HW 3 DIRAC EQUATION PART II

PHIL SAAD
1. Problem 1
[J

, J

] = i(g

+ g

)
{

} = 2g

I
nn
1.1. Commutation Relations of L and K.
L
i
=
1
2

ijk
J
jk
, K
i
= J
0i
[L
1
, L
2
] =
1
4

1jk

2lm
[J
jk
, J
lm
]
=
i
4

1jk

2lm
(g
kl
J
jm
g
jl
J
km
g
km
J
jl
+ g
jm
J
kl
)
=
i
4

1jk

2lm
(
kl
J
jm

jl
J
km

km
J
jl
+
jm
J
kl
)
=
i
4
(
1jk

2km
J
jm

1jk

2jm
J
km

1jk

2lk
J
jl
+
1jk

2lj
J
kl
)
=
i
4
(
1m

2j
J
jm
+
1m

2k
J
km
+
1l

2j
J
jl
+
1l

2k
J
kl
= iJ
21
= iJ
3
By cyclic permutations of this
[L
i
, L
j
] = i
ijk
L
k
[L
1
, K
2
] =
1
2

1lm
[J
lm
, J
02
] =
i
2

1lm
(g
m0
J
l2
g
l0
J
m2
g
m2
J
l0
+ g
l2
J
m0
)
=
i
2

1lm
(g
m2
J
l0
+ g
l2
J
m0
) =
i
2
(
1l2
J
l0
+
12m
J
m0
) = iJ
03
By cyclic permutations of this
[L
i
, K
j
] = i
ijk
K
k
[K
i
, K
j
] = [J
0i
, J
0j
] = i(g
i0
J
0j
g
00
J
ij
g
ij
J
00
+ g
0j
J
i0
)
i and j run from one to three, so all the terms with g
0i
and such drop out. J is antisymmetric so
J
00
= 0, so this is equal to
J
ij
=
ijk
L
k
Date: July 1st 2013.
1
2 PHIL SAAD
1.2. Lorentz Transformations in terms of L and K.
D(1 + ) = 1
i
2
(

)
By the antisymmetry of and J
= 1 i(
01
J
01
+
02
J
02
+
03
J
03
+
12
J
12
+
23
J
23
+
31
J
31
)
= 1 i(
01
K
1
+
02
K
2
+
03
K
3
+
12
L
3
+
23
L
1
+
31
L
2
)
= 1 i L i K
Provided we choose
0i
=
i
,
ij
=
ijk

k
1.3. Vector Transformation Matrices.
(J

= i(g

)
(K
i
)

= i(g
0

g
1

)
K
1
=
_

_
0 i 0 0
i 0 0 0
0 0 0 0
0 0 0 0
_

_
, K
2
=
_

_
0 0 i 0
0 0 0 0
i 0 0 0
0 0 0 0
_

_
, K
3
=
_

_
0 0 0 i
0 0 0 0
0 0 0 0
i 0 0 0
_

_
(L
3
)

= i(g
1

g
2

)
(L
2
)

= i(g
3

g
1

)
(L
1
)

= i(g
2

g
3

)
L
3
=
_

_
0 0 0 0
0 0 i 0
0 i 0 0
0 0 0 0
_

_
, L
2
=
_

_
0 0 0 0
0 0 0 i
0 0 0 0
0 i 0 0
_

_
, K
3
=
_

_
0 0 0 0
0 0 0 0
0 0 0 i
0 0 i 0
_

_
= exp
_

_
0
1

1
0
3

3
0
1

1
0
_

_
It is clear that these reduce to the familiar matrices for rotations and boosts
1.4. Showing that The Lorentz Generators Can Be Written as Generators of SU(2).
A
i
=
1
2
(L
i
+ iK
i
), B
i
=
1
2
(L
i
iK
i
)
[A
i
, A
j
] =
1
4
[L
i
+ iK
i
, L
i
+ iK
i
] =
1
4
([L
i
, L
j
] + i[L
i
, K
j
] + i[K
i
, L
j
] [K
i
, K
j
])
=
1
4
(i
ijk
L
k

ijk
K
k
+ i
jik
K
k
i
ijk
L
k
)
=
1
2
(i
ijk
L
k

ijk
K
k
) = i
ijk
A
k
[B
i
, B
j
] =
1
4
[L
i
iK
i
, L
i
iK
i
] =
1
4
([L
i
, L
j
] i[L
i
, K
j
] i[K
i
, L
j
] [K
i
, K
j
])
=
1
4
(i
ijk
L
k

