 z). (74)
We take q = u, d, c, s, and Q
u,c
= +2/3, Q
d,s
= 1/3. The factor 3 in the A-P equations
above takes account of 3 colors. Since no more leptons appear in nal states other than
original e
+
e
, they are not included in the photon structure. With the initial condition
f
(x, Q
0
) = (1x) and f
q, q
(x, Q
0
) = 0 where Q
0
= 0.5GeV, these distribution functions
can be solved from the equations above to the rst order in , to be
f
q
(x, Q) = f
q
(x, Q) =
3Q
2
q
2
log
Q
2
Q
2
0
_
x
2
+ (1 x)
2
, (75)
f
(x, Q) =
_
1
q
Q
2
q
log
Q
2
Q
2
0
_
(1 x). (76)
(b) The formulation of deep inelastic scattering from a photon is similar to the one
for the proton, as described in Peskin & Schroeder. The process can be formulated as
11
a two-photon scattering, with one photon being hard and the other one play the role of
proton, which has the internal structure as shown in (a). Therefore, we can write down
the corresponding current product as
W
= i
_
d
4
xe
iqx
|T{J
(x)J
(0)}|, (77)
which can be again expanded in terms of scalar form factors,
W
=
_
g
+
q
q
2
_
W
1
+
_
P
P q
q
2
__
P
P q
q
2
_
W
2
. (78)
After operator product expansion, the form factor W
2
can be expressed as
W
2
= 3
q
Q
2
q
n
8
Q
2
(2q P)
n2
(Q
2
)
n2
A
n
q
(Q
2
), (79)
and A
n
q
(Q
2
) is a scale-dependent quantity, whose scaling behavior is dictated by the
anomalous dimension matrix . This matrix can be evaluated again from the diagrams
in Fig. 18.13 in Peskin & Schroeder and in gures of last problem. The only dierence is
that we should replace the gluon eld with photon eld. Therefore it is straightforward
to see that a
n
= 0. For a
n
qq
and a
n
q
, we should take away the group factor C
2
(N) = 4/3,
while for a
n
q
, we should take away the factor tr (t
a
t
b
) =
ab
/2. In addition, we should
also include the factor Q
2
q
corresponding to electric charge of each quark. Then we have,
a
n
qq
=2Q
2
f
_
1 + 4
n
j=2
1
j

2
n(n + 1)
_
, (80)
a
n
q
=
8Q
2
f
(n
2
+ n + 2)
n(n + 1)(n + 2)
, (81)
a
n
q
=
4Q
2
f
(n
2
+ n + 2)
n(n
2
1)
, (82)
a
n
= 0. (83)
(c) The n = 2 moment photon structure function can be worked out through the
moment sum rules (18.154) in Peskin & Schroeder, where the matrix elements A
n
q
in
our case is a scale-dependent quantity. This dependence can be found by evaluating
the anomalous dimension matrix of operator O
(2)
q
as is done below (18.185) of Peskin &
Schroeder, but with dierent entries, given by
=

4
_
_
_
a
2
uu
0 3 2a
2
u
0 a
2
dd
3 2a
2
u
a
2
u
a
2
d
a
2
_
_
_ =

4
_
_
_
64
27
0
32
27
0
16
27
8
27
64
27
16
27
0
_
_
_. (84)
(d) As can be inferred from (a), the photon structure function f
(x, Q) is originally
peaked at x = 1 for Q = Q
0
, and the peak shifts toward smaller x and the peak goes
lower and broader, as Q goes large from Q
0
.
12
Chapter 19
Zhong-Zhi Xianyu

1 Fermion number nonconservation in parallel E and
B elds
(a) In this problem we investigate the eect of chiral anomaly on the (non)conservation
of fermion number with denite chirality. Let us begin with the Adler-Bell-Jackiw
anomaly equation,
j
5
=
e
2
16
2
. (1)
Integrating the left hand side over the whole spacetime, we get the dierence between
the numbers of right-handed fermions N
R
and of left-handed fermions N
L
, namely,
_
d
4
x
j
5
=
_
d
4
x
(j
R
j
L
) =
_
d
3
x(j
0
R
j
0
L
)
t2
t1
= N
R
N
L
, (2)
where we assume that the integral region for time is [t
1
, t
2
] and that
i
j
i
integrates
to zero with suitable boundary conditions (i.e. vanishing at spatial innity or periodic
boundary condition). On the other hand,
= 4
0ijk
F
0i
F
jk
= 8F
0i
_
1
2

ijk
F
jk
_
= 8E B. (3)
Therefore, the ABJ anomaly equation gives,
N
R
N
L
=
e
2
2
2
_
d
4
xE B. (4)
(b) The Hamiltonian for massless charged fermions with background electromagnetic
eld is given by
H =
_
d
3
x
_
D
0
L
_
=
_
d
3
xi

i
D
i
, (5)
where = i
is the canonical conjugate momentum of , L = i
/ D is the Lagrangian
for the fermion, and D
+ ieA
is the covariant derivative. Now we expand the
1
Hamiltonian in the chiral basis,
H =
_
d
3
x
_
R
_
_
i D 0
0 i D
__
R
_
=
_
d
3
x
_
L
(i D)
L
R
(i D)
R
_
. (6)
(c) Now we focus on the eigenvalue problem of the right-handed fermion
R
, namely
the equation i D
R
= E
R
. To be denite, we set the background electromagnetic
potential to be A
= (0, 0, Bx
1
, A) with B and A two constants. To seek the eigenfunc-
tion of the form
R
=
_
1
(x
1
),
2
(x
1
)
_
T
e
i(k2x
2
+k3x
3
)
, we substitute it into the equation
above and get
1
= (k
2
eBx
1
)
1
+ i(E + k
3
eA)
2
, (7a)
2
= i(E k
3
+ eA)
1
(k
2
eBx
1
)
2
. (7b)
Eliminating
2
from these two equations, we get a single dierential equation in the
form of the harmonic oscillator,
1

_
e
2
B
2
_
x
1
k
2
eB
_
2
E
2
+ (k
3
eA)
2
eB
_
1
= 0. (8)
(d) Now we specify the spatial boundary condition to be the box of length L in each
side and periodic boundary condition. Then the condition
R
(x
1
, x
2
, x
3
) =
R
(x
1
, x
2
+
L, x
3
) =
R
(x
1
, x
2
, x
3
+ L) implies that k
2
and k
3
are quantized according to k
i
=
2n
i
/L (i = 2, 3). On the other hand, k
2
also has an upper bound since (8) shows that
the center of the oscillator would be out of the box if k
2
is too large. This condition
implies that k
2
/eB < L, which further gives the maximum value of n
2
to be (n
2
)
max
=
eBL
2
/2. Note also that the energy eigenvalue does not depend on k
2
, thus each
energy level consists of eBL
2
/2 degenerate states. Furthermore, we can also write
down explicitly the energy eigenvalue associated with the state labeled by (n
1
, n
3
):
E =
_
_
2n
3
L
eA
_
2
(n
1
+
3
2
)eB
_
1/2
. (9)
(e) Now we consider the case with n
1
= 0 for simplicity. Then the spectrum reads
E = 2n
3
/L eA. Suppose that the background potential changes by A = 2/eL.
Then it is easy to see that all state with energy marked by n
3
will turn to states with
energy marked by n
3
1. Note that each energy eigenvalue is eBL
2
/2-degenerate,
thus the net change of right-handed fermion number is eBL
2
/2. Similar analysis
shows that the left-handed fermion number get changed by eBL
2
/2. Therefore the
total change is N
R
N
L
= eBL
2
/.
2 Weak decay of the pion
(a) In this problem we study the decay of charged pion. So let us work out the
amplitude for
+
, with the eective four-fermion interaction
L =
4G
F
2
(
L
)( u
L
d
L
) + h.c. (10)
2
and the relation
0|j
5a
(x)|
b
(p) = ip
ab
e
ipx
(11)
as inputs. Firstly we recall that
j
a
=

Q
L
a
Q
L
+

Q
R
a
Q
R
, (12a)
j
5a
=

Q
a
Q
L
+

Q
R
a
Q
R
, (12b)
where Q
L
= (u
L
, d
L
)
T
and Q
R
= (u
R
, d
R
)
T
. Thus,
1
2
(j
1
+ ij
2
j
51
ij
52
) =

Q
L
(
1
+ i
2
)Q
L
= u
L
d
L
. (13)
Then we nd the decay amplitude M
_
+
(p)
+
(k)(q)
_
to be
iM=
4iG
F
2
u(q)
_
1
5
2
_
v(k)
1
2
f
ip
= G
F
f
u(q)
/
p(1
5
)v(k). (14)
(b) Now let us calculate the decay rate of the charged pion. We note that the amplitude
above can be further simplied to
iM= G
F
f
u(q)(
/
q + / k)(1
5
)v(k) = G
F
f
u(q)(1 +
5
)v(k). (15)
Therefore we have
|M|
2
= G
2
F
f
2
m
2
tr
_
/
q(1 +
5
)(/ k m
)(1
5
)
_
= 8G
2
F
f
2
m
2
q k, (16)
where the summation goes over all nal spins. We choose the momenta to be
p = (m
, 0, 0, 0), k = (E
k
, 0, 0, k), q = (E
q
, 0, 0, k). (17)
Then the kinematics can be easily worked out to be
E
k
=
m
2
+ m
2
2m
2
, E
q
= k =
m
2
m
2
2m
(18)
The decay rate then follows straightforwardly,
=
1
2m
_
d
16
2
k
2
E
k
E
q
_
k
E
k
+
k
E
q
_
1
8G
2
F
f
2
m
2
(q k)
=
G
2
F
f
2
4m
_
m
_
2
(m
2
m
2
)
2
, (19)
and we have the ratio between two decay channels,
(
+
e
+
e
)
(
+
)
=
m
2
e
(m
2
m
2
e
)
2
m
2
(m
2
m
2
)
2
10
4
. (20)
Thus to determine the pion decay constant f
, we can consider the channel

+
only
as a good approximation. With the lifetime of charged pion
= 2.6 10
8
sec as well
as m
and m
, we nd that
f
4m
G
2
F
_
m
_
(m
2
m
2
)
1
90.6MeV. (21)
3
3 Computation of anomaly coecients
(a) By denition, A
abc
= tr [t
a
, {t
b
, t
c
}]. Then for the product representation r
1
r
2
,
we have
A
abc
(r
1
r
2
) =tr
r1r2
_
t
a
1 + 1 t
a
,
_
t
b
1 + 1 t
b
, t
c
1 + 1 t
c
_
_
=tr
r1r2
_
t
a
1 + 1 t
a
, {t
b
, t
c
} 1 + t
b
t
c
+ t
c
t
b
+ 1 {t
b
, t
c
}
_
=tr
r1r2
_
[t
a
{t
b
, t
c
}] 1 + [t
a
, t
b
] t
c
+ [t
a
, t
c
] t
b
+ t
b
[t
a
, t
c
] + t
b
[t
a
, t
c
] + 1 [t
a
, {t
b
, t
c
}]
_
= tr
r1
[t
a
, {t
b
, t
c
}] tr
r2
(1) + tr
r2
[t
a
, {t
b
, t
c
}] tr
r1
(1)
= A
abc
(r
1
)d(r
2
) +A
abc
(r
2
)d(r
1
). (22)
On the other hand, as we decompose the representation r
1
r
2
into a direct product of
irreducible representations
i
r
i
, we have
A
abc
_
i
r
i
_
=tr
r
_
i
t
a
i
,
_
j
t
b
j
,
k
t
c
k
__
= tr
r
_
k
[t
a
i
, {t
b
j
, t
c
k
}]
_
=
i
tr
ri
[t
a
i
, {t
b
i
, t
c
i
}] =
i
A
abc
(r
i
) (23)
Note that A
abc
(r) =
1
2
A(r)d
abc
where d
abc
is the unique symmetric gauge invariant.
Then equating the two expressions above, we get
d(r
2
)A(r
1
) + d(r
1
)A(r
2
) =
i
A(r
i
). (24)
(b) In this part we show that the representation (3 3)
a
of SU(3) is equivalent to

