Вы находитесь на странице: 1из 7

# 221A HW 9

1
1.1. We have a set of nine scalar elds,
ia
, with i and a running from one to three. We are given the
Lagrangian
L =
1
2

ia

ia
+

6

ijk

abc

ia

jb

kc
L =
1
2
Tr
_
()
2
_
+ det
_

_
We see immediately that the kinetic term is that of 9 massless scalar elds. This gives us the propagator
D
ia,jb
=
ij

ab

F
This can also be seen by looking at the generating function
Z[J] =
_

i,a
D
ia
exp
_
i
_
d
4
x
1
2

ia

ia
+

6

ijk

abc

ia

jb

kc
+ J
ia

ia
_
We rewrite this as
Z[J] =
_

i,a
D
ia
exp
_
i
_
d
4
x
ij

ab
1
2

ia

jb
+

6

ijk

abc

ia

jb

kc
+ J
ia

ia
_
Doing the usual integration by parts and pulling the interaction term out
Z[J] = exp
_
i

6
_
d
4
x
ijk

abc
1
i

J
ia
1
i

J
jb
1
i

J
kc
_ _

i,a
D
ia
exp
_
i
_
d
4
x
1
2

ij

ab

ia

jb
+ J
ia

ia
_
Doing the usual completing the square and integrating over the gaussian function of gives us
Z[J] = exp
_
i

6
_
d
4
x
ijk

abc
1
i

J
ia
1
i

J
jb
1
i

J
kc
_
Z
0
 exp
_
i
1
2
_
d
4
xd
4
yJ
ia
(x)D
ia,jb
(x y)J
jb
(y)
_
From this it is also pretty clear that we have the momentum space vertex factor of
i
ijk

abc
But this looks like maybe its a bit of a funky theory so Im gonna make sure. Using some shorthand whose
meaning is hopefully pretty clear, our rst order in lambda term is,
i

6
1
i
3
_
x

ijk

abc

iax

jbx

kcx
Z
0
[J]
= i

6
1
i
3
i
_
x

ijk

abc

iax

jbx
_
y
D
kc,ld,xy
J
ldy
Z
0
[J]
= i

6
1
i
3
i
_
x

ijk

abc

iax
D
kc,jb,xx
Z
0
[J] + i

6
1
i
3
i
2
_
x

ijk

abc

iax
_
y
D
kc,ld,xy
J
ldy
_
z
D
jb,me,xz
J
mez
Z
0
[J]
Really, I dont think its necessary to continue from here. Well get a diagram that that gives another factor
of i and an integral over a source and a propagator to the second term, which is our three source, one vertex
diagram that has a symmetry factor of 1/6.
Date: November 29 2013.
1
i
6

ijk

abc
_
d
4
wd
4
xd
4
yd
4
zJ
ld
(y)J
me
(z)J
nf
(w)D
(x y)D
jm,be
(x z)D
kn,cf
(x w)
The problem here is that I have Z[J], not an amplitude. Im summing over all the dierent sources that
could contribute. To get the vertex factor, I need to nd the vev of a product of three elds, which involves
taking three derivatives. I expect this to give me the 3! factor needed to cancel the 1/6. And im running
out of letters.
| T
i

,a
(x
1
)
j

,b
(x
2
)
k

,c
(x
3
) | =
1
i
3

x1

x2

x3
Z[J]

J=0
The rst order in lambda term is
i
ijk

abc
_
d
4
xD
ii

,aa
(x x
1
)D
jj

,bb
(x x
2
)D
kk

,cc
(x x
3
)
= i
i

k

a

_
d
4
x
F
(x x
1
)
F
(x x
2
)
F
(x x
3
)
Bam! so our vertex factor is exactly what we expected.
Now I want to nd some scattering amplitudes. I assume this is to lowest order in lambda. The diagrams
I need to calculate for
ia

jb

kc

ld
scattering are
and the other two (t and u) diagrams. This one, the s channel, is given by
i
abe

ijm
D
em,fn
i
nkl

fcd
= i
2

abe

ijm

nkl

fcd

ef

mn
1
s
= i
2

abe

ijm

mkl

ecd
1
s
= i
2
_

ac

bd

bc
__

ik

jl

il

jk
_
1
s
By doing the same thing for the t and u channel diagrams, we get
iT
ia jbkc ld
= i
2
_
_

ac

bd

bc
__

ik

jl

il

jk
_
s
+
_

ab

cd

bc
__

ij

kl

il

jk
_
t
+
_

ac

bd

ab

dc
__

ik

jl

ij

lk
_
u
_
This was not nearly as bad as it looked. NICE!
221A HW 9 3
2
2.1. From your homework three solutions (thanks!) and some help from Srednicki, we know that for a
complex scalar eld
a(k) = i
_
d
3
xe
ikx
_
i
k
(x)

