Quantum Field Theory

Part I: Motivation and Foundation

Simon Crase (compiler)
[email protected]
November 12, 2023

Abstract
This document contains derivations of equations from [?, Part I: Mo-
tivation and Foundation].

Contents
1.1 Path Integral Formulation 1

1.2 From Field to Particle to Force 8

1.3 Coulomb and Newton 9

1.4 Feynman Diagrams 12

Bibliography 12

1.1 Path Integral Formulation

Theorem 1 (Some useful integrals).
Z ∞ √
(1)
1 2
dxe− 2 x = 2π
−∞
Z ∞  1
2π 2
dxe − 12 ax2
= (2)
−∞ a
Z ∞   12
2π J2
(3)
a 2
dx e− 2 x +Jx
= e 2a
−∞ a
Z ∞   12
2π J2
(4)
a 2
dx e− 2 x +iJx
= e− 2a
−∞ a

1
Proof.
Z ∞ Z ∞ Z ∞ Z ∞
1 2
2 2
1 1
+y 2 )
dxe− 2 x dye− 2 y dxdye− 2 (x


=
−∞ −∞ −∞ −∞
Z Z ∞
2π
−r 2
= dθ dr r e 2 , substituting x = r cos θ,y = r sin θ
0 0
Z ∞
−r2
= − 2π due−u , substituting u =
0 2
∞
= − 2πe−u
0
=2π

Taking square roots of both side, we get (1).

Z ∞ Z ∞
1 √
dx0 e− 2 x , substituting x0 = ax
1 2 1 02
dxe− 2 ax = √
−∞ a −∞
  12
2π
= , from (1), which gives (2)
a

Z ∞ Z ∞ 

2 
2
x2 −2
, completing the square
a 2 a J J a J
dx e− 2 x +Jx
= dx e− 2 a x+ a e2 a

−∞ −∞
Z ∞ 2
J

J2
, now substitute x0 = x −
a
x− J
=e 2a dxe− 2 a

−∞ a
Z ∞
J2 a 02
=e 2a dx0 e− 2 x
−∞

Substituting (2) gives (3). Replacing J by iJ gives (4).

Theorem 2 (Feynman Path Integral–Exercise I.2.1 - eq(5)). If the Hamiltonian

is given by

p̂2
H= + V (q̂), and the corresponding Lagrangian is (5)
2m
p̂2
L= − V (q̂), then (6)
2m
Z
, where (7)
RT
hqF |H|qI i = Dq(t)ei 0
dtL

N −1 Z
−im
N2
Z
(8)
Y

Dq(t) , lim dqj
N →∞ 2πδt j=1

Proof. For some N, define δt = T /N and qj = q(jδt).

2
 N −1 Z −1
Y
NY
hqF |H|qI i = dqj hqj+1 |e−iHT |qj i
j=1 i=0
N −1 Z N −1
p̂2


|qj i , from (5)
Y Y −i 2m +V (q̂) δt


= dqj hqj+1 |e
j=1 i=0
N −1 Z −1

NY p̂2
hqj+1 |e−i 2m δt e−iV (q̂)δt |qj i , we’ll expand in eigenvectors of q̂
Y
= dqj
j=1 i=0
N −1 Z N −1 Z
p̂2
dq hqj+1 |e−i 2m δt |qi hq|e−iV (q̂)δt |qj i , since |qi hq| = I
Y
Y
= dqj
j=1 i=0
(9)

We will evaluate the integrand in two steps, starting with the potential energy.

hq|e−iV (q̂)δt |qj i = heiV (q̂)δt q|qj i

= heiV (q)δt q|qj i , since |qi is an eigenvector, with eigenvalue q
= heiV (q)δt q|qj i
=e−iV (q)δt hq|qj i (10)

Now for the kinetic energy.

