Homework 6

Armando De La Cruz Rangel Pantoja

December 16, 2024

6 Solutions
6.1 QCD color factors
In homework 1 we showed that any n × n unitray matrix U can be written as U = eiH where H is an n × n hermitian
matrix, and if det U = +1 then H is traceless; as a result of this, we saw that the SU (n) group has dimension n2 − 1,
i.e. it has n2 − 1 independent parameters. Hence any matrix U ∈ SU (n) can be written as

U = exp (iθa T a ), a = 1, 2, · · · , n2 − 1, (1)

where θa are the n2 − 1 parameters and T a are the n2 − 1 generators. The Lie algebra of a group SU (n) is

(2)
 a b
T , T = if abc T c ,

where f abc are the structure constants of the group. Let TRa be the generators of the group in the representation R.
First, let’s see that the operator TRa TRa commutes with all the generators

(3)
 a a b  a b a
TR TR , TR = TR , TR TR + TRa TRa , TRb = if abc TRc TRa + if abc TRa TRc = if abc {TRc , TRa } = 0,
 

where we have used the properties of the commutator and the fact that the structure constants of a simple and
compact Lie algebra (as the algebra of SU (n) is) are completely antisymmetric. Now, Schur’s lemma states that a
group element that commutes with all the other group elements in any irreducible representation must be proportional
to 1, hence
TRa TRa = C2 (R)1; (4)
and this operator is called the quadratic Casimir operator. The conventional inner product on a matrix space is
defined as hA, Bi = Tr[A† B]. In this case, the inner product of generators is hTRa , TRb i = Tr[TRa† TRb ] = Tr[TRa TRb ], and
the generators can always be chosen so that

hTRa , TRb i = Tr[TRa TRb ] = T (R)δ ab (5)

, where the number T (R) is known as the index of the representation. Setting b = a in this equation and summing
over a we get
Tr[TRa TRa ] =T (R)δ aa
Tr[C2 (R)1] =T (R)δ aa
(6)
C2 (R)Tr[1] =T (R)δ aa
C2 (R)d(R) =T (R)d(G),

1
where d(R) is the dimension of the representation and d(G) is the dimension of the group. From this expression it’s
clear that the quadratic Casimir C2 (R) is given by

T (R)d(G) T (R)(n2 − 1)
C2 (R) = = . (7)
d(R) d(R)

For the fundamental (F ) and the adjoint (A) representations, their dimension and the normalization of the generators
are
1 1
F : d(F ) =n, Tr[TFa TFb ] = δ ab ⇒ T (F ) = (8)
2 2
A: d(A) =n2 − 1, Tr[TAa TAb ] =nδ ab ⇒ T (A) = n, (9)

hence
(a)
T (F )(n2 − 1) n2 − 1
C2 (F ) = = (10)
d(F ) 2n

(b)
T (A)(n2 − 1) n(n2 − 1)
C2 (A) = = = n. (11)
d(A) n2 − 1

In particular, for SU (3)

4
C2 (F ) = , C2 (A) =3. (12)
3

6.2 Cross section for three jets events

(a) The three-body phase space is

d3 k1 d3 k2 d3 k3
Z Z
ρ = dΠ3 = (2π)4 δ (4) (q − k1 − k2 − k3 ), (13)
(2π)3 2E1 (2π)3 2E2 (2π)3 2E3

where k1 , k2 and k3 are the momentum of the quarks q, q̄ and the gluon g, respectively. We can rewrite the k3
integral by making use of the expression

d3 p
Z Z
f (p) = f (p)δ[p2 − m2 ]θ(Ep )d4 p, (14)
2Ep

where θ is the Heaviside step function. The integral is then

d3 k1 d3 k2 d3 k3
Z
ρ= (2π)4 δ (4) (q − k1 − k2 − k3 )
(2π)3 2E1 (2π)3 2E2 (2π)3 2E3
d3 k1 d3 k2 d4 k3
Z
= δ[k 2 − µ2 ]θ(E3 )(2π)4 δ (4) (q − k1 − k2 − k3 )
(2π)3 2E1 (2π)3 2E2 (2π)3 3
(15)
(2π)4 d3 k1 d3 k2
Z
δ (q − k1 − k2 )2 − µ2 θ(E3 )
 
= 3 3 3
(2π) (2π) 2E1 (2π) 2E2
(2π)4 d3 k1 d3 k2
Z
δ (q − k1 − k2 )2 − µ2 ,
 
= 3 3 3
(2π) (2π) 2E1 (2π) 2E2

2
where we have used that E3 = Eq − E1 − E2 > 0, hence θ(E3 ) = 1. Now, in the center of momentum frame
(CM): q µ = (Eq , q = 0), then we get

(q − k1 − k2 )2 =(Eq − E1 − E2 )2 − (−k1 − k2 )2
=Eq2 − 2Eq (E1 + E2 ) + E12 + 2E1 E2 + E22 − k12 − 2k1 · k2 − k22
(16)
=Eq2 − 2Eq (E1 + E2 ) + 2E1 E2 − 2k1 · k2 + 2m2
=Eq2 − 2Eq (E1 + E2 ) + 2E1 E2 − 2|k1 ||k2 | cos θ12 + 2m2

where m is the mass of the quarks q and q̄. The delta inside the integral is then (assuming |k1 | 6= 0, |k2 | 6= 0)

δ (q − k1 − k2 )2 − µ2 =δ Eq2 − 2Eq (E1 + E2 ) + 2E1 E2 − 2|k1 ||k2 | cos θ12 + 2m2 − µ2

   

=δ −2|k1 ||k2 | cos θ12 + Eq2 − 2Eq (E1 + E2 ) + 2E1 E2 + 2m2 − µ2

 
" # (17)
1 Eq2 − 2Eq (E1 + E2 ) + 2E1 E2 + 2m2 − µ2
= δ − cos θ12 + .
2|k1 ||k2 | 2|k1 ||k2 |

The volume element d3 k2 in spherical coordinates is

d3 k2 = |k2 |2 d|k2 |dΩ = |k2 |2 d|k2 | sin θ12 dθ12 dφ, (18)

where we have taken θ12 as the polar angle. From the mass-shell relation E22 − k22 = m2 we have |k2 |d|k2 | =
E2 dE2 and hence

|k2 |2 d|k2 | =|k2 |E2 dE2 ⇒ d3 k2 =|k2 |2 d|k2 | sin θ12 dθ12 dφ = |k2 |E2 dE2 sin θ12 dθ12 dφ (19)

Substituting these expressions back into (15)

(2π)4 d3 k1 d3 k2
Z
δ (q − k1 − k2 )2 − µ2
 
ρ=
(2π)3 (2π)3 2E1 (2π)3 2E2
" #
(2π)4 d3 k1 |k2 |E2 dE2 sin θ12 dθ12 dφ Eq2 − 2Eq (E1 + E2 ) + 2E1 E2 + 2m2 − µ2
Z
1
= δ − cos θ12 +
(2π)3 (2π)3 2E1 (2π)3 2E2 2|k1 ||k2 | 2|k1 ||k2 |
" #
1 (2π)4 d3 k1 |k2 |E2 dE2 sin θ12 dθ12 dφ Eq2 − 2Eq (E1 + E2 ) + 2E1 E2 + 2m2 − µ2
Z
= δ − cos θ12 +
2 (2π)3 (2π)3 2E1 |k1 | (2π)3 2E2 |k2 | 2|k1 ||k2 |
" #
1 (2π)4 d3 k1 Eq2 − 2Eq (E1 + E2 ) + 2E1 E2 + 2m2 − µ2
Z
= dE2 dφd(− cos θ12 )δ − cos θ12 +
4 (2π)6 (2π)3 2E1 |k1 | 2|k1 ||k2 |
" #
1 (2π)5 d3 k1 Eq2 − 2Eq (E1 + E2 ) + 2E1 E2 + 2m2 − µ2
Z
= dE2 d(− cos θ12 )δ − cos θ12 +
4 (2π)6 (2π)3 2E1 |k1 | 2|k1 ||k2 |
5 Z 3
1 (2π) d k1
= dE2 ,
4 (2π)6 (2π)3 2E1 |k1 |
(20)
in the same way E12 − k12 = m2 , then |k1 |d|k1 | = E1 dE1 and hence

|k1 |2 d|k1 | =|k1 |E1 dE1 ⇒ d3 k1 =|k1 |2 d|k1 | sin θdθdφ̃ = |k1 |E1 dE1 sin θdθdφ̃, (21)

3
and ρ is
1 (2π)5 d3 k1 1 (2π)5 |k1 |E1 dE1 sin θdθdφ̃
Z Z
ρ= dE 2 = dE2
4 (2π)6 (2π)3 2E1 |k1 | 4 (2π)6 (2π)3 2E1 |k1 |
1 (2π)5 1 (2π)6
Z Z
= dE 1 sin θdθd φ̃dE 2 = dE1 sin θdθdE2 (22)
8 (2π)9 8 (2π)9
Z Z
1 1 1
= dE 1 dE 2 = dE1 dE2 .
4 (2π)3 32π 3
The Lorentz invariant variables x1 , x2 , x3 are defined as
2k1 · q 2k2 · q 2k3 · q
x1 = , x2 = , x3 = , (23)
q2 q2 q2
notice that they satisfy
(k1 + k2 + k3 ) · q q2
x1 + x2 + x3 = 2 = 2 = 2. (24)
q2 q2
In the CM frame, they become
2k1 · q 2E1 Eq 2E1 2k2 · q 2E2 Eq 2E2 2k3 · q 2E3 Eq 2E3
x1 = 2
= 2
= , x2 = 2
= 2
= , x3 = 2
= 2
= . (25)
q Eq Eq q Eq Eq q Eq Eq

In differential form these equations are

2dE1 2dE2 2dE3
dx1 = , dx2 = , dx3 = , (26)
Eq Eq Eq

which in turn imply

1 2
dE1 dE2 = E dx1 dx2 . (27)
4 q
Substituting this result into (22) we get

Eq2 [q 2 ]CM
Z Z Z Z
1
ρ= dΠ3 = dE1 dE2 = dx1 dx2 = dx1 dx2 . (28)
32π 3 128π 3 128π 3
Remember that this result was calculated in the CM frame. From this expression and the fact that x1 , x2 are
Lorentz-invariant variables, we can infer that the general expression for ρ in an arbitrary frame is given by

q2
Z Z
ρ = dΠ3 = dx1 dx2 . (29)
128π 3
Let’s now find the region of integration. When both quarks are massless we have

(30)
p
E1 =|k1 |, E2 =|k2 |, E3 = |k3 |2 + µ2 .

