You are on page 1of 15

VILNIAUS UNIVERSITETAS

TEORINĖS FIZIKOS KATEDRA

Thomas Gajdosik

Using the Pathintegral


to derive the Feynman Rules
for the ABC-Theory

notes for the lecture Modernioji Teorinė Fizika

Vilnius 2012
Leidini˛ T. Gajdosik. ”Using the Pathintegral to derive the Feynman Rules for the ABC-
Theory” apsvarstė ir rekomendavo spaudai VU teorinės fizikos katedra (2012-06-25, pro-
tokolas Nr. 07-2012).

Recenzavo: Prof. dr. Egidijus Norvaišas

Prof. dr. Darius Abramavičius

Contents
1 Introduction 3

2 Lagrangian and action 3

3 Pathintegral 5

4 Generating Functionals 6

5 ABC-Theory example 10

References 15
MTF: Feynman Rules for the ABC-Theory October 3, 2012 3

1 Introduction
This note is written mainly for students of the course Modernoji Teorinė Fizika at VU. The
goal of these notes is to illustrate how the methods of Quantum Field Theory (QFT) are
used to obtain the Feynman rules for the ABC-Theory, which is in a certain way the sim-
plest possible toy model in 3+1 dimensions. The ABC-Theory describes three interacting
scalar fields, that have only a singe intereaction. It was introduced for teaching purposes
by David Griffiths [1] and then extensively used by I. J. R. Aitchison and A. J. G. Hey [2]
to explain preturbative expansions, regularization and renormalisation in QFT.

2 Lagrangian and action


The simplest quantity in space is a point without any structure, i.e. the real scalar. So let
us first consider the Lagrangian of a single real scalar field φ(x) without interactions:

Lφ = 1
2
(∂ µ φ(x)) (∂µ φ(x)) − 21 m2φ φ2 (x) . (1)

The action is the integral over the Lagrangian. So the action Sφ of the free, i.e. non
interacting, real scalar field φ = φ(x), can be written as
Z Z
Sφ = d x Lφ = d4 x 12 (∂ µ φ)(∂µ φ) − 21 m2φ φ2
4

Z
= − 2 d4 x φ(∂ 2 + m2φ )φ + 21 [φ(∂µ φ)]at x→∞ ,
1
(2)

where we used partial integration. The last piece can usually be set to zero, as fields that
do not vanish rapidly enough at infinity are not localized enough to be treated as particles.
Using the Fourier transformation on the field φ

d4 k −ikx
Z Z
φ(x) = e φ̃(k) and φ̃(k) = d4 xeikx φ(x) (3)
(2π)4

with the understanding, that kx = kµ xµ , we can rewrite the action

d4 k d4 p 1 µ −ikx
Z Z Z
4
Sφ = dx (∂ e φ̃(k))(∂µ e−ipx φ̃(p)) − 12 m2φ e−ikx φ̃(k)e−ipx φ̃(p)
(2π)4 (2π)4 2
d4 k d4 p
Z Z Z
4
1
= 2 dx [(−ik µ )(−ipµ ) − m2φ ]e−i(k+p)x φ̃(k)φ̃(p) . (4)
(2π)4 (2π)4

The integral over x gives the Dirac delta function δ(k + p), which can be used to perform
the k-integral by replacing k with (−p):

d4 k d4 p
Z Z
Sφ = − 2 1
φ̃(k)[k µ pµ + m2φ ](2π)4 δ(k + p)φ̃(p)
(2π)4 (2π)4
d4 p
Z
= −2 1
φ̃(−p)(−p2 + m2φ )φ̃(p) . (5)
(2π)4
MTF: Feynman Rules for the ABC-Theory October 3, 2012 4

When Fourier transforming back to φ(x)


d4 p i(−p)x
Z Z Z
4 4
1
Sφ = − 2 d x dy e φ(x)(−p2 + m2φ )eipy φ(y)
(2π)4
d4 p −ip(x−y)
Z Z Z 
1 4 4 2 2
= −2 d x d y φ(x) e (−p + mφ ) φ(y) . (6)
(2π)4
we could go back to the same for where we started, but we can also use the opportunity
to define the Feynman propagator
d4 p
Z
i
DF (x, y) := DF (x − y) = e−ip(x−y) , (7)
(2π)4 p2 − m2φ + iǫ
where the infinitesimal iǫ gives the prescription, how we should calculate the integral, when
p2 → m2φ . Its inverse DF−1 (x, y) can be found from
Z Z
4
δ(x − z) = d y DF (x, y)DF (y, z) = d4 y DF−1 (x, y)DF (y, z)
−1
(8)