(x)
_
= i
_
d
3
xe
ikx

0
(x)
Since

=
0
. Also,
a

(k) = i
_
d
3
xe
ikx

(x)
The b operators should be the same as the a operators but the role of phi and phi dagger switched
b(k) = i
_
d
3
xe
ikx

(x)
b

(k) = i
_
d
3
xe
ikx

0
(x)
We make a creation operator for some particle localized around k
1
a

1
=
_
d
3
kf
1
(k)a

(k)
a

1
(+) a

1
() =
_
+

dt
0
a

1
(t)
= i
_
d
3
kf
1
(k)
_
d
4
x
0
_
e
ikx

(x)
_
= i
_
d
3
kf
1
(k)
_
d
4
xe
ikx
_

2
0
+
2
_

(x)
= i
_
d
3
kf
1
(k)
_
d
4
xe
ikx
_

2
0
+ k
2
+ m
2
_

(x)
= i
_
d
3
kf
1
(k)
_
d
4
xe
ikx
_

2
0

2
+ m
2
_

(x)
= i
_
d
3
kf
1
(k)
_
d
4
xe
ikx
_

2
0

2
+ m
2
_

(x)
= i
_
d
3
kf
1
(k)
_
d
4
xe
ikx
_

2
+ m
2
_

(x)
Thus we have
a

1
() = a

1
(+) + i
_
d
3
kf
1
(k)
_
d
4
xe
ikx
_

2
+ m
2
_

(x)
a
1
(+) = a
1
() + i
_
d
3
kf
1
(k)
_
d
4
xe
ikx
_

2
+ m
2
_
(x)
b

1
() = b

1
(+) + i
_
d
3
kf
1
(k)
_
d
4
xe
ikx
_

2
+ m
2
_
(x)
b
1
(+) = b
1
() + i
_
d
3
kf
1
(k)
_
d
4
xe
ikx
_

2
+ m
2
_

(x)
For a scattering process, we have our initial and nal states as
|i =

n,m
b

m
()a

n
() |
f| =

j,k
| b
j
(+)a
k
(+)
So we have
f|i =

jknm
| Tb
j
(+)a
k
(+)b

m
()a

n
() |
f|i = i
n+n

_

jknm
d
4
x
j
d
4
x
k
d
4
x
n
d
4
x
m
e
ikxn
_

2
xn
+ m
2
_
e
ikxm
_

2
xm
+ m
2
_
e
ikxj
_

2
xj
+ m
2
_
e
ikx
k
_

2
x
k
+ m
2
_
| T

(x
j
)(x
k
)(x
m
)

(x
n
) |
3
3.1. This rst part is essentially done in Peskin and Schroeder. Given the action
S =
_
d
4
x
1
2
_

_
2
+ V ()
Under a translation, the elds transform as

## Due to the fact that it is a scalar, the Lagrangian changes as

L L

L = L

L
_
By varying the elds we also get
L L
L

Integrating by parts

L
=

_
L

_
L

_
The rst term vanishes if the elds satisfy the equations of motion. We are left with the two variations in
the action
S =
_
d
4
x

_
L

_
=
_
d
4
x

L
_
Since our theory is translation invariant, then the variation in the action is zero. Also, this is true for all ,
so we have

_
L

_
=

L
_

_
L

L
_
= 0
T

=
L

L
Thus for our theory,
T

+ g

_
1
2

+ V ()
_
Well this is up to a sign. Im gonna check this in the case of a massive free particle. T
00
should be the energy
density. If you plug in = = 0 you get the negative of the Hamiltionian density (Im using the -+++
metric if it hasnt been clear yet), so actually
T