3.
Let
i
be the base vectors of 3 representation. Then, a set of base vectors of (33)
a
can
be chosen to be
ijk
k
. From the transformation rule
i
U
ij
j
, we know that the
(3 3)
a
base vectors transform according to
ijk
k

imn
U
mj
U
nk
k
. Now, it is
easy to show that
mn
U
i
U
mj
U
nk
is totally antisymmetric, and thus is proportional to
ijk
. Let us write
mn
U
i
U
mj
U
nk
= C
ijk
, then taking U = I shows that C = 1. Now we
multiply both sides of this equality by (U
)
ip
. Since U is unitary, (U
)
ip
= (U
1
)
ip
, so we
get
pmn
U
mj
U
nk
=
ijk
(U
)
ip
. That is, the base vector
ijk
U
j
U
k
transforms according
to
ijk
U
j
U
k
(U
)
i
jk
k
= (U
)
i
jk
k
, which is exactly the transformation
rule of

3.
Now from A(3) = 1, it follows that A(
3) = 1. Therefore A((3 3)
a
) = 1, and
by using the equation derived in (a), we have A((3 3)
s
) = 6 (1) = 7.
(c) Now we compute the anomaly coecients for a and s representations of the SU(N)
group. As indicated in Peskin & Schroeder, it is enough to consider an SU(3) subgroup
of SU(N). Then the fundamental representation N is decomposed into a direct sum of
irreducible representations when restricted to SU(3), that is, N = 3 + (N 3)1. This
decomposition is easily justied by considering the upper-left 3 3 block of a matrix in
fundamental representation of SU(N). When this block is treated as a transformation
of SU(3), the rst three components of the vector on which the matrix acts form a
4
fundamental representation vector of SU(3), while the other (N 3) components of the
column vector are obviously invariant. With this known, we have,
NN =
_
3 + (N 3)1
_
_
3 + (N 3)1
_
= 3 3 + 2(N 3)3 + (N 3)
2
1. (25)
On the other hand, we know that N = s + a while s and a are irreducible. Then we
have, by (a), 2N A(N) = A(s) + A(a). But we already know that A(N) = 1. Thus
A(s) + A(a) = 2N. Now, to compute A(a), we make use of the SU(3) restriction,
(NN)
a
= (3 3)
a
+ (N 3)3 +
1
2
(N 3)(N 4)1. (26)
Then,
A(a) = A
_
(3 3)
a
_
+ (N 3)A(3) = A(3) + (N 3)A(3) = N 4, (27)
and A(s) = 2N A(a) = N + 4.
Now consider totally antisymmetric rank-j tensor representation. Again we decom-
pose the fundamental representation as N = 3 + (n 3)1. Then the rank-j totally
antisymmetric tensor can be decomposed as
(N
j
)
a
=
(N j) (N j + 1)
(j 3)!
(3 3 3)
a
+
(N 3) (N j)
(j 2)!
(3 3)
a
+
(N 3) (N j 1)
(j 1)!
3 +1s. (28)
Therefore,
A(a) =
(N 3) (N j)
(j 2)!
+
(N 3) (N j 1)
(j 1)!
=
(N 3) (N j)(N 2j)
(j 1)!
. (29)
4 Large fermion mass limits
In this problem we study the chiral anomaly and the trace anomaly in triangle
diagrams with Pauli-Villars regularization.
(a) and (c) Firstly we evaluate the expectation value of the divergence of the chiral
current j
5
between the vacuum and the two-photon state, namely the matrix element
p, k|j
5
|0. This matrix element receives contributions at 1-loop level from the following
two diagrams:
+
In momentum space, the divergence of the rst diagram reads
iq
1
= (1)(ie)
2
_
d
4
(2)
4
_
tr
_
/
q
5
i
/ / k
i
/
i
/ +
/
p
_
5
tr
_
/
q
5
i
/ / k M
i
/ M
i
/ +
/
p M
__
(30)
The integral is nite, thus we are allowed to shift the integral variable. For the rst
trace and the second trace above, we rewrite the
/
q
5
factors, respectively, as follows,
/
q
= (/ +
/
p / + / k)
= (/ +
/
p)
5
+
5
(/ / k),
/
q
= (/ +
/
p M / + / k + M)
= (/ +
/
p M)
5
+
5
(/ / k M) + 2M
5
.
Then, the loop integral becomes
iq
1
= e
2
_
d
4
(2)
4
_
tr
_
5
1
/ / k
1
/
+
5
1
/
1
/ +
/
p
_
tr
_
5
1
/ / k M
1
/ M
+
5
1
/ M
1
/ +
/
p M
_
+ 2M tr
_
5
1
/ / k M
1
/ M
1
/ +
/
p M
__
(31)
In the expression above, the rst and the second lines are canceled by the corresponding
terms from the second diagram with (k, p, ), while the third line is doubled.
Therefore the sum of two diagrams gives
iq
= 4e
2
M
_
d
4
(2)
4
tr
_
5
1
/ / k M
1
/ + M
1
/ +
/
p M
_
= 4e
2
M
_
d
4
(2)
4
_
1
0
dx
_
1x
0
dy
2N
1
_
( xk + yp)
2
3
=
4ie
2
MN
1
(4)
2
_
1
0
dx
_
1x
0
dy
1
(32)
with
N
1
= tr
_
5
(/ / k + M)
(/ M)
(/ +
/
p + M)
= 4iM
,
= M
2
x(1 x)k
2
y(1 y)p
2
2xyk p.
Then the integral can be carried out directly in the M
2
limit, to be
iq
=
e
2
2
2
, (33)
as expected.
(b) and (d) For scale anomaly, the diagrams are the same. Now the relevant matrix
element is given by p, k|M

|0. Then the rst diagram reads
iM
(p)
(k) = ie
2
M
_
d
4
(2)
4
__
1
/ / k
/
(k)
1
/ k
/
(p)
1
/ +
/
p
_
tr
_
1
/ / k M
/
(k)
1
/ M
/
(p)
1
/ +
/
p M
__
(34)
The rst trace vanishes upon regularization, then,
iM
(p)
(k) =ie
2
M
_
d
4
(2)
4
_
1
0
dx
_
1x
0
dy
2N
2
(
2
)
3
, (35)
6
where
= xk + yp, = M
2
2xyk p, and the trace in the numerator is
N
2
= tr
_
(/ k + M)/
(k)(/ + M)/
(p)(/ +
/
p + M)
= 4M
_
M
2
(k)
(p) +
_
(k) p
__
(p) k
_
(k)
(p)
__
k p
_
+ 4
_
(k)
__
(p)
_
(k)
(p)
_
= 4M
_
M
2
(k)
(p) + (1 4xy)
_
(k) p
__
(p) k
_
(1 2xy)
_
(k)
(p)
_
(k p) +
_
4
d
1
__
(k)
(p)
_
,
where we used the transverse condition k
(k) = p
(p) = 0, and in the last equality, the

substitution
1
d
g
2
. We also dropped all terms linear in
in the last equality.

The integral is then divergent, and we regularize it by dimensional regularization. Then
after carrying out the loop integral, we get
iM
(p)
(k) =
e
2
4
2
__
(k)
(p)
_
(k p)
_
(k) p
__
(p) k
_
_
1
0
dx
_
1x
0
dy
(1 4xy)M
2
M
2
2xyk p
(36)
Then, taking M
2
limit, we nd
iM
(p)
(k) =
e
2
12
2
__
(k)
(p)
_
(k p)
_
(k) p
__
(p) k
_
(37)
The second diagram is obtained, again, by the exchange (k, p, ), which gives the
identical result. Therefore we nally get
iM
(p)
(k) =
e
2
6
2
__
(k)
(p)
_
(k p)
_
(k) p
__
(p) k
_
. (38)
7
Chapter 20
Zhong-Zhi Xianyu

1 Spontaneous breaking of SU(5)
We consider two patterns of spontaneous breaking of SU(5) gauge symmetry, with
an adjoint-representation scalar eld picking up vacuum expectation values
= Adiag(1, 1, 1, 1, 4), = Bdiag(2, 2, 2, 3, 3), (1)
respectively. The kinetic term of the scalar eld in the Lagrangian is
L
kin.
= tr
_
(D
(D
)
_
= tr
_
_
+g[A
, ]
_
+g[A
, ]
_
_
. (2)
Then the mass term of gauge bosons after symmetry breaking is given by
L = g
2
tr
_
[A
, ]
[A
, ]
_
= g
2
A
a
A
b
tr
_
[T
a
, ][T
b
, ]
_
. (3)
To analyze the gauge bosons spectrum, we note that there are 24 independent generators
for SU(5) group, each of which can be represented as a 55 traceless hermitian matrix.
Then, for the rst choice of = diag(1, 1, 1, 1, 4), we see that for the generators of
the form
T =
_
T
(4)
0
_
and T =
1
2
10
diag(1, 1, 1, 1, 4),
where T
(4)
is a 4 4 matrix being any generator of SU(4) group, the commutators
vanish. That is, a subgroup SU(4) U(1) remains unbroken in this case. Then, for the
rest of the generators, namely
1
2
_
_
_
_
_
_
_
0 0 0 0 1
0 0 0 0 0
0 0 0 0 0
0 0 0 0 0
1 0 0 0 0
_
_
_
_
_
_
_
,
1
2
_
_
_
_
_
_
_
0 0 0 0 i
0 0 0 0 0
0 0 0 0 0
0 0 0 0 0
i 0 0 0 0
_
_
_
_
_
_
_
,
1
2
_
_
_
_
_
_
_
0 0 0 0 0
0 0 0 0 1
0 0 0 0 0
0 0 0 0 0
0 1 0 0 0
_
_
_
_
_
_
_
,
1
etc, it is easy to calculate the commutators to get the trace equal to 25A
2
/2. Thus
the corresponding components of gauge bosons acquire mass M
A
= 5gA. In the same
way, we can also analyze the case of = diag(2, 2, 2, 3, 3). This time the unbroken
subgroup is SU(3) SU(2) U(1), and the remaining 12 components of gauge bosons
acquire a mass equal to M
A
= 5gB, as can be found by evaluating the corresponding
commutators.
2 Decay modes of the W and Z bosons
(a) The relevant interaction term in the Lagrangian reads
L =
1
2
gW
+
i

iL
e
iL
+
j,c
u
c
jL
d
c
jL
_
, (4)
where the sum on i goes over all three generations of leptons, the sum on j goes over
the rst two generations of quarks, since m
t
> m
W
, and the sum on c is due to 3 colors.
Now consider the decay of W
+
boson. The amplitude of the decay into a pair of
fermions is
iM=
ig
(k) u(p
1
)
_
1
5
2
_
v(p
2
), (5)
where
is the polarization vector for W