with the ansatz


d4 q −iq(x−y) ˜
Z
DF−1 (x, y) = e d(q) (9)
(2π)4
and the help of Fourier transformations
Z
eikz
= d4 x eikx δ(x − z)
d4 p d4 q −iq(y−z) ˜
Z Z Z Z
4 ikx 4 i −ip(x−y)
= d xe dy e e d(q)
(2π)4 p2 − m2φ + iǫ (2π)4
d4 p d4 q
Z Z Z Z
4 i ˜
= 4 4
d xe i(k−p)x
d4 y ei(p−q)y 2 2
eiqz d(q)
(2π) (2π) p − mφ + iǫ
d4 p d4 q
Z Z
i ˜
= (2π)4 δ(k − p)(2π)4 δ(p − q) 2 eiqz d(q)
(2π) 4 (2π) 4 p − m2φ + iǫ
˜
ieikz d(k)
= 2 , (10)
k − m2φ + iǫ
˜ = −i(k 2 − m2 + iǫ), where the iǫ is no longer necessary, since the function
or simply d(k) φ
does not have any singularity. So we can write
d4 q −iqz ˜ d4 q −iqz
Z Z
−1
DF (z) = e d(q) = i e (−k 2 + m2φ ) . (11)
(2π)4 (2π)4
Here is also the best place to note, that DF (x, y) and DF−1 (x, y) are symmetric under the
exchange of x and y:
Z +∞ D Z −∞ D
d p ie−ip(y−x) D d k ieik(y−x)
DF (y, x) = 4 2 2
= (−1) 4 2 2
−∞ (2π) p − mφ + iǫ +∞ (2π) (−k) − mφ + iǫ
Z +∞ D
2D d k ie−ik(x−y)
= (−1) 4 2 2
= (−1)2D DF (x, y) , (12)
−∞ (2π) k − mφ + iǫ
MTF: Feynman Rules for the ABC-Theory October 3, 2012 5

for all integer dimensions D.


For the action we can now write
d4 p −ip(x−y)
Z Z  Z 
1 4 4 2 2
Sφ = 2 d x d y φ(x)i i e (−p + mφ ) φ(y)
(2π)4
Z Z
= 2 d x d4 y φ(x)iDF−1 (x − y)φ(y) = 21 φx (iDF−1 )xy φy ,
1 4
(13)

where the last rewriting of the action emphasizes that it is bilinear in the scalar field.

3 Pathintegral
The pathintegral describes the vacuum to vacuum transition by summing over all quantum
mechanically possible trajectories in the presence of a source function J(x). This generating
functional can now be written as
Z Z Z
1
Z(J) = Dφ exp{iSφ + i d x J(x)φ(x)} = Dφ e− 2 φx (DF )xy φy +iJx φx ,
4 −1
(14)

which is a Gaussian integral of the form


Z +∞ r
−bx2 +cx π c2
e dx = e 4b (15)
−∞ b
1
with b = 2
and c = iJ. So
1 i
Z(J) = N × e− 2 Jx (DF )xy Jy = N × e 2 Ja (iDF )ab Jb , (16)
with an undetermined normalization factor N. When moving to the logarithm
W (J) := −i ln Z(J) = −i(ln N + 2i Ja (iDF )ab Jb ) (17)
we find that the propagator is exactly the second functional derivative of W :
δ δ
W (J)
δJx δJy
δ δ δ
= (−i ln N + 21 Ja (iDF )ab Jb ) = (0 + 21 δay (iDF )ab Jb + 12 Ja (iDF )ab δby )
δJx δJy δJx
1 x x 1
= 2 ((iDF )yb δb + δ (iDF )ay ) = 2 ((iDF )yx + (iDF )xy ) = (iDF )xy (18)
So we get an exact description of a non-interacting theory.
Including now an interaction
LI = gABC (19)
of three different scalar fields A, B, and C, which are in their free form identical to our
previous field φ, we get the total action
S = SA + SB + SC + SI Z
1 1 1
= A (iDF−1A )xy Ay
2 x
+ B (iDF−1B )xy By
2 x
+ C (iDF−1C )xy Cy
2 x
+g Az Bz Cz (20)
z
MTF: Feynman Rules for the ABC-Theory October 3, 2012 6

and the pathintegral


Z
Z(JA , JB , JC ; g) = DA DB DC exp{iS + iJAx Ax + iJBx Bx + iJCx Cx } . (21)