In the CM frame we have k3 = −k1 − k2 , then |k3 | = |k1 + k2 | and so

p p p
E3 = |k3 |2 + µ2 = |k1 + k2 |2 + µ2 = |k1 |2 + |k2 |2 + 2|k1 ||k2 | cos θ12 + µ2
q (31)
= E12 + E22 + 2E1 E2 cos θ12 + µ2 .

4
Hence, the minimum and maximum values E3 can take are
q
(32)
p
θ12 =π : (E3 )min = E12 + E22 − 2E1 E2 + µ2 = (E1 − E2 )2 + µ2 ,
q
(33)
p
θ12 =0 : (E3 )max = E12 + E22 + 2E1 E2 + µ2 = (E1 + E2 )2 + µ2 .
Conservation of energy states
E1 + E2 + E3 =Eq . (34)
The values of E3 go from (E3 )min to (E3 )max , which imply that the maximum and minimum values of E1 + E2
are
(E1 + E2 )min + (E3 )max =Eq ⇒ (E1 + E2 )min − Eq = − (E3 )max (35)
(E1 + E2 )max + (E3 )min =Eq ⇒ (E1 + E2 )max − Eq = − (E3 )min (36)
Equation (35) implies that the region of integration is bounded from below by the expression
(E1 + E2 − Eq )2 =(E1 + E2 )2 + µ2
(E1 + E2 )2 − 2(E1 + E2 )Eq + Eq2 =(E1 + E2 )2 + µ2
−2(E1 + E2 )Eq + Eq2 =µ2
2(E1 + E2 )Eq =Eq2 − µ2 (37)
2E1 + 2E2 µ2
=1 − 2
Eq Eq
µ2
 
[x1 + x2 ]CM = 1 − 2 .
q CM
While equation (36) implies that the region of integration is bounded from above by the expression
(E1 + E2 − Eq )2 =(E1 − E2 )2 + µ2
(E1 + E2 )2 − 2(E1 + E2 )Eq + Eq2 =(E1 − E2 )2 + µ2
4E1 E2 − 2(E1 + E2 )Eq + Eq2 =µ2
4E1 E2 2E1 + 2E2 µ2 (38)
− = 2 −1
Eq2 Eq Eq
 2 
µ
[x1 x2 − (x1 + x2 )]CM = 2 − 1 .
q CM

The variables x1 , x2 , x3 and µ2 , q 2 are Lorentz invariant, then these equations hold in any frame and so the
region of integration is that bounded by the equations
µ2 µ2
below: x1 + x2 =1 − , above: x1 x2 − (x1 + x2 ) = − 1. (39)
q2 q2
µ2 µ2 1
x2 =1 − 2 − x1 , x2 =1 − 2 , (40)
q q 1 − x1
and these equations intersect at x1 = 0 and x1 = 1 − µ2 /q 2 . Hence, the expression (29) for ρ with limits of
integration is
Z 1−µ2 /q2 Z 1− µ22 1−x
1
q2
Z
(41)
q 1
ρ = dΠ3 = 3
dx 1 dx2 .
128π 0 µ 2
1− 2 −x1
q

5
 (b) First we are going to calculate the amplitude M for the process e− (p1 )+e+ (p2 ) → q(k1 )+ q̄(k2 )+g(k3 ) working
under the assumption that the gluon is described by a massless abelian vector field Bµ and its interaction with
the quarks is given by (see1 for more details on this approach)
Z
∆H = d3 xg ψ̄f i γ µ ψf i Bµ . (42)

where f is the flavor index of the quarks (u, d, s, c, etc.) and i is the color index. After that we’re going to
calculate the amplitude M for the same process but working now with QCD. At the end we will compare the
results from both approaches and see what are the differences between them.
The diagramas that contribute to the amplitude M for the process e− (p1 ) + e+ (p2 ) → q(k1 ) + q̄(k2 ) + g(k3 )
are

e− (p) p q(k1 ) e− (p) p k1 q(k1 )

k1 + k3 q
g(k3 )
γ
1) 2)
γ
q g(k3 )
+ 0 0 + 0 0
k2 + k3
e (p ) p k2 q̄(k2 ) e (p ) p q̄(k2 )

The amplitude M1 for the first diagram is

 
µ −igµν α iδij ν ∗
iM1 = [v̄(p2 , s2 )(−ieγ )u(p1 , s1 )] 2 ūi (k1 , r1 )(igγ ) (ieQq γ )vj (k2 , r2 )α (k3 , λ)
q + i k/1 + k/3
  (43)
gµν 1
2 µ
=ie gQq [v̄(p2 , s2 )γ u(p1 , s1 )] 2 ūi (k1 , r1 )γ α
γ vi (k2 , r2 ) ∗α (k3 , λ)
ν
q + i k/1 + k/3

The amplitude M2 for the second diagram is

 
−igµν −iδij ∗
iM2 = [v̄(p2 , s2 )(−ieγ µ )u(p1 , s1 )] ū (k
i 1 1 , r )(ieQ q γ ν
) (igγ α
)v (k , r
j 2 2 α 3) (k , λ)
q 2 + i k/2 + k/3
  (44)
gµν −1
=ie2 gQq [v̄(p2 , s2 )γ µ u(p1 , s1 )] 2 ūi (k1 , r1 )γ ν γ α vi (k2 , r2 ) ∗α (k3 , λ),
q + i k/2 + k/3

where the minus sign in the fermion propagator comes from the fact that the momentum flow direction is in
the opposite direction of the fermion charge flow. The total amplitude M is then
gµν
iM = iM1 + iM2 =ie2 gQq [v̄(p2 , s2 )γ µ u(p1 , s1 )] [ūi (k1 , r1 )Λαν vi (k2 , r2 )] ∗α (k3 , λ)
q2
+ i
1 (45)
M =e2 gQq [v̄(p2 , s2 )γµ u(p1 , s1 )] 2 [ūi (k1 , r1 )Λαµ vi (k2 , r2 )] ∗α (k3 , λ)
q + i
where
1 1 k/1 + k/3 µ k/2 + k/3 α
Λαµ = γ α γµ − γµ γα = γα 2
γ − γµ γ . (46)
k/1 + k/3 k/2 + k/3 (k1 + k3 ) (k2 + k3 )2
1 M.E. Peskin and D.V. Schroeder. An Introduction To Quantum Field Theory. Frontiers in Physics. Avalon Publishing, 1995. isbn:

9780813345437. url: [Link]

6
Thus, we arrive at the expression the exercise asks for, i.e. (45) with (46). Notice that
† † † †
k/1 + k/3 α† 0 k/2 + k/3 µ† 0
(Λαµ )† γ 0 =γ µ† 2
γ γ − γ α† γ γ
(k1 + k3 ) (k2 + k3 )2
† † † †
k/1 + k/3 0 α k/2 + k/3 0 µ
=γ µ† γ γ − γ α† γ γ
(k1 + k3 )2 (k2 + k3 )2 (47)
k/1 + k/3 α k/2 + k/3 µ
=γ µ† γ 0 γ − γ α† γ 0 γ
(k1 + k3 )2 (k2 + k3 )2
k/1 + k/3 α k/2 + k/3 µ
=γ 0 γ µ γ − γ0γα γ = γ 0 Λµα .
(k1 + k3 )2 (k2 + k3 )2

It will be useful to have the expression for M†

† 1 †
M† =e2 gQq [v̄(p2 , s2 )γµ u(p1 , s1 )] [ūl (k1 , r1 )Λαµ vl (k2 , r2 )] α (k3 , λ)
q 2 + i
† 1 h † i†
=e2 gQq v † (p2 , s2 )γ 0 γµ u(p1 , s1 ) 0 αµ

u (k 1 , r 1 )γ Λ v l (k 2 , r 2 ) α (k3 , λ)
q 2 + i l
 1 h † i
=e2 gQq u† (p1 , s1 )㵆 γ 0 v(p2 , s2 ) 2 vl (k2 , r2 )Λᵆ γ 0 ul (k1 , r1 ) α (k3 , λ) (48)

q + i
1 h † i
=e2 gQq u† (p1 , s1 )γ 0 γµ v(p2 , s2 ) 2 vl (k2 , r2 )γ 0 Λµα ul (k1 , r1 ) α (k3 , λ)
 
q + i
2 1
=e gQq [ū(p1 , s1 )γµ v(p2 , s2 )] 2 [v̄l (k2 , r2 )Λµα ul (k1 , r1 )] α (k3 , λ).
q + i
For non-polarized experiments, we have to average over the initial spin states and sum over the final state

7
polarizations, hence the squared amplitude |M̄|2 is
1X1X X 1 X X 1 X X
|M̄|2 = |M|2 = |M|2 = MM†
2 s 2 s 4 s ,s 4 s ,s
1 2 r1 ,r2 ,λ 1 2 r1 ,r2 ,λ 1 2 r1 ,r2 ,λ

1 X X 1
= e2 gQq [v̄(p2 , s2 )γµ u(p1 , s1 )] 2 [ūi (k1 , r1 )Λαµ vi (k2 , r2 )] ∗α (k3 , λ)
4 s ,s q + i
1 2 r1 ,r2 ,λ

2 1  νβ

×e gQq [ū(p1 , s1 )γν v(p2 , s2 )] 2 v̄l (k2 , r2 )Λ ul (k1 , r1 ) β (k3 , λ)
q + i
e4 g 2 Q2q 1 X X
= 2 2
{[v̄(p2 , s2 )γµ u(p1 , s1 )] [ū(p1 , s1 )γν v(p2 , s2 )] ∗α (k3 , λ)
4 (q + i) s ,s
1 2 r1 ,r2 ,λ

× v̄l (k2 , r2 )Λνβ ul (k1 , r1 ) [ūi (k1 , r1 )Λαµ vi (k2 , r2 )] β (k3 , λ)

 

e4 g 2 Q2q 1
(49)
X X
= {v̄(p2 , s2 )γµ [u(p1 , s1 )ū(p1 , s1 )] γν v(p2 , s2 )∗α (k3 , λ)
4 (q 2 + i)2 s ,s
1 2 r1 ,r2 ,λ

×v̄l (k2 , r2 )Λνβ [ul (k1 , r1 )ūi (k1 , r1 )] Λαµ vi (k2 , r2 )β (k3 , λ)
e4 g 2 Q2q 1 X X
= {Tr {[v(p2 , s2 )v̄(p2 , s2 )] γµ [u(p1 , s1 )ū(p1 , s1 )] γν } ∗α (k3 , λ)
4 (q 2 + i)2 s1 ,s2 r1 ,r2 ,λ

×Tr [vi (k2 , r2 )v̄l (k2 , r2 )] Λνβ [ul (k1 , r1 )ūi (k1 , r1 )] Λαµ β (k3 , λ)

