Solutions to Peskin and Schroder – Andrzej Pokraka

Problem 10.1: One-loop structure of QED

In Section 10.1 we argued form general principles that the photon one-point and three-point functions
vanish, while the four-point function is finite.
(a) Verify directly that the one-loop diagram contributing to the one-point function vanishes.
There are two Feynman diagrams contributions to the three-point function at one-loop order.
Show that these cancel. Show that the diagrams contributing to any n-point photon amplitude,
for n odd, cancel in pairs.
(b) The photon four-point amplitude is a sum of six diagrams. Show explicitly that the potential
logarithmic divergences of these diagrams cancel.

10.1 part (a)

The one-point function is given by the diagram

iΠµ1-pt (q 2 ) = µ (1)
q
k

Applying the Feynman rules we have

d4 k
Z 
µ i(k
+ me )
iΠµ1-pt (q 2 ) = Tr (−ie)γ = 0. (2)
(2π)4 k 2 − m2e + i
kν
The trace of me γ µ and the integral d4 k f (k
R
2 ) vanish giving the result (2).

At one-loop, the three-point function is

β β

k + p2 p2 −k − p2 p2
iΠαβγ
3-pt = α k + α −k (3)
p1 p3 p1 p3
k − p3 −k + p3
γ γ

where the momentum flow in the loop is counter-clockwise and the momentum of the external
photons is directed inward towards the loop. The first graph in equation (4) contains electrons while
the second graph contains positrons. In the second graph we use minus the momentum of the first
graph so that k is the same for both graphs – this is due to the fact that the momentum flow is
in the same direction as in the first diagrams and that the trace is taken against the fermion flow,
which is reversed relative to the first diagram.
Applying the QED Feynman rules we find
d4 k (−ie)3 i3
Z
iΠαβγ
3-pt =
(2π) ((k + p2 ) − me ) (k 2 − m2e ) ((k − p3 )2 − m2e )
4 2 2

× Tr γ α k
+ p

β
γ


2 + me γ
k
+ me γ
k
− p
3 + me
−γ α k
− p γ β



3 − me γ k
− me γ k
+ p
2 − me . (4)

1
 Solutions to Peskin and Schroder – Andrzej Pokraka

The first term in the trace of (4) is

Tr γ α (k
+ p β γ 2
   α β γ α β γ α β γ


2 )γ k
γ (k
− p
2 )γ γ + γ γ k
γ + γ γ γ (k
− p

3 ) + me Tr γ (k
+ p
3 ) (5)
while the second term is
−Tr γ α (k
− p γ β 2
   α γ β α γ β α γ β


3 )γ k
γ (k
+ p
2 ) − me Tr γ (k
−p
3 )γ γ + γ γ k
γ + γ γ γ (k
+ p
2)
β γ α
− m2e Tr γ β γ γ (k
− p α β γ α β γ α
   
= −Tr (k
+ p
2 )γ k
γ (k
− p

3 )γ
3 )γ + γ k
γ γ + (k
+ p

2 )γ γ γ

= −Tr γ α (k
+ p β γ 2
   α β γ α β γ α β γ


2 )γ k
γ (k
− p
3 ) − me Tr γ γ γ (k
− p
3 ) + γ γ k
γ + γ (k
+ p
2 )γ γ . (6)
Since (6) is the exact negative of (5), the three-point function (4) vanishes. Note that in the second
line of (6) we have used the fact that the trace of a product of gamma matrices is the same as the
trace of the reversed product: Recall that the charge conjugation matrix, C = γ 0 γ 2 , satisfies
C2 = 1 and Cγ µ C = −(γ µ )T . (7)
Therefore,
Tr [γ α1 γ α2 . . . γ αn ] = Tr [CCγ α1 CCγ α2 CC . . . CCγ αn CC]
= −(1)n Tr C(γ α1 )T (γ α2 )T . . . (γ αn )T C
 