For zero coupling, i.e. g = 0, we can still solve this integral exactly

Z(JA , JB , JC ; 0)
Z
= DA DB DC exp{iSA + iJAx Ax + iSB + iJBx Bx + iSC + iJCx Cx }

= N × exp{ 2i JAx (iDF A )xy JAy + 2i JBx (iDF B )xy JBy + 2i JCx (iDF C )xy JCy } . (22)

Taking the interaction term and writing the exponential as its powerseries

x
X xk
e = (23)
k=0
k!

and noting that we can write


Z Z
δ
DA A(z) exp{iSA + iJAx Ax } = DA exp{iSA + iJAx Ax } (24)
iδJAz

we can reformulate the pathintegral as Z(JA , JB , JC ; g)


Z
= DA DB DC exp{iSI } × exp{iSA + iJAx Ax + iSB + iJBx Bx + iSC + iJCx Cx }
"∞ k #
Z X 1  Z
= DA DB DC ig Az Bz Cz × eiSA +iJAx Ax +iSB +iJBx Bx +iSC +iJCx Cx
k! z
" k=0
∞  k #
1
Z Z
X δ δ δ
= DA DB DC ig × eiSA +iJAx Ax +iSB +iJBx Bx +iSC +iJCx Cx
k=0
k! z iδJAz iδJBz iδJCz
"∞  k # Z
X 1 Z
δ δ δ
= −g × DA DB DC eiSA +iJAx Ax +iSB +iJBx Bx +iSC +iJCx Cx
k=0
k! z δJAz δJBz δJCz

δ3
 Z 
= exp −g × Z(JA , JB , JC ; 0) . (25)
z δJAz δJBz δJCz

4 Generating Functionals
When we go to the logarithm we can no longer split the exponents and solve exactly, as
δ
δJAz
does not commute with JAx (iDF A )xy JAy and has only a meaning when it acts on the
functional.
But it still allows the definition of the full propagator of the interacting theory as
δ δ
(iDF )xy := W (J; g) . (26)
δJx δJy
MTF: Feynman Rules for the ABC-Theory October 3, 2012 7

Therefore W (J; g) is also called the generating functional of all connected Greensfunctions.
We will write out the full propagator with the definition of the pathintegral later and see,
that it contains the averaged correlator of two fields at the points x and y: hφ(x)φ(y)i.
As the next step we define the classical field
δ δ −i δZ(J; g)
ϕ(x) := W (J; g) = [−i ln Z(J; g)] =
δJ(x) δJx Z δJx
iSφ +iJx φx
R
Dφ iφx e
= −i R = hφx i , (27)
Dφ eiSφ +iJx φx

which is the average over all quantum states generated by the classical source function J(x).
It allows to perform the Legendre transform W (J) → Γ(ϕ) from the generating functional
of all connected Greensfunctions to the generating functional of all one-particle-irreducible
(1PI) vertices

Γ := W − Jx ϕx = Γ(ϕ; g) . (28)

which is also called effective potential.


We will first investigate some features of and relations between the two generating
functionals. The first is the normal property of the Legendre transformation:
δ δ δ δ δ
Γ(ϕ; g) = (W − Jz ϕz ) = W −( Jz )ϕz − Jz ( ϕz )
δϕ(x) δϕx δϕx δϕx δϕx
δJy δ δJz δJy δJz
= W −( )ϕz − Jz δzx = ϕy − ϕz − Jx = −Jx (29)
δϕx δJy δϕx δϕx δϕx
But this property has important implications. One is, that the effective potential is as a
functional an extremum in the classical field when there is no external source. This is also
reflected by the Euler Lagrange equations: the classical solution is the minmum of the
action. And it holds also with the inclusion of quantum corrections for the effective action
— hence the name.
There is also an important connection of the second derivatives of Γ and W . Taking
the above equation and performing the derivative with respect to the source we get
δ2Γ
  
z δ δ δ δϕy δ δ δ δW
δx = Jx = (− Γ) = (− Γ) = −
δJz δJz δϕx δJz δϕy δϕx δJz δJy δϕy δϕx
 2  2