(" # " # )
e4 g 2 Q2q 1
Tr
X X
= v(p2 , s2 )v̄(p2 , s2 ) γµ u(p1 , s1 )ū(p1 , s1 ) γν
4 (q 2 + i)2 s2 s1
(" # " # )
× Tr
X X X
vi (k2 , r2 )v̄l (k2 , r2 ) Λ νβ
ul (k1 , r1 )ūi (k1 , r1 ) Λ αµ
∗α (k3 , λ)β (k3 , λ),
r2 r1 λ

now we can use the following relations for the spinors u(p, s) and v(p, s)

(50)
X X
u(p, s)ū(p, s) =(p
/ + m), / − m)
v(p, s)v̄(p, s) =(p
s s

(51)
X X
ul (p, s)ūi (p, s) =(p
/ + ml )δli , / − mi )δil ,
vi (p, s)v̄l (p, s) =(p
s s

assuming massless electrons (me = 0) and massless quarks (mi = 0) to simplify the last expression for |M̄|2 as
(" # " # )
e4 g 2 Q2q 1
Tr
X X
2
|M̄| = v(p2 , s2 )v̄(p2 , s2 ) γµ u(p1 , s1 )ū(p1 , s1 ) γν
4 (q 2 + i)2 s2 s1
(" # " # )
× Tr
X X X
vi (k2 , r2 )v̄l (k2 , r2 ) Λ νβ
ul (k1 , r1 )ūi (k1 , r1 ) Λ αµ
∗α (k3 , λ)β (k3 , λ)
r2 r1 λ
(52)
e4 g 2 Q2q 1 n o 
Tr p /1 γν Tr k/2 Λ k/1 Λ
X
= 2 2 / 2 γµ p
νβ αµ
δli δil ∗α (k3 , λ)β (k3 , λ)
4 (q + i)
λ
4 2 2
e g Qq 1 n o
Tr p /1 γν Tr k/2 Λ k/1 Λ
X
νβ αµ
∗α (k3 , λ)β (k3 , λ),

=3 / 2 γµ p
4 (q 2 + i)2
λ

8
where we have used that δli δil = δll = 3 (this factor accounts for the three colors of quarks). In the Feynman
gauge, the sum over the polarizations of the vector is

(53)
X
∗α (k3 , λ)β (k3 , λ) = −gαβ ,
λ

with this we get

e4 g 2 Q2q 1 n o 
|M̄|2 = − 3 Tr p
/ γ µ p
/ γν Tr k/2 Λνβ k/1 Λαµ gαβ
4 (q 2 + i)2 2 1

e4 g 2 Q2q 1
(54)
n o
=−3 Tr p
/ γ p
/
2 µ 1 ν
γ Tr {/
k 2 Λνα k/1 Λα µ }
4 (q 2 + i)2
e4 g 2 Q2q 1
≡−3 Lµν (p1 , p2 )H µν (k1 , k2 , k3 ),
4 (q 2 + i)2
where
n o
Lµν (p1 , p2 ) =Tr p /1 γν = 4(p1µ p2ν + p1ν p2µ − p1 · p2 gµν ),
/ 2 γµ p k 2 Λνα k/1 Λα µ } .
H µν (k1 , k2 , k3 ) =Tr {/ (55)

In QCD the calculation is analogous. The diagrams are the same

e− (p) p q(k1 ) e− (p) p k1 q(k1 )

k1 + k3 q
g(k3 )
γ
1) 2)
γ
q g(k3 )
+ 0 0 + 0 0
k2 + k3
e (p ) p k2 q̄(k2 ) e (p ) p q̄(k2 )

The amplitude M1 for the first diagram is

 
µ −igµν α a iδkj ν ∗
iM1 = [v̄(p2 , s2 )(−ieγ )u(p1 , s1 )] 2 ūi (k1 , r1 )(igγ Tik ) (ieQq γ )vj (k2 , r2 )α (k3 , λ)
q + i k/1 + k/3
  (56)
gµν 1
2 µ
=ie gQq [v̄(p2 , s2 )γ u(p1 , s1 )] 2 ūi (k1 , r1 )γ α
γ vj (k2 , r2 ) Tija ∗α (k3 , λ)
ν
q + i k/1 + k/3

The amplitude M2 for the second diagram is

 
µ −igµν ν −iδik α a ∗
iM2 = [v̄(p2 , s2 )(−ieγ )u(p1 , s1 )] 2 ūi (k1 , r1 )(ieQq γ ) (igγ Tkj )vj (k2 , r2 )α (k3 , λ)
q + i k/2 + k/3
  (57)
gµν −1
2 µ
=ie gQq [v̄(p2 , s2 )γ u(p1 , s1 )] 2 ūi (k1 , r1 )γ ν
γ vj (k2 , r2 ) Tija ∗α (k3 , λ),
α
q + i k/2 + k/3

where the minus sign in the fermion propagator comes from the fact that the momentum flow direction is in
the opposite direction of the fermion charge flow. The total amplitude M is then
gµν
iM = iM1 + iM2 =ie2 gQq [v̄(p2 , s2 )γ µ u(p1 , s1 )] [ūi (k1 , r1 )Λαν vj (k2 , r2 )] Tija ∗α (k3 , λ)
q 2 + i
1 (58)
M =e2 gQq [v̄(p2 , s2 )γµ u(p1 , s1 )] 2 [ūi (k1 , r1 )Λαµ vj (k2 , r2 )] Tija ∗α (k3 , λ)
q + i

9
where Λαµ is the same as in (46). It will be useful to have the expression for M†
1 † † a∗
M† =e2 gQq [v̄(p2 , s2 )γµ u(p1 , s1 )] [ūl (k1 , r1 )Λαµ vk (k2 , r2 )] Tlk α (k3 , λ)
q 2 + i
† 1 h † i†
=e2 gQq v † (p2 , s2 )γ 0 γµ u(p1 , s1 ) 0 αµ a∗

u (k 1 , r 1 )γ Λ v k (k 2 , r 2 ) Tlk α (k3 , λ)
q 2 + i l
 1 h † i
=e2 gQq u† (p1 , s1 )㵆 γ 0 v(p2 , s2 ) 2 vk (k2 , r2 )Λᵆ γ 0 ul (k1 , r1 ) Tlk a∗
(59)

α (k3 , λ)
q + i
 1 h † i
=e2 gQq u† (p1 , s1 )γ 0 γµ v(p2 , s2 ) 2 vk (k2 , r2 )γ 0 Λµα ul (k1 , r1 ) Tlk a∗

α (k3 , λ)
q + i
1
=e2 gQq [ū(p1 , s1 )γµ v(p2 , s2 )] 2 [v̄k (k2 , r2 )Λµα ul (k1 , r1 )] Tlk a∗
α (k3 , λ).
q + i
For non-polarized experiments, we have to average over the initial spin states and sum over the final state
polarizations, hence the squared amplitude |M̄|2 is (in the following expressions there is no sum over the gluon
color index a)
1X1X X 1 X X 1 X X
|M̄|2 = |M|2 = |M|2 = MM†
2 s 2 s 4 s ,s 4 s ,s
1 2 r1 ,r2 ,λ 1 2 r1 ,r2 ,λ 1 2 r1 ,r2 ,λ

1 X X 1
= e2 gQq [v̄(p2 , s2 )γµ u(p1 , s1 )] 2 [ūi (k1 , r1 )Λαµ vj (k2 , r2 )] Tija ∗α (k3 , λ)
4 s ,s q + i
1 2 r1 ,r2 ,λ

1 
×e2 gQq [ū(p1 , s1 )γν v(p2 , s2 )] νβ
 a∗
v̄ k (k 2 , r 2 )Λ ul (k 1 , r 1 ) Tlk  β (k 3 , λ)
q 2 + i
e4 g 2 Q2q 1 X X 
= 2 2
[v̄(p2 , s2 )γµ u(p1 , s1 )] [ū(p1 , s1 )γν v(p2 , s2 )] Tija ∗α (k3 , λ)
4 (q + i) s ,s
1 2 r1 ,r2 ,λ

× v̄k (k2 , r2 )Λνβ ul (k1 , r1 ) [ūi (k1 , r1 )Λαµ vj (k2 , r2 )] Tlk a∗

 
β (k3 , λ)
4 2 2
e g Qq 1 X X 
= 2 2
v̄(p2 , s2 )γµ [u(p1 , s1 )ū(p1 , s1 )] γν v(p2 , s2 )Tija ∗α (k3 , λ)
4 (q + i) s ,s
1 2 r1 ,r2 ,λ

×v̄k (k2 , r2 )Λνβ [ul (k1 , r1 )ūi (k1 , r1 )] Λαµ vj (k2 , r2 )Tlk
a∗
β (k3 , λ)
e4 g 2 Q2q 1
Tr {[v(p2 , s2 )v̄(p2 , s2 )] γµ [u(p1 , s1 )ū(p1 , s1 )] γν } Tija ∗α (k3 , λ)
X X 
=
4 (q 2 + i)2 s1 ,s2 r1 ,r2 ,λ

×Tr [vj (k2 , r2 )v̄k (k2 , r2 )] Λνβ [ul (k1 , r1 )ūi (k1 , r1 )] Λαµ Tlk a∗

β (k3 , λ)
(" # " # )
e4 g 2 Q2q 1
Tr
X X
= v(p 2 , s 2 )v̄(p2 , s 2 ) γ µ u(p1 , s1 )ū(p1 , s1 ) γν
4 (q 2 + i)2 s2 s1
(" # " # )
× Tr
X X X
vj (k2 , r2 )v̄k (k2 , r2 ) Λ νβ
ul (k1 , r1 )ūi (k1 , r1 ) Λ αµ a∗ a
Tlk Tij ∗α (k3 , λ)β (k3 , λ),
r2 r1 λ
(60)

10
 now we can use the following relations for the spinors u(p, s) and v(p, s)

(61)
X X
u(p, s)ū(p, s) =(p
/ + m), / − m)
v(p, s)v̄(p, s) =(p
s s

(62)
X X
ul (p, s)ūi (p, s) =(p
/ + ml )δli , / − mi )δil ,
vi (p, s)v̄l (p, s) =(p
s s

assuming massless electrons (me = 0) and massless quarks (mi = 0) to simplify the last expression for |M̄|2 as
(" # " # )
e4 g 2 Q2q 1
Tr
X X
2
|M̄| = v(p2 , s2 )v̄(p2 , s2 ) γµ u(p1 , s1 )ū(p1 , s1 ) γν
4 (q 2 + i)2 s2 s1
(" # " # )
× Tr
X X X
vj (k2 , r2 )v̄k (k2 , r2 ) Λνβ ul (k1 , r1 )ūi (k1 , r1 ) Λαµ Tlk a∗ a
Tij ∗α (k3 , λ)β (k3 , λ)
r2 r1 λ
(63)
4 2
e g Q2q 1 n o 
Tr Tr k/2 Λνβ k/1 Λαµ δli δjk Tlk
X
= 2 2
p/ 2
γ µ p
/ 1
γ ν
a∗ a
Tij ∗α (k3 , λ)β (k3 , λ)
4 (q + i)
λ
e4 g 2 Q2q 1 n o 
Tr p /1 γν Tr k/2 Λ k/1 Λ
X
= / 2 γµ p
νβ αµ
Tija∗ Tija ∗α (k3 , λ)β (k3 , λ).
4 (q 2 + i)2
λ