+
, and the labels for momenta are shown in

Fig. 1. Thus the squared amplitude with initial polarizations averaged is
k
p
2
p
1
W
+
f

f
k
p
2
p
1
Z
0
f

f
Figure 1: The decay of W
+
and Z
0
into fermion-antifermion pairs. All initial momenta go
inward and all nal momenta go outward.
1
3
spin
|iM|
2
=
g
2
6
_
g
+
k
m
2
W
_
tr
_
/
p
2
_
1
5
2
_
/
p
1
_
1
5
2
_
_
=
g
2
3
_
p
1
p
2
+ 2
k p
1
k p
2
m
2
W
_
. (6)
The momenta in the center-of-mass frame can be taken to be
k = (m
W
, 0, 0, 0), p
1
= (p, 0, 0, p), p
2
= (p, 0, 0, p), (7)
and energy conservation requires that p = m
W
/2. Thus we get
1
3
spin
|iM|
2
=
1
3
g
2
m
2
W
, (8)
and the decay rate
_
d =
1
2m
W
_
d
3
p
1
d
3
p
2
(2)
6
2E
1
2E
2
_
1
3
g
2
m
2
W
_
(2)
4
(4)
(k p
1
p
2
) =
m
W
12 sin
2
w
, (9)
2
where we have used g = e/ sin
w
and = e
2
/4. For each quark nal state we multiply
the result by a QCD correction factor
_
1 +
s
_
. Then, taking account of 3 generations
of leptons and 2 generations of quarks with 3 colors, we get the partial decay rate of
W
+
into fermions,
(W
+
e
+
i

i
) =
m
W
12 sin
2
w
0.23GeV; (10)
(W
+
u
j

d
j
) =
m
W
4 sin
2
w
_
1 +

s
_
0.70GeV; (11)
(W
+
fermions) =
m
W
12 sin
2
w
_
9 + 6

s
_
2.08GeV. (12)
and also the branching ratios BR(W
+
e
+
i

i
) = 0.11%, and BR(W
+
u
j

d
j
) = 0.34%.
Note that the ne structure constant at m
W
is (m
W
) 1/129.
(b) In the same way, we can also calculate the decay rate of Z fermions. The
relevant term in the Lagrangian is
L =
g
cos
w
Z
f
i
_
I
3
i
sin
2
w
Q
i
_
f
i
, (13)
where the sum goes over all left- and right-handed fermions, including 3 generations
of leptons, and the rst two generations of quarks with 3 colors, while I
3
i
and Q
i
are
associated 3-component of the weak isospin and the electric charge, respectively.
Then we can write down the amplitudes of the decay of Z
0
into a pair of fermions
f

f with specic I
3
and Q, as illustrated in Fig. 3,
iM=
ig
cos
w
(k) u(p
1
)
_
_
I
3
sin
2
w
Q
_
_
1
5
2
_
sin
2
w
Q
_
1 +
5
2
_
_
v(p
2
)
=
ig
cos
w
(k) u(p
1
)
_
I
3
_
1
5
2
_
sin
2
w
Q
_
v(p
2
), (14)
the squared matrix elements,
1
3
spin
|iM|
2
=
g
2
3 cos
2
w
_
g
+
k
m
2
Z
_
tr
_
/
p
2
_
1
2
I
3
(1
5
) sin
2
w
Q
_
/
p
1
_
1
2
I
3
(1
5
) sin
2
w
Q
_
_
=
4g
2
3 cos
2
w
_
_
1
2
I
3
sin
2
w
Q
_
2
+
_
1
2
I
3
_
2
_
_
p
1
p
2
+
2k p
1
k p
2
m
2
Z
_
=
4g
2
m
2
Z
3 cos
2
w
_
_
1
2
I
3
sin
2
w
Q
_
2
+
_
1
2
I
3
_
2
_
, (15)
and the partial decay rate,
(Z
0
f

f) =
m
Z
3 sin
2
w
cos
2
w
_
_
1
2
I
3
sin
2
w
Q
_
2
+
_
1
2
I
3
_
2
_
. (16)
We should also multiply the result for quarks by the QCD factor
_
1 +
s
_
. Now we
list the numerical results of partial width and the branching ratios for various decay
products as follows.
3
f

f (f

f)/GeV BR(f

f)
e

e
,
0.17 6.7%
e
e
+
,
+
,
+
0.08 3.4%
u u, c c 0.30 11.9%
d
d, s s, b
b 0.39 15.4%
All fermions 2.51 100%
3 e
+
e
hadrons with photon-Z

0
interference
(a) It is easier to work with amplitudes between initial and nal fermions with denite
chirality. In this case the relevant amplitude is given by
iM= (ie)
2
v(k
2
)
u(k
1
)
i
q
2
u(p
1
)
Q
f
v(p
2
)
_
Q
f
+
(I
3
e
+ s
2
w
)(I
3
f
s
2
w
Q
f
)
s
2
w
c
2
w
q
2
q
2
m
2
Z
_
,
(17)
where I
3
e
= 1/2 or 0 when the initial electron is left-handed or right-handed, so as I
3
f
to the nal fermion. The momenta is labeled as shown in Fig. 2. Then we can nd
k
1
p
2
k
2
q
p
1
e
e
+
f

f
Figure 2: The process of e
+
e
f

f via the exchange of a photon/Z
0
in s-channel. The
directions of kis and pis are inward and outward, respectively.
associated dierential cross section to be
d
d cos
(e
+
R
e
L

f
R
f
L
) =

2
2s
(1 + cos )
2
F
LL
(f), (18a)
d
d cos
(e
+
R
e
L

f
L
f
R
) =

2
2s
(1 cos )
2
F
LR
(f), (18b)
d
d cos
(e
+
L
e
R

f
R
f
L
) =

2
2s
(1 cos )
2
F
RL
(f), (18c)
d
d cos
(e
+
L
e
R

f
L
f
R
) =

2
2s
(1 + cos )
2
F
RR
(f), (18d)
in which is the ne structure constant, s = q
2
is the center-of-mass energy, and the
F factors are dened as follows:
F
LL
(f) =
Q
f
+
(
1
2
s
2
w
)(I
3
f
s
2
w
Q
f
)
s
2
w
c
2
w
s
s m
2
Z
+ im
Z
2
, (19)
F
LR
(f) =
Q
f

(
1
2
s
2
w
)Q
f
c
2
w
s
s m
2
Z
+ im
Z
2
, (20)
F
RL
(f) =
Q
f

(I
3
f
s
2
w
Q
f
)
c
2
w
s
s m
2
Z
+ im
Z
2
, (21)
In this problem we simplify the notation by sw sinw and cw cos w.

4
F
RR
(f) =
Q
f
+
s
2
w
Q
f
c
2
w
s
s m
2
Z
+ im
Z
2
, (22)
where we have added the correction from resonance by using the Breit-Wigner formula.
Summing up the four expressions in (18), averaging the initial spins, and integrating
over the angle , we get nally the unpolarized cross section
(f

f) =

2
3s
_
F
LL
(f) +F
LR
(f) +F
RL
(f) +F
RR
(f)
_
. (23)
When the nal state particle f is a quark, one should multiply the result by 3
_
1 +
s
_
where 3 is the color factor, and
_
1 +
s
_
is the 1-loop QCD correction.
For the nal fermion being muon (I
3
f
= 1/2, Q
f
= 1), up quark (I
3
f
= 1/2,
Q = 2/3), and down quark (I
3
f
= 1/2, Q
f
= 1/3), we plot the corresponding cross
section as a function of center-of-mass energy E
CM
=
s in Fig. 3.
50 100 150 200
10
100
1000
10
4
E
CM
GeV
p
b
Figure 3: The cross section (e
+
e
f

f) as a function of center-of-mass energy ECM. The
black, blue, and red curves correspond to f

f =
+
, u u and d
d, respectively.
(b) Now we calculate the forward-backward asymmetry A
f
FB
, dened to be
A
f
FB
=

F

B
F
+
B
=
_ _
1
0

_
0
1
_
d cos (d/d cos )
_ _
1
0
+
_
0
1
_
d cos (d/d cos )
. (24)
Then from (18), we nd
F
=

2
24s
_
7F
LL
(f) +F
LR
(f) +F
RL
(f) + 7F
RR
(f)
_
, (25)
B
=

2
24s
_
F
LL
(f) + 7F
LR
(f) + 7F
RL
(f) + F
RR
(f)
_
. (26)
Thus
A
f
FB
=
3
4

F
LL
(f) F
LR
(f) F
RL
(f) +F
RR
(f)
F
LL
(f) +F
LR
(f) +F
RL
(f) +F
RR
(f)
. (27)
Again, we plot A
f
FB
, as a function of E
CM
, for f =
, u, d, in Fig. 4.
5
M
Z
40 60 80 100 120 140
0.6
0.4
0.2
0.0
0.2
0.4
0.6
E
CM
GeV
A
F
B
Figure 4: The forward-backward asymmetry A
f
FB
as a function of center-of-mass energy ECM.
The black, blue, and red curves correspond to f

f =
+
, u u and d
d, respectively.
(c) Recall the denition of Fs, we nd, on the Z
0
resonance (s = m
Z
),
F
LL
(f)
_
_
1
2
s
2
w
__
I
3
f
s
2
w
Q
f
_
s
2
w
c
2
w
m
Z
Z
_
2
, F
LR
(f)
_
_
1
2
s
2
w
_
Q
f
c
2
w
m
Z
Z
_
2
,
F
RL
(f)
_
_
I
3
f
s
2
w
Q
f
_
c
2
w
m
Z
Z
_
2
, F
RR
(f)
_
s
2
w
Q
f
c
2
w
m
Z
Z
_
2
,
therefore,
A
f
FB
=
3
4

__
1
2
s
2
w
_
2
s
4
w
__
I
3
f
s
2
w
Q
f
_
2
(s
2
w
Q
f
)
2
__
1
2
s
2
w
_
2
+s
4
w
__
I
3
f
s
2
w
Q
f
_
2
+ (s
2
w
Q
f
)
2
=
3
4
A
e
LR
A
f
LR
. (28)
(d)
peak
=

2
3m
2
Z
1
s
4
w
c
4
w
m
2
Z
2
Z
_
_
1
2
s
2
w
_
2
+s
4
w
__
_
I
3
f
s
2
w
Q
f
_
2
+ (s
2
w
Q
f
)
2
_
=
12
m
2
Z
2
Z
_
m
Z
6s
2
w
c
2
w
_
_
1
2
s
2
w
_
2
+s
4
w
_
__
m
Z
6s
2
w
c
2
w
_
_
I
3
f
s
2
w
Q
f
_
2
+ (s
2
w
Q
f
)
2
_
_
=
12
m
2
Z
(Z
0
e
+
e
)(Z
0
f

f)
2
Z
. (29)
4 Neutral-current deep inelastic scattering
(a) In this problem we study the neutral-current deep inelastic scattering. The process
is mediated by Z
0
boson. Assuming m
Z
is much larger than the energy scale of the
scattering process, we can write down the corresponding eective operators, from the
neutral-current Feynman rules in electroweak theory,
L =
g
2
4m
2
W
_