δ W δ Γ
= − , (30)
δJz δJy δϕy δϕx
which basically states, that the second derivatives are inverse to each other:
−1
δ2W δ2Γ

= − , (31)
δJx δJy δϕy δϕx
and including the definition of the propagator
−1
δ2W δ2Γ

(iDF )xy = = − . (32)
δJx δJy δϕy δϕx
MTF: Feynman Rules for the ABC-Theory October 3, 2012 8

This allows us to write the derivative with respect to a source fully in terms of the effective
action and the classical fields:
δ δϕy δ δ δW δ δ2W δ δ
= =( ) =( ) = (iDF )xy , (33)
δJx δJx δϕy δJx δJy δϕy δJx δJy δϕy δϕy

or in the opposite direction

δ δJy δ δ δΓ δ δ2Γ δ δ
= =[ (− )] = (− ) = [−(iDF )−1
xy ]
δϕx δϕx δJy δϕx δϕy δJy δϕx δϕy δJy δJy
δ
= (iDF−1 )xy . (34)
δJy

As the next step we investigate the connection of the higher derivatives of W , Z and
Γ. From the direct writing of the pathintegral we get:

iδ 2 iδ 2 δ 1 δZ 1 δ2Z 1 δZ δZ
W = [−i ln Z] = [ ]= − 2
δJx δJy δJx δJy δJx Z δJy Z δJx δJy Z δJx δJy
= [hφx φy i − hφx ihφy i] = hφx φy iconnected (35)

and
iδ 3 δ 1 δ2Z 1 δZ δZ
W = [ − 2 ]
δJx δJy δJz δJx Z δJy δJz Z δJy δJz
1 δ3Z 1 δZ δ 2 Z 1 δ 2 Z δZ 1 δZ δ 2 Z
= − 2 − 2 − 2
Z δJx δJy δJz Z δJx δJy δJz Z δJx δJy δJz Z δJy δJx δJz
1 δZ δZ δZ
+2 3
Z δJx δJy δJz
= [hφx φy φz i − hφx ihφy φz i − hφx φy ihφz i − hφy ihφx φz i + 2hφx ihφy ihφz i]
= hφx φy φz iconnected (36)

and
iδ 4
W
δJw δJx δJy δJz
δ 1 δ3Z 1 δZ δ 2 Z 1 δ 2 Z δZ 1 δZ δ 2 Z 1 δZ δZ δZ
= [ − 2 − 2 − 2 +2 3 ]
δJw Z δJx δJy δJz Z δJx δJy δJz Z δJx δJy δJz Z δJy δJx δJz Z δJx δJy δJz
1 δ4 Z 1 δZ δ3 Z 1 δ3 Z δZ 1 δZ δ3 Z
= Z δJw δJx δJy δJz
− Z 2 δJx δJw δJy δJz
− Z 2 δJw δJx δJy δJz
− Z 2 δJy δJw δJx δJz
2 2 1 δ2 Z δ2 Z δ2 Z δ2 Z
− Z12 δJδw δJ
Z δ Z
x δJy δJz
− Z 2 δJx δJy δJw δJz
− 1
Z 2 δJw δJy δJx δJz
2 2 2
+2 Z13 δJδw δJ
Z δZ δZ
x δJy δJz
+ 2 Z13 δJ
δZ δ Z δZ
x δJw δJy δJz
+ 2 Z13 δJ
δZ δZ δZ
x δJy δJw δJz
3 2 δ2 Z δZ δZ δ2 Z
− Z12 δJ
δZ δ Z
w δJx δJy δJz
+ 2 Z13 δJ
δZ δZ δ Z
[
w δJx δJy δJz
+ δJx δJy δJz
+ δJy δJx δJz
]
−6 Z14 δJ
δZ δZ δZ δZ
w δJx δJy δJz
(37)
MTF: Feynman Rules for the ABC-Theory October 3, 2012 9

which gives

iδ 4
W
δJw δJx δJy δJz
= hφw φx φy φz i − hφw ihφx φy φz i − hφx ihφw φy φz i − hφy ihφw φx φz i − hφz ihφw φx φy i
−hφw φx ihφy φz i − hφw φy ihφx φz i − hφw φz ihφx φy i
+2hφw ihφx ihφy φz i + 2hφw ihφy ihφx φz i + 2hφw ihφz ihφx φy i
+2hφx ihφy ihφw φz i + 2hφx ihφz ihφw φy i + 2hφy ihφz ihφw φx i
−6hφw ihφx ihφy ihφz i
= hφw φx φy φz iconnected (38)

and so forth.