= −(1)n Tr (γ αn . . . γ α2 γ α1 )T
 

= −(1)n Tr [γ αn . . . γ α2 γ α1 ]
= Tr [γ αn . . . γ α2 γ α1 ] (8)
where in the last line we have used the fact that n must be even to get a non-zero trace.
The n-point function for n odd is given by

α2 α2
... ...
p2 p2
k2 −k2
k1 p1 −k1 p1
α1 ...αn
iΠn-pt = α1 + α1 + . . . (9)
kn −kn
pn pn
... ...
αn αn

Pi
where kn ≡ k and ki = k + l=1 pl for 1 ≤ i ≤ n − 1. Applying the Feynman rules we obtain
d4 k (−ie)n in
Z
α1 ...αn
iΠn-pt =
(2π)4 (k12 − m2e ) . . . (kn2 − m2e )
× Tr (k
n + me )γ αn . . . (k
1 + me )γ α1 + γ α1 (−k
1 + me ) . . . γ αn (−k
n + me ) + . . . .
 

(10)
We can show, for odd n, that the trace from the second diagram of (9) is the negative of the trace
from the first diagram in (9):
Tr γ α1 (−k
1 + me ) . . . γ αn (−k
n + me ) = Tr CCγ α1 CC(−k
1 + me )CC . . . CCγ αn CC(−k
n + me )CC
   

= Tr C(−γ α1 )T (k
T1 + me ) . . . (−γ αn )T (k
Tn + me )C

 
h
T i
= (−1)n Tr (k
n + me )γ αn . . . (k
1 + me )γ α1
= −Tr (k
n + me )γ αn . . . (k
1 + me )γ α1 .
 
(11)

2
 Solutions to Peskin and Schroder – Andrzej Pokraka

Therefore the first two terms in the n-point function vanish for odd n.
There are more diagrams other than those shown explicitly in equation (9). However, each
diagram has a pair where the particles in the loop are antiparticles and vice versa. In exactly the
same way that the first two diagrams of (9) cancel, each other pair of diagrams cancel. Thus, the
n-point function for odd n vanishes.

10.1 part (b)

The diagrams for the 4-pt function are given in Fig. 1. To calculate the divergent part of each
diagram we only need to retain the terms in the numerator with highest power of loop momentum,
k. Furthermore, the trace of the diagram with particles in the loop is the same as the trace for
antiparticles in the loop and therefore add. Therefore, the total amplitude is

M = 2(M1 + M3 + M5 ). (12)

α2 α3 α2 α3 α2 α3
p3 p3
p2 p3
p2 p2
k2 −k2 k2

k1 k3 −k1 −k3 k1 k − p3

k −k k
p1 p4
p1 p4 p1 p4
α1 α4 α1 α4 α1 α4

α2 α3 α2 α3 α2 α3
p3 p
k1 + p3 −k1 − p3 3
p2 p3 p2 p2
−k2

−k1 −k + p3
k1 k3 −k1 −k3
−k
p1 p4 p4 p4
p1 k p1 −k
α1 α4 α1 α4 α1 α4
Figure 1: Diagrams contributing to the 4-pt function. To keep track of the diagrams let us label
them by Mi where i = 1, . . . , 6.

We begin by evaluating the divergent part of M1 by series expanding the denominators and
keeping the O(k −4 ) term
 
d4 k Tr k
γ α4 k
γ α3 k
γ α2 k
γ α1
Z
4
M1 = e + finite terms. (13)
(2π)4 (k 2 − m2e )4

Note that because we know that there will be no divergence we can work in 4 dimensions to make

3
 Solutions to Peskin and Schroder – Andrzej Pokraka

the Dirac algebra in the trace simpler

k4
Tr k
γ α4 k
γ α3 k
γ α2 k
γ α1 → (gβ1 β2 gβ3 β4 + gβ1 β3 gβ2 β4 + gβ1 β4 gβ2 β3 ) Tr γ β4 γ α4 γ β3 γ α3 γ β2 γ α2 γ β1 γ α1
   