_
1
2

+ V ()
_
We get the conserved charges by integrating T
0
over space
P
0
=
_
d
3
x
1
2
(
t
)
2
+
1
2
_
)
2
+ V () = H
P
i
=
_
d
3
x
t

221A HW 9 5
3.2. Under an innitesimal Lorentz transformation, we have

We also have
L L

L
L =

L =

(x

L) +

L(

)
=

(x

L) +

n
=

(x

L)
=

(x

L)
By the antisymmetry of . We also have, due to the variation in the elds (already performing the integration
by parts),
L =
_
L

_
+

_
L

_
The rst part drops out if the elds satisfy the equations of motion.
L =

_
L

_
Identifying
L

= T

+ g

L
We get
L =

_
x
n
T

L
_
Setting our two variations in the Lagrangians equal to each other

_
x
n
T

L
_
=

(x

L)

_
x

_
= 0
Due to the antisymmetry of , we can only say that, for all

_
x
[
T
]
_
= 0

_
x

_
= 0
Thus we get the set of 6 conserved currents for Lorentz transformations
_
J

= x

## The charges are given by

Q

=
_
d
3
x
_
J

_
0
We have
Q
12
=
_
d
3
x
_
x
1
T
20
x
2
T
10
_
=
_
d
3
x
_
x
y
y
x
_
= L
z
As would be expected. Now what are our boost charges?
Q
01
=
_
d
3
x
_
x
0
T
10
x
1
T
00
_
=
_
d
3
x
_
t
x
xH
_
Okay sure I guess. I didnt really expect it to have a clear interpretation in terms of things I already know
anyways.
4
4.1. We have the action
S =
_
d
4
x
1
2
_

_
2
+
1
2
m
2

2
+

4!

4
And we have the innitesimal variations
x

= x

, =
We recognize these as the innitesimal versions of the variations
e

, x

We can see rather quickly (using the fact that the jacobian is e
4
and the partial derivatives give e

) that
what makes this not a symmetry is the mass term. That makes sense because the xed parameter that has
dimension is the mass term. (lambda is dimensionless). I guess thats expected.
Now to nd that current. Instead of messing around with varying coordinates and stu, Im gonna rewrite
the problem in a way that makes it a bit more familiar. I rewrite the action as
S =
_
d
4
x

g
_
1
2
g

+
1
2
m
2

2
+

4!

4
_
And our variations are rewritten as
e

= (1 + O(
2
))
g

e
2
g

= (1 + 2 + O(
2
))g

e
2
g

= (1 2 + O(
2
))g

With g

always proportional to

.
Just to make it clear why this all works, Ill go through this a bit. The root of the determinant takes care
of the variations of the coordinates, which would pick up a jacobian. The metric in the derivative part takes
care of the changes of the derivatives that Id get if I did the change in coordinates stu. Remember that
my covariant derivative in this case is just the partial derivative, so it is metric compatible. I used the fact
that

## , which using the metric compatability of the derivative, gives us

= g

.
Under this transformation, the action changes as
S =
_
d
4
x

gL
g

g
+
L

+
L

We take a hint from GR and x up that rst term, recongnizing it as involving the energy momentum tensor.
We also integrate by parts on the second term and say that the eld equations are satised.
S =
_
d
4
x

g
_

1
2
T

_
L

_
_
Substituting in the expressions for the variations, but in their innitesimal forms,
S =
_
d
4
x

g
_
T

_
L

_
_
Rewriting g

as

, we get
S =
_
d
4
x

g
_
T

_
L

_
_
Integrating by parts and using the fact that the energy momentum tensor is divergenceless, and doing the
derivative in the second term,
S =
_
d
4
x

_
T

_
Thus
J

= T

## Now to check that it is conserved in the massless case

= x

+ T

+
2

221A HW 9 7
The rst term vanishes. Lets write out the energy momentum tensor and equation of motion explicitely
T

_
1
2

+
m
2
2

2
+

4!

4
_

2
= m
2
+

3!

3
So the second term, which is the trace of the energy momentum tensor, is

2m
2

3!

4
We have

= 2m
2

3!

4
+
2

## Using the equation of motion

= m
2

2
Which, as expected, is zero in the massless case.