Interpreting the derivative with respect to the source as a starting


point for a field, we can conclude that W generates diagrams where
all starting points are connected and all the unconnected ones are
substracted.

We can interpret the third derivative of W also in another way. We can write
" −1 #
δ3 δ δ2W δ −δ 2 Γ
W = = −
δJx δJy δJz δJx δJy δJz δJx δϕz δϕy
" −1 #
δ δ2Γ
= (iDF )xa (39)
δϕa δϕz δϕy

which we can evaluate using the matrix identity ∂M −1 = −M −1 (∂M)M −1 , obtained from

0 = ∂(1) = ∂(M · M −1 ) = (∂M)M −1 + M(∂M −1 ) , (40)

and write
−1  −1
δ3 −δ 2 Γ −δ 2 Γ −δ 2 Γ
   
δ
W = (iDF )xa −
δJx δJy δJz δϕz δϕc δϕa δϕc δϕb δϕb δϕy
2
−1  2
−1  2
−1 
δ3Γ
 
−δ Γ −δ Γ −δ Γ
=
δϕx δϕa δϕy δϕb δϕz δϕc δϕa δϕb δϕc
δ3Γ
= (iDF )xa (iDF )yb (iDF )zc . (41)
δϕa δϕb δϕc
Rewriting the same derivative as
−1
−δ 2 Γ δ3Γ

δ δ
(iDF )yz = = (iDF )xa (iDF )yb (iDF )zc . (42)
δJx δJx δϕx δϕa δϕa δϕb δϕc
MTF: Feynman Rules for the ABC-Theory October 3, 2012 10

δ4 W
we get for the four point function δJw δJx δJy δJz

δ3W δ3Γ
 
δ δ
= = (iDF )xa (iDF )yb (iDF )zc
δJw δJx δJy δJz δJw δϕa δϕb δϕc
3
δ3Γ
 
δ Γ
= (iDF )wd (iDF )xe (iDF )af (iDF )yb (iDF )zc
δϕd δϕe δϕf δϕa δϕb δϕc
3
δ3Γ
 
δ Γ
+(iDF )xa (iDF )wd(iDF )ye (iDF )bf (iDF )zc
δϕd δϕe δϕf δϕa δϕb δϕc
3
δ3Γ
 
δ Γ
+(iDF )xa (iDF )yb (iDF )wd (iDF )ze (iDF )cf
δϕd δϕe δϕf δϕa δϕb δϕc
4
δ Γ
+(iDF )xa (iDF )yb (iDF )zc (iDF )wd , (43)
δϕa δϕb δϕc δϕd

where the brackets [. . . ] indicate the derivatives of the propagators (iDF ). This equation
already shows, that the connected diagrams can be constructed by glueing vertices together
with propagators.
And since the propagators are only the inverse of the second derivative of Γ, we need
only calculate the derivatives of Γ to the required accuracy to obtain all information about
the investigated process.

5 ABC-Theory example
n δ Γ
So let us calculate now (δϕ) n for the ABC-theory. Here we have to go back and express

everything in terms of the generating functional Z, since this is the only quantity that we
can calculate directly in perturbation theory. We will use the form

δ3
 Z 
Z(JA , JB , JC ; g) = exp −g × Z(JA , JB , JC ; 0) . (44)
z δJAz δJBz δJCz

We have to treat the classical field A as depending on its source JA . The quantum fields
do not appear anymore, as they are integrated out and the sources are independent from
each other:
δJj (x)
= δkj δ(x − y) , where j, k ∈ {A, B, C} . (45)
δJk (y)

In the same way, the propagator is assumed to be independent of the sources:

δ(iDF j )xy δ(iDF−1j )xy


= =0 , where j, k ∈ {A, B, C} , (46)
δJk (z) δJk (z)

although the propagator is the second derivative of W or the inverse of the second derivative
of Γ. These relations tell, how to expand the full propagator into subdiagrams.
MTF: Feynman Rules for the ABC-Theory October 3, 2012 11