Z
p̂2 dp p̂2
hqj+1 |e−i 2m δt |qi = hqj+1 |e−i 2m δt |pi hp|qi
2π
Z
dp −i p2 δt
= e 2m hqj+1 |pi hp|qi , since |pi is an eigenvector of p̂
2π
Z
dp −i p2 δt ipqj+1 −ipq
= e 2m e e
2π

Using (4) with a = iδt

m and J = qj+1 − q:
Z
p̂2 dp −iδt p2 ip[qj+1 −q]
hqj+1 |e−i 2m δt |qi = e 2m e
2π
1 2πm
12 − m(qj+1 −q)2
= e 2iδt
2π iδt
−im
12 iδt m(qj+12−q)2
= e 2δt (11)
2πδt

3
Substituting (10) and (11) in (9):
N −1 Z −1
−im
N2
NY −im
12 m(qj+1 −q)2
Y Z
hqF |H|qI i = dqj dqeiδt 2δt2 e−iV (q)δt hq|qj i δ(q − qj )
2πδt j=1 i=0
2πδt | {z }
= m2
 
m(qj+1 −qj )2
N −1 Z −1
−im
N2 Y
NY iδt 2δt2
−V (qj )
= dqj e
2πδt j=1 j=0
 
m(qj+1 −qj )2
N −1 Z PN −1
−V (qj )
−im
N2
i i=0 δt 2δt2
(12)
Y
= dqj e
2πδt j=1

Now, as N → ∞
N
X −1 Z
δt → dt
i=0
m(qj+1 − qj )2 mq̇
→ , so, substituting (6) and (8), (12) becomes
2δt2 Z
2
hqF |H|qI i → Dq(t)ei 0 dtL i.e. (7)
RT

Theorem 3.
R∞ 1 2 Qn
dxe− 2 ax x2n


−∞ k=1 2k − 1
R∞ 1 =
−∞
dxe− 2 ax
2
an

Proof. The proof by mathematical induction starts by defining the proposition

P (n):
R ∞ 1 2 Qn
dxe− 2 ax x2n


−∞ k=1 2k − 1
P (n) ≡ R ∞ 1 =
−∞
dxe− 2 ax
2
an

Lemma 4. P(0) is true.

1 2
dxe− 2 ax
R∞
Proof. P (0) reduces to −∞
1 2 =1
dxe− 2 ax
R∞
−∞

Lemma 5.
Z ∞ Z ∞
d 1 2 ∂  − 1 αx2 2n 
∀α > 0, dxe− 2 αx x2n = dx e 2 x
dα −∞ −∞ ∂α

4
Proof.
Z ∞ Z ∞
1 2 1 2 1 2
∀a > 0, ∀α0 ∈ (0, α) dxe− 2 αx x2n = dxe− 2 α0 x e− 2 (α−α0 )x x2n
a a

Now define M (x) =e − 12 α0 x2

We can choose a large enough that e− 2 (α−α0 )x x2n is monotone decreasing for
1 2

x > a. Clearly
1 2 1 2
e− 2 α0 x e− 2 (α−α0 )x x2n ∈ C
1 2
e− 2 α0 x ∈ C
e− 2 α0 x e− 2 (α−α0 )x x2n <M (x) ∀x > a and
1 2 1 2

Z ∞
dxM (x) <∞
a

for α in closed interval [A,B] that includes α0 ; the Weierstrass M-Test [?, Chap-
ter 10, 6.1] shows that the integral converges uniformly for α ∈ [A, B]. Hence
we can differentiate under the integral sign [?, Chapter 10, 8.3]. Now
Z ∞ Z ∞ Z a
− 12 αx2 2n − 12 αx2 2n 1 2
dxe x =2 dxe x + dxe− 2 αx x2n
−∞ a −a

Noting that e− 2 αx x2n ∈ C 1 for x in [-a,a], we can also differentiate the second
1 2

integral.

Lemma 6. P (0) =⇒ P (1)

Proof. Since (2) converges uniformly, Lemma 5 allows us to differentiate under
the integral sign[?].
Z ∞  1
d 1 2 d 2π 2
−2 dxe− 2 ax = − 2
da −∞ da a
Z ∞   12
1 2 2π
dxe− 2 ax x2 = 3
−∞ a
  12
2π 1
=
a a
Z ∞
1 2 1
= dxe− 2 ax
−∞ a

Lemma 7. P (n) ∧ P (1) =⇒ P (n + 1)

