Lecture 8 8.

324 Relativistic Quantum Field Theory II Fall 2010

8.324 Relativistic Quantum Field Theory II

MIT OpenCourseWare Lecture Notes
Hong Liu, Fall 2010

Lecture 8
In the last lecture, we showed, for a general interacting theory:
ˆ ∞
−iZ i
GF (p) = 2 + dµ2 σ(µ2 ) 2 , (1)
p + m − iϵ
2
4m2 p + µ2 − iϵ

where the first term is the contribution from single-particle states, and the second term is the contribution from
multi-particle states. From this, we have that
∑
Im(iGF (p)) = πδ(p2 + m2i )Zi + πσ(−p2 ). (2)
i

This is the spectral function, picking out the physical on-shell states. There is one more sum rule we wish to observe.
Begin with the canonical quantization:
[ ]
ϕ̇(t, ⃗x), ϕ(t, ⃗y ) = −iδ(⃗x − ⃗y ). (3)

As this operator is just a complex number, we can equate it with its expectation value:
[ ]
⟨0| ϕ̇(t, ⃗x), ϕ(t, ⃗y ) |0⟩ = − iδ(⃗x − ⃗y )
= ∂t ⟨0| ϕ(t, ⃗x)ϕ(t′ , ⃗y ) |0⟩|t′ →t − ∂t ⟨0| ϕ(t′ , ⃗y )ϕ(t, ⃗x) |0⟩|t′ →t
= ∂t G+ (x − y)|t′ →t − ∂t G+ (y − x)|t′ →t
ˆ ∞ [ ]
(0) (0)
= dµ2 ρ(µ2 ) ∂t G+ (x − y; µ2 ) ′ − ∂t G+ (y − x; µ2 )
t →t t′ →t
ˆ ∞
0
(0)
=2 dµ2 ρ(µ2 ) ∂t G+ (x − y; µ2 ) ′ .
0 t →t

Recall, in the free theory, we have that

(0) i
∂t G+ (x − y) = − δ(⃗x − ⃗y ), (4)
t′ →t 2
and so, we have that ˆ ∞
1= dµ2 ρ(µ2 ). (5)
0
∑
Because ρ(µ2 ) = σ(µ2 ) + i Zi δ(µ2 − mi2 ), where both terms are greater than or equal to zero for all values of µ2 ,
we have that ˆ ∞
dµ2 σ(µ2 ) < 1, Zi < 1. (6)
4m2
In reality, the above argument may not always hold due to possible ultraviolet divergences. The same discussion
can be generalized to a spinor, as seen in the problem set, and to a vector, as we will discuss later.

2.2: AN EXPLICIT EXAMPLE

Consider a scalar Lagrangian with a ϕ3 interaction:
1 1 1
L = − ∂µ ϕ∂ µ ϕ − m20 ϕ2 + gϕ3 . (7)
2 2 6

1
Lecture 8 8.324 Relativistic Quantum Field Theory II Fall 2010

This leads to the Feynman rules:

−i
= ,
p p2 + m2 − iϵ

= ig, (8)

and by dimensional analysis, we find

d−2
[ϕ] = ,
2
[x] = − 1,
d
[g] =3 − ,
2
where we work in a general space-time dimension d. We note that g is dimensionless in d = 6.

GF (p) = + +

+ + + ··· (9)

Note that we do not consider diagrams involving tadpoles:

⟨0| ϕ |0⟩ ≡ ϕ0 = + + ··· (10)

as we can always shift the field definition to ϕ = ϕ0 + ϕ,˜ so that ⟨0| ϕ̃ |0⟩ = 0, and the sum of the tadpole sub­
diagrams gives zero. Now, we define 1PI, or one-particle irreducible diagrams, as the diagrams which cannot be
separated into two disconnected parts by cutting one propagator. We denote the sum of 1PI diagrams by

i�(p) =
p

= + +

+ + ··· , (11)

and so, we have that

GF (p) = + + + ··· . (12)

2
Lecture 8 8.324 Relativistic Quantum Field Theory II Fall 2010

That is,
(0) (0) (0) (0) (0) (0)
GF (p) =GF + GF i�GF + GF i�GF i�GF + · · ·
(0) 1
=GF (0)
1 − i�GF
1
=( )−1
(0)
GF − i�
−i
=
p2 + m20 − �(p) − iϵ

where �(p) is the self-energy. We note that � is, in fact, a function of p2 only, by Lorentz symmetry. Before
evaluating �(p2 ) explicitly to lowest order, we make two remarks:
1. The physical mass, the pole of GF (p), is given by p2 + m20 − �(p2 ) = 0. That is, the physical mass m2
satisfies
m2 − m20 + �(−m2 ) = 0 (13)