The first derivative


δΓ δ
= [W − JAz Az − JBz Bz − JCz Cz ]
δAx δAx
δW δJAz δ
= (iDF−1A )xy − JAz δ(x − z) − (iDF−1A )xy Az − (iDF−1A )xy [JBz Bz + JCz Cz ]
δJAy δJAy δJAy
δW δBz δCz
= −JAx + (iDF−1A )xy [ − δ(y − z)Az − JBz − JCz ] (47)
δJAy δJAy δJAy
δW
is proportional to sources when the definition of the classical fields ϕ = δJ
are plugged in:
δΓ δW δW δ δW δ δW
= −JAx + (iDF−1A )xy [ − δ(y − z) − JBz − JCz ]
δAx δJAy δJAz δJAy δJBz δJAy δJCz
δ −i δZ δ −i δZ
= −JAx − (iDF−1A )xy [JBz ( ) + JCz ( )]
δJAy Z δJBz δJAy Z δJCz
= 0 for Jφ = 0 . (48)
We get a similar result for the other fields B and C:

δΓ δΓ δΓ
= = =0 . (49)
δAx Jφ =0 δBx Jφ =0 δCx Jφ =0
I kept here the formal derivation and did not do the normal expansion into powers of the
coupling, as this first derivative can be done in a short enough way. I will still discuss
the second and third derivative on general grounds in order to show, that only the higher
derivatives of the generating functional have to be considered for the Feynman rules.
From the second derivatives
δ2Γ δ2
= [W − JAz Az − JBz Bz − JCz Cz ]
δBy δAx δBy δAx
δ2W δ 2 JAz δ 2 Az δ 2 (JBz Bz ) δ 2 (JCz Cz )
= − Az − JAz − − (50)
δBy δAx δBy δAx δBy δAx δBy δAx δBy δAx
we see, that
δ 2 JAz δ δJAz δ
Az = [(iDF−1A )xw ]Az = [(iDF−1A )xw δ(z − w)]Az = 0 (51)
δBy δAx δBy δJAw δBy
and
δ 2 Az δ
JAz = JAz [δ(z − x)] = 0 (52)
δBy δAx δBy
and similar for the other parts, so that only the derivative of W remains:
δ2Γ −1 −1 δ2W −1 −1 δ 2 ln Z
= (iDF A )xu (iDF B )yw = −i(iDF A )xu (iDF B )yw
δBy δAx δJBw δJAu δJBw δJAu
δ 1 δZ
= −i(iDF−1A )xu (iDF−1B )yw [ ]
δJBw Z δJAu
1 δ2Z 1 δZ δZ
= −i(iDF−1A )xu (iDF−1B )yw [ − 2 ] (53)
Z δJBw δJAu Z δJBw δJAu
MTF: Feynman Rules for the ABC-Theory October 3, 2012 12

and
δ2Γ 1 δ2Z 1 δZ δZ
= −i(iDF−1A )xu (iDF−1A )yw [ − 2 ] . (54)
δAy δAx Z δJAw δJAu Z δJAw δJAu
Using
R δ3 i i i
−g
Z = Ne z δJAz δJBz δJCz
× e 2 JAx (iDF A )xy JAy + 2 JBx (iDF B )xy JBy + 2 JCx (iDF C )xy JCy (55)

we get for δJδZAu always a source down from the exponent, that stays and is set to zero at
the end. The contribution to O(g k ) gives
δZ X
∝ JA2n−k−1 JB2n−k JC2n−k . (56)
δJAu n

There are no integers k and n such that

2n − k − 1 = 0 and 2n − k = 0 (57)

and hence

δZ
=0 . (58)
δJAu J=0
δ2 Z
A similar analysis holds for δJBw δJAu
. The contribution to O(g k ) gives

δ2Z X
∝ JA2n−k−1 JB2n−k−1 JC2n−k , (59)
δJBw δJAu n

which gives the same equations as before and hence


δ 2 Z

=0 . (60)
δJBw δJAu J=0
δ2 Z
But for δJAw δJAu
we get the contribution to O(g k )

δ2Z X
∝ JA2n−k−2 JB2n−k JC2n−k , (61)
δJAw δJAu n

which has vanishing exponents, and hence a finite value, for k = 2n + 2. So the lowest
order for a one-particle-irreducible (1PI) two-point function is k = 2:

δ 2 Γ[2] δ2Z
 
−1 −1 1 1 δZ δZ
= −i(iDF A )xu (iDF A )yw −
δAy δAx J=0 Z δJAw δJAu Z 2 δJAw δJAu J=0, O(g2 )
−1
= −i(iDF−1A )xu (iDF−1A )yw Z(0; 0) + Z(0; g 2)


