Teoria Quântica dos Campos

1st semester, 2nd period 2024-25

Prof. João Paulo Silva

Problem Set 4b
Green’s functions and vector boson polarizations
Self study.
December 9, 2024

1. Let Γ be an oriented closed curve on a complex plane and f (z) some function of the complex variable
z with poles at ak (k = 1 · · · n) inside the area defined by Γ. Cauchy’s residue theorem states that
I n
X
f (z)dz = ± 2πi Res(f )|z=ak , (1)
Γ k=1

where the “−” (“+”) sign is used when Γ is oriented clockwise (counter-clockwise). If ak is a simple pole,
then

Res(f )|z=ak = lim (z − ak ) f (z) , (2)

z→ak

and close to ak the function may be approximated by

Res(f )|z=ak
. (3)
z − ak

a) Consider a function of the type

 
−iz(t′ −t) 1 1
f (z) = e ··· . (4)
z − a1 z − a2

One wishes to know the integral along the Re(z) axis. This will be obtained using Cauchy’s residue
theorem by either completing the real line with a semi-circle on the upper half plane (UHP) or by
completing the line with a semi-circle on the lower half plane (LHP). Show that, if t′ − t < 0, then
Z Re(z)=+∞ X
f (z)dz = +2πi Res(f )|z=ak in UHP . (5)
Re(z)=−∞

Conversily, if t′ − t > 0, then

Z Re(z)=+∞ X
f (z)dz = −2πi Res(f )|z=ak in LHP . (6)
Re(z)=−∞

b) Consider now the function

g(z) ′ 1 1
f (z) = ≃ h(z) e−iz(t −t) , (7)
z2 2
− E + 2iEη z + E − iη z − E + iη
 where E, η > 0 and, at the end, η and ǫ = 2Eη will be taken to zero, justifying the ≃ in (7). The
′
function g(z) = h(z)e−iz(t −t) is analytic. By appropriate use of Cauchy’s residue theorem, show
that
Z Re(z)=+∞ ′
dz e−iz(t −t) h(z) ′ h(E)
i 2 2
= e−iE(t −t) if t′ − t > 0 ,
Re(z)=−∞ 2π z − E + iǫ 2E
Z Re(z)=+∞ ′
dz e−iz(t −t) h(z) ′ h(−E)
i = e+iE(t −t) if t′ − t < 0 . (8)
Re(z)=−∞ 2π z 2 − E 2 + iǫ 2E

2. The Klein Gordon Green’s function ∆F (x′ − x) is defined by

x′ + m2 ∆F (x′ − x) = −iδ4 (x′ − x) . (9)

a) Fourier transform by using

Z
d4 p −ip.(x′ −x)
∆F (x′ − x) = e ∆F (p) ,
(2π)4
Z
4 ′ d4 p −ip.(x′ −x)
i δ (x − x) = i e , (10)
(2π)4

to prove that
i
∆F (p) = . (11)
p2 − m2
p
Viewed as a function of p0 , ∆F (p) has poles at p0 = ±E, with E = + |~
p|2 + m2 .

b) Using Feynman’s “+iǫ” prescription to deal with poles along the real line discussed in the previous
problem, prove that
Z
d4 p −ip.(x′ −x) i
∆F (x′ − x) = e
(2π)4 p 2 − m2
Z h i
d3 p ′ −ip.(x′ −x) ′ ip.(x′ −x)
= θ(t − t) e + θ(t − t ) e . (12)
(2π)3 2E

3. Consider the field operator

Z h i
f â(p)e−ip·x + ↠(p)eip·x ,
φ̂(~x, t) = dp (13)

where
d3 p p
f≡
dp , p0 = Ep ≡ + |~
p|2 + m2 , (14)
(2π)3 2Ep
e − q) ≡ (2π)3 2Ep δ3 (~
a|0i = 0, and [a(p), a† (q)] = δ(p p − ~q). Define

D(x − y) = h0|φ̂(x)φ̂(y)|0i . (15)

a) Show that
Z
D(x − y) = f e−ip·(x−y) .
dp (16)
 b) Show that

∆F (x′ − x) = θ(t′ − t) DF (x′ − x) + θ(t − t′ ) DF (x − x′ )

= θ(t′ − t) h0|φ̂(x′ )φ̂(x)|0i + θ(t − t′ ) h0|φ̂(x)φ̂(x′ )|0i
≡ h0|T φ̂(x)φ̂(x′ )|0i . (17)

The last line defines the so-called time ordered product.