24
k4
= (Tr [γ α4 γ α3 γ α2 γ α1 ] − g α4 α2 Tr [γ α3 γ α1 ])
3
4k 4 α4 α3 α2 α1
= (g g − 2g α4 α2 g α3 α1 + g α4 α1 g α3 α2 ) . (14)
3
The trace in M3 is the same as the trace in M1 with α4 ↔ α3 and the trace in M5 is the same
as the trace in M1 with α3 ↔ α2 . Therefore,

M ∝ g α4 α3 g α2 α1 − 2g α4 α2 g α3 α1 + g α4 α1 g α3 α2
+ g α4 α3 g α2 α1 − 2g α3 α2 g α4 α1 + g α3 α1 g α4 α2
+ g α4 α2 g α3 α1 − 2g α4 α3 g α2 α1 + g α4 α1 g α2 α3
=0 (15)

Problem 10.2: Renormalization of the Yukawa Lagrangian

Consider the pseudo scalar Yukawa Lagrangian,
1 1
L= (∂µ φ)2 − m2φ φ2 + ψ̄(i∂
− mψ )ψ − ig ψ̄γ 5 ψφ (16)
2 2
where φ is a real scalar field and ψ is a Dirac fermion. Notice that this Lagrangian is invariant under
the parity transformation ψ(t, x) → γ 0 ψ(t, −x), φ(t, x) → −φ(t, −x), in which the field φ carries
odd parity.
(a) Determine the superficially divergence amplitudes and work out the Feynman rules for renor-
malized perturbation theory for Lagrangian. Include all necessary counter term vertices. Show
that the theory contains a superficially divergent φ4 . This means that the theory cannot be
renormalized unless one includes a scalar self-interaction,
λ 4
δL = φ , (17)
4!
and a counter term of the same form. It is of course possible to set the renormalized value
of this coupling to zero, but that is not a natural choice, since the counter term will still be
nonzero. Are there any further interactions required?

(b) Compute the divergent part (the pole as d → 4) of each counter term, to the one-loop order
of perturbation theory, implementing a sufficient set of renormalization condition. You need
not worry about finite parts of the counter terms. Since the divergent parts must have a
fixed dependence on the external momenta, you can simplify this calculation by choosing the
momenta in the simplest possible way.

4
 Solutions to Peskin and Schroder – Andrzej Pokraka

Part (a)
The Feynman rules for the pseudo scalar Yukawa theory are listed in Figs. 2 and 3.

φ φ i ψ ψ̄ i
= p2 −m2φ
, = ,
p p
p−mψ

Figure 2: Feynman propagators for the Lagrangian (16).

= gγ 5

Figure 3: Feynman vertices for the Lagrangian (16).

D=2 D=1

D=0 D=0

Figure 4: All one-loop diagrams along with their superficial divergence.

Our first task is to determine all the superficially divergent diagrams at one-loop (Fig. 4). Note
that since the Lagrangian is invariant under parity, the interaction cannot change the parity of the
initial state. This means that scattering amplitudes with an odd number of external pseudo scalars
and no external fermions must vanish. Furthermore, since there is a divergent φ4 diagram in Fig. 4,
we must include a φ4 interaction,
λ 4
δL = − φ , (18)
4!
and a corresponding counter term to renormalize the theory. The new Lagrangian becomes
1 1 λ
L= (∂µ φ)2 − m2φ φ2 + ψ̄(i∂
− mψ )ψ − ig ψ̄γ 5 ψφ − φ4 . (19)
2 2 4!