5
Proof.

P (n) =⇒
Z ∞ n Z ∞
− 12 ax2 2n
Y 1 2 1
dxe− 2 ax


dxe x = 2k − 1
−∞ −∞ an
k=1
∞ n

d ∞
Z Z
d 1 2 Y 1 2 1 
−2 dxe− 2 ax x2n = − 2 2k − 1 dxe− 2 ax n
da −∞ da −∞ a
k=1
Z ∞ n  Z ∞ Z ∞ 
1 2 Y 1 2 1 1 1 2 n
dxe− 2 ax x2(n+1) dxe− 2 ax (− x2 ) n − dxe− 2 ax n+1


=−2 2k − 1
−∞ −∞ 2 a −∞ a
k=1
n  Z ∞ Z ∞ 
Y 1 2 1 1 2 n
dxe− 2 ax x2 n + 2 dxe− 2 ax n+1


= 2k − 1
−∞ a −∞ a
k=1 | {z }
Now we apply P (1)
n Z ∞ Z ∞ 
Y 1 2 1 1 1 2 2n
dxe− 2 ax dxe− 2 ax


= 2k − 1 +
−∞ a an −∞ an+1
k=1
n Z ∞
Y
2n + 1 1 2
= 2k − 1 n+1 dxe− 2 ax
a −∞
k=1
n+1 Z ∞
Y
1 1 2
= 2k − 1 n+1 dxe− 2 ax
a −∞
k=1
≡P (n + 1)

Summarizing:

Lemma 4 =⇒ P (0)
P (0) ∧ Lemma 6 =⇒ P (1)
P (0) ∧ P (1) ∧ P (n) ∧ Lemma 7 =⇒ P (n + 1)

Theorem 8. If A is a symmetric matrix

∞ 1
(2π)N 2 1 J·A
Z 
(13)
1 ~ ~ −1 J~
dx1 dx2 ...dxN e− 2 ~xA~x+J·~x = e2
−∞ |A|

6
Proof. Since A is symmetric, there exists an orthogonal matrix O such that
A = OT DO, where D is a diagonal matrix–[?]. Define
X
yi = Oij xj
j
Z ∞  N Z ∞
− 12 ~ ~x 1 1 T ~
dx1 dx2 ...dxN e xA~
x+J·~
= dy1 dy2 ...dyN e− 2 ~xO DO~x+J·~x
−∞ |O| −∞
Z ∞
1 T ~
= dy1 dy2 ...dyN e− 2 (~xO )D(O~x)+J·~x
−∞
Z ∞
1 ~
= dy1 dy2 ...dyN e− 2 ~yD~y+(OJ)·~y
−∞
N Z ∞
Y 1 2
= dyi e− 2 Dii yi +(OJ)i yi
i=1 −∞
N   12

2π [(OJ)i ]2
(14)
Y
= e 2Dii

i=1
Dii

But
N 
(2π)N

Y 2π
=
i=1
Dii |D|
(2π)N
= , since O is orthogonal (15)
|A|

7
And
N
[(OJ)i ]2 [(OJ)i ]2
Y
PN

e 2Dii
=e i=1 2Dii

i=1
N N
X [(OJ)i ]2
1 X
= (OJ)i (D−1 )ii (OJ)i
i=1
2Dii 2 i=1
N
1 X
= Oik Jk (D−1 )ij Ojl Jl
2
i,j,k,l=1
N
1 X
= T
Jk Oki (D−1 )ij Ojl Jl
2
i,j,k,l=1
N N
 X 
1 X
= Jk T
Oki (D−1 )ij Ojl Jl
2
k,l=1 k,l=1

N
1 X
Jk A−1 kl Jl


=
2
k,l=1
1
= JA−1 J (16)
2
Substituting (15) and (16) in (14), we obtain (13).