The hermiticity of the generators (T b† )ij = Tijb implies Tji b

= Tijb∗ and, in the Feynman gauge, the sum over the
polarizations of one single external gluon can be replaced by (see2 )

(64)
X
∗α (k3 , λ)β (k3 , λ) = −gαβ ,
λ

with this we get

e4 g 2 Q2q 1 n o 
|M̄|2 = − 2 2
Tr p /2 γµ p/1 γν Tr k/2 Λ k/1 Λ
νβ αµ a a
Tji Tij gαβ
4 (q + i)
e4 g 2 Q2q 1
(65)
n o
=− Tr p
/ γµ p
/ γ ν Tr {/
k 2 Λνα k/1 Λα µ } Tr{T a T a }
4 (q 2 + i)2 2 1

e4 g 2 Q2q 1
=− Lµν (p1 , p2 )H µν (k1 , k2 , k3 )Tr{T a T a }
4 (q + i)2
2

If now we sum over the gluon color index a we get (see (12)) Tr{T a T a } = Tr{C2 (F )1} = C2 (F )Tr{1} = (4/3)·3
  4 2 2
4 e g Qq 1
|M̄|2 = − 3 Lµν (p1 , p2 )H µν (k1 , k2 , k3 ). (66)
3 4 (q 2 + i)2

Comparing (66) with (54) we see that the correct result for the amplitude in QCD can be obtained from the
expression (54) just by making the replacement

4 4 g2
g2 → g2 or αg → αg , where αg = . (67)
3 3 4π
2 M. Kachelrieß and M. N. Malmquist. “Using Covariant Polarisation Sums in QCD”. in: The European Physical Journal Plus 137.1

(Jan. 2022), p. 89. issn: 2190-5444. doi: 10.1140/epjp/s13360-021-02273-3. url: [Link]

3.

11
(c) The cross-section σ for a process a(p1 ) + b(p2 ) → c(k1 ) + d(k2 ) + f (k3 ) is

d3 k1 d3 k2 d3 k3 (2π)4 δ (p1 + p2 − i ki )
Z Z P
σ = dσ = |M|2 . (68)
(2π)3 2E1 (2π)3 2E2 (2π)3 2E3 4 (p1 · p2 )2 − m21 m22
p

The cross section for the process e− (p1 ) + e+ (p2 ) → q(k1 ) + q̄(k2 ) + g(k3 ) is then (rememeber that we are
working with me = 0, mi = 0)

d3 k1 d3 k2 d3 k3 (2π)4 δ (p1 + p2 − i ki )
Z Z P
σ = dσ = |M̄|2
(2π)3 2E1 (2π)3 2E2 (2π)3 2E3
p
4 (p1 · p2 )2
d3 k1 d3 k2 d3 k3
Z
1
= (2π)4 δ (q − k1 − k2 − k3 ) |M̄|2
4(p1 · p2 ) (2π)3 2E1 (2π)3 2E2 (2π)3 2E3
Z (69)
1 2
= dΠ3 |M̄|
4(p1 · p2 )
  4 2 2
e g Qq
Z
4 1 1
=− 3 Lµν (p1 , p2 ) dΠ3 H µν (k1 , k2 , k3 )
3 4 4(p1 · p2 ) (q 2 + i)2

where we have used (13) for the three-body phase space integral and (66) for the squared amplitude. Let’s
start by calculating q µ Lµν (p1 , p2 ), from (55) we have

q µ Lµν (p1 , p2 ) =4(pµ1 + pµ2 )(p1µ p2ν + p1ν p2µ − p1 · p2 gµν )

=4pµ1 (p1µ p2ν + p1ν p2µ − p1 · p2 gµν ) + 4pµ2 (p1µ p2ν + p1ν p2µ − p1 · p2 gµν )
(70)
=4(p21 p2ν + p1 · p2 p1ν − p1 · p2 p1ν ) + 4(p1 · p2 p2ν + p1ν p22 − p1 · p2 p2ν )
=4p21 p2ν + 4p22 p1ν = 4m2e (p1ν + p2ν ) = 0,

where the last equality comes from the fact that we are working under the assumption me = 0. Lµν (p1 , p2 ) is

12
symmetric, then q µ Lµν = 0 = q ν Lµν . Now let’s calculate H µν (k1 , k2 , k3 )

k 2 Λνα k/1 Λα µ }
H µν =Tr {/
    
ν k/1 + k/3 α α k/2 + k/3 ν k/1 + k/3 µ µ k/2 + k/3
=Tr k/2 γ γ −γ γ k/1 γα γ −γ γα
(k1 + k3 )2 (k2 + k3 )2 (k1 + k3 )2 (k2 + k3 )2
  
ν k/1 + k/3 α α k/2 + k/3 ν k/1 + k/3 µ µ k/2 + k/3
=Tr /
k2 γ γ − k2 γ / γ /
k 1 γα γ − k1 γ / γα
(k1 + k3 )2 (k2 + k3 )2 (k1 + k3 )2 (k2 + k3 )2
  
k/1 + k/3 α k/1 + k/3 µ k/2 + k/3
=Tr k/2 γ ν γ k/1 γα γ − k/1 γ µ γα
(k1 + k3 )2 (k1 + k3 )2 (k2 + k3 )2
 
α k/2 + k/3 ν k/1 + k/3 µ µ k/2 + k/3
−/k2 γ γ k 1 γα/ γ − k1 γ / γα
(k2 + k3 )2 (k1 + k3 )2 (k2 + k3 )2

k/1 + k/3 α k/1 + k/3 µ k/1 + k/3 α k/2 + k/3
=Tr k/2 γ ν γ k/1 γα γ − k/2 γ ν γ k/1 γ µ γα
(k1 + k3 )2 (k1 + k3 )2 (k1 + k3 )2 (k2 + k3 )2

α k/2 + k/3 ν k/1 + k/3 µ α k/2 + k/3 ν µ k/2 + k/3
−/k2 γ /
γ k 1 γα γ + k2 γ/ /
γ k1 γ γα
(k2 + k3 )2 (k1 + k3 )2 (k2 + k3 )2 (k2 + k3 )2

k/1 + k/3 k/1 + k/3 µ k/1 + k/3 k/2 + k/3 µ
=Tr −2/ k2 γ ν k/ γ + 2/ k2 γ ν γ k/1
(k1 + k3 )2 1 (k1 + k3 )2 (k1 + k3 )2 (k2 + k3 )2

k/2 + k/3 k/1 + k/3 µ α k/2 + k/3 µ k/2 + k/3
+2/k 2 k/1 γ ν γ + /
k 2 γ γ ν
/
k 1 γ γ α
(k2 + k3 )2 (k1 + k3 )2 (k2 + k3 )2 (k2 + k3 )2
   
k/1 + k/3 k/1 + k/3 µ ν k/1 + k/3 k/2 + k/3 µ
= − 2Tr k/2 γ ν /
k γ + 2Tr /
k γ γ /
k
(k1 + k3 )2 1 (k1 + k3 )2 2
(k1 + k3 )2 (k2 + k3 )2 1
(71)
   
k/2 + k/3 k/1 + k/3 µ α k/2 + k/3 µ k/2 + k/3
+ 2Tr k/2 k/1 γ ν γ + Tr γ /
k
α 2 γ γ ν
/
k 1 γ
(k2 + k3 )2 (k1 + k3 )2 (k2 + k3 )2 (k2 + k3 )2
   
k/1 + k/3 k/1 + k/3 µ ν k/1 + k/3 k/2 + k/3 µ
= − 2Tr k/2 γ ν /
k γ + 2Tr /
k γ γ /
k
(k1 + k3 )2 1 (k1 + k3 )2 2
(k1 + k3 )2 (k2 + k3 )2 1
   
k/2 + k/3 k/1 + k/3 µ k/2 + k/3 ν µ k/2 + k/3
+ 2Tr k/2 k/1 γ ν γ − 2Tr /
k 2 γ /
k 1 γ
(k2 + k3 )2 (k1 + k3 )2 (k2 + k3 )2 (k2 + k3 )2
   
k/1 + k/3 k/1 + k/3 µ ν k/1 + k/3 k/2 + k/3 µ
= − 2Tr k/2 γ ν /
k γ + 2Tr /
k /
k γ γ
(k1 + k3 )2 1 (k1 + k3 )2 1 2
(k1 + k3 )2 (k2 + k3 )2
   
k/2 + k/3 k/1 + k/3 µ k/2 + k/3 k/2 + k/3 ν
+ 2Tr k/2 k/1 γ ν γ − 2Tr /
k γ /
k γ µ
(k2 + k3 )2 (k1 + k3 )2 (k2 + k3 )2 2 (k2 + k3 )2 1
   
k/1 + k/3 k/1 k/3 ν k/1 + k/3 k/2 + k/3 µ
= − 2Tr k/2 γ ν γ µ
+ 2Tr k/ /
k
1 2 γ γ
(k1 + k3 )2 (k1 + k3 )2 (k1 + k3 )2 (k2 + k3 )2
   
k/2 + k/3 k/1 + k/3 µ k/2 + k/3 k/2 k/3
+ 2Tr k/2 k/1 γ ν γ − 2Tr γ ν
/
k 1 γ µ
(k2 + k3 )2 (k1 + k3 )2 (k2 + k3 )2 (k2 + k3 )2
Tr {/ k 2 γ ν k/3 k/1 k/3 γ µ } Tr {/ k 1 k/2 γ ν (k/1 + k/3 )(k/2 + k/3 )γ µ }
=−2 2 2
+2
(k1 + k3 ) (k1 + k3 ) (k1 + k3 )2 (k2 + k3 )2
Tr {/ k 2 k/1 γ ν (k/2 + k/3 )(k/1 + k/3 )γ µ } Tr {/ k 1 γ µ k/3 k/2 k/3 γ ν }
+2 2 2
−2 ,
(k2 + k3 ) (k1 + k3 ) (k2 + k3 )2 (k2 + k3 )2

where we have used the assumption of massless quarks k12 = 0, k22 = 0. To evaluate the traces we can use the