5
 Solutions to Peskin and Schroder – Andrzej Pokraka

Next, we rescale the fields by the field strength

p
ψ → Zψ ψ
p
φ → Zφ φ (20)
to get
1 1 λ
Zφ (∂µ φ)2 − Zφ m2φ φ2 − Zφ2 φ4 + Zψ ψ̄(i∂
− mψ )ψ − igZψ Zφ ψ̄γ 5 ψφ.
p
L= (21)
2 2 4!
Defining
Zψ = 1 + δZψ
Zφ = 1 + δZφ
m2ψ Zψ → m2ψ + δmψ
m2φ Zφ → m2φ + δmφ
Zφ2 λ → λ + δλ,
p
g Zφ Zψ → g + δg. (22)
where mψ , mφ , λ and g are now the physical masses and coupling constants, the Lagrangian becomes
L = L0 + δL,
1 1 λ
L0 = (∂µ φ)2 − m2φ φ2 + ψ̄(i∂
− mψ )ψ − ig ψ̄γ 5 ψφ − φ4 ,
2 2 4!
1 1 1
2 2 4
δL = δZφ (∂µ φ) − δmφ φ − δλ φ + δZψ ψ̄(i∂
)ψ − δmψ ψ̄ψ − iδg ψ̄γ 5 ψφ. (23)
2 2 4!
In addition to the Feynman rules of Figs. 2 and 3, the counter term Lagrangian adds the vertices of
Fig. 5.

= i p2 δZφ − δmφ
Zψ − δmψ
= i pδ
p p p p

= δg γ 5 = −iδλ

Figure 5: Counter term vertices.

Part (b)
Now we compute the infinite part of the counter terms. We will use an on-shell scheme. We rescale
the of the fields by the field strength in such away that we set the reside of the Fourier transformed

6
 Solutions to Peskin and Schroder – Andrzej Pokraka

propagator to 1. Furthermore, to give meaning to the mass and coupling counter terms we must
define the physical mass and physical coupling constants. The physical mass is defined as the location
of the pole of the Fourier transformed propagator and the physical coupling constants are defined
to the magnitude of the scattering amplitude at zero momentum

i
= + terms regular at p2 = m2φ , (24)
p2 − mφ

i
= + terms regular at p

= mψ , (25)
p −

mψ
 
 
 
 
= gγ 5 , (26)
 
 
 
 
 
amputated
 
 
 
 
= −iλ at s = 4m2φ , t = u = 0. (27)
 
 
 
 
 
amputated

The renormalization conditions (24) and (25) can be stated more clearly by defining the self
energies

1PI = −iM 2 (p2 ), (28)

1PI = −iΣ(p).

(29)

Then the full two-point function is given by the geometric series

= 1PI + 1PI 1PI + ... (30)

i
= , (31)
p2 − m2φ − M 2 (p2 )

= 1PI + 1PI 1PI + ... (32)

i
= . (33)
p

− m ψ − Σ(p)

7
 Solutions to Peskin and Schroder – Andrzej Pokraka

The renormalization conditions (24) and (25) can then be stated as

M 2 (p2 )|p2 =m2φ = 0, (34)
Σ(p)|


p=mψ = 0, (35)
2
2 

dM (p ) 
= 0, (36)
dp2 p2 =m2
φ

dΣ(p) 

 = 0. (37)
dp 


p=mψ

Computation of δZφ and δmφ

At one-loop the scalar self-energy is

−iM 2 (p2 ) = + + . (38)

p p p p p p

We evaluate the first two diagrams above, keeping only the divergent terms, and then determine the
divergent parts of the counter terms. The first diagram is

dd k
Z
i
= (−iλ) 2
p p (2π) d k − m2φ

dd k
Z
1
= λ
(2π)d k 2 − m2φ
!1− d2
−iλ 1
= d 2 Γ(1 − d/2)
(4π) 2 mφ
−iλm2φ
∼ . (39)
16π 2
where ∼ denotes the equality of the divergent terms. The second diagram is

"
#
dd k 5 i (k

+ p) + mψ
Z
5 i(k

+ mψ )
= Tr (gγ ) 2 (gγ )

p p (2π)d k − m2ψ (k + p)2 − m2ψ

dd k k 2 + k · p − m2ψ
Z Z
= dg 2 dx
(2π)d (k 2 + 2xk · p + xp2 − m2ψ )2
dd ` `2 − x(1 − x)p2 − m2ψ
Z Z
→ dg 2 dx (40)
(2π)d (`2 − ∆)2