_
P
L
_
u
_
_
1
4
3
s
2
w
_
P
L
4
3
s
2
w
P
R
_
u
+

d
_
(1
2
3
s
2
w
)P
L
2
3
s
2
w
P
R
_
d
_
+ h.c., (30)
where P
L
= (1
5
)/2 and P
R
= (1 +
5
)/2 are left- and right-handed projectors,
respectively. Compare the eective operator with the charged-operator in (17.31) of Pe-
skin & Schroeder, we can write down directly the dierential cross section by modifying
6
(17.35) in Peskin & Schroeder properly, as
d
2
dxdy
(p X) =
G
2
F
sx
4
__
_
1
4
3
s
2
w
_
2
+
16
9
s
4
w
(1 y
2
)
_
f
u
(x)
+
_
_
1
2
3
s
2
w
_
2
+
4
9
s
4
w
(1 y
2
)
_
f
d
(x)
+
_
16
9
s
4
w
+
_
1
4
3
s
2
w
_
2
(1 y
2
)
_
f
u
(x)
+
_
4
9
s
4
w
+
_
1
2
3
s
2
w
_
2
(1 y
2
)
_
f
d
(x)
_
, (31)
d
2
dxdy
( p X) =
G
2
F
sx
4
__
16
9
s
4
w
+
_
1
4
3
s
2
w
_
2
(1 y
2
)
_
f
u
(x)
+
_
4
9
s
4
w
+
_
1
2
3
s
2
w
_
2
(1 y
2
)
_
f
d
(x)
+
_
_
1
4
3
s
2
w
_
2
+
16
9
s
4
w
(1 y
2
)
_
f
u
(x)
+
_
_
1
2
3
s
2
w
_
2
+
4
9
s
4
w
(1 y
2
)
_
f
d
(x)
_
. (32)
(b) For the neutrino scattering from a nucleus A with equal numbers of protons and
neutrons, we have f
u
= f
d
and f
u
= f
d
. Then the dierential cross sections reads
d
2
dxdy
(A X) =
G
2
F
sx
_
_
1
2
s
2
w
+
5
9
s
4
4
+
5
9
s
4
4
(1 y
2
)
f
u
(x)
+
_
5
9
s
4
w
+
_
1
2
s
2
w
+
5
9
s
4
w
_
(1 y
2
)
f
u
(x)
_
, (33)
d
2
dxdy
( p X) =
G
2
F
sx
_
_
5
9
s
4
w
+
_
1
2
s
2
w
+
5
9
s
4
w
_
(1 y
2
)
f
u
(x)
+
_
1
2
s
2
w
+
5
9
s
4
4
+
5
9
s
4
4
(1 y
2
)
f
u
(x)
_
. (34)
Recall that for charged-current neutrino deep inelastic scattering, the dierential cross
sections are given by (17.35) in Peskin & Schroeder. Thus it is easy to nd that
R
=
d
2
/dxdy(A X)
d
/dxdy(A
X)
=
1
2
s
2
w
+
5
9
s
4
w
_
1 +
f
u
(x)(1 y
2
) +f
u
(x)
f
u
(x) +f
u
(x)(1 y)
2
_
, (35)
R
u
=
d
2
/dxdy( A X)
d
/dxdy( A
+
X)
=
1
2
s
2
w
+
5
9
s
4
w
_
1 +
f
u
(x) +f
u
(x)(1 y)
2
f
u
(x)(1 y
2
) +f
u
(x)
_
, (36)
where
f
u
(x)(1 y
2
) +f
u
(x)
f
u
(x) +f
u
(x)(1 y)
2
= r. (37)
(c) The plot Weinbergs Nose with r = 0.4 is shown in Figure 5.
5 A model with two Higgs elds
(a) The gauge boson mass matrix comes from the kinetic term of scalar elds,
(D
1
)
(D
1
) + (D
2
)
(D
2
),
with D
1,2
=
_
i
2
gA
a
i
2
g
1,2
. After
1,2
acquire the vacuum expectation
value
1
2
_
0
v1,2
_
, we observe that each of the kinetic terms gives rise to mass terms for
gauge bosons similar to the ones in the standard electroweak theory. Thus it is straight-
forward that the masses of gauge bosons in this model is given by the replacement
v
2
v
2
1
+v
2
2
.
7
sin
2
w
0.23
0.0 0.1 0.2 0.3 0.4 0.5
0.2
0.4
0.6
0.8
1.0
R
Figure 5: Weinbergs nose with r = 0.4. See problem 21.4.

(b) The statement that the conguration
1
2
_
0
v1,2
_
is a locally stable minimum, is
equivalent to that all particle excitations generated above this solution have positive
squared mass m
2
. Thus we investigate the mass spectrum of the theory with the vacuum
chosen to be
1
2
_
0
v1,2
_
. Firstly, we parameterize two scalar doublets as
i
=
_

+
i
,
1
2
(v
i
+h
i
+ i
0
i
)
_
, (i = 1, 2) (38)
and substitute this parameterization into the potential,
V =
2
1
2
2
2
+
1
(
1
)
2
+
2
(
2
)
2
+
3
(
1
)(
2
) +
4
(
2
)(
1
) +
5
_
(
2
)
2
+ h.c.
_
. (39)
Then the mass term of various scalar components can be extracted, as follows.
L
mass
= (
4
+ 2
5
)v
1
v
2
_
2
_
_
v
2
/v
1
1
1 v
1
/v
2
__
+
1
+
2
_
+ 2
5
v
1
v
2
_
0
1

0
2
_
_
v
2
/v
1
1
1 v
1
/v
2
__
0
1
0
2
_
v
1
v
2
_
h
1
h
2
_
_

1
(v
1
/v
2
)
3
+
4
+ 2
5
3
+
4
+ 2
5

2
(v
2
/v
1
)
__
h
1
h
2
_
. (40)
The eigenvalues of these matrices are easy to be found. For charged components, there
is a zero mode corresponding two broken directions in SU(2), and the mass of the other
charged scalar is given by m
2
c
= (
4
+ 2
5
)(v
2
1
+ v
2
2
). For pseudoscalar components,
there is also a zero mode corresponding to the rest one direction of broken SU(2), and
the mass of the other pseudoscalar is m
2
p
= 4
5
(v
2
1
+v
2
2
). Finally, for neutral scalars,
the two mass eigenvalues are given by the roots of following equation,
m
2
n
(
1
v
2
1
+
2
v
2
2
)m
2
n
+
_
2
(
3
+
4
+ 2
5
)
2
= 0. (41)
Therefore, to make m
2
c
> 0, m
2
p
> 0 and m
2
n
> 0, it is sucient that
4
+ 2
5
< 0,
5
< 0,
1
,
2
> 0,
1
2
> (
3
+
4
+ 2
5
)
2
. (42)
8
(c) From the mass terms in (b) we can diagonalize the charged scalar mass matrix
with the rotation matrix
_
+
_
=
_
cos sin
sin cos
__
+
1
+
2
_
, (43)
where
+
is the Goldstone mode and
+
is a physical charged scalar. Given that
+
to get the physical mass, it is easy to see that the rotation angle can be chosen to be
tan = v
2
/v
1
.
(d) Assuming that the Yukawa interactions between quarks and scalars take the fol-
lowing form,
L
m
=
_
u
L

d
L
_
_
d
_

+
1
1
2
v
1
_
d
R
+
u
_
1
2
v
2
_
u
R
_
+ h.c., (44)
where we have suppressed avor indices and neglected neutral scalar components. We
focus on charged component only. Then, with Peskin & Schroeders notation, we make
the replacement u
L
U
u
u
L
, d
L
U
d
d
L
, u
R
W
u
u
R
, and d
R
W
d
d
R
. Then,
together with
d
= U
d
D
d
W
d
and
u
= U
u
D
u
W
u
where D
d
and D
u
are diagonal matrix,
we have
L
m
=
1
2
_
v
1

d
L
D
d
d
R
+v
2
u
L
D
u
u
R
_
uV
CKM
D
d
d
R
+
1
+

d
L
V
CKM
D
u
u
R
2
+ h.c. (45)
From the rst line we see that the diagonal mass matrix for quarks are given by m
u
=
(v
1
/
2)D
u
and m
d
= (v
2
/
2)D
d
. We further dene v =
_
v
2
1
+v
2
2
and note that
+
1
=
+
sin ,
+
2
=
+
cos + , then the Yukawa interactions between charged
boson and quarks can be written as
L
m

2
v
1
_
u
L
V
CKM
m
d
d
R
+
1
+

d
L
V
CKM
m
u
u
R
2
_
+ h.c.
2
v
_
u
L
V
CKM
m
d
d
R
+
tan +

d
L
V
CKM
m
u
u
R
cot
_
+ h.c.. (46)
9
Chapter 21
Zhong-Zhi Xianyu

1 Weak-interaction contributions to the muon g 2
In this problem we study the weak-interaction corrections to the muons anomalous
magnetic moment (AMM). The relevant contributions come from the W-neutrino loop
and Z-muon loop, together with the diagrams with the gauge bosons replaced with the
corresponding Goldstone bosons. Here we will evaluate the W-neutrino loop diagram
with Feynman-t Hooft gauge and general R
gauge in part (a) and part (b) respectively,

and Z-muon diagram in part (c).
(a) Now we come to the W-neutrino loop diagram and the corresponding Goldstone
boson diagrams, shown in Fig. 1.
q
k
q + k
+
(a)
q
+
(b)
q
+
(c)
q
+
(d)
Figure 1: The weak-interaction contributions to muons EM vertex. These four diagrams
contain neutrino internal lines in the loops.
The Fig. 1(a) with W-neutrino loop reads
(a)
(q) =
(ig)
2
2
_
d
4
k
(2)
4
_
g
(2k + q)
+ g
(2q k)
+ g
(q k)
ig
k
2
m
2
W
ig
(q + k)
2
m
2
W
u(p
_
1
5
2
_
i
/
p
+ / k
_
1
5
2
_
u(p)
=
ig
2
2
_
d
4
k
(2)
4
_
1
0
dx
_
1x
0
dy
2
(k
2
)
3
u(p
)
_
(2k + q)
(
/
p
+ / k)
+ (/ k 2
/
q)(
/
p
+ / k)
(
/
p
+ / k)(
/
q / k)
__
1
5
2
_
u(p), (1)
1
where
k
= k + xq + yp
,
= (1 y)m
2
W
x(1 x)q
2
y(1 y)p
2
+ 2xyq p
.
To extract the form factor F
2
(q
2
), recall that the total diagram can be written as a
linear combination of (p
+ p)
, q
, and parity-violating terms containing

5
. Only
the (p
+p)
terms contribute to F
2
(q
2
) through the Gordon identity. With this in mind,
now we try to simplify the expression in the square bracket in (1), during which we will
drops terms proportional to q
or
freely, and totally ignore the

5
terms.
_
(2k + q)
(
/
p
+ / k)
_
+
_
(/ k 2
/
q)(
/
p
+ / k)
_
+
_
(
/
p
+ / k)(
/
q / k)
_
=
_
2
_
2k
+ (1 2x)q 2yp
_
/ k
x
/
q + (1 y)
/
p
_
_
+
_
_
/ k
(2 x)
/
q + y
/
p
__
/ k
x
/
q + (1 y)
/
p
_
+
_
_
/ k
x
/
q + (1 y)
/
p
__
/ k
+ (1 + x)
/
q + y
/
p
_
_
_
4y(1 y)mp
_
+
_
2(x + 2y 2)mp
_
+
_
2(1 x + y)mp
_
(1 y)(3 2y)m(p
+ p)
2(1 y)(3 2y)m

2
2m
.
The steps of this calculation is basically in parallel with the one of Problem 7.2. Here
we have written the mass of muon as m instead of m
to avoid confusions. Thus the

contribution to the muons AMM from Fig. 1(a) is
ig
2
2
_
d
4
k
(2)
4
_
1
0
dx
_
1x
0
dy
2
(k
2
)
3

1
2
2(1 y)(3 2y)m
2
7
3

g
2
m
2
64
2
m
2
W
=
7
3

G
F
m
2
8
2
2
, (2)
where we have used the approximation m
W
m, and set q
2
= 0 in the second line.
The Fermi constant G
F
/
2 = g
2
/8m
2
W
.
Fig. 1(b) and 1(c) read
(b)
(q) =
ig
m
W
2

i
2gm
m
W
_
d
4
k
(2)
4
g
i
k
2
m
2
W
ig
(q + k)
2
m
2
W
u(p
)
_
1
5
2
_
i
/
p
+ / k
_
1
5
2
_
u(p). (3)
(c)
(q) =
ig
m
W
2

i
2gm
m
W
_
d
4
k
(2)
4
g
ig
k
2
m
2
W
i
(q + k)
2
m
2
W
u(p
_
1
5
2
_
i
/
p
+ / k
_
1 +
5
2
_
u(p). (4)
Through the calculation similar to that of Fig. 1(a), it is easy to show that these two
diagrams contribute the same to the AMM, which reads
1
2

G
F
m
2
8
2
2
. (5)
Finally, Fig. 1(d) reads
(d)
(q) =
_
i
2gm
m
W
_
2
_
d
d
k
(2)
d
(2k q)
i
k
2
m
2
W
i
(q + k)
2
m
2
W
2
u(p
)
_
1
5
2
_
i
/
p
+ / k
_
1 +
5
2
_
u(p). (6)
But it is not dicult to see that the contribution to the muons AMM from this diagram
is proportional to (m/m
W
)
4
, which can be omitted in the limit m
W
m, compared
with the other three diagrams. Therefore we conclude that the AMM of the muon
contributed by W-neutrino and corresponding Goldstone bosons 1-loop diagrams is
a
() =
_
7
3
+
1
2
+
1
2
+O
_
m
2
m
2
W
_
_
G
F
m
2
8
2
10
3

G
F
m
2
8
2
2
. (7)
(c) Now we come to the second set of diagrams as shown in Fig. 2.
q
+
(a)
q
+
(b)
Figure 2: The weak-interaction contributions to muons EM vertex. These two diagrams
contains no neutrino internal lines.
Firstly the Fig. 2(a) reads
(a)
Z