13
result

Tr[γ α γ ν γ β γ ρ γ σ γ µ ] = + 4(g αβ g µρ g νσ − g αβ g µσ g νρ − g αβ g µν g ρσ − g αµ g βσ g νρ + g αµ g βρ g νσ + g αµ g βν g ρσ
+ g ασ g βµ g νρ + g αν g βµ g ρσ − g αν g βσ g µρ + g αν g βρ g µσ + g αρ g βν g µσ + g αρ g βσ g µν (72)
αρ βµ νσ ασ βρ µν ασ βν µρ
−g g g −g g g −g g g );

that was obtained in Mathematica with the help of the FeynCalc package, the Mathematica notebook is
attached to this homework. The expression we get for H µν (k1 , k2 , k3 ) is

H µν (k1 , k2 , k3 )
Tr {/k 2 γ ν k/3 k/1 k/3 γ µ } Tr {/k 1 k/2 γ ν (k/1 + k/3 )(k/2 + k/3 )γ µ }
=−2 2 2
+2
(k1 + k3 ) (k1 + k3 ) (k1 + k3 )2 (k2 + k3 )2
Tr {/k 2 k/1 γ ν (k/2 + k/3 )(k/1 + k/3 )γ µ } Tr {/
k 1 γ µ k/3 k/2 k/3 γ ν }
+2 2 2
−2
(k2 + k3 ) (k1 + k3 ) (k2 + k3 )2 (k2 + k3 )2
[µ2 (k1 · k2 ) − 2(k1 · k3 )(k2 · k3 )]g µν − µ2 [k1µ k2ν + k1ν k2µ ] + 2(k1 · k3 )[k2ν k3µ + k2µ k3ν ]
=−8
(k1 + k3 )2 (k1 + k3 )2
[µ2 + 2(k1 · k3 ) + 2(k2 · k3 ) + 2(k1 · k2 )](k1 · k2 ) µν (73)
+ 16 g
(k1 + k3 )2 (k2 + k3 )2
(k1 · k2 )
− 16 [2k µ k ν + 2k1ν k2µ + k1µ k3ν + k1ν k3µ + k2ν k3µ + k2µ k3ν ]
(k1 + k3 )2 (k2 + k3 )2 1 2
(k1 · k3 ) (k2 · k3 )
− 16 [k1µ k2ν + k1ν k2µ − 2k2µ k2ν ] − 16 [k µ k ν + k1ν k2µ − 2k1µ k1ν ]
2
(k1 + k3 ) (k2 + k3 ) 2 (k1 + k3 )2 (k2 + k3 )2 1 2
[µ2 (k1 · k2 ) − 2(k1 · k3 )(k2 · k3 )]g µν − µ2 [k1µ k2ν + k1ν k2µ ] + 2(k2 · k3 )[k1µ k3ν + k1ν k3µ ]
−8 ,
(k2 + k3 )2 (k2 + k3 )2

where we have used k12 = 0, k22 = 0, and added a mass µ to the gluon k32 = µ2 to regularize infrared divergences.
From this expression it’s evident that H µν is symmetric H νµ = H µν . Additionally, in the attached Mathematica
notebook it’s demonstrate that

qµ H µν (k1 , k2 , k3 ) = (k1µ + k2µ + k3µ )H µν (k1 , k2 , k3 ) = 0 (74)

(it’s hard to do this calculation by hand, as you can see from (73), so I decided to do it with Mathematica;
as I said before, the Mathematica notebook is attached to this homework). H µν (k1 , k2 , k3 ) is symmetric, then
qµ H µν = 0 = qν H µν . The result of the phase space integral in the cross section (69)
Z
dΠ3 H µν (k1 , k2 , k3 ) (75)

can depend only on q µ because the momenta k1 , k2 , k3 are integrated. Then, as a consequence of (74), this
integral must have the form
qµ qν
Z  
dΠ3 H µν (k1 , k2 , k3 ) = g µν − 2 H, (76)
q
where H is a scalar that can be determined from contracting this expression with gµν

qµ qν
Z   Z
1
dΠ3 gµν H µν (k1 , k2 , k3 ) = gµν g µν − gµν 2 H = 3H ⇒ H= dΠ3 gµν H µν (k1 , k2 , k3 ), (77)
q 3

14
substituting back this result into (76) we have

qµ qν qµ qν
Z    Z
1
µν µν
dΠ3 H (k1 , k2 , k3 ) = g − 2 H= µν
g − 2 dΠ3 gαβ H αβ (k1 , k2 , k3 ). (78)
q 3 q

Substituting this result into the cross section (69) we get

  4 2 2
e g Qq
Z
4 1 1
σ =− 3 Lµν (p1 , p2 ) dΠ3 H µν (k1 , k2 , k3 )
3 4 4(p1 · p2 ) (q 2 + i)2
  4 2 2
e g Qq qµ qν
 Z
4 1 1 1
=− 3 L (p
µν 1 2 , p ) g µν
− dΠ3 gαβ H αβ (k1 , k2 , k3 ) (79)
3 4 4(p1 · p2 ) (q 2 + i)2 3 q2
  4 2 2
4 e g Qq
Z
1 1 µν
=− g L (p
µν 1 2 , p ) dΠ3 gαβ H αβ (k1 , k2 , k3 ),
3 4 4(p1 · p2 ) (q 2 + i)2

where we have used that q µ Lµν = 0. Making use of the expression (29) for the phase space integral the last
expression becomes
  4 2 2
4 e g Qq
Z
1 1
σ =− g Lµν (p1 , p2 ) dΠ3 gαβ H αβ (k1 , k2 , k3 )
µν
3 4 4(p1 · p2 ) (q 2 + i)2
  4 2 2
4 e g Qq q2
Z
1 1
=− g µν
Lµν (p 1 , p 2 ) dx1 dx2 gαβ H αβ (k1 , k2 , k3 ) (80)
3 4 4(p1 · p2 ) (q 2 + i)2 128π 3
  4 2 2
4 e g Qq
Z
1 1 1 µν
=− g L (p
µν 1 2 , p ) dx1 dx2 gαβ H αβ (k1 , k2 , k3 ).
128π 3 3 4 4(p1 · p2 ) (q 2 + i)

This means that the differential cross section is

  4 2 2
dσ 1 4 e g Qq 1
=− 3
g µν Lµν (p1 , p2 )gαβ H αβ (k1 , k2 , k3 )
dx1 dx2 128π 3 4 4(p1 · p2 )q 2
  4 2 2 (81)
1 4 e g Qq 1 µν
=− 3
g Lµν (p1 , p2 )gαβ H αβ (k1 , k2 , k3 ),
128π 3 4 2q 4

in the last step we have used the fact that, in the massless limit, q 2 = (p1 + p2 )2 = p21 + 2p1 · p2 + p22 = 2p1 · p2 .
On one hand, the factor g µν Lµν is

g µν Lµν (p1 , p2 ) =4g µν (p1µ p2ν + p1ν p2µ − p1 · p2 gµν )

(82)
=4(p1 · p2 + p1 · p2 − 4p1 · p2 ) = −8p1 · p2 = −4q 2

(in the last step remember that we are taking p21 = p22 = me = 0) and, on the other hand, the factor gαβ H αβ
is (see the Mathemtica notebook for the calculation)

32(k1 · k3 )(k2 · k3 ) 32(k1 · k3 )(k2 · k3 ) 64(k1 · k2 )2

gαβ H αβ (k1 , k2 , k3 ) = + + +
(2k1 · k3 + µ2 )2 (2k2 · k3 + µ2 )2 (2k1 · k3 + µ2 )(2k2 · k3 + µ2 )
64(k1 · k2 )(k1 · k3 ) 64(k1 · k2 )(k2 · k3 )
+ + (83)
(2k1 · k3 + µ2 )(2k2 · k3 + µ2 ) (2k1 · k3 + µ2 )(2k2 · k3 + µ2 )
16µ2 (k1 · k2 ) 16µ2 (k1 · k2 ) 64µ2 (k1 · k2 )
− − + .
(2k1 · k3 + µ2 )2 (2k2 · k3 + µ2 )2 (2k1 · k3 + µ2 )(2k2 · k3 + µ2 )

15
We can express this result in terms of x1 , x2 and q 2 by inverting their defining equations

2k1 · q 2k1 · (k1 + k2 + k3 ) k1 · k2 + k1 · k3

x1 = = =2 (84)
q2 q2 q2
2k2 · q 2k2 · (k1 + k2 + k3 ) k2 · k1 + k2 · k3
x2 = 2 = 2
=2 (85)
q q q2
2k3 · q 2k3 · (k1 + k2 + k3 ) k3 · k1 + k3 · k2 + µ2
x3 = 2 = = 2 , (86)
q q2 q2
then
1
x1 q 2 =k1 · k2 + k1 · k3 (87)
2
1
x2 q 2 =k2 · k1 + k2 · k3 (88)
2
1
x3 q 2 − µ2 =k3 · k1 + k3 · k2 , (89)
2
in matrix form   1 2
  
1 1 0 k1 · k2 2 x1 q
1 0 1 k1 · k3  =  12 x2 q 2 
1 2 2
0 1 1 k2 · k3 2 x3 q − µ
 1
x1 q 2
   
k1 · k2 1 1 −1
k1 · k3  = 1  1 −1 1   1 x2 q 2 
2
2 (90)
2 1 2 2
k2 · k3 −1 1 1 2 x3 q − µ
 1 2 2


4 2µ + q (x1 + x2 − x3 )

1
=  4 q (x1 − x2 + x3 ) − 2µ2

2
1 2 2
4 q (−x1 + x2 + x3 ) − 2µ

(again, this calculations were done in Mathematica). Using the fact that

2k1 · q 2k2 · q 2k3 · q 2(k1 + k2 + k3 ) · q 2q · q

x1 + x2 + x3 = 2
+ 2
+ 2
= 2
= = 2, (91)
q q q q q2
the previous relations are
  1 2 2

  2
q (1 − x3 ) + µ2
 
2k1 · k2 2 2µ + q (x1 + x2 − x3 )

2k1 · k3  =  1 q 2 (x1 − x2 + x3 ) − 2µ2  = q 2 (1 − x2 ) − µ2  .
2 (92)
1
q 2 (1 − x1 ) − µ2
2 2