8
 Solutions to Peskin and Schroder – Andrzej Pokraka

where ∆ = −x(1 − x)p2 + m2ψ . Performing the momentum integration yields

" #
idg 2 −2∆ + x(1 − x)p2 + m2ψ −γE
Z  
4πe
− ∆ + −2∆ + x(1 − x)p2 + m2ψ log


− 2 dx + O()
4π  ∆
"
ig 2 −2∆ + x(1 − x)p2 + m2ψ
Z
=− dx + 3∆ − 2x(1 − x)p2 − 2m2ψ
16π 2 
4πe−γE
  
2 2


+ −2∆ + x(1 − x)p + mψ log + O()
∆
ig 2
Z
dx 3x(1 − x)p2 − m2ψ


∼− 2
4π 
ig 2
 2 
p 2
=− 2 − mψ (41)
4π  2
Therfore, the self energy is
iλm2φ ig 2 p2
 
2 2
− m2ψ + i p2 δZφ − δmφ .


−iM (p ) ∼ − − (42)
16π 2 4π 2  2
Next we evaluate the derivative of the self-energy
dM 2 (p2 ) ig 2
−i ∼ − + iδZφ . (43)
dp2 8π 2 
Then, the renormalization conditions imply
g2
δZφ ∼ ,
8π 2 
1
λm2φ + 4g 2 m2φ − m2ψ .



δmφ ∼− (44)
16π 2

Computation of δZψ and δmψ

At one-loop the fermion self-energy is

−iΣ(p) = + . (45)

p p p p

We start by evaluating the first diagram

dd k igγ 5 i(k
+ p 5
Z
=
+ mψ )gγ
p p (2π)d ((k + p)2 − m2ψ )(k 2 − m2φ )

dd k
− mψ
Z Z
k
+ p
= g2 dx
(2π) ((k + xp) + x(1 − x)p2 − xm2ψ − (1 − x)m2φ )2
d 2

dd `
Z Z
1
→ g 2 dx [(1 − x)p − mψ ] (46)

(2π) (` − ∆0 )2
d 2

9
 Solutions to Peskin and Schroder – Andrzej Pokraka

where ∆0 = −x(1 − x)p2 + xm2ψ + (1 − x)m2φ . Performing the momentum integral we obtain

ig 2
  Z
4π
Γ() dx [(1 − x)p
− mψ ]
16π 2 ∆0
ig 2 4πe−γE
    
p

− mψ 1
= + log + O()
16π 2 2  ∆0
ig 2
 
p
∼ 2

− mψ . (47)
16π 2

Therefore, up to finite terms, the fermion self energy is

ig 2
 
p

−iΣ(p)

∼ 16π 2 2 − mψ + i pδ



Zψ − δmψ . (48)

Its derivative is then

dΣ(p) ig 2
−i
∼ + iδZψ . (49)
dp

32π 2
The renormalization conditions then imply

g2
δZψ ∼ − ,
32π 2
g 2 mψ
δmψ ∼ . (50)
16π 2

Computation of δδg
Since the tree level amplitude already satisfies the renormalization condition, we require the one-loop
contribution to the 3-point ψψφ amplitude to vanish
 
 
= + = 0. (51)
 

 p 0
p 
one-loop, amputated

We start by evaluating the following loop diagram

dd k γ 5 (k
+ p 5

+ mψ )γ (k
− p
0 5
Z
= −ig 3
+ mψ )γ
(2π) ((k + p) − mψ )((k − p ) − mψ )(k − m2φ )
d 2 2 0 2 2 2