(q) =
_
ig
4c
w
_
2
_
d
d
k
(2)
d
ig
(p
+ k)
2
m
2
Z
u(p
_
4s
2
w
1
5
_
i
/ k m
/
q / k m
_
4s
2
w
1
5
_
u(p)
ig
2
16c
2
w
_
d
d
k
(2)
d
_
1
0
dx
_
1x
0
dy
2
(k
2
)
3
u(p
_
(/ k + m)
(/ k +
/
q + m)
+ (4s
2
w
1)
2
(/ k m)
(/ k +
/
q m)
_
u(p), (8)
where we have omitted terms proportional to
5
, as indicated by sign, and
k
= k + xq + yp
,
= (1 y)m
2
+ ym
2
Z
x(1 x)q
2
y(1 y)p
2
+ 2xyq p
.
We will again focus only on the terms proportional to (p
+ p)
. Then the spinor part

can be reduced to
u(p
_
(/ k + m)
(/ k +
/
q + m) + (3s
2
w
c
2
w
)
2
(/ k m)
(/ k +
/
q m)
_
u(p)
_
2y(3 + y) (4s
2
w
1)
2
2y(1 y)
_
2m
2
u(p
)
i
2m
u(p).
Thus the AMM contribute by this diagram is
ig
2
16c
2
w
_
d
4
k
(2)
4
_
1
0
dx
_
1x
0
dy
2 2m
2
[2y(3 + y) (4s
2
w
1)
2
2y(1 y)]
(k
2
)
3
=
G
F
m
2
8
2
1
3
_
(4s
2
w
1)
2
5
_
. (9)
3
On the other hand, the Fig. 2(b) only contributes terms of order m
4
/m
4
W
that can be
omitted, as can be seen from the coupling between the Goldstone boson and the muon.
Thus we conclude that the total contribution to a
(Z) from the two diagrams in Fig. 2

at the leading order is given by (9).
2 Complete analysis of e
+
e
W
+
W
In this problem we calculate the amplitude for the process e

+
e
W
+
W
at tree
level in standard electroweak theory. There are 3 diagrams contributing in total, as
shown in Figure 3.
k
1
p
2
k
2
p
1
e
e
+
W
+
W
Z
0
e
e
+
W
+
W
e
+
W
+
W
Figure 3: The process e
e
+
W
+
W
at tree level. All initial momenta go inward and all

nal momenta go outward.
We will evaluate these diagrams for denite helicities for initial electrons as well
as denite polarizations for nal W bosons. The initial and nal momenta can be
parameterized as
k
1
= (E, 0, 0, E), p
1
= (E, p sin , 0, p cos ),
k
2
= (E, 0, 0, E), p
2
= (E, p sin , 0, p cos ), (10)
with E
2
= p
2
+ m
2
W
, and electron mass ignored. For initial electron and positron, the
spinors with denite helicities can be chosen to be
u
L
(k
1
) =
2E(0, 1, 0, 0)
T
, v
L
(k
2
) =
2E(1, 0, 0, 0)
T
,
u
R
(k
1
) =
2E(0, 0, 1, 0)
T
, v
R
(k
2
) =
2E(0, 0, 0, 1)
T
. (11)
For nal W bosons, the polarization vectors are
(p
1
) =
1
2
(0, cos , i, sin),
(p
2
) =
1
2
(0, cos , i, sin),
+
(p
1
) =
1
2
(0, cos , i, sin ),
+
(p
2
) =
1
2
(0, cos , i, sin),
L
(p
1
) =
1
mW
(p, E sin, 0, E cos ),
L
(p
2
) =
1
mW
(p, E sin , 0, E cos ). (12)
It is easy to see that for initial electron-positron pair, only two helicity states e
L
e
+
R
and e
R
e
+
L
contribute nonzero amplitudes. This is because the rst two diagrams with
s-channel gauge bosons vanish for the other two possibilities e
L
e
+
L
and e
R
e
+
R
due to
angular momentum conservation, while the third diagram vanishes since the weak cou-
pling vanishes for right-handed electron and left-handed positron. With this known, we
can write down the amplitudes for e
L
e
+
R
and e
R
e
+
L
initial states, as follows. Generally
the amplitude reads
iM(e
L
e
+
R
W
+
W
)
4
=
__
(ie)
i
(k
1
+ k
2
)
2
(ie) +
ie(
1
2
+ s
2
w
)
c
w
s
w
i
(k
1
+ k
2
)
2
m
2
Z
(igc
w
)
_
v
L
(k
2
)
u
L
(k
1
)
_
(p
2
p
1
)
(p
1
2p
2
)
(2p
1
+ p
2
)
+
_
ig
2
_
2
v
L
(k
2
)
i
/ k
1
/
p
2
u
L
(k
1
)
_
(p
1
)
(p
2
)
= ie
2
_
m
2
Z
s(s m
2
Z
)

1
2s
2
w
1
s M
2
Z
_
v
L
(k
2
)
_
(p
1
)
(p
2
)(
/
p
2
/
p
1
)
(p
1
+ 2p
2
)
(p
1
)/
(p
2
) + (2p
1
+ p
2
)
(p
2
)/
(p
1
)
_
u
L
(k
1
)
ie
2
2s
2
w
1
u
v
L
(k
2
)/
(p
1
)(/ k
1
/
p
2
)/
(p
2
)u
L
(k
1
), (13)
and,
iM(e
R
e
+
L
W
+
W
)
=
_
(ie)
i
(k
1
+ k
2
)
2
(ie) +
ies
2
w
c
w
s
w
i
(k
1
+ k
2
)
2
m
2
Z
(igc
w
)
_
v
R
(k
2
)
u
R
(k
1
)
(p
2
p
1
)
(p
1
2p
2
)
(2p
1
+ p
2
)
(p
1
)
(p
2
)
= ie
2
m
2
Z
s(s m
2
Z
)
v
R
(k
2
)
_
(p
1
)
(p
2
)(
/
p
2
/
p
1
) (p
1
+ 2p
2
)
(p
1
)/
(p
2
)
+ (2p
1
+ p
2
)
(p
2
)/
(p
1
)
_
u
R
(k
1
), (14)
In what follows we need the inner products among some of these vectors, as listed below.
p
1
p
2
= E
2
+ p
2
p
1

0
(p
2
) = p
2

0
(p
1
) =
2Ep
m
W
,
+
(p
1
)
+
(p
2
) =
(p
1
)
(p
2
) = 1,
0
(p
1
)
0
(p
2
) =
E
2
+ p
2
m
2
W
. (15)
We also need
v
L
(k
2
)
/
p
1
u
L
(k
1
) = v
L
(k
2
)
/
p
2
u
L
(k
1
) = 2Ep sin, (16)
v
L
(k
2
)/
(p
1
)u
L
(k
1
) = 2E
_
1 + cos
2
_
, (17)
v
L
(k
2
)/
(p
2
)u
L
(k
1
) = 2E
_
1 cos
2
_
, (18)
v
L
(k
2
)/
0
(p
1
)u
L
(k
1
) = v
L
(k
2
)/
0
(p
2
)u(k
1
) =
2E
2
sin
m
W
, (19)
v
R
(k
2
)
/
p
1
u
R
(k
1
) = v
R
(k
2
)
/
p
2
u
R
(k
1
) = 2Ep sin, (20)
v
L
(k
2
)/
(p
1
)u
L
(k
1
) = 2E
_
1 + cos
2
_
, (21)
v
L
(k
2
)/
(p
2
)u
L
(k
1
) = 2E
_
1 cos
2
_
, (22)
v
L
(k
2
)/
0
(p
1
)u
L
(k
1
) = v
L
(k
2
)/
0
(p
2
)u(k
1
) =
2E
2
sin
m
W
. (23)
We rst consider e
L
e
+
R
W
+
W
. In this case we take u(k

1
) = u
L
(k
1
) =
2E(0, 1, 0, 0)
T
and v(k
2
) = v
L
(k
2
) =
2E(0, 0, 1, 0). Then each of the nal W particle can have polar-
ization (+, , 0), which gives 9 possible combinations for (W
+
, W
). Now we evaluate
the corresponding amplitudes in turn.
iM(e
L
e
+
R
W
+
(0)
W
(0)
)
5
= ie
2
_
m
2
Z
s(s m
2
Z
)

1
2s
2
w
1
s M
2
Z
_
_
4Ep(E
2
+ p
2
)
m
2
W
+
16E
3
p
m
2
W
_
sin
+
ie
2
2s
2
w
1
u

2E(3E
2
p + p
3
2E
3
cos ) sin
m
2
W
=ie
2
s
4m
2
W
_
m
2
Z
s m
2
Z
(3
2
)
1
2s
2
w
_
_
2
1 +
2
+ 2 cos

s
s m
2
Z
_
(3
2
) +
4 cos
1 +
2
+ 2 cos
__
sin (24)
iM(e
L
e
+
R
W
+
(0)
W
()
) = iM(e
L
e
+
R
W
+
()
W
(0)
)
= ie
2
_
m
2
Z
s(s m
2
Z
)

1
2s
2
w
1
s M
2
Z
_
_
8E
2
p
m
W
1 + cos
2
_
ie
2
2s
2
w
1
u

2E
m
W
(E
2
(2 cos 1) + 2Ep p
2
)
1 + cos
2
= ie
2
_
m
2
Z
s m
2
Z

1
2s
2
W
_
s
s m
2
Z
+
1 2 cos 2
2
1 +
2
+ 2 cos
__
s
m
W
1 + cos
2
(25)
iM(e
L
e
+
R
W
+
()
W
()
)
= ie
2
_
m
2
Z
s(s m
2
Z
)

1
2s
2
w
1
s M
2
Z
_
_
4Ep sin
_
+
ie
2
2s
2
w
1
u
2E(p + E cos ) sin
= ie
2
_
m
2
Z
(s m
2
Z
)
+
1
2s
2
w
_
s
s M
2
Z

2( + cos )
(1 +
2
+ 2 cos )
__
sin (26)
iM(e
L
e
+
R
W
+
()
W
()
)
=
ie
2
2s
2
w
1
u
2E
2
(1 + cos ) sin =
ie
2
2s
2
w
2(1 cos ) sin
(1 +
2
+ 2 cos )
. (27)
Though not manifest, these expressions have correct high energy behavior. To see this,
we note that 1 2m
2
W
/s when s m
2
W
. Then, for instance, the amplitude for two
longitudinal W nal state becomes
iM(e
L
e
+
R
W
+
(0)
W
(0)
) = ie
2
s
4m
2
W
_
m
2
Z
s m
2
Z
(3
2
)
1
2s
2
w
_
_
2
1 +
2
+ 2 cos

s
s m
2
Z
_
(3
2
) +
4 cos
1 +
2
+ 2 cos
__
sin
=
ie
2
2s
2
W
(1 + 2 cos ) sin
1 + cos
+O(1/s). (28)
Then we can plot the azimuthal distribution of the corresponding dierential cross
section at s = (1000GeV)
2
, as shown in Figure 4.
Next we consider the other case with e
R
e
+
L
initial state. Now there is no contribution
from u-channel neutrino exchange. The amplitudes for various polarizations of nal W
pairs can be worked out to be
iM(e
R
e
+
L
W
+
(0)
W
(0)
) = ie
2
s
s m
2
Z
m
2
Z
4m
2
W
(
2
3) sin , (29)
iM(e
R
e
+
L
W
+
(0)
W
()
) = iM(e
R
e
+
L
W
+
()
W
(0)
)
= ie
2
m
2
Z
s m
2
Z
s
m
W
1 cos
2
(30)
6
1.0 0.5 0.0 0.5 1.0
0.01
0.1
1
10
100
cos
d
d
c
o
s
total
0,0
,
,
,00,
,00,
Figure 4: The dierential cross section of e
e
+
W
+
W
with denite helicity as a