δ2
 
2
× (Z(J; 0) + Z(J; g )) (62)
δJAw δJAu J=0, O(g 2 )
MTF: Feynman Rules for the ABC-Theory October 3, 2012 13

where
i i i
Z(J; 0) = N × e 2 JAa (iDF A )ab JAb + 2 JBa (iDF B )ab JBb + 2 JCa (iDF C )ab JCb
1 δ3 δ3
Z Z
2 2
Z(J; g ) = (−g) × Z(J; 0) (63)
2! 1 δJA1 δJB1 δJC1 2 δJA2 δJB2 δJC2

and hence Z(0; 0) = N and


δ3 δ3

1
Z Z
2 2

Z(0; g ) = (−g) × Z(J; 0)
2! 1 δJA1 δJB1 δJC1 2 δJA2 δJB2 δJC2 J=0
2 Z
g
= −N (DF A )12 (DF B )12 (DF C )12 . (64)
2 1,2
Further
δ2
 
Z(J; 0) = Ni(iDF A )wu = −N(DF A )wu (65)
δJAw δJAu J=0
and
δ2
 
2
Z(J; g )
δJAw δJAu
Z J=0
δ2 g2 δ3 δ3
Z

= × Z(J; 0)
δJAw δJAu 2 1 δJA1 δJB1 δJC1 2 δJA2 δJB2 δJC2 J=0
2 4
1 i
Z
g δ
= N (−1) (iDF B )12 (iDF C )12 ( 2 JAa (iDF A )ab JAb )2 (66)
2 1,2 δJAw δJAu δJA1 δJA2 2!
g2
Z
= N (DF B )12 (DF C )12 [(DF A )wu (DF A )12 + (DF A )w1 (DF A )u2 + (DF A )w2 (DF A )u1 ] .
2 1,2
So
δ 2 Γ[0+2]

δAy δAx J=0


−1
g2
Z
= i(DF−1A )xu (DF−1A )yw
N −N (DF A )12 (DF B )12 (DF C )12
2 1,2
g2
h Z
× −N(DF A )wu + N (DF B )12 (DF C )12 [(DF A )wu (DF A )12
2 1,2
i
+(DF A )w1 (DF A )u2 + (DF A )w2 (DF A )u1 ]
= i(DF−1A )xu (DF−1A )yw (−(DF A )wu )
g2
Z
−1 −1
+i(DF A )xu (DF A )yw (DF B )12 (DF C )12 [(DF A )w1(DF A )u2 + (DF A )w2(DF A )u1 ]
2 1,2
= −i(DF−1A )xu δ(y − u)
g2
Z
+i (DF B )12 (DF C )12 [δ(y − 1)δ(2 − x) + δ(y − 2)δ(1 − x)]
2 1,2
1
= −(iDF−1A )xy + ig 2 [(DF B )yx (DF C )yx + (DF B )xy (DF C )xy ]
2
= −(iDF−1A )xy − ig 2 (iDF B )xy (iDF C )xy , (67)
MTF: Feynman Rules for the ABC-Theory October 3, 2012 14

where the first term is the definition of the propagator and the second term is the 1PI
two-point function to order g 2 , which is just two other propagators connecting the two
points.
With a similar argumentation we find, that the only non-vanishing three-point function
is
δ3Γ δ3W
= (iDF−1A )xu (iDF−1B )yw (iDF−1C )zt (68)
δCz δBy δAx δJCt δJBw δJAu
and we again only need to investigate
δ3Z X
∝ JA2n−k−1 JB2n−k−1 JC2n−k−1 , (69)
δJCt δJBw δJAu n

which has solutions for k = 2n + 1 with the lowest order k = 1. The calculation is even
simpler, since
δ3
Z
1

Z(0, g ) = (−g) × Z(J; 0) =0 . (70)
1 δJA1 δJB1 δJC1 J=0

Now
δ 3 Z [1] δ3 δ3
Z

= N (−g) × Z(J; 0)
δJCt δJBw δJAu δJCt δJBw δJAu 1 δJA1 δJB1 δJC1 J=0
Z
= −Ng(−i) (iDF A )u1 (iDF B )w1 (iDF C )t1 (71)
1

and so
δ 3 Γ[1]
Z
= (iDF−1A )xu (iDF−1B )yw (iDF−1C )zt [N]−1 × Ng (DF A )u1 (DF B )w1 (DF C )t1
δCz δBy δAx 1
Z
= −ig δ(x − 1)δ(y − 1)δ(z − 1) . (72)
1