2k2 · k3 2 q (−x1 + x2 + x3 ) − 2µ

16
Substituting into (83)

32(k1 · k3 )(k2 · k3 ) 32(k1 · k3 )(k2 · k3 ) 64(k1 · k2 )2

gαβ H αβ (k1 , k2 , k3 ) = + + +
(2k1 · k3 + µ2 )2 (2k2 · k3 + µ2 )2 (2k1 · k3 + µ2 )(2k2 · k3 + µ2 )
64(k1 · k2 )(k1 · k3 ) 64(k1 · k2 )(k2 · k3 )
+ +
(2k1 · k3 + µ2 )(2k2 · k3 + µ2 ) (2k1 · k3 + µ2 )(2k2 · k3 + µ2 )
16µ2 (k1 · k2 ) 16µ2 (k1 · k2 ) 64µ2 (k1 · k2 )
− − +
(2k1 · k3 + µ2 )2 (2k2 · k3 + µ2 )2 (2k1 · k3 + µ2 )(2k2 · k3 + µ2 )
2 2 2 2
[q (1 − x2 ) − µ ][q (1 − x1 ) − µ ] [q 2 (1 − x2 ) − µ2 ][q 2 (1 − x1 ) − µ2 ]
=+8 + 8
[q 2 (1 − x2 )]2 [q 2 (1 − x1 )]2
2 2 2
[q (1 − x3 ) + µ ]
+ 16 2
[q (1 − x2 )][q 2 (1 − x1 )]
[q 2 (1 − x3 ) + µ2 ][q 2 (1 − x2 ) − µ2 ] [q 2 (1 − x3 ) + µ2 ][q 2 (1 − x1 ) − µ2 ]
+ 16 + 16
[q 2 (1 − x2 )][q 2 (1 − x1 )] [q 2 (1 − x2 )][q 2 (1 − x1 )]
2 2 2 2 2 2
µ [q (1 − x3 ) + µ ] µ [q (1 − x3 ) + µ ] µ2 [q 2 (1 − x3 ) + µ2 ]
−8 − 8 + 32
[q 2 (1 − x2 )]2 [q 2 (1 − x1 )]2 [q 2 (1 − x2 )][q 2 (1 − x1 )]
2 2



8 x1 + x2 8 2x1 x2 − 3x1 + 2x1 x2 − 4x1 x2 + 4x1 − 3x22 + 4x2 − 2 µ2
2 2 2
= +
(1 − x1 )(1 − x2 ) (1 − x1 )2 (1 − x2 )2 q2
 2 2
16 µ
+
(1 − x1 )(1 − x2 ) q 2
 2
 2

16 1 + µq2 x1 + x2 − 1 + µq2 
1 1

µ2

= +8 + (1 − x 1 )(1 − x 2 ) −
(1 − x1 )(1 − x2 ) (1 − x1 )2 (1 − x2 )2 q2
(93)
in the last steps we have substituted x3 = 2 − x1 − x2 (see the calculation in the Mathemathica notebook).
Now we take the gluon mass as zero

[q 2 (1 − x1 )] [q 2 (1 − x2 )] [q 2 (1 − x3 )]2
gαβ H αβ (k1 , k2 , k3 ) = + 8 + 8 + 16
[q 2 (1 − x2 )] [q 2 (1 − x1 )] [q 2 (1 − x2 )][q 2 (1 − x1 )]
2 2
[q (1 − x3 )] [q (1 − x3 )]
+ 16 2 + 16 2
[q (1 − x1 )] [q (1 − x2 )]
(1 − x1 ) (1 − x2 ) (1 − x3 )2 (1 − x3 ) (1 − x3 )
=+8 +8 + 16 + 16 + 16
(1 − x2 ) (1 − x1 ) (1 − x2 )(1 − x1 ) (1 − x1 ) (1 − x2 )
(1 − x1 )2 + (1 − x2 )2 (1 − x3 )2 (1 − x3 )(1 − x2 ) + (1 − x3 )(1 − x1 )
=+8 + 16 + 16
(1 − x2 )(1 − x1 ) (1 − x2 )(1 − x1 ) (1 − x1 )(1 − x2 )
x21 + x22
=8 .
(1 − x1 )(1 − x2 )
(94)

17
 With these results, the differential cross section (81) is finally
  4 2 2
dσ 1 4 e g Qq 1 µν
=− g Lµν (p1 , p2 )gαβ H αβ (k1 , k2 , k3 )
dx1 dx2 128π 3 3 4 2q 4
  4 2 2
1 4 e g Qq 1 x21 + x22
=− 3 4
(−4q 2 )8
128π 3 4 2q (1 − x1 )(1 − x2 )
2 2
4π16π 2 4
  4 2
x21 + x22
 
8 4 4 2 2 1 x1 + x2 e g 2 1
= e g Q q = Qq 2 (95)
128π 3 3 2
2q (1 − x1 )(1 − x2 ) 16π 3 3 16π 4π 2 2q (1 − x1 )(1 − x2 )
2 2
x21 + x22
   
4 1 x1 + x2 4  αg  1
=4 α2 αg Q2q 2 = 4(2π) α2 Q2q 2
3 2q (1 − x1 )(1 − x2 ) 3 2π 2q (1 − x1 )(1 − x2 )
2 2 2 2
x21 + x22
   
4 4πα 2 αg   x1 + x2 4 4πα  αg 
= 2
Qq = 3Q2q ,
3 q 2π (1 − x1 )(1 − x2 ) 3 3s 2π (1 − x1 )(1 − x2 )

where s = q 2 . The result (95) is, up to a factor 4/3, the one shown in page 261 of3 . This 4/3 factor is the color
factor C2 (F ) of SU (3) and it doesn’t appear in Peskin’s expression because they’re using the abelian approach
(42) to describe the gluon interactions. The correct result is (95) and it can be obtained from Peskin’s result
by making the replacement (67), as it’s stated in the book.
(d) The differential cross section in the limit µ2 → 0 is

4 4πα2 2  αg  x21 + x22

 
dσ
= 3Qq . (96)
dx1 dx2 3 3s 2π (1 − x1 )(1 − x2 )

This expression is divergent at x1 = 0 and x2 = 0. From the expressions (25) we have

2E1 2E2 2E3

x1 = , x2 = , x3 = , (97)
Eq Eq Eq

while from expressions (30) and (31) we have that when µ2 → 0 the energies are
q
E1 =|k1 |, E2 =|k2 |, E3 =|k3 | = E12 + E22 + 2E1 E2 cos θ12 . (98)

From the expressions (92) the following relations hold in the CM frame (with µ2 → 0)

2k1 · k3 =q 2 (1 − x2 ) = Eq2 (1 − x2 ) 2k2 · k3 =q 2 (1 − x1 ) = Eq2 (1 − x1 ) (99)

then
1 2
k1 · k3 =E1 E3 − k1 · k3 = E1 E3 (1 − cos θ13 ) = E (1 − x2 ) (100)
2 q
1
k2 · k3 =E2 E3 − k2 · k3 = E2 E3 (1 − cos θ23 ) = Eq2 (1 − x1 ). (101)
2
From these expressions it’s evident that

if x1 =1 ⇒ cos θ23 =1 ⇒ θ23 =0 ⇒ gluon parallel to antiquark (102)

if x2 =1 ⇒ cos θ13 =1 ⇒ θ13 =0 ⇒ gluon parallel to quark. (103)
3 Peskin and Schroeder, An Introduction To Quantum Field Theory.

18
 It’s possible to see the divergence explicitly as follows. If we keep the gluon mass µ2 6= 0, then, using (93), the
differential cross section is
  4 2 2
dσ 1 4 e g Qq 1 µν
=− 3
g Lµν (p1 , p2 )gαβ H αβ (k1 , k2 , k3 )
dx1 dx2 128π 3 4 2q 4
  4 2 2
1 4 e g Qq 1
=− (−4q 2 )
128π 3 3 4 2q 4
  2
 2
 
 16 1 + µq2 x1 + x2 − 1 + µq2 
1 1

µ 2 

× +8 + (1 − x1 )(1 − x2 ) − 2
 (1 − x1 )(1 − x2 ) (1 − x1 )2 (1 − x2 )2 q 
  4 2 2
8 4 e g Qq
= 3
128π 3 2q 2
  2
 2
 
 2 1 + µq2 x1 + x2 − 1 + µq2 
1 1

µ2 

× + + (1 − x1 )(1 − x2 ) − 2
 (1 − x1 )(1 − x2 ) (1 − x1 )2 (1 − x2 )2 q 

4 4πα2 2  αg 
 
= 3Qq
3 3s 2π
  2
 2
 
 2 1 + µq2 x1 + x2 − 1 + µq2 
1 1

µ 2 

× + + (1 − x1 )(1 − x2 ) − 2
 (1 − x1 )(1 − x2 ) (1 − x1 )2 (1 − x2 )2 q 
(104)
The total cross section σ, recalling the limits of integration in (41), is
2
Z 1−µ2 /q 2 Z 1− µ 1
q2 1−x1 dσ
σ= dx1 dx2
0 1− µ
2
−x1 dx1 dx2
q2

2 2 Z 1− µ22 1−x
1
4 4πα2 2  αg  1−µ /q
  Z
q 1
= 3Qq dx1 dx2 (105)
3 3s 2π 0 µ2
1− q2 −x1
  2
 2
 
 2 1 + µq2 x1 + x2 − 1 + µq2 
1 1

µ2 

× + + (1 − x1 )(1 − x2 ) − 2 .
 (1 − x1 )(1 − x2 ) (1 − x1 )2 (1 − x2 )2 q 

Doing the integral in Mathematica, we get the final result

4 4πα2 2  αg 
    2  2 
µ µ π
σ= 3Qq log2 + 3 log − + 5 + · · · , (106)
3 3s 2π q2 q2 3

where · · · are finite terms. From this expression it’s evident that there is a divergence in the infrared region as
we take µ → 0.