1.2 From Field to Particle to Force

Z Z
1
W (J) = − d4 xd4 yJ(x)D(x − y)J(y)
2
d4 k eik(x−y)
Z
D(x − y) =
(2π) k − m2 + i
4 2

J(x) =J1 (x) + J2 (x), where

Ja (x) =δ (3) (~x − x~a )

8
Considering only the cross terms

eik(x−y)
Z Z Z
1 1
W (J) = − 4
2 d4 xd4 yd4 kδ (3) (~x − x~1 ) 2 δ (3) (~y − x~2 ))
(2π) 2 k − m2 + i


0 0 0 ~
eik (x −y )−ik·(~x−~y) (3)
Z Z Z Z Z Z
1 0 0 0 ~ (3)
=− dx d~x dy d~
y dk dkδ (~x − x
~ 1 ) δ (~y − x~2 ))
(2π)4 k 2 − m2 + i
0 0 0 ~
eik (x −y )−ik·(~x−x~2 )
Z Z Z Z Z
1 0 0 0 ~ (3)
=− dx d~
x dy dk dkδ (~
x − x~1 )
(2π)4 k 2 − m2 + i
0 0 0 ~
eik (x −y )−ik·(x~1 −x~2 )
Z Z Z Z
1
=− 4
dx0 dy 0 dk 0 d~k
(2π) k 2 − m2 + i
i~ x−~
dk 0 ik0 (x0 −y0 ) k·(~ y)
Z Z Z Z
1 0 0 ~k e
=− dx dy e d
(2π)3 2π k 2 − m2 + i
| {z }
δ(x0 − y 0 )
Z Z i~ x−~
k·(~ y)
1 e
=− dx0 d~k
(2π)3 k 2 − m2 + i

1.3 Coulomb and Newton

The Lagrangian Density is given by:
1 1
L = − Fµν F µν + m2 Aµ Aµ + Aµ J µ , where
4 2
Fµν =∂µ Aν − ∂ν Aµ and
∂µ J ν =0. We define the action (17)
Z
S(A) = d4 L
Z  
4 1 µν 1 2 µ µ
= d x − Fµν F + m Aµ A + Aµ J
4 2

Expanding the first term in the Lagrangian

Fµν F µν =g µρ g νσ Fµν Fρσ

=g µρ g νσ ∂µ Aν − ∂ν Aµ ∂ρ Aσ − ∂σ Aρ

=g µρ g νσ ∂µ Aν ∂ρ Aσ − ∂µ Aν ∂σ Aρ − ∂ν Aµ ∂ρ Aσ + ∂ν Aµ ∂σ Aρ
 

=g µρ g νσ ∂µ Aν ∂ρ Aσ − g µρ g νσ ∂µ Aν ∂σ Aρ − g µρ g νσ ∂ν Aµ ∂ρ Aσ + g µρ g νσ ∂ν Aµ ∂σ Aρ
| {z } | {z }
substitute: µνρσ → νµσρ µνρσ → νµσρ
µρ νσ µρ νσ νσ µρ νσ µρ
=g g ∂µ Aν ∂ρ Aσ − g g ∂µ Aν ∂σ Aρ − g g ∂µ Aν ∂σ Aρ + g g ∂µ Aν ∂ρ Assigma
=2 g µρ g νσ ∂µ Aν ∂ρ Aσ − g µρ g νσ ∂µ Aν ∂σ Aρ
 

9
So the first term in the action is
Z Z
1 2
d4 xFµν F µν = − d4 x g µρ g νσ ∂µ Aν ∂ρ Aσ − g µρ g νσ ∂µ Aν ∂σ Aρ . Integrating by parts gives
 
−
4 4
Z
1
d4 x g µρ g νσ Aν ∂µ ∂ρ Aσ − g µρ g νσ Aν ∂µ ∂σ Aρ
 
=
2
Z
1
d4 x g νσ Aν ∂ 2 Aσ − Aν ∂ ρ ∂ ν Aρ
 
=
2
Z
1
d4 xAν g νσ ∂ 2 − ∂ σ ∂ ν Aσ
 
=
2
So the action becomes
Z Z  
4 4 1  νσ 2 σ ν
 1 2 µ µ
d L = d x Aν g ∂ − ∂ ∂ Aσ + m Aµ A + Aµ J
2 2
Z  
4 1  νσ 2 2 σ ν
 µ
= d x Aν g (∂ + m ) − ∂ ∂ Aσ + Aµ J
2

νλ
Dµν D−1 =δµλ

From Lecture. How do we get a repulsive force? This is tied up with spin. If
we want particles fro field to have spin 1, we need a vector–Aµ . We need to
reduce Aµ to 3 degrees of freedom (polarization), and (17) is the only Lorentz
invariant way to do this. We will start with the physics–Figure 1.