ijk
K
k
i
jik
K
k
+ i
ijk
L
k
)
=
1
2
(i
ijk
L
k
+
ijk
K
k
) = i
ijk
A
k
HW 3 DIRAC EQUATION PART II 3
[A
i
, B
j
] =
1
4
[L
i
+ iK
i
, L
i
iK
i
] =
1
4
([L
i
, L
j
] i[L
i
, K
j
] + i[K
i
, L
j
] + [K
i
, K
j
])
=
1
4
(i
ijk
L
k

ijk
K
k
i
jik
K
k
i
ijk
L
k
)
= 0
1.5. Transformation Laws for Spinor and Vector Representations.
A =
1
2
(L + iK), B =
1
2
(L iK)
L = A+B, K = i(AB)

L
(
1
2
, 0) : A =

2
, B = 0, L =

2
, K = i

L

_
1 i L i K
_

L

_
1

2
(i + )
_

R
(0,
1
2
) : A = 0, B =

2
, L =

2
, K = i

R

_
1 i L i K
_

R

_
1

2
(i )
_

R
To show how
V
transforms we must show how
T
L

2
transforms.

L

_
1

2
(i + )
_

T
L

T
L
_
1

T
2
(i + )
_

T
L

2

T
L
_
1

T
2
(i + )
_

2
=
T
L

2
_
1

T
2
(i + )
_

T
L

2

T
L

2
_

2
_
1

T
2
(i + )
_

2
_

T
L

2

T
L

2
_
1

2

2
2
(i + )
_

2
= so

T
L

2

T
L

2
_
1 +

2
(i + )
_
Thus by arranging
V
as
_

R
,
T
L

2
_
, we obtain the transformation law

V

_
1

2
(i )
_

V
(1 +

2
(i + )
_
4 PHIL SAAD
1.6. Showing that (1/2,1/2) is the Vector Representation. We can put the components of
a two by two matrix that transforms under a right handed spinor rotation from the left and an
inverse left handed transformation from the right. We use this to show that an object in the (1/2,
1/2) representation is a vector. Notice that the matrix is hermitian and has determinant of the
mod squared of the vector. This transformation preserves these properties, as can be seen from the
fact that the right handed and inverse left handed transformations are hermitian conjugates, and
they both have determinant 1, as they are products of a unitary and the exponential of a traceless
hermitian matrix, which are elements of SL(2,C). The fact that multiplication on the left and on the
right by the dagger of an element of SL(2,C) preserves the length of vectors suggests that SO(3,1)
is isomorphic to SL(2,C); however, the same transformation is accomplished by the negative of
a transformation, thus SO(3,1) is isomorphic to SL(2,C) / Z2, which is isomporphic to R3 X S3
/Z2 since any element of SL(2,C) can be decomposed into the product of a special unitary and
exponential of a traceless hermitian matrix, as mentioned before. The space of traceless hermitian
matrices is isomorphic to R3, and SU(2) is isomorphic to S3. From examining the quotient space of
S3/Z2, which is not simply connected, we can see a topological reason as to why spinors transoform
under 2 pi rotations up to a negative sign. Thanks Weinberg for this cool info.

V
=
_
V
0
+ V
3
V
1
iV
2
V
1
+ iV
2
V
0
V
3
_
= V

_
1

2
(i )
_
V

(1 +

2
(i + )
_
= V

2
(i
j
+
j
)(

j
)
V

2
(i
j

j
)(
j

)
= V

+
V

2
_
i
j
[

,
j
] +
j
{

,
j
}
_
= V

+
V
0
2
_
i
j
[I,
j
] +
j
{I,
j
}
_
+
V
i
2
_
i
j
[
i
,
j
] +
j
{
i
,
j
}
_
= V

+ V
0
+
V
i
2
_
2
j

ijk

k
+ 2
j

ij
_
= V

+ V
0
+ V
0

ijk
V
i

k
= V

+
i
_
g
0

i
g
i

_
V

ijk
V
i

k
= V

+
i
_
g
0

i
g
i

_
V

+
1
2

ijk

k
g
i
V
j

1
2

ijk
g
j

V
i
= V

+
i
_
g
0

i
g
i

_
V

+
1
2

ijk

k
_
g
i

g
j

_
V

i
i
(J
0i
)

i
2

ijk

k
(J
ij
)

=
_

i
2

(J

_
V

= V

Thus an object in the

_
1
2
,
1
2
_
representation transforms as a vector.
HW 3 DIRAC EQUATION PART II 5
2. Problem 2
2.1. Showing S Satises the Lorentz Algebra. Let
J