4. Recall that the inhomogeneous Maxwell’s equations, which read,

∂µ F µν = J ν , F µν = ∂ µ Aν − ∂ ν Aµ , (18)

are invariant under the gauge transformations

Aµ → Aµ + ∂ µ λ , (19)

where λ(x) is any function of spacetime.

a) The Lorenz gauge is defined by

∂φ ~ ~
0 = ∂µ Aµ = +∇·A (Lorenz gauge) . (20)
∂t
Is this gauge Lorentz invariant? Does one remain in the Lorenz gauge under the most general
transformation (19)?

b) Show that there is a residual gauge freedom

Aµ → A′µ = Aµ + ∂µ λ̃ , (21)

with  λ̃ = 0 such that, if one starts in a Lorenz gauge, one remains in a Lorenz gauge.

c) Find a simplified expression for (18) in Lorenz gauge and show that in vacuum it reduces to

 Aµ = 0 . (22)

Comparing with the usual KG equation, what do you conclude about the mass of the photon?

d) Show by direct substitution that (22) admits the plane wave solutions

Aµ = ǫµ (k)e−ik.x , (23)

as long as k2 = 0, where we introduced the polarization vector ǫµ (k). Confirm your earlier conclusion
about the mass of the photon.

We wish to show in two steps that instead of 4, ǫµ only has 2 independent degrees of freedom (dof).

e) Imposing the Lorenz gauge condition on (23), conclude that

k.ǫ = 0 , (24)

reducing the dof in ǫµ by one unit.

 f ) Consider

λ̃ = iae−ik.x , (25)

for some generic constant a. Show that it qualifies as a residual gauge freedom; that is,  λ̃ = 0.
Show that, under this residual gauge freedom, one concludes that the same photon can be described
by

ǫµ or ǫµ + ak µ . (26)

This reduces the dof in ǫµ to only two independent components.

g) Within Lorenz gauge, one can use the residual gauge freedom in order to set ǫ0 = 0, leading to

~ǫ · ~k = 0 . (27)
~ ·A
This is known as the Coulomb or transverse gauge. Indeed, show that it is equivalent to ∇ ~ = 0.
Is this gauge choice Lorentz invariant?

5. (Halzen and Martin 6.11) Choosing ~k along the z axis and using k2 = 0, k = (k, 0, 0, k). Notice the
three different uses of k: as the 4-vector; as the magnitude of the 4-vector; and as the magnitude of the
3-vector. Get used to it. One can chose to describe the two independent dof in ǫ as

ǫ(1) = (0, 1, 0, 0) = (0,~ǫ1 )

(linear polarization) . (28)
ǫ(2) = (0, 0, 1, 0) = (0,~ǫ2 )

Alternatively, one may choose

ǫ(R) = − √12 (~ǫ1 + i ǫ2 ) = √1 (0, −1, −i, 0)

2
= (0,~ǫR )
(circular polarization) . (29)
ǫ(L) = √1 (~
ǫ − i ǫ2 ) = √1 (0, 1, −i, 0) = (0,~ǫL )
2 1 2

a) Show that,

ǫµ (λ) ǫ∗µ (λ) = −1 (transverse polarization) . (30)

More generally, show that

ǫµ (λ) ǫ∗µ (λ′ ) = −δλλ′ . (31)

b) Under a rotation about the z axis by the angle θ, the linear polarization vectors transform as