Figure 1: Electromagnetic Field with source and sink

Jφ →Jµ Aµ
3
1 1
(18)
X
2 2
→ 2 (a) (a)
µ (k)ν (k)
k +m k + m2 a=1
|{z}
3 polarizations
| {z }
A Lorentz tensor, −Gµν say

We will evaluate in the rest frame of the particle, where k = (m, 0, 0, 0), and
polarization is the 3 directions in Cartesian space.

(1)
µ (k) =(0, 1, 0, 0)

(2)
µ (k) =(0, 0, 1, 0)

(3)
µ (k) =(0, 0, 0, 1)
k µ µ =0, since k = (m, 0, 0, 0) (19)

10
(19) is Lorentz Invariant! c.f. ∂µ Aµ = 0. We will establish the tensor Gµν
in (18) using symmetry and Lorentz invariance. There are only two things to
construct Gµν from, kµ and gµν .
−Gµν =Akµ kν + Bgµν
Gµν k µ =0 from (19), whence
Akµ kν + Bgµν k µ =0

k 2 +B kν =0


A |{z}
= m2
B
A=−
m2
kµ kν

−Gµν = − gµν − . Normalization fixed by rest frame.
m2
µ ν k k
1 −gµν + m 2
→
k 2 + m2 k 2 + m2
Imagine two stationary charges: then J i = 0
kµ kν
d4 k µ ∗ −gµν + m
Z
2
W (J) = J (k) 2 2
J ν (k)
(2π)4 k +m
∂µ J µ =0 implies
kµ J µ =0
d4 k 0 ∗ −g00
Z
W (J) = J (k) 2 J 0 (k)
(2π)4 k + m2
d4 k 0 ∗ −1
Z
= J (k) 2 J 0 (k)–Like charges repel
(2π)4 k + m2
Spin 2 has two indices hµν . Two sign switches cancel!
Even spin can exchange lumps, but not odd.
Theorem 9.
δ 2 + m2 Aµ =0 and (20)

∂µ A =0 are equivalent to
µ
(21)
µν 2 µ ν 2 µ
(22)


g ∂ −∂ ∂ Aν + m A =0
Proof: (20) ∧ (21) =⇒ (22). From (21)
∂ ν Aν =0, whence
∂ µ ∂ ν Aν =0. Now (20) can be rearranged (23)
2 µν 2 µ


∂ g Aν + m A =0
g ∂ Aν + m2 Aµ =0. Now, using (23)
µν 2

g µν ∂ 2 Aν − ∂ µ ∂ ν Aν + m2 Aµ =0, which rearranges to (22)

11
(22) =⇒ (20) ∧ (21). From (22)

m2 Aµ = ∂ µ ∂ ν − g µν ∂ 2 Aν

m2 ∂µ Aµ =∂µ ∂ µ ∂ ν − g µν ∂ 2 Aν

=∂µ ∂ µ ∂ ν Aν − g µν ∂µ ∂ 2 Aν
=∂ 2 ∂ ν Aν − ∂ ν ∂ 2 Aν
=0, which is (21). Now (22) becomes
g µν 2
∂ ∂ µ
Aν −  ∂ ν
Aν + m A =0, which is (20)
2 µ

1.4 Feynman Diagrams

Z  
d4 x 1 2 2 2 λ 4
R
Z(J, λ) = Dφ e i 2 [(∂φ) −m φ ]− 4! φ +Jφ

Z   λ 4
d4 x 1 2 2 2
e−i 4! φ
R
= Dφ ei 2 [(∂φ) −m φ ]+Jφ

The functional derivative is defined in [?]:

F (ρ + φ) − f [ρ]
Z
δF (x)
φ(x)dx , lim
δρ →0 
 
dF (ρ + φ)
=
dρ =0

Theorem 10.
4
− iλ δ
R
dx (δJ) 4
Z(J, λ) =e 4!
Z(J, 0)

Proof. After [?]...

12