1
2
= S

=
i
4
[

]
[S

, S

] = [
i
4
[

],
i
4
[

]] =
1
16
[[

], [

]]
=
1
16
(

)
Using the anticommutation relations on the inner two matrices in each product, we get this.
Notice every factor of 2g

is cancelled by another.
=
1
16
(

)
Using the identity {A, B}[C, D] + [A, B]{C, D} = 2(ABCD BADC)
=
1
32
({

}[

] + [

]{

} {

}[

] [

]{

}
{

}[

] [

]{

} +{

}[

] + [

]{

})
Since {

} is proportional to the identity, it commutes with the commutators. Thus we can

simplify to
=
1
16
({

}[

] {

}[

] {

}[

] +{

}[

])
By the anticommutation relations
=
1
8
(g

] g

] g

] + g

])
= i(g

+ g

)
Thus S

satises the Lorentz algebra

2.2. Proving That The Gamma Matrices Must Be 4 Dimensional. To prove that the
matrices satsifying the Cliord algebra in Minkowski spacetime must be four dimensional, we rst
prove that they must be even dimensional.
For =

, det(

) = det(

)
det(

)det(

) = det(I)det(

)det(

)
det(I) = 1
This is only true for even dimensions.
To show that two dimensions does not work, we must show that the pauli matrices plus another
matrix cannot work. This is done by showing that the pauli matrices plus the identity form a basis
for two by two complex matrices. Then any other matrix that we suppose anticommutes with the
pauli matrices must be a linear combination of the the pauli matrices and the identity.
Consider an abritary complex two by two matrix, and some linear combination of the identity
and the Pauli matrices.
_
a b
c d
_
,
_
+ i
+ i
_
= I +
1
+
2
+
3
6 PHIL SAAD
We can solve for , , , and in terms of a, b, c, and d and write any two by two matrix as a linear
combination of the sigma matrices and the identity. Now to show that this implies that there are
no other matrices that anticommute with all three Pauli matrices.
Let A be an arbitrary two by two matrix
{
i
, A} = {
i
, I} + {
i
,
1
} + {
i
,
2
} + {
i
,
3
}
With , , , and not all zero.
= 2(
i
+
i1
+
i2
+
i3
)
If is zero, than A is just a linear combination of the sigma matrices, which obviously does not
work. If is nonzero and the rest are zero, than A is just the identity and that obviously doesnt
work. If and the rest are nonzero, than we still cannot get this anticommutator to zero, since no
linear combination of the identity and each sigma can be zero. Thus two dimensions does not work
and the gamma matrices in Minkowski spacetime must be at least four dimensional.
2.3. Unitarily Equivalent Gamma Matrices. Let a

= U
1

U be a set of four matrices.

Then
{a

, a

} =

= U
1

UU
1

U + U
1

UU
1

U
= U
1

U + U
1

U = U
1
({

})U
= U
1
(2g

I)U = 2g

I
Thus any matrices related to the gammas by a unitary transformation satisfy the Cliord algebra
We can actually make a stronger statement: any matrix represtation of the Cliord algebra is
related to the weyl representation by a similarity transformation. The crux of this proof is in Schurs
lemma, which states that if

and

m
are two n and n dimensional irreducible represntiations
such that

S = S

for some n by n matrix S, than S is invertible or the zero matrix. We begin the proof by
discussing the sixteen gamma matrices that form a basis for 4 by 4 matrices. We call these the
r

r
= 1,

( = ),
5

,
5
With
5
=
0

3
. For each class of matrices in the
r
s, the square is proportional to the
identity, with the proportionality constant being plus or minus one, depending only on the class.
We now need to use the fact that the trace of two anticommuting matrices is zero. This is easily
proven, using the linearity of the trace and the fact that it is unchanged under cyclic rotations of
the matrices. With A and B anticommuting,
Tr(AB) = Tr(BA) = Tr(AB) = Tr(AB) Tr(AB) = 0
We now want to show that all the
r
s except 1 are traceless. This can be done by writing them all as
products of anticommuting matrices. We already have

and
5

as products of anticommuting
matrices. We can also write, with =

) = (

And

5
=
0
(
1

3
) = (
1

3
)
0
Thus all the
r
s except 1 are traceless.
HW 3 DIRAC EQUATION PART II 7
Now we want to show that they are linearly independent. Suppose