2. The field renormalization Z, the residue of the pole, is given by expanding around the pole to lowest
order:
1
iGF (p)|p2 ≈−m2 =
p2 + m2 − �′ (−m2 )(p2 + m2 ) − iϵ
1 1
= ,
p2 + m2 − iϵ 1 − d�
dp2
p2 =−m2

and so Z −1 = 1 − d�
dp2 .
p2 =−m2

We now proceed to evaluate �(p2 ) to the lowest order in g:

q+p

i�(p2 ) = . (14)
p
q
Using our Feynman rules, we have
ˆ
1 dd q (0) (0)
i�(p2 ) = (ig)2 G (q)GF (q + p)
2 (2π)d F
ˆ
g2 dd q 1
= .
2 (2π)d (q 2 + m20 − iϵ)((q + p)2 + m20 − iϵ)

We now evaluate the integral explicitly, using a series of tricks. Firstly, we note the identity, due to Feynman:
ˆ 1
1
= dx1 · · · dxn δ(x1 + · · · + xn − 1)(n − 1)!(x1 a1 + · · · + xn an )−n . (15)
a1 · · · an 0

Thus, our integrand can be rewritten as

ˆ 1
1 1
= dx
(q + m0 − iϵ)((q + p)2 + m20 − iϵ)
2 2
0 [x((q + p)2 + m20 )
2
+ (1 − x)(q 2 + m02 )]
ˆ 1
1
= dx 2
0 [(q + xp)2 + D]

with D = m20 +x(1−x)p2 . Next, we perform a Wick rotation. In Lorentzian signature, the integral is not convenient
to evaluate, because of the poles near the integration paths. Thus, we rotate the q0 contour to the ´ imaginary
´ axis
along the direction shown in figure 1. So, we let q0 = iqd , and therefore q 2 = q12 +. . .+qd2 = qE
2
, and dd q = i dd qE .

3
Lecture 8 8.324 Relativistic Quantum Field Theory II Fall 2010

Im(q) 0

Re(q)
0

Figure 1: Illustration of the Wick rotation of the variable q0 .

Combining our two results, we have the self-energy in the form

ˆ ˆ d
ig 2 1 d qE 1
i�(p2 ) = dx 2 + D)2 . (16)
2 0 (2π)d (qE

We now proceed to the final evaluation. We observe that (16) is convergent only for d < 4. We may evaluate it for
d < 4, then analytically continue its value for d ≥ 4, treating d as a complex variable. We make use of the general
formula ˆ d
d qE (qE )a �(b − a − d2 )�(a + d2 ) −(b−a− d )
= d D 2 . (17)
(2π)d (qE + D)b (4π) 2 �(b)�( d ) 2
In this case, we have b = 2, a = 0. And so,
ˆ
g 2 �(2 − d2 ) 1
1
�(p2 ) = dx . (18)
2 (4π) d2 0 (m20 + x(1 − x)p2 )2− 2
d

Example 1: d=3

ˆ 1
g2 1
�(p2 ) = dx 1 . (19)
16π 0 (m20 + x(1 − x)p2 ) 2
First, we consider the physical mass to O(g 2 ). We wish to find the solution to (13). Note that the self-energy is of
order g 2 , and so the solution is given to O(g 2 ) by

m2 =m20 − �(−m20 )
ˆ 1
g2 1 1
=m0 −
2
dx 1
16π m0 0 (1 − x(1 − x)) 2
g2
=m20 − log 3.
16πm0
The second term is the mass renormalization due to the interaction, to lowest order. Now, we consider the field
renormalization, Z −1 = 1 − dp
d�
2 , and note that again, as �(p2 ) is of order g 2 , to lowest order we have
p2 =−m2

d�
Z −1 =1 −
dp2 p2 =−m20
ˆ
g 2
1 1 1
x(1 − x)
=1 − (− ) dx
16π 2 m30 0 (1 − x(1 − x)) 2
3

1
= 0.23g 2
< 1,
1 + 32�m3
0

4
Lecture 8 8.324 Relativistic Quantum Field Theory II Fall 2010

where the integral in dx has been evaluated explicitly. Finally, we note that for −p2 ≥ 4m20 , we have that m20 +
x(1 − x)p2 is smaller than zero for a range of x between 0 and 1. Therefore, �(p2 ) becomes complex. It is convenient
to consider ˆ 1
g2 1
�(s) = dx 1 (20)
16π 0 (m20 − x(1 − x)s) 2
as a function of a complex variable s, with �(p2 ) = �(s = −p2 + iϵ). �(s) has a branch point at s = 4m20 . This is
precisely the multiple-particle cut predicted in the general formalism last lecture. Note that m2 = m20 + O(g 2 ). We
can now understand the physical interpretation of this result:

i�(p2 ) = (21)

to lowest order, and when −p2 > 4m20 , both the intermediate particles can simultaneously go on-shell. Im(iGF ) =
πσ(−p2 ) becomes non-zero and Im(−p2 ) is just the Feynman diagram evaluated with both the intermediate particles
on shell, giving a factor of δ(p2 + m2 ) for each propagator.

5
MIT OpenCourseWare
http://ocw.mit.edu

8.324 Relativistic Quantum Field Theory II

Fall 2010

For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.