~ǫ′1 = ~ǫ1 cos θ + ~ǫ2 sin θ ,

~ǫ′2 = −~ǫ1 sin θ + ~ǫ2 cos θ . (32)

Show that, correspondingly,

~ǫ′R = ~ǫR e−iθ ,

~ǫ′L = ~ǫL eiθ . (33)

Compare with the effect of the unitary operator U (θ) = e−iθJ3 to conclude that ~ǫR and ~ǫL describe
a photon of helicity +1 and −1, respectively.
Notice that, to simplify this problem, we have suppressed the polarization’s explicit dependence on the
4-vector k.
6. (Halzen and Martin 6.12) Clearly, any 3-vector may be written in the ǫR , ǫL , k̂ ≡ ~k/k basis. Use
this fact to prove that
X
ǫ(λ)∗i ǫ(λ)j = δij − k̂i k̂j . (34)
λ=L,R

7. (Polarization sum for real photons.) The problem with (34) is that a specific basis has been chosen
to describe ~k, ǫR , and ǫL . Define
X
P µν (k) = ǫµ (k, λ) ǫν∗ (k, λ) . (35)
λ=1,2

For real polarization vectors, P µν is a second rank tensor. With the usual choice
kµ = (k, 0, 0, k) , ǫµ (k, 1) = (0, 1, 0, 0) , ǫµ (k, 2) = (0, 0, 1, 0) , (36)

P 11 = P 22 = 1, with all other entries zero. Can one find a covariant formula for P µν ? We start by
showing that such a covariant formula does not exist unless something else is introduced.

a) Since the polarizations have been summed over, a second rank symmetric tensor must be of the
form

P µν (k) = a gµν + b k µ kν , (37)

for some generic coefficients a and b. Take in turn (µ, ν) = (0, 0) and (µ, ν) = (3, 3) in the frame
where k = (k, 0, 0, k), in order to conclude that P µν cannot have the form in (37).

Let us choose one auxiliary timelike 4-vector η, such that

k · η 6= 0 , η2 = 1 , η · ǫ(k, λ) = 0 (for λ = 1, 2) . (38)

[For example, in the usual frame one could take η = (1, 0, 0, 0).]

b) We now try

P µν (k) = ag µν + bk µ kν + c(η µ kν + η ν kµ ) . (39)

Use (35) and (39) to calculate ǫµ (λ)P µν in two different ways. You might need (31). Conclude that
a = −1. Repeat for kµ P µν and for ηµ P µν . Thus, conclude that
X kµ kν ηµ kν + ην kµ
ǫµ (k, λ)ǫν∗ (k, λ) = −gµν − + . (40)
(k · η)2 k·η
λ=1,2

8. (Gauge invariance and processes with external photons.) Consider a sub-process (part of a bigger
picture) where a fermion f with charge eQf has initial momentum p, emits a real photon with momentum
k, acquiring a final momentum p′ .1 The amplitude for this process may be written as

M = Jµ ǫµ (k, λ) , (41)
where λ = 1, 2, and
Jµ = eQf ū(p′ )γµ u(p) . (42)
1
By itself, such a process cannot exist, as it violates conservation of 4-momentum (prove it!). However, it can be part of
a bigger process. For example, we might think of d → dγ as part of the hadronic decay ρ0 → π 0 γ.
 a) Due to the residual gauge invariance, it does not matter whether we describe the photon’s polar-
ization as ǫµ or as ǫµ + ak µ . Conclude that

kµ Jµ = 0 . (43)

b) Generalize the reasoning for many external photons

M = ǫµ (k1 ) ǫν (k2 ) · · · Tµν··· (k1 , k2 , · · · ) , (44)

and conclude that the amplitude must vanish if the polarization of an external photon is substituted
by its momentum.

c) Consider again (41) and (43). Imagine that one wishes to calculate the unpolarized cross section,
proportional to
X X
|M|2 = ǫµ ǫ∗ν J µ J ν∗ = (−gµν + kµ and/or kν terms)J µ J ν∗ . (45)
λ=1,2 λ=1,2