r
c
r

r
= 0
Then, multiplying by
s
and taking the trace

r
c
r
Tr(
r

s
) = 0
For s = 1 this is equal to
c
1
Tr(
s
) +

r=1
c
r
Tr(
r

s
) = 0
Since the
r
s are traceless and anticommute, what remains is the trace of the square of
s
, and
the square of the
r
s are plus or minus one times the identity. Thus
c
s
Tr(I) = 0 c
s
= 0
For s = 1, this reduces to
c
1
Tr(I) +

r=1
c
r
Tr(
r
) = 0
c
1
Tr(I) = 0 c
1
= 0
Thus the
r
s are linearly independent and form a basis for 4 by 4 matrices. Now we continue with
the proof by dening the matrix S
S =

r
F
r
(1)
r
Where F is an arbitrary matrix and thr 1 is the sign of the square of the matrices for r. We now
mulitply on one side by

s
and on the other by
s

s
S =

r
F
r
(1)
r
, S
s
=

r
F
r

s
(1)
r
From the anticommutitivity of the
r
s and the fact that any combination is another
r

r
=
sr

t
,

r
=
sr

t
With being the antisymmetric symbol. It is also true that

t
(1)
t
=
1
t
Since
t
squares to (1)
t
. Also

t
=
sr

r

1
t
=
sr

1
r

1
s
=
sr
(1)
r

1
s
We then obtain

s
S =

r
F
r
(1)
r
=

t
sr

t
F
r
(1)
r
And
S
s
=

r
F
r

s
(1)
r
=

t
F
t

s
(1)
t
=

t
F(
t
(1)
t
)
s
=

t
F(
1
t
)
s
=

t
F
sr
(1)
r

1
s
)
s
=

sr

t
F
r
(1)
r
Thus

s
S = S
s
8 PHIL SAAD
And since F is arbitrary and the
r
s are linearly independent, we can always make it nonzero.
Thus by Schurs lemma, S is invertible and the irreducible representations of the Cliord algebra
in 4 dimensions are equivalent.
2.4. Proving That The Weyl Representation Works. Suppose

0
=
_
0 1
1 0
_
,
i
=
_
0
i

i
0
_
Then
{
0
,
0
} = 2
0

0
= 2
_
0 1
1 0
_ _
0 1
1 0
_
= 2g
00
I
44
{
i
,
j
} =
_
0
i

i
0
_ _
0
j

j
0
_
+
_
0
j

j
0
_ _
0
i

i
0
_
=
_
0
i

j
0
_
+
_
0
j

i
0
_
=
_
0 {
i
,
j
}
{
i
,
j
} 0
_
= 2
ij
I
44
= 2g
ij
I
44
{
0
,
j
} =
_
0 1
1 0
_ _
0
i

i
0
_
+
_
0
i

i
0
_ _
0 1
1 0
_
=
_

i
0
0
i
_
+
_

i
0
0
i
_
= 0 = 2g
0i
I
44
3
3.1. Gamma Matrix Transformations.
[

, S

] =
i
4
[

, [

]] =
i
4
(

)
=
i
4
({

} {

})
Since the anticommutator is proportional to the identity
i
2
({

) = i(g

) = i(g

= (J

So then

1
1
2

1
2
= (1 + i

+ O(
2
)
2
)

(1 i

+ O(
2
)
2
)
=

2
[

, S

] + O(
2
) =

2
(J

+ O(
2
)
=

HW 3 DIRAC EQUATION PART II 9

3.2. Inverse Spinor Transformations. It is clear that in the Dirac representation,

0
=
0
,
i
=
i
So

0
S
0i

0
=
i
4

0
[
0
,
i
]

0
=
i
4
(
0

0
) =
i
4
(
i

i
)
=
i
4
[
0

i
] = S
0i

0
S
ij

0
=
i
4

0
[
i
,
j
]

0
=
i
4
(
0

0
)
=
i
4
(
i

0
) =
i
4
[
0

j
] = S
ij
Where from the third line to the fourth, I anticommuted the
0
past both
i
and
j
, picking up
two negative signs. Thus

0
S

0
= S

So

1
2

0
=
0
(1 + i

2
+ O(
2
))
0
= 1 + i

0
S

0
2
+ O(
2
) = 1 + i

]S

2
+ O(
2
)
=
1
1
2
4. Peskin and Schroeder 3.4 -Majorana Field
4.1. Relativistic Invariance of the Majorana Equation And Connection to the Klein
Gordon Equation. From problem 1 we know that
R