function of azimuthal angle at s = (1000GeV)
2
.
iM(e
R
e
+
L
W
+
()
W
()
) = ie
2
m
2
Z
s m
2
Z
sin (31)
iM(e
R
e
+
L
W
+
()
W
()
) = 0. (32)
3 Cross section for d u W
k
1
p
2
k
2
p
1
d
u
k
1
p
2
p
1
k
2
d
u
k
1
p
1
k
2
p
2
d
u
Figure 5: The process d u W
at tree level. All initial momenta go inward and all nal

momenta go outward.
In this problem we compute the tree amplitude of d u W
at high energies so
that the quark masses can be ignored. In this case the left-handed and right-handed
spinors decouple and only the amplitudes with d
L
u
R
initial state do not vanish. To
calculate it, we rstly work out the kinematics as follows.
k
1
= (E, 0, 0, E), p
1
= (p, p sin, 0, p cos ),
k
2
= (E, 0, 0, E), p
2
= (E
W
, p sin, 0, p cos ), (33)
where p = Em
2
W
/4E and E
W
= E+m
2
W
/4E. The initial spinors of denite helicities
are given by
u
L
(k
1
) =
2E(0, 1, 0, 0)
T
, v
L
(k
2
) =
2E(1, 0, 0, 0)
T
, (34)
while the polarization vectors for nal photon and W
read
(p
1
) =
1
2
(0, cos , i, sin),
(p
2
) =
1
2
(0, cos , i, sin),
7
L
(p
2
) =
1
mW
(p, E
W
sin, 0, E
W
cos ). (35)
Then the amplitude is given by
iM(d
L
u
R
W
) =
ie
2
2s
w
N
s
s m
2
W
ie
2
3
2s
w
_
N
t
t
+
2N
u
u
_
, (36)
where
N
s
= v
L
(k
2
)
_
(p
1
)
(p
2
)(
/
p
1
/
p
2
) + (p
1
+ 2p
2
)
(p
1
)/
(p
2
)
(2p
1
+ p
2
)
(p
2
)/
(p
1
)
u
L
(k
1
), (37)
N
t
= v
L
(k
2
)/
(p
2
)(/ k
1
/
p
1
)/
(p
1
)u
L
(k
1
), (38)
N
u
= v
L
(k
2
)/
(p
1
)(/ k
1
/
p
2
)/
(p
2
)u
L
(k
1
). (39)
Now, using the physical conditions
(p
i
) p
i
= 0, / k
1
u
L
(k
1
) = 0 and v
L
(k
2
)/ k
2
= 0, we
can show that N
s
= N
t
N
u
. In fact,
N
s
= v
L
(k
2
)
_
2
(p
1
)
(p
2
)
/
p
1
+ 2p
2

(p
1
)/
(p
2
) 2p
1

(p
2
)/
(p
1
)
u
L
(k
1
),
N
t
= v
L
(k
2
)
_
2k
1

1
/
2
+ 2
(p
1
)
(p
2
)
/
p
1
/
(p
1
)/
(p
2
)
/
p
1
u
L
(k
1
),
N
u
= v
L
(k
2
)
_
2k
2

1
/
2
+ 2p
1

(p
2
)/
(p
1
) /
(p
1
)/
(p
2
)
/
p
1
u
L
(k
1
).
Then N
s
= N
t
N
u
is manifest. Note further that s m
2
W
= (t + u), we have
iM(d
L
u
R
W
) =
ie
2
2s
w
_
N
t
N
u
t + u

N
t
3t
+
2N
u
3u
_
=
ie
2
2s
w
(2t u)
3(t + u)
_
N
t
t
+
N
u
u
_
=
ie
2
6
2s
w
(1 3 cos )
_
N
t
t
+
N
u
u
_
. (40)
One can see clearly from this expression that all helicity amplitudes vanish at cos = 1/3.
(Note that the denition of scattering angle is dierent from the one in Peskin &
Schroeder, which, in our notation, is .) Then, by including all helicity combinations
(6 in total), we nd the dierential cross section, as a function of s and , to be
d
d cos
=

2
32s
2
w
_
1 cos
sin
_
2
x
3
+ 18x
2
+ 9x + 24 (x
3
14x
2
+ 9x 8) cos 2
36(s m
2
W
)
, (41)
where x m
2
W
/s.
4 Dependence of radiative corrections on the Higgs
boson mass
(a) We rst analyze the radiative corrections to decay process at 1-loop level with
the Higgs boson in the loop. It is easy to see that if the internal Higgs boson line is
attached to one of the external fermions, the resulted vertex will contribute a factor
of m
f
/v which can be ignored. Therefore only the vacuum polarization diagrams are
relevant, and they should sum to a gauge invariant result.
8
(b) Now we compute the vacuum polarization amplitudes of W
, Z
0
and photon
with Higgs contribution. We will only consider the pieces proportional to g
, namely
WW
(q
2
),
ZZ
(q
2
),
(q
2
) and
Z
(q
2
). It is easy to show that
(q
2
) and
ZZ
(q
2
)
receive no contribution from Higgs boson at 1-loop level, while
WW
(q
2
) and
ZZ
(q
2
)
can be found by computing the following three diagrams:
h
Z
0
Z
0
W
+
W
+
h
0
h
Now we compute these three diagrams in turn for W
+
. The rst diagram reads
(igm
W
)
2
g
_
d
4
k
(2)
4
i
k
2
m
2
h
i
(q k)
2
m
2
W
=
i
(4)
2
g
2
m
2
W
g
_
1
0
dx
(2
d
2
)
2d/2
(m
2
W
, q
2
)
i
(4)
2
g
2
m
2
W
g
_
E +
_
1
0
dx log
M
2
(m
2
W
, q
2
)
_
, (42)
where (m
2
W
, q
2
) = xm
2
W
+(1 x)m
2
h
x(1 x)q
2
, E = 2/ +log 4 log M
2
, and
M
2
is the subtraction scale. The second one reads
(ig/2)
2
_
d
4
k
(2)
4
i
k
2
m
2
h
i
(q k)
2
m
2
W
(2k p)
(2k p)
g
2
4
g
_
d
4
k
(2)
4
_
1
0
dx
(4/d)k
2
(k
2
(m
2
W
, q
2
))
2
i
(4)
2
g
2
4
g
_
1
0
dx2(m
2
W
, q
2
)
_
E + 1 + log
M
2
(m
2
W
, q
2
)
_
, (43)
in which we have ignored terms proportional to q
. Then, the last diagram reads

1
2
(ig
2
/2)g
_
d
4
k
(2)
4
i
k
2
m
2
h
=
g
2
4
g
_
d
4
k
(2)
4
1
k
2
m
2
h
(q k)
2
m
2
W
(q k)
2
m
2
W
=
g
2
4
g
_
d
4
k
(2)
4
_
1
0
dx
k
2
+ (1 x)
2
q
2
m
2
W
(k
2
(m
2
W
, q
2
))
2
i
(4)
2
g
2
4
g
_
1
0
dx
_
_
2(m
2
W
, q
2
) m
2
W
+ (1 x)
2
q
2
_
E
+
_
2(m
2
W
, q
2
) m
2
W
+ (1 x)
2
q
2
_
log
M
2
(m
2
W
, q
2
)
+ (m
2
W
, q
2
)
_
. (44)
Thus we have, when the three diagrams above are taken into account only,
WW
(q
2
) =
g
2
4(4)
2
_
_
3m
2
W
+
1
3
q
2
_
E
+
_
1
0
dx
_
(m
2
W
, q
2
)
_
3m
2
W
+ (1 x)
2
q
2
] log
M
2
(m
2
W
, q
2
)
__
. (45)
Now we extract Higgs mass contribution from this expression in the large Higgs limit, and
also x the subtraction point at M
2
= m
2
W
. In this limit we may take (m
2
W
, q
2
) xm
2
h
,
and log(M
2
/) log(m
2
h
/m
2
W
). We also throw divergent terms with E, which should
be canceled out in the nal expression of zeroth order natural relation after including
9
completely loop diagrams with W, Z, and would-be Goldstone boson internal lines.
Then we have
WW
(q
2
) =
g
2
4(4)
2
_
1
2
m
2
h
+
_
3m
2
W
+
1
3
q
2
_
log
m
2
h
m
2
W
_
. (46)
Similarly, we have, for
ZZ
(q
2
),
ZZ
(q
2
) =
g
2
4(4)
2
cos
2
w
_
1
2
m
2
h
+
_
3m
2
Z
+
1
3
q
2
_
log
m
2
h
m
2
Z
_
. (47)
(c) Now, we derive the zeroth order natural relation given in (21.134) of Peskin &
Schroeder, in the large Higgs mass limit. Note that
=
Z
= 0. Thus,
s
2
sin
2
2
0
=
sin
2
w
cos
2
w
cos
2
w
sin
2
w
_

ZZ
(m
2
Z
)
m
2
Z

WW
(0)
m
2
W
_
=

48
1 + 9 sin
2
w
cos
2
w
sin
2
w
log
m
2
h
m
2
W
, (48)
s
2
W
s
2
=

WW
(m
2
W
)
m
2
Z
+
m
2
W
m
2
Z
ZZ
(m
2
Z
)
m
2
Z
=
5
24
log
m
2
h
m
2
W
. (49)
10
Final Project III
Zhong-Zhi Xianyu

In this nal project, we calculate partial widths of various decay channels of the
standard model Higgs boson. Although a standard-model-Higgs-like boson has been
found at the LHC with mass around 125GeV, it is still instructive to treat the mass of
the Higgs boson as a free parameter in the following calculation.
The main decay modes of Higgs boson include h
0
f

f with f the standard model
fermions, h
0
W
+
W
, h
0
Z
0
Z
0
, h
0
gg and h
0
. The former three
processes appear at the tree level, while the leading order contributions to the latter two
processes are at one-loop level. We will work out the decay widths of these processes in
the following.
In this problem we only consider the two-body nal states. The calculation of decay
width needs the integral over the phase space of the two-body nal states. By momentum
conservation and rotational symmetry, we can always parameterize the momenta of two
nal particles in CM frame to be p
1
= (E, 0, 0, p) and p
2
= (E, 0, 0, p), where E =
1
2
m
h
by energy conservation. Then the amplitude Mwill have no angular dependence. Then
the phase space integral reads
_
d
2
|M|
2
=
1
4
p
m
h
|M|
2
. (1)
Then the decay width is given by
=
1
2m
h
_
d
2
|M|
2
=
1
8
p
m
2
h
|M|
2
. (2)
In part (d) of this problem, we will also be dealing with the production of the Higgs
boson from two-gluon initial state, thus we also write down the formula here for the
cross section of the one-body nal state from two identical initial particle. This time,
the two ingoing particles have momenta k
1
= (E, 0, 0, k) and k
2
= (E, 0, 0, k), with
E
2
= k
2
+ m
2
i
and 2E = m
f
where m
i
and m
f
are masses of initial particles and nal
particle, respectively. The nal particle has momentum p = (m
f
, 0, 0, 0). Then, the
cross section is given by
=
1
2s
_
d
3
p
(2)
3
1
2E
p
|M|
2
(2)
4
(4)
(p k
1
k
2
)
1
Notes by Zhong-Zhi Xianyu Solution to P&S, Final Project III (draft version)
=
1
4m
f
s
|M|
2
(2)(2k m
f
) =

m
2
f
|M|
2
(s m
2
f
), (3)
where =
_
1 (4m
i
/m
f
)
2
is the magnitude of the velocity of the initial particle in
the center-of-mass frame.
(a) The easiest calculation of above processes is h
0
f

f, where f represents all
quarks and charged leptons. The tree level contribution to this process involves a single
Yukawa vertex only. The corresponding amplitude is given by
iM(h
0
f

f) =
im
f
v
u
(p
1
)v(p
2
). (4)
Then it is straightforward to get the squared amplitude with nal spins summed to be
|M(h
0
f

f)|
2
=
m
2
f
v
2
tr
_
(
/
p
1
+ m
f
)(
/
p
2
m
f
)
=
2m
2
f
v
2
(m
2
h
4m
2
f
). (5)
In CM frame, the nal states momenta can be taken to be p
1
= (E, 0, 0, p) and p
2
=
(E, 0, 0, p), with E =
1
2
m
h
and p
2
= E
2
m
2
f
. Then the decay width is given by
(h
0
f

f) =
1
8
p
m
2
h
|M|
2
=
m
h
m
2
f
8v
2
_
1
4m
2
f
m
2
h
_
3/2
. (6)
This expression can be expressed in terms of the ne structure constant , the mass of
W boson m
w
and Weinberg angle sin
w
, as
(h
0
f

f) =
m
h
8 sin
2
w
m
2
f
m
2
W
_
1
4m
2
f
m
2
h
_
3/2
. (7)
(b) Next we consider the decay of h
0
to massive vector bosons W
+
W
and Z
0
Z
0
.
The amplitude for the process h
0
W
+
W
is given by
iM(h
0
W
+
W
) =
ig
g
2
v
2