Use (40) and (43) to conclude that one can effectively do

X
ǫµ (k, λ)ǫν∗ (k, λ) → −gµν . (46)
λ=1,2

9. (Polarization sum for virtual photons.) Virtual photons are interpretations of Feynman diagrams
which, in turn, describe terms in a perturbative expansion of a scattering matrix. A virtual photon with
4-momentum k does not obey k2 = 0, and, thus, one needs all four possible polarizations in order to
describe it. Taking kµ = (k0 , 0, 0, |~k|), the polarizations are usually taken as
ǫµ (k, 1) = (0, 1, 0, 0)
(transverse polarizations)
ǫµ (k, 2) = (0, 0, 1, 0)
ǫµ (k, 3) = (0, 0, 0, 1) (longitudinal polarization)
µ
ǫ (k, 0) = (1, 0, 0, 0) (timelike polarization) (47)
The form of these polarizations in a generic frame can be obtained by an appropriate LT.

a) Show the obvious:

ǫ(k, 1)2 = ǫ(k, 2)2 = ǫ(k, 3)2 = −1 , ǫ(k, 0)2 = +1 . (48)

Moreover, show that on can write

ǫµ (k, λ) = δµλ . (49)

b) Generalize the results, proving that

′
ǫ(k, λ) · ǫ∗ (k, λ′ ) = ǫµ (k, λ) ǫ∗µ (k, λ′ ) = gλλ , (50)
X
gλλ ǫµ (k, λ) ǫν∗ (k, λ) = gµν . (51)
λ

10. (Polarization sum for massive vector bosons.) The E-L eqs. for massive vector fields read

( + M 2 )B µ − ∂ µ (∂ν B ν ) = 0 . (52)
 a) Contracting with ∂µ , show that ∂µ B µ = 0 always. (Recall that ∂µ Aµ = 0 is the Lorenz gauge choice
for the photon; here there is no alternative.)

b) Conclude that the equations may be written as

( + M 2 )B µ = 0 , (53)

and show that a plane wave solution is

B µ = ǫµ e−iq.x , (54)

where q 2 = M 2 .

c) Show that the polarizations must obey

q µ ǫµ = 0 . (55)

How many polarizations are independent?

d) Taking q = (q 0 , 0, 0, |~
q |), show that a suitable choice is

1
ǫµ (q, +) = − √ (0, 1, i, 0) ,
2
µ 1
ǫ (q, −) = √ (0, 1, −i, 0) ,
2
1

ǫµ (q, L) = q |, 0, 0, q 0 .
|~ (56)
M

e) Use the explicit forms (56), to show that the polarization sum yields
X qµqν
ǫµ ∗ (q, λ) ǫν (q, λ) = −gµν + . (57)
M2
λ

f ) One can show (57) in a much simpler way by going into the rest frame, where q µ = (M, 0, 0, 0)
and the polarizations consistent with ǫ.p = 0 and ǫ2 = −1 may be chosen as ǫµ (q, 1) = (0, 1, 0, 0),
ǫµ (q, 2) = (0, 0, 1, 0), and ǫµ (q, 3) = (0, 0, 0, 1). Show by explicit calculation that
3
X
ǫµ ∗ (q, λ) ǫν (q, λ) = diag(0, 1, 1, 1) ,
λ=1
qµqν
= diag(1, 0, 0, 0) , (58)
M2
thus recovering (57) .

Note: It is known from the theory of Green’s functions that one can construct a Green’s function for
some operator from its eigenvectors and eigenvalues. In the case of propagators, this means that one
can get the propagator for some particle by multiplying the scalar field propagator by an appropriate
polarization (or spin) completeness relation. Thus, Eq. (57) implies that the propagator for a massive
vector boson in momentum space is (in unitary gauge)
 
µν −i µν qµ qν
D (q) = 2 g − . (59)
q − M2 M2
Confirm that this coincides with the result found in Problem Set 3 for the propagator of a massive gauge
boson.