If we want to write now the Feynman rules not in position space but momentum space, we
have for each field the plane-wave factor eipx . Writing these factors down, by writing the
propagators with their Fourier transformations, we can integrate over the arbitrary point
1 and obtain
Γ[qA , qB , qC ; g 1 ]
δ 3 Γ[1]
Z Z Z
4 iqA x 4 iqB y 4 iqC z
= d xe d ye d ze
δCz δBy δAx
Z Z Z Z Z Z Z
4 4 4 iqC z
= −ig d xe iqA x
d ye iqB y
d ze d t d u d w d4 1
4 4 4

Z 4
d pA −ipA (x−u)
4
ie (−p2A + m2A )
(2π)
Z 4 Z 4
d pB −ipB (y−w) 2 2 d pC −ipC (z−t)
4
ie (−pB + mB ) 4
ie (−p2C + m2C )
(2π) (2π)
Z 4 −ikA (u−1) Z 4 −ikB (w−1) Z
d kA ie d kB ie d4 kC ie−ikC (t−1)
(73)
(2π)4 kA2 − m2A + iǫ (2π)4 kB 2
− m2B + iǫ (2π)4 kC2 − m2C + iǫ
MTF: Feynman Rules for the ABC-Theory October 3, 2012 15

and further

Γ[qA , qB , qC ; g 1 ]
Z 4 Z 4 Z 4
d pA 2 2 d pB 2 2 d pC 2
= −ig 4
(pA − mA ) 4
(pB − mB ) 4
(pC − m2C )
(2π) (2π) (2π)
Z 4 Z 4 Z 4
d kA 1 d kB 1 d kC 1
2 2 2 2
4
(2π) kA − mA + iǫ 4
(2π) kB − mB + iǫ (2π) kC − m2C + iǫ
4 2
Z Z Z Z Z Z Z
4 i(qA −pA )x 4 i(qB −pB )y 4 i(qC −pC )z
d xe d ye d ze d t d u d w d4 1
4 4 4

i(pA −kA )u i(pB −kB )w i(pC −kC )t i(kA +kB +kC )1


Ze 4 e e
Z 4
e Z 4
d pA 2 2 d p B 2 2 d pC 2
= −ig 4
(p A − m A ) 4
(p B − mB ) 4
(pC − m2C )
(2π) (2π) (2π)
Z 4 Z 4 Z 4
d kA 1 d kB 1 d kC 1
2 2 2 2
4
(2π) kA − mA + iǫ 4
(2π) kB − mB + iǫ (2π) kC − m2C + iǫ
4 2

(2π)4 δ(qA − pA )(2π)4 δ(qB − pB )(2π)4 δ(qC − pC )


(2π)4 δ(pA − kA )(2π)4δ(pB − kB )(2π)4 δ(pC − kC )(2π)4 δ(kA + kB + kC )
Z 4
d pA (p2A − m2A )
Z 4
d pB (p2B − m2B )
Z 4
d pC (p2C − m2C )
= −ig
(2π)4 p2A − m2A + iǫ (2π)4 p2B − m2B + iǫ (2π)4 p2C − m2C + iǫ
(2π)4 δ(qA − pA )(2π)4 δ(qB − pB )(2π)4 δ(qC − pC )(2π)4 δ(pA + pB + pC )
= −ig(2π)4 δ(qA + qB + qC ) , (74)

which is just the Feynman rule for the simple vertex in momentum space: vertex factor
times momentum conservation.

Acknowledgements
I want to thank Prof. dr. Darius Abramavičius for the suggestions, how to make the
manuscript better readable.

References
[1] David Griffiths,
Introduction to Elementary Particles
John Wiley & Sons, Inc.; ISBN 0-471-60386-4 (1987)

[2] I. J. R. Aitchison and A. J. G. Hey,


Gauge theories in particle physics: A practical introduction.
Vol. 1: From relativistic quantum mechanics to QED,
Bristol, UK: IOP (2003) 406 p
Vol. 2: Non-Abelian gauge theories: QCD and the electroweak theory,
Bristol, UK: IOP (2004) 454 p

You might also like