(p
1
)
(p
2
). (8)
Then the squared amplitude with nal polarizations summed reads
|M|
2
=
g
4
v
2
4
_
g

p
1
p
1
m
2
W
__
g
2
p
2
m
2
W
_
=

sin
2
w
m
4
h
m
2
W
_
1
4m
2
W
m
2
h
+
12m
4
W
m
4
h
_
. (9)
Therefore the decay width is
(h
0
W
+
W
) =
1
8
p
1
m
2
h
|M|
2
=
m
3
h
16m
2
W
sin
2
w
(1 4
1
W
+ 12
2
W
)(1 4
1
W
)
1/2
,
(10)
where we have dened
W
(m
h
/m
W
)
2
for brevity. For h
0
Z
0
Z
0
process, it can be
easily checked that nothing gets changed in the calculation except that all m
W
should be
replaced with m
Z
, while an additional factor 1/2 is needed to account for the identical
particles in nal state. Therefore we have
(h
0
Z
0
Z
0
) =
m
3
h
32m
2
Z
sin
2
w
(1 4
1
Z
+ 12
2
Z
)(1 4
1
Z
)
1/2
, (11)
where
Z
(m
h
/m
Z
)
2
.
2
(c) Now we come to the process h
0
gg. The leading order contribution comes from
diagrams with one quark loop.
The amplitude reads
iM(h
0
gg) =
im
q
v
(ig
s
)
2
(p
1
)
(p
2
) tr (t
a
t
b
)
_
d
d
q
(2)
d
_
(1) tr
_
i
/
q m
q
i
/
q +
/
p
2
m
q
i
/
q
/
p
1
m
q
_
+ (1) tr
_
i
/
q m
q
i
/
q +
/
p
1
m
q
i
/
q
/
p
2
m
q
__
(12)
The rst trace in the integrand can be simplied through standard procedure,
tr
_
i
/
q m
q
i
/
q +
/
p
2
m
q
i
/
q
/
p
1
m
q
_
=
i tr
_
(
/
q + m
q
)(
/
q +
/
p
2
m
q
)(
/
q
/
p
1
m
q
)
(q
2
m
2
q
)
_
(q + p
2
)
2
m
2
q
_
(q p
1
)
2
m
2
q
= 2i
_
1
0
dx
_
1x
0
dy
N
(q
2
)
3
, (13)
where
q
= q
xp
1
+ yp
2
, (14)
= m
2
q
x(1 x)p
2
1
y(1 y)p
2
2
2xyp
1
p
2
= m
2
q
xym
2
h
,
N
= 4m
q
_
p
1
p
2
p
1
p
2
+ 2p
2
q
2p
1
q
+ 4q
+ (m
2
q
p
1
p
2
q
2
)
_
(15)
Then we can reexpress N
in terms of q
, p
1
and p
2
and drop o all terms linear
in q which integrates to zero. It is most easy to work with denite helicity states for
nal gluons. Then the result gets simplied if we dot N
with polarization vectors

as N
(p
1
)
(p
2
). Note that
(p
i
) p
j
= 0 with i, j = 1, 2. Note also the on-shell
condition p
2
1
= p
2
2
= 0, p
1
p
2
=
1
2
m
2
h
. then
N
(p
1
)
(p
2
) = 4m
q
_
m
2
q
+
_
xy
1
2
_
m
2
h
+
_
4
d
1
_
q
2
_
(p
1
)
(p
2
). (16)
The same calculation shows that the second trace in the integrand of (12) gives identical
result with the rst trace. To check the gauge invariance of this result, one can simply
replace
(p
1
) with p
1
in the expression above, then it is straightforward to nd that
N
p
1
(p
2
) = 0. Similarly, it can also be checked that N
(p
1
)p
2
= 0.
Then the amplitude (12) now reads
iM(h
0
gg) =
2g
2
s
m
q
v

ab
_
1
0
dx
_
1x
0
dy
_
d
d
q
(2)
d
N
(p
1
)
(p
2
)
(q
2
)
3
, (17)
3
where the relation tr (t
a
t
b
) =
1
2

ab
in fundamental representation is also used. The
momentum integration is nite as d 4 under dimensional regularization, and can now
be carried out directly to be
iM(h
0
gg) =
2ig
2
s
m
2
q
(4)
2
v

ab
(p
1
)
(p
2
)
_
1
0
dx
_
1x
0
dy
(1 4xy)m
2
h
m
2
q
xym
2
h
=
i
s
m
2
h
6v

ab
(p
1
)
(p
2
)I
f
(
q
), (18)
where
q
(m
h
/m
q
)
2
, and
I
f
(
q
) 3
_
1
0
dx
_
1x
0
dy
1 4xy
1 xy
q
Note that the inner product between two polarization vectors is nonzero only for
and
+
. Therefore the squared amplitude with nal states polarizations, color indices
summed (
ab
ab
= 8) is,
|M(h
0
gg)|
2
= |M
+
(h
0
gg)|
2
+|M
+
(h
0
gg)|
2
=
4
2
s
m
4
h
9
2
v
2
|I
f
(
q
)|
2
, (19)
and the decay width is
(h
0
gg) =
_
m
h
8 sin
2
w
_
m
2
h
m
2
w

2
s
9
2
|I
f
(
q
)|
2
, (20)
where an additional factor 1/2 should be included in (2) when calculating (h
0
gg) because the two gluons in nal states are identical particles. This result is easily
generalized for N
q
copies of quarks to be
(h
0
gg) =
_
m
h
8 sin
2
w
_
m
2
h
m
2
W

2
s
9
2

q
I
f
(
q
)
2
, (21)
(d) Now we calculate the cross section for the Higgs production via gluon fusion at the
leading order. The amplitude is simply given by the result in (c), namely (18). When
we take the square of this amplitude, an additional factor (
1
8

1
2
)
2
should be included,
to average over helicities and color indices of initial gluons. Then, comparing (3) with
(2), we nd that
(gg h
0
) =

2
8m
h
( s m
2
h
)(h
0
gg), (22)
where the hatted variable s is the parton level center-of-mass energy. We note again
that the correct formula is obtained by including an factor of (
1
8

1
2
)
2
in (gg h
0
)
to average over the initial degrees of freedom of two gluons, and an factor of 1/2 in
(h
0
gg) to count the identical particles in the nal state. Then, from (21), it is
straightforward to nd
(gg h
0
) =

2
s
576 sin
2
m
2
h
m
2
W
q
I
f
(
q
)
2
( s m
2
h
). (23)
Then the proton-level cross section of Higgs boson production via gluon-gluon fusion is
given by
GGF
_
p(P
1
)p(P
2
) h
0
)
4
=
_
1
0
dx
1
_
1
0
dx
2
f
g
(x
1
)f
g
(x
2
)
_
g(x
1
P
1
)g(x
2
P
2
) h
0
_
=
_
dM
2
dY
(x
1
, x
2
)
(M
2
, Y )
f
g
(x
1
)f
g
(x
2
)
_
g(x
1
P
1
)g(x
2
P
2
) h
0
_
=
_
dM
2
dY
1
M
2
x
1
f
g
(x
1
)x
2
f
g
(x
2
)
_
g(x
1
P
1
)g(x
2
P
2
) h
0
_
, (24)
where M
2
= x
1
x
2
s is the center-of-mass energy of two initial gluons, while s is the
center-of-mass energy of two initial protons, and Y , given by expY =
_
x
1
/x
2
, is
the rapidity of the produced Higgs boson relative to the center-of-mass frame of the
proton system. (Note that in our case M
2
= m
2
h
.) The relations between M
2
, Y and
the momentum fractions x
1
, x
2
can be inverted to give x
1
= (M/
s)e
Y
and x
2
=
(M/
s)e
Y
. Furthermore, f
g
is the parton distribution function of the gluon in a
proton, which we will take to be f
g
= 8(1 x)
7
/x in the following calculations. Then
the cross section can be evaluated to be
GGF
_
p(P
1
)p(P
2
) h
0
)
=

2
s
9 sin
2
1
m
2
W
q
I
f
(
q
)
2
_
Y0
Y0
dY
_
1
m
h
s
e
Y
_
7
_
1
m
h
s
e
Y
_
7
, (25)
where Y
0
, given by coshY
0
=

s/2m
h
is the largest possible rapidity of a produced
Higgs boson. We plot this cross section as a function of the center-of-mass energy
s of
the pp pair, with the Higgs bosons mass taken to be m
h
= 30GeV and m
h
= 125GeV,
respectively, in Figure
1 2 5 10 20 50
1
2
5
10
20
50
s GeV
C
r
o
s
s
S
e
c
t
i
o
n
p
b
m
h
30 GeV
m
h
125 GeV
(e) Next we consider the process h
0
2. The contribution to this decay channel at
the leading (1-loop) level is from two types of diagrams, one with a fermion loop and
the other with a W boson (and related would-be Goldstone boson) loop. The former
contribution is easy to nd by virtue of the result in (c) for h
0
gg. The calculation
here is in fully parallel, except that we should include the factor for the electric charges
of internal fermions Q
f
, take away the color factor tr (t
a
t
b
), change the strong coupling
g
s
by the electromagnetic coupling e, and sum over all charged fermions. Note that the
color factor enters the expression of the decay width as | tr (t
a
t
b
)|
2
=
1
2

ab 1
2

ab
= 2,
then it is straightforward to write down the fermion contribution to the h
0
2 to be
iM(h
0
2)
f
=
_
m
h
8 sin
2
w
_
m
2
h
m
2
w

2
s
18
2

f
Q
2
f
N
c
(f)I
f
(
f
)
2
, (26)
5
where N
c
(f) is the color factor, equal to 3 for quarks and 1 for charged leptons.
(f ) Now we come to the W-loop contributions to h
0
2. In Feynman-t Hooft
gauge, we should also include the corresponding Goldstone loop diagrams. Then there
are 13 diagrams in total. We compute them as follows,
(a) (b)
iM
(a)
=
1
2
ig
g
2
v
2
(ie
2
)(2
(p
1
)
(p
2
)
_
d
d
q
(2)
d
D
W
(q)D
W
(k q)
=
2i
(4)
d/2
e
2
m
2
W
v

(p
1
)
(p
2
)(d 1)(2
d
2
)
_
1
0
dx
[m
2
W
x(1 x)m
2
h
]
2d/2
, (27)
iM
(b)
=
1
2
(2iv)(2ie
2
)
(p
1
)
(p
2
)
_
d
d
q
(2)
d
D
s
(q)D
s
(k q)
=
i
(4)
d/2
e
2
m
2
h
v

(p
1
)
(p
2
)(2
d
2
)
_
1
0
dx
[m
2
W
x(1 x)m
2
h
]
2d/2
. (28)
(c) (d)
iM
(c)
= iM
(d)
=
ig
2
sin
w
2

ig
2
v sin
w
2

(p
1
)
(p
2
)
_
d
d
q
(2)
d
D
s
(q)D
W
(p
2
q)
=
i
(4)
d/2
e
2
m
2
W
v

(p
1
)
(p
2
)(2
d
2
)
1
(m
2
W
)
2d/2
. (29)
(e) (f) (g)
6
(h) (i) (j)
(k) (l) (m)
iM
(e)
=
ig
2
v
2
(ie)
2
(p
1
)
(p
2
)
_
d
d
q
(2)
d
D
W
(q)D
W
(q p
1
)D
W
(q + p
2
)
(2q p
1
)
(2p
1
q)
(p
1
+ q)
(2q + p
2
)
(q p
2
)
(2p
2
+ q)
=
i
(4)
d/2
e
2
m
2
W
v