0 0 0 2
dd k (−k
k
+ k
p

−p
k
+ p

p
− 2mψ k
− mψ (p
−p
) − mψ ) γ 5
Z Z Z
= −ig 3 2! dx dy
(2π)d ((k + xp − yp0 )2 − ∆(p, p0 ; x, y))3
(52)

where ∆(p, p0 ; x, y) = −2xyp · p0 − x(1 − x)p2 − y(1 − y)p02 + (x + y)m2ψ + (1 − x − y)m2φ . Shifting
the loop momentum to ` = k + xp − yp0 we obtain

dd k −`2 + f (p, p0 ; x, y) 5
Z Z Z
−ig 3 2! dx dy γ (53)
(2π)d (`2 − ∆(p, p0 ; x, y))3

10
 Solutions to Peskin and Schroder – Andrzej Pokraka

where f (p, p0 ; x, y) = x(1+x)p2 +y(1+y)p02 +2xyp·p0 +(1−x−y)p
p

0

+mψ (2x−1)p−m


0 2

−mψ .
ψ (2y−1)p
Since the renormalization conditions are specified for on-shell fermions, f is to be sadwitched between
the fermion spinors,

ū(p0 )f (p, p0 ; x, y)u(p) = ū(p0 ) x(4 + x) + y 2 − 2 m2ψ + 2(1 − x − y + xy)p · p0 u(p).

(54)

Furthermore, we are only interested in the limit where p = p0 . Therefore, we take

ū(p0 )f (p, p0 ; x, y)u(p) → ū(p)f (p, p; x, y)u(p)

= ū(p) (x(2 + x) − y(2 − y)) m2ψ u(p) (55)

and

∆(p, p0 ; x, y) → ∆(p, p; x, y)
= (x − y)2 m2ψ + (1 − x − y)m2φ . (56)

With these considerations we obtain

dd k −`2 + f (p, p; x, y) 5
Z Z Z
−2ig 3 dx dy γ
(2π)d (`2 − ∆(p, p; x, y))3
Z Z "  2−d/2  3−d/2 #
3 i d 1 i 1
= −2ig dx dy − Γ(2 − d/2) − Γ(3 − d/2)f γ 5
(4π)d/2 2 ∆ (4π)d/2 ∆
 2−d/2  
Γ(2 − d/2) 1
Z Z
d f
= −2g 3 dx dy − (2 − d/2) γ5
(4π)d/2 ∆ 2 ∆
2g 3 4πe−γE
Z Z    
2 f
=− dx dy + 2 log − + O() γ5
16π 2  ∆ ∆
g3
Z Z
∼ − 2 γ 5 dx dy
4π 
g3
= − 2 γ5. (57)
4π 
Thus, the divergent part of the one-loop contributions to the total amplitude, evaluated for
on-shell fermions at zero momentum, becomes
 
  g3 5
∼ − γ + δg γ 5 . (58)
 
p0  4π 2 
 
 p
one-loop, amputated

The renormalization conditions then imply

g3
δg ∼ . (59)
4π 2 

Computation of δδλ
Since the tree level amplitude already satisfies the renormalization condition, we require the one-loop
contribution to the 4-point scalar amplitude to vanish

11
 Solutions to Peskin and Schroder – Andrzej Pokraka

p2 p3
 
 
= + +
 
 
 

p1 p4 one-loop, amputated
= 0. (60)

We start by evaluating the following box diagram

dd k Tr γ 5 (k
+ mψ )γ 5 (k
+ p
 5 5


1 + p

2 + p

3 + mψ )γ (k
+ p

1 + p

2 + mψ )γ (k
+ p

1 + mψ )
Z
4 4
= i g
(2π)d (k 2 − m2ψ )((k + p1 + p2 + p3 )2 − m2ψ )((k + p1 + p2 )2 − m2ψ )((k + p1 )2 − m2ψ )

dd k Tr (−k
+ mψ )(k
+ p
 

1 + p

2 + p
3 + mψ )(−k
− p
1 − p
2 + mψ )(k
+ p

1 + mψ )
Z
4
= g (61)
(2π)d (k 2 − m2ψ )((k + p1 + p2 + p3 )2 − m2ψ )((k + p1 + p2 )2 − m2ψ )((k + p1 )2 − m2ψ )