6.3 Gauge coupling running at two-loops

In QCD, the beta function β at two loops has the form

β(g) = −β0 g 3 − β1 g 5 + · · · (107)

19
where
   
1 2 1 38
β0 = 11 − Nf β1 = 102 − Nf . (108)
(4π)2 3 (4π)4 3

(a) The effective coupling ḡ is defined by

dḡ(g, t) 1 Q2
=β(ḡ) with ḡ(g, 0) =g, t = ln 2 . (109)
dt 2 Λ
First we solve the equation considering terms up to order g 3 in β(g)

dḡ(g, t) dḡ
= − β0 ḡ 3 ⇒ = − β0 dt, (110)
dt ḡ 3
integrating this equation we get
Z ḡ(g,t) Z t
dḡ
=− β0 dt
ḡ(g,0)=g ḡ 3 0
 ḡ(g,t)
1
− 2 = − β0 t
2ḡ g

− 2
1 1
+ 2 = − β0 t (111)
2ḡ (g, t) 2g
1 1
=β0 t +
2ḡ 2 (g, t) 2g
1
2ḡ 2 (g, t) = 1 ,
β0 t + 2g

using t = 1
2 ln Q2 /µ2 we have

1 1 2g
2ḡ 2 (g, t) = 1 = Q2
= 2 (112)
β0 t + 2g β0 12 ln µ2 + 1
2g β0 g ln Q
µ2 + 1

thus
g g
ḡ 2 (g, t) = = 2 . (113)
2β0 gt + 1 β0 g ln Q
µ2 + 1

Now we solve the equation considering terms up to order g 5 in β(g)

dḡ(g, t) dḡ
= − β0 ḡ 3 − β1 ḡ 5 ⇒ = − β0 dt, (114)
dt β0 ḡ 3 + β1 ḡ 5
integrating this equation with the use of the integral


b ln a + bx2
Z
dx b ln(x) 1
= − − +C (115)
ax3 + bx5 2a2 a2 2ax2

20
 we get
Z ḡ(g,t) Z t
dḡ
3 5
= − β0 dt
ḡ(g,0)=g β0 ḡ + β1 ḡ 0
"
#ḡ(g,t)
β1 ln β0 + β1 ḡ 2 β1 ln ḡ 1
− − = − β0 t
2β02 β02 2β0 ḡ 2
g
"
# "
#
β1 ln β0 + β1 ḡ 2 β1 ln ḡ 1 β1 ln β0 + β1 g 2 β1 ln g 1 (116)
− − − − − = − β0 t
2β02 β02 2β0 ḡ 2 2β02 β02 2β0 g 2
"
# "
#
β1 ln β0 + β1 ḡ 2 β1 ln ḡ 2 1 β1 ln β0 + β1 g 2 β1 ln g 2 1
− − − − − = − β0 t
2β02 2β02 2β0 ḡ 2 2β02 2β02 2β0 g 2
β0 + β1 ḡ 2 β0 + β1 g 2
       
β1 1 β1 1
ln − − ln − = − β0 t,
2β02 ḡ 2 2β0 ḡ 2 2β02 g2 2β0 g 2

now we can use the approximation

β0 + β1 ḡ 2 Q2 Q2
     
β0
ln = ln β 1 + ' ln β 1 + β 2
0 ln ' ln ln (117)
ḡ 2 ḡ 2 Λ2 Λ2

substituting this approximation we get the equation

β 0 + β1 g 2 Q2
   
1 β1 1 β1
2
=β 0 t − 2 ln 2
− 2
+ 2 ln ln 2
2β0 ḡ 2β g 2β0 g 2β0 Λ
 0  2
  2
1 β1 β 0 + β1 g 1 β1 Q
=2β02 t − ln − 2 + ln ln 2
ḡ 2 β0 g2 g β0 Λ
2 2 2
   
1 2 Q β1 Q β1 β0 + β1 g 1
=β0 ln 2 + ln ln 2 − ln − 2
ḡ 2 Λ β0 Λ β0 g2 g
(118)
( )
Q2 Q2 β0 + β1 g 2
  
1 2 1 β1 1 β1 1
=β0 ln 2 1 + 2 ln ln 2 − 2 ln − 2
ḡ 2 Λ β02 ln Q β0 Λ β02 ln Q β0 g2 g
Λ2 Λ2
( )
Q2 Q2 β0 + β1 g 2
   
1 1 β1 1 β1 1
=β0 ln 2 1 + Q2 2 ln ln 2 − 2 ln − 2
ḡ 2 Λ ln Λ2 β0 Λ β02 ln Q β0 g2 g
Λ2
( 2 )−1
Q
β0 + β1 g 2
   
1 β1 ln ln Λ2 1 β1 1
ḡ 2 = Q 2 1 + 2 Q 2 − Q 2 ln 2
− 2
β0 ln 2 β0 ln 2 β0 ln 2 β0
2 g g
Λ Λ Λ

now we can use the series expansion (1 + x)−1 = 1 − x + · · · and get

Q2
( )
1 β ln ln
1
(119)
2
ḡ 2 = 2 1− 2 Λ
+ ···
β0 ln Q2 β0 ln Q22
Λ Λ

6.4 Combining fundamental representations

(a) The Lie algebra for the SU (2) group is
[Ii , Ij ] = iijk Ik , (120)

21
 where Ii are the generators. The Casimir operator is I 2 = I12 + I22 + I32 , as can be seen from

[I 2 , Ii ] =[Ij Ij , Ii ] = Ij [Ij , Ii ] + [Ij , Ii ]Ij = ijik Ij Ik + ijik Ik Ij

(121)
=ijik Ij Ik + ikij Ij Ik = ijik Ij Ik − ijik Ij Ik = 0.

The common eigenvectors of I 2 and Iz , denoted as |I, I3 i, satisfy

1
I 2 |I, I3 i =I(I + 1) |I, I3 i , I =0, , 1, · · · (122)
2
Iz |I, I3 i =I3 |I, I3 i , I3 = − I, −I + 1, · · · , I − 1, I. (123)

Thus, the irreps of SU (2) are defined by I = 0, 12 , 1, · · · . The ladder operators I± are defined as

I± = Ix ± iIy (124)

and their action on the eigenstates |I, I3 i is

(125)
p
I± |I, I3 i = (I ∓ I3 )(I ± I3 + 1) |I, I3 ± 1i .

The irrep I = 0 is 1-dimensional and it’s denoted by 1, this is the trivial irrep. The irrep I = 12 is 2-dimensional
because I3 = 12 , − 12 , it’s denoted by 2, and the eigenvectors are | 12 , + 12 i and | 12 , − 12 i satisfying
1 1 3 1 1 1 1 3 1 1
I2 | , + i = | , + i , I2 | , + i = | , − i , (126)
2 2 4 2 2 2 2 4 2 2
1 1 1 1 1 1 1 1 1 1
Iz | , + i = | , + i , Iz | , + i = − | , − i . (127)
2 2 2 2 2 2 2 2 2 2
The irrep I = 1 is 3-dimensional because I3 = 1, 0, −1, it’s denoted by 3 and the eigenvectors are |1, 1i , |1, 0i,
and |1, −1i satisfying

I 2 |1, 1i =2 |1, 0i I 2 |1, 0i =2 |1, 0i I 2 |1, −1i =2 |1, −1i (128)

Iz |1, 1i =1 |1, 0i Iz |1, 0i =0 |1, 0i Iz |1, −1i = − 1 |1, −1i . (129)

For the case of the SU (2)-isospin group the states belonging to the irrep 2, are |ui and |di respectively
1 1 1 1
|ui = | , + i , |di = | , − i , (130)
2 2 2 2
and we have

I− |ui = |di , I+ |di = |ui (131)

I+ |ui =0, I− |di =0. (132)

The dual representation of 2, denoted by 2∗ has the basis

¯
B2∗ = {|ūi , |di}. (133)

It’s possible to show that the fundamental representation 2 of SU (2) and its dual 2∗ are equivalent(see4 , page
226), and the elements of their basis are related as

|ūi = − |di , ¯ = |ui .

|di (134)
4 N. Jeevanjee. An Introduction to Tensors and Group Theory for Physicists. Springer International Publishing, 2015. isbn:
9783319147949. url: [Link]

22
Thus, the tensor product 2 ⊗ 2∗ is equivalent to 2 ⊗ 2, and its tensor product basis is

BS⊗S ={|ui |ui , |ui |di , |di |ui , |di |di} ≡ {|uui , |udi , |dui , |ddi}. (135)

Let’s compute the action of I 2 and Iz on |uui

I 2 |uui =I 2 (|ui |ui) = (I 2 |ui) |ui + 2(Ii |ui)(Ii |ui) + |ui (I 2 |ui)
3 1 i2 1 3
= |ui |ui + |di |di + |di |di + |ui |ui + |ui |ui
4 4 4 4 4
=2 |ui |ui = 1(1 + 1) |uui (136)
1 1
Iz |uui =Iz (|ui |ui) = (Iz |ui) |ui + |ui (Iz |ui) = |ui |ui + |ui |ui
2 2
= |uui ,

thus we see that this state has eigenvalues I = 1 and Iz = 1, i.e it belongs to the irrep 3. To find the rest of
the states in 2 ⊗ 2 with I = 1 we apply the ladder operator I− to |uui

I− |uui = I− (|ui |ui) = (I− |ui) |ui + |ui (I− |ui) = |di |ui + |ui |di = |dui + |udi , (137)

which has eigenvalues I = 1 and Iz = 0 due to the way we’re constructing it. Now we can apply the ladder
operator I− to this state

I− (|dui + |udi) =I− (|di |ui + |ui |di) = I− (|di |ui) + I− (|ui |di)
=(I− |di) |ui + |di (I− |ui) + (I− |ui) |di + |ui (I− |di) = |di |di + |di |di (138)
=2 |ddi ,

which has eigenvalues I = 1 and Iz = −1 by construction. Finally we can construct a fourth state by requiring
ortogonalithy with all the rest of the states cosntructed so far. This states turns out to be |udi − |dui and its
egienvalues are

I 2 (|udi − |dui) =0(|udi − |dui), Iz (|udi − |dui) =0(|udi − |dui), (139)

this means that this state belongs to the 1 irrep. In summary we have that the tensor product 2 ⊗ 2 has a
triplet 3 and a singlet 1
   
1 1
irrep 3: |1, 1i = |uui , |1, 0i = √ (|udi + |dui), |1, −1i = |ddi irrep 1: |0, 0i = √ (|udi − |dui) , (140)
2 2
this fact is expressed as
2 ⊗ 2 = 1 ⊕ 3. (141)
Writing the results (140) in terms of the dual space basis by making use of (133) and the relations (134) we get
   
¯ , |1, 0i = √1 (|ddi
irrep 3: |1, 1i = |dui ¯ − |ūui), |1, −1i = − |ūdi 1 ¯
irrep 1: |0, 0i = √ (|ddi + |ūui) .
2 2
(142)

The states in the triplet 3 are the pions (up to a phase)

1
¯ ,
|π + i = − |1, 1i = − |dui ¯
|π 0 i = |1, 0i = √ (|ūui − |ddi), |π − i = |1, −1i = − |ūdi . (143)
2

23
(b) The group SU (3) has 8 generators T a . In the fundamental representation they are T a = 12 λa , where λa are
the Gell-Mann matrices. The Cartan subalgebra of this group is {T 3 , T 8 }. The ladder operators are
1 1 1
I± = √ (T1 ± iT2 ), V± = √ (T4 ± iT5 ), U± = √ (T6 ± iT7 ). (144)
2 2 2
In this case the basis of the irrep 3 and its dual 3̄ are
¯ , |s̄i . (145)

B3 = {|ui , |di , |si} , B3̄ = |ūi , |di

The action of the ladder operators on these states is

I− |ui = |di ¯ = − |ūi

I− |di V− |ui = |si V− |s̄i = |ūi U− |si = |di ¯ = − |s̄i
U− |di (146)
I+ |di = |ui ¯
I+ |ūi = − |di V+ |si = |ui V+ |ūi = |s̄i U+ |di = |si ¯,
U+ |s̄i = − |di (147)

and any other action independent of these is zero. The eigenvalues of the basis states for the Cartan subalgebra
operators {T 3 , T 8 } are
     