(p
1
)
(p
2
)
_ _
dxdy
(5 x y + 4xy)m
2
h
m
2
W
xym
2
h
+ 6(d 1)(2
d
2
)
_
dxdy
(m
2
W
xym
2
h
)
2d/2
_
, (30)
iM
(f)
= (2i)(ie)
2
(p
1
)
(p
2
)
_
d
d
q
(2)
d
(2q p
1
)
(2q + p
2
)
D
s
(q)D
s
(q p
1
)D
s
(q + p
2
)
=
i
(4)
d/2
e
2
m
2
h
v

(p
1
)
(p
2
)(2
d
2
)
_
2dxdy
(m
2
W
xym
2
h
)
2d/2
, (31)
iM
(g)
=
_
im
2
W
v
_
(ie)
2
(p
1
)
(p
2
)
_
d
d
q
(2)
d
(1)(q p
1
)
D
s
(q)D
s
(q p
1
)D
s
(q + p
2
)
=
i
(4)
d/2
e
2
m
2
W
v

(p
1
)
(p
2
)(2
d
2
)
_
dxdy
(m
2
W
xym
2
h
)
2d/2
, (32)
iM
(h)
= iM
(i)
=
ig
2
ig
g
2
v sin
w
2
(ie)
(p
1
)
(p
2
)
_
d
d
q
(2)
d
(q p
1
k)
(2q + p
2
)
(q p
2
)
(2p
2
+ q)
D
W
(q)D
s
(q p
1
)D
W
(q + p
2
)
=
i
(4)
d/2
e
2
m
2
W
v

(p
1
)
(p
2
)
_ _
dxdy
(1 x)(1 + y)m
2
h
m
2
W
xym
2
h
1
2
(d 1)(2
d
2
)
_
dxdy
(m
2
W
xym
2
h
)
2d/2
_
, (33)
iM
(j)
=
ig
2
v
2
_
ig
2
v sin
w
2
_
2
(p
1
)
(p
2
)
_
d
d
q
(2)
d
D
s
(q)D
W
(q p
1
)D
W
(q + p
2
)
7
=
i
(4)
d/2
e
2
m
2
W
v

(p
1
)
(p
2
)
_
dxdy
2m
2
W
m
2
W
xym
2
h
, (34)
iM
(k)
= iM
(l)
=
ig
2
ig
2
v sin
w
2
(ie)
(p
1
)
(p
2
)
_
d
d
q
(2)
d
(p
1
+ 2p
2
+ q)
(2q + p
2
)
D
s
(q)D
W
(q p
1
)D
s
(q + p
2
)
=
i
(4)
d/2
e
2
m
2
W
v

(p
1
)
(p
2
)(2
d
2
)
_
dxdy
(m
2
W
xym
2
h
)
2d/2
, (35)
iM
(m)
= (2i)
_
ig
2
v sin
w
2
_
2
(p
1
)
(p
2
)
_
d
d
q
(2)
d
D
W
(q)D
s
(q p
1
)D
s
(q + p
2
)
=
i
(4)
d/2
e
2
m
2
h
v

(p
1
)
(p
2
)
_
dxdy
m
2
W
m
2
W
xym
2
h
. (36)
The results can be summarized as,
iM
(X)
=
i
(4)
d/2
e
2
m
2
W
v

(p
1
)
(p
2
)
_
A (2
d
2
) + B
_
, (X = a, b, , m) (37)
with the coecients A and B for each diagram listed in Table.
Diagrams A B
(a) 2(d 1)J
1
0
(c)+(d) 2(m
2
W
)
d/22
0
(e) 6(d 1)J
2
J
3
(g) J
2
0
(h)+(i) (d 1)J
2
2J
4
(j) 0 2(m
W
/m
h
)
2
J
5
(k)+(l) 2J
2
0
(m) 0 J
5
(b) (m
h
/m
W
)
2
J
1
0
(f) 2(m
h
/m
W
)
2
J
2
0
where
J
1
=
_
1
0
dx
1
[m
2
W
x(1 x)m
2
h
]
2d/2
= 1

2
_
1
0
dxlog
_
m
2
W
x(1 x)m
2
h
_
+O(
2
), (38)
J
2
=
_
1
0
dx
_
1x
0
dy
1
(m
2
W
xym
2
h
)
2d/2
=
1
2

2
_
1
0
dx
_
1x
0
dy log
_
m
2
W
xym
2
h
_
+O(
2
), (39)
J
3
=
_
1
0
dx
_
1x
0
dy
(5 x y + 4xy)m
2
h
m
2
W
xym
2
h
, (40)
J
4
=
_
1
0
dx
_
1x
0
dy
(1 x)(1 + y)m
2
h
m
2
W
xym
2
h
, (41)
J
5
=
_
1
0
dx
_
1x
0
dy
m
2
h
m
2
W
xym
2
h
. (42)
8
To see that the divergences of all diagrams cancel among themselves, it just needs to
show that sum of all A-coecients is of order . This is straightforward by noting that
J
1
= 1 +O() and J
2
= 1/2 +O().
Before reaching the complete result, let us rst nd out the W-loop contribution in
the limit m
2
h
m
2
W
, although it seems unlikely to be true within our current knowledge.
To nd the amplitude in this limit, we expand the ve integrals J
1
, , J
5
in terms of
m
h
/m
W
,
J
1
1

2
log m
2
W
+

12
m
2
h
m
2
W
, J
2

1
2

4
log m
2
W
+

48
m
2
h
m
2
W
,
J
3

7
3
m
2
h
m
2
W
, J
4

11
24
m
2
h
m
2
W
, J
5

1
2
m
2
h
m
2
W
+
1
24
_
m
2
h
m
2
W
_
2
.
Then the amplitude can be recast into
iM=
ie
2
m
2
W
(4)
2
v

(p
1
)
(p
2
)
_
C
_
2
+ log 4
_
+ D log m
2
W
+ E + F
m
2
h
m
2
W
_
(43)
Diagrams C D E F
(a) 6 3 4 1
(c)+(d) 2 1 0 0
(e) 9 9/2 6 37/12
(g) 1/2 1/4 0 1/24
(h)+(i) 3/2 3/4 1 19/24
(j) 0 0 1 1/12
(k)+(l) 1 1/2 0 1/12
(m) 0 0 0 1/2
(b) (m
h
/m
W
)
2
(m
h
/m
W
)
2
/2 0 0
(f) (m
h
/m
W
)
2
(m
h
/m
W
)
2
/2 0 0
sum 0 0 0 7/2
Therefore, the amplitude in the limit m
2
h
m
2
W
is given by
iM(h
0
2)
W
= 2
7
2
im
2
h
4v

(p
1
)
(p
2
), (44)
where the factor 2 counts the identical contributions from the diagrams with two nal
photons changed. Now we sum up the fermion-loop contribution found in (e) and the
result here to get the h
0
2 amplitude in the light Higgs limit,
iM=
im
2
h
3v
_
f
Q
2
f
N
c
(f)
21
4
_
(p
1
)
(p
2
). (45)
Then the corresponding partial width is given by
(h
0
2) =
_
m
h
8 sin
2
w
_
m
2
h
m
2
w

2
18
2

f
Q
2
f
N
c
(f)
21
4
2
, (46)
Now we retain m
h
as a free variable. Then the various diagrams sum into the
following full expression for the W-loop contribution to h
0
2,
iM(h
0
2)
W
=
im
2
h
2v

(p
1
)
(p
2
)I
W
(
W
), (47)
9
where the factor I
W
(
W
), as a function of
W
(m
h
/m
W
)
2
, is given by
I
W
(
W
) =
1
W
_
6I
1
(
W
) 8I
2
(
W
) +
W
_
I
1
(
W
) I
2
(
W
)
_
+ I
3
(
W
)
_
, (48)
where
I
1
(
W
)
_
1
0
dx log
_
1 x(1 x)
W
, (49)
I
2
(
W
) 2
_
1
0
dx
_
1x
0
dy log
_
1 xy
W
_
, (50)
I
3
(
W
)
_
1
0
dx
_
1x
0
dy
(8 3x + y + 4xy)
W
1 xy
W
. (51)
Then the full expression for the partial width of h
0
2 at one-loop is
(h
0
2) =
_
m
h
8 sin
2
w
_
m
2
h
m
2
w

2
18
2

f
Q
2
f
N
c
(f)I
f
(
f
) I
W
(
W
)
2
, (52)
(h) Collecting all results above (expect the channel, which is quite small
), we plot
the total width and decay branching fractions of the Higgs boson in Figures 1 and 2,
respectively.
100 200 300 400 500
10
4
0.001
0.01
0.1
1
10
m
h
GeV
T
o
t
a
l
W
i
d
t
h
G
e
V
Figure 1: The total width of the Higgs boson as a function of its mass.
but very important!

10
100 200 300 400 500
0.001
0.005
0.010
0.050
0.100
0.500
1.000
m
h
GeV
B
r
a
n
c
h
i
n
g
R
a
t
i
o
s
W
Z
0
Z
0
tt
gg
bb
cc
Figure 2: The Higgs decay branching fractions of t
t, b
b, c c,
+
, WW, ZZ and gg
channels, as functions of Higgs mass.
11

Peskin QFT Solutions

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Peskin QFT Solutions

Uploaded by

Copyright:

Available Formats

Solutions to Peskin & Schroeder

Institute of Modern Physics and Center for High Energy Physics,

(y)] = i(x y), (11)

Institute of Modern Physics and Center for High Energy Physics,

= 0, 1/2, 1, 3/2, 2, are labels of irreducible repre-

is in fact a Lorentz vector.

is called a Lorentz vector, if it satises the following transformation

in its innitesimal form. We further note that:

is invariant under the simultane-

is a right-handed spinor that transforms as

vanishes, we make use of the equations of motion:

. (3) Integrating over d

and aa terms vanish, while the aa

a terms remain. (5)

, we get the corresponding equation of motion:

, we calculate the trace in Weyl representation without loss

) (no sum on ). For = 0, we have tr (

with C a commuting number, which

(t, x). (62)

(t, x). (64c)

(t, x)P = (1)

(t, x), (65a)

(t, x)T = (1)

(t, x); (65b)

(t, x). (65c)

. But it is easy to show that any operator of this

Institute of Modern Physics and Center for High Energy Physics,

= 0, thus we see that O(p

amplitude is still given by

(x). Then the interaction Hamiltonian is:

. Taking all these delta functions into account, we get:

(q)u(p) with q = p k. (57)

Institute of Modern Physics and Center for High Energy Physics,

(q)u(p) with q = p k. (2)

. At the tree level, it consists

can be determined by projecting out the other one

back, we nally get the

2. In the following, a hatted amplitude

2, which is illustrated in Figure 3

since the temporal component of the polarization

always vanishes, we get a much more simplied expression,

= cos . We can use these

2 for triplet and singlet positronium, respectively, as done in last part,

M, we only need the terms linear in k, as listed in (67). Let us

B) and the decay rate (B

is the polarization of photon while e

is the polarization for B. Now we square

(k) of the photon can be taken to be

2 in the massless limit. The

k] vanish. Square this amplitude, we get

Institute of Modern Physics and Center for High Energy Physics,

)/E. Now let

which is logarithmically divergent as 0.

p) can be thrown away by Ward identity q

Institute of Modern Physics and Center for High Energy Physics,

into a suitable form. The calculation is basically in parallel

/ k contributes to the nal result. Thus we

which contribute nothing to nal

into a linear combination of

. Terms proportional to (pp

into a linear combination of

into the following form:

, which is similar to the corresponding QED correction:

by gamma matrix relations and equations of motion, as