Following the advice given in the question we specify to a specific frame to simplify the momentum.
Let us choose p1 + p2 = 0 = p3 + p4 . Then this diagram becomes

dd k Tr (−k
+ mψ )(k
+ p
 
3 + mψ )(−k
+ mψ )(k
+ p 1 + mψ )
Z
4
g


(2π)d (k 2 − m2ψ )2 ((k + p3 )2 − m2ψ )((k + p1 )2 − m2ψ )
dd k
Z Z Z
f (k, p1 , p3 )
g 4 3! dx (1 − x) dy
(2π) (k + 2(yp3 + (x − y)p1 ) · k + yp23 + (x − y)p21 − m2ψ )4
d 2

dd k
Z Z Z
f (k, p1 , p3 )
6g 4 dx (1 − x) dy d 4 (62)
(2π)

2
(k + yp3 + (x − y)p1 ) − ∆(p1 , p3 ; x, y)

where

f (k, p1 , p3 ) = 4k 4 + 4k 2 k · (p1 + p3 ) − p1 · p3 − 2m2ψ

+ 8k · p1 k · p3 − 4m2ψ (k · (p1 + p3 ) − p1 · p3 ) + m4ψ , (63)

and

∆(p1 , p3 ; x, y) = −y(1 − y)p23 − (x − y) (1 − (x − y)) p21 + 2y(x − y)p3 · p1 + m2ψ . (64)

Shifting the loop-momentum to ` = k + yp3 + (x − y)p1 , imposing the on-shell condition, and setting
p3 = p1 as required of the renormalization condition, we obtain

dd `
Z Z Z
f (`, p1 , p1 ; x, y)
6g 4 dx (1 − x) dy (65)
(2π)d (`2 − ∆(p1 , p3 ; x, y))4

where
8 − 16x + 16x2
 
f (`, p1 , p1 ; x, y) = 4`4 + m2ψ `2 − 12 − 8x + 8x2 + m4ψ 5 + 8x − 4x2 − 8x3 + 4x4


d
(66)

12
 Solutions to Peskin and Schroder – Andrzej Pokraka

Only the first term of f contributes to the divergent part of the amplitude,

dd ` `4
Z Z Z
4
∼ 24g dx (1 − x) dy
(2π)d (`2 − ∆(p1 , p3 ; x, y))4
dd ` `4
Z Z
= 24g 4 dx (1 − x)
(2π) (`2 − ∆(p1 , p3 ; x, y))4
d

Z  2−d/2
i 1
= 24g 4 dx (1 − x) Γ(2 − d/2)
(4π)d/2 ∆
3ig 4
∼ (67)
4π 2 
Next we evaluate the scalar bubble diagram

dd k
Z
(−iλ)i(−iλ)i
=
(2π)d (k 2 − m2φ )((k + p1 + p2 )2 − m2φ )

dd k
Z Z
1
= λ2 dx
(2π)d (k 2 + 2xk · (p1 + p2 ) + xp21 + xp22 − m2φ )2
dd k
Z Z
1
= λ2 dx
(2π) (k + x(p1 + p2 )) + x(1 − x)(p21 + p22 ) − 2x2 p1 · p2 − m2φ )2
d 2 2

dd `
Z Z
1
= λ2 dx
(2π) (` + x(1 − x)(p1 + p22 ) − 2x2 p1 · p2 − m2φ )2
d 2 2

Z !2−d/2
2 i 1
= λ dx Γ(2 − d/2)
(4π)d/2 −x(1 − x)(p21 + p22 ) + 2x2 p1 · p2 + m2φ
iλ2
∼ . (68)
16π 2
Finally, the divergent part of the full one-loop contribution to the amplitude is
p2 p3
 
  3ig 4 iλ2
∼ + − iδλ . (69)
 
4π  16π 2
2
 
 

p1 p4 one-loop, amputated
The renormalization conditions then imply

3g 4 λ2
δλ ∼ + . (70)
4π  16π 2
2

13