1 1 1 1 1
T 3 , T 8 |ui = T 3 , T 8 |di = − , √ T 3 , T 8 |si = 0, − √ (148)




, √ |ui |di |si
2 2 3 2 2 3 3
     
1 1 ¯ = 1, − √ 1 1
T 3 , T 8 |ūi = − , − √ T 3 , T 8 |di ¯ T 3 , T 8 |s̄i = 0, √ (149)




|ūi |di |s̄i
2 2 3 2 2 3 3
The basis for the tensor product 3̄ ⊗ 3 has nine elements and is
¯ , |ddi
¯ , |dsi
¯ , |ūui , |ūdi , |ūsi , (150)

B3̄⊗3 = |ūui , |ūdi , |ūsi , |dui

whose eigenvalues for the Cartan subalgebra operators {T 3 , T 8 } are (they are additive when acting on the
tensor product space, for example T 3 |ābi = (T 3 |āi) |bi + |āi (T 3 |bi) = t3ā |āi |bi + t3b |āi |bi = (t3a + t3b ) |ābi)
√ !
1 3
3 8 3 8 3 8
|ūsi (151)




T , T |ūui = (0, 0) |ūui T , T |ūdi = (−1, 0) |ūdi T , T |ūsi = − , −
2 2
√ !
1 3
¯ = (1, 0) |dui
T 3 , T 8 |dui ¯ ¯ = (0, 0) |ddi
T 3 , T 8 |ddi ¯ T 3 , T 8 |dsi¯ = ¯ (152)




,− |dsi
2 2
√ ! √ !
1 3 1 3
3 8 3 8
T 3 , T 8 |s̄si = (0, 0) |s̄si . (153)




T , T |s̄ui = , |s̄ui T , T |s̄di = − , |s̄di
2 2 2 2

In figure 1 it’s shown a diagramtic representation of these eigenvalues. We can use the ladder operators
I± , V± , U± to get a state with eigenvalues (0, 0) by starting from the external states as follows (see figure 1)
¯ = |ūui − |ddi
I− |−dui ¯ ¯ + |ūui
I+ |ūdi = − |ddi (154)
V− |s̄ui = |ūui + |s̄si V+ |ūsi = |s̄si + |ūui (155)
¯ = |s̄si − |ddi
U− |−dsi ¯ ¯ + |s̄si ,
U+ |s̄di = − |ddi (156)

we can see that there are only three independent states with eigenvalues (0, 0). Among these, we can recognize
the state |π 0 i that belongs to the isospin triplet
1 ¯
|π 0 i = √ (|ūui − |ddi). (157)
2

24
 Figure 1: Representation of the Cartan subalgebra eigenvalues for the tensor product states

and the states orthogonal to this one also with eigenvalues (0, 0) are
1 1
¯ − 2 |s̄si),
|ηi = √ (|ūui + |ddi ¯ + 2 |s̄si).
|η 0 i = √ (|ūui + |ddi (158)
6 6
The state |η 0 i is annihilated by all the ladder operators

I± |η 0 i =0, V± |η 0 i =0, U± |η 0 i =0, (159)

and then is a singlet. The remaining eight states are related to each other by the action of the ladder operators
{I± , V± , U± }, for example, we can also use them to move between the external states. Therefore, these eight
states form an octet whose elements can be identified with the physical particles known as pions, kaons and
eta as follows
1
|K + i = |s̄ui ¯
|π + i = − |dui ¯
|K̄ 0 i = − |dsi ¯
|π 0 i = √ (|ūui − |ddi) (160)
2
1
|K − i = |ūsi |π − i = |ūdi |K 0 i = |s̄di ¯ − 2 |s̄si).
|ηi = √ (|ūui + |ddi (161)
6

This decomposition of 3 ⊗ 3̄ into a singlet and an octet is expressed as

3 ⊗ 3̄ = 1 ⊕ 8. (162)

25
6.5 Scalar fields with color charge
The QCD bare Lagrangian is
1 1
L = − (∂µ Aa0ν − ∂ν Aa0µ )2 − (∂ µ Aa0µ )2 + ψ̄0i (i∂/ − m0 )ψ0i − c̄a0 ∂ 2 ca0
4 2ξ0
+ g0 Aa0µ ψ̄0i γ µ Tija ψ0j − g0 f abc (∂µ Aa0ν )Aµ0 Acν
0
(163)
1 2 eab a b ecd c d µ a abc b c
− g0 (f A0µ A0µ )(f A0µ A0µ ) + g0 (∂ c̄0 )f A0µ c0
4
where the subindex 0 denotes the bare fields and parameters. Introducing the physical fields

(164)
1/2 1/2 1/2
Aa0µ =Z3 Aaµ ψ0i =Z2 ψi ca0 =Z3c ca ,

the Lagrangian is split into physical terms and counterterms as

1 1
L = − (∂µ Aaν − ∂ν Aaµ )2 − (∂ µ Aaµ )2 + ψ̄i (i∂/ − m)ψi − c̄a ∂ 2 ca
4 2ξ
+ gAaµ ψ̄i γ µ Tija ψj − gf abc (∂µ Aaν )Abµ Acν
1
− g 2 (f eab Aaµ Abµ )(f ecd Acµ Adµ ) + g(∂ µ c̄a )f abc Abµ cc
4 (165)
1
− δ3 (∂µ Aaν − ∂ν Aaµ )2 + ψ̄i (iδ2 ∂/ − δm )ψi − δ3c c̄a ∂ 2 ca
4
+ gδ1 Aaµ ψ̄i γ µ Tija ψj − gδA3 f abc (∂µ Aaν )Abµ Acν
1
− g 2 δA4 (f eab Aaµ Abµ )(f ecd Acµ Adµ ) + gδ1c (∂ µ c̄a )f abc Abµ cc
4
where ξ = ξb /Z3 and

δ2 =Z2 − 1 δ3 =Z3 − 1 δ3c =Z3c − 1 δm =Z2 m0 − m (166)

g0 g0 3/2 g0 g0
(167)
1/2 1/2
δ1 = Z2 Z3 − 1 δA3 = Z3 − 1 δA4 = Z32 − 1 δ1c = Z3c Z3 − 1.
g g g g
The Lagrangian for the interaction of a gluon with a colored scalar triplet field Φ is given by
1 µ a 2 1 1
L =(Dµ Φ)† (Dµ Φ) − m2 Φ† Φ − (∂ Aµ ) = (∂µ Φ − igAaµ T a Φ)† (∂µ Φ − igAaµ T a Φ) − F aµν Faµν − (∂ µ Aaµ )2
2ξ 4 2ξ
µ † a a µa a † a a 2 † 1 aµν 1 µ a 2
=(∂ Φ) (∂µ Φ − igAµ T Φ) + (−igA T Φ) (∂µ Φ − igAµ T Φ) − m Φ Φ − F Faµν − (∂ Aµ )
4 2ξ
1 1
=(∂ µ Φ)† (∂µ Φ) + (∂ µ Φ)† (−igAaµ T a Φ) + (−igAµa T a Φ)† (∂µ Φ) + (−igAµa T a Φ)† (−igAbµ T b Φ) − m2 Φ† Φ − F aµν Faµν − (∂ µ A
4 2ξ
µ † 2 † µ † a a † a µa 2 † a µa b b 1 aµν 1 µ a 2
=∂ Φ ∂µ Φ − m Φ Φ − ig∂ Φ Aµ T Φ + igΦ T A ∂µ Φ + g Φ T A Aµ T Φ − F Faµν − (∂ Aµ ) .
4 2ξ
(168)

26
The Feynman rules for this theory are

q
 µ ν

p iδij µ ν −iδ ab η µν − (1 − ξ) q q2q
=P Φ(p)ij = , = ,
p2 − m2 + i q 2 + i
µ ν
µ
aµ 0 abµν
=V3ij (p , p) = −igTija (p0 + p)µ , =V4ij = 2ig 2 Tik
a b µν
Tkj g ,
p p0 p p0

(a) The contribution to the gluon self-energy comes from the diagrams

l
l+q
q q q q
µ ν µ ν
−iΠ(2)
µν (q) = + (169)
l

In dimensional regularization this amplitude is

dD l
Z 
i
iΠab
µν (q) = D
[−igTik a
(2l + q)µ ]
(2π) (l + q) − m2 + i
2

b i
× [−igTkj (2l + q)ν ] 2
l − m2 + i

i
+ 2ig 2 Tik (170)
a b
 
Tkj 2
l − m2 + i
dD l (2l + q)µ (2l + q)ν
Z
a b 2
=Tik Tkj g
(2π) (l + q) − m + i l − m2 + i
D 2 2 2

dD l
Z 
a b 2 1
− 2Tik Tkj g
(2π)D l2 − m2 + i

Computing this integrals with the help of Feyncalc (see the Mathematica notebook attached) and extracting
the UV divergences we get
ig 2
Div[iΠab q µ q ν − q 2 g µν Tik (171)

a b
µν (q)] = 2
Tkj
48π ε
thus, the contribution to δ3 is given by
g2
δ3 = TF . (172)
48π 2 ε
(b) The contribution to the A3 vertex comes from the diagrams

27
 In dimensional regularization this amplitude is

dD k   a
Z
Tr V3µ (−q1 + k, k)P Ψ(k)V3νb
(k, q2 + k)P Ψ(q2 + k)
(2π)D
(173)
c
V3α (q2 + k, −q1 + k)P Ψ(−q1 + k)}]
Tr V3µ (q1 + k, k)P Ψ(k)V3ν
 a b
(k, −q2 + k)P Ψ(−q2 + k)
c
V3α (−q2 + k, q1 + k)P Ψ(q1 + k)}] ,

where the trace is over the color indices of the scalar triplet. Computing this integral with FeynCalc we find
that it has no divergences. This is the analogous to the case of QED in that these contrbutions are finite too
due to Ward identity.
(c) Taking into account the contribution to δ3 coming from this scalar triplet, we have that the δ3 with ns scalar
fields and nf fermion fields is (we have used the result of the book Schwartz for writing the contribution of
gluons and fermions)  2  
1 g 10 8 1
δ3 = C A − n T
f F − n s F ,
T (174)
ε 16π 2 3 3 3
The beta function can be calculated from the equation
  
d ε d 1
β(gR ) = µ gR = gR − −µ δ1 − δ2 − δ3 + ···
dµ 2 dµ 2

that can be preoperatively solved and produces

 
ε ε 2 ∂ 1
β(gR ) = − gR + gR δ1 − δ2 − δ3
2 2 ∂gR 2

With this expression, the β function is

g3
 
11 4 1
β(g) = − CA − TF nf − TF ns . (175)
(4π)2 3 3 3

(d) For CA = 3 and TF = 1/2, this beta function vanishes when

2 1
11 − nf − ns =0 ⇒ ns = 66 − 4nf . (176)
3 6

28