1
L
=

. This applies for all

V so

1
L
=

1
L
=

= (
1
)

L
We use this fact to show that the Majorana equation
i im
2

= 0
is relativisticly invariant.
Under a lorentz transformation,
(x)
L
(
1
x)

(x)

(
1
x) =
_
exp
_

2
(i + )
_
_

(
1
x)

(x)
2
_
exp
_

2
(i + )
_
_

(
1
x)
Since
2

=
2

(x)
_
exp
_

2
(i )
_
_

(
1
x)
10 PHIL SAAD
=
R

(
1
x)
Lets look at how the rst part of the equation transforms.
i (x) i
_

L
(
1
x)
_
= i

(
1
)

L
(
1
x)
= i
_

(
1
)

L
_

(
1
x)
=
R
_
i (
1
x)
_
Now lets look at how the whole equation transforms
i im
2

im
2

=
R
_
i (
1
x)
_

R
i
2

(
1
x)
=
R
_
i

im
2

_
= 0
Now we must show that this equation implies the Klein-Gordon equation. Taking the complex
conjugate of the majorana equation, we get
_
i im
2

= i

+ im
2
= 0
Multiplying by
2
on both sides yields
i
2

im = 0
Now we solve for

from the original Majorana equation

i = im
2

=
1
m

2

Plugging this into the complex conjugate equation,
i
2

_
1
m

2

_
im = 0

m
2
= 0

2
m
2
= 0

=
_
1,
1
,
2
,
3
), so

2 =
_

2
,
2

1
,
2

2
,
2

3
_
=
2

Then,

2
m
2
= 0

+ m
2
= 0
Since partial derivatives commute

(

)

+ m
2
= 0
(
2
+ m
2
) = 0
HW 3 DIRAC EQUATION PART II 11
4.2. Action for the Majorana Equation. The action for the Majorana equation can be written
as
S =
_
d
4
x
_

i +
im
2
_

_
_
To show that the action is real, we take the complex conjugate of it
S

=
_
d
4
x
_
_

i
_

im
2
_

_
Lets look at the second part
_

2
=
_

_
Thus the second part of the action is real.
Lets look at the rst part
_

i
_

=
_
_

2
_
i
_
( )
1
( )
1
__

= i
_

1
( )
1
+

2
( )
2
_

= i
_
( )

1
+ ( )

2
_
= i( )

= i

Where in the last line I used the fact that the sigma matrices are hermitian.
Now lets look at the dierence between the action and its conjugate
S S

=
_
d
4
x
_

i + i

_
=
_
d
4
x i

_
= 0
So the action is real. Now we must show that this action yields the Majorana equation. We
want to use the Euler Lagrange equations for

, so lets write the second part of the second part

of the action in components, as the rst part of the second part contains no terms with

=
_

2
_
_
i

2
i

1
_
= i

2
i

1
We can rewrite this for taking derivatives in these two forms
= 2i

2
, = 2i

1
Now using the Euler Lagrange equations for

1
and

2
L

1
= (i )
1
m

2
= (i )
1
im(
2

)
1
= 0
L

2
= (i )
2
+ m

1
= (i )
2
im(
2

)
2
= 0
Thus we recover the Majoraja equation
i im
2

= 0
12 PHIL SAAD
4.3. Dirac Lagrangian With Majorana Spinors. We can rewrite the Dirac spinor as two left
handed spinors
=
_

R
_
=
_

1
i
2

2
_
So we can rewrite the Dirac lagrangian as
L =

(i
/
m) =
_

1
i
T
2

2
_
_
0 1
1 0
__
m i
i m
__

1
i
2

2
_
=
_

1
i
T
2

2
_
_
i m
m i
__

1
i
2

2
_
=
_

1
i
T
2

2
_
_
i

1
im
2

2
m
1
_
= i

1
+ i
T
2

2
im(

T
2

1
)
= i

1
+ i
T
2

2
im(

T
2

1
)
Lets check out the second term there. We can use integration by parts to rewrite it as (excluding
a divergence)
i
T
2

2
= i

T
2

2
= i

2
Where the last step involved splitting into components, anticommuting them, and rewriting. So
the Lagrangian is
L = i

1
+ i

2
im(

T
2

1
)
Looking at the mass term, we can see that it is very similar to the mass term in the Majorana
Lagrangian if we set
1
=
2
. In fact, this is just twice the Majorana Lagrangian if
1
=
2
4.4. Symmetries. We can see that the action for part 3 is invariant under the transformations

1
e
i

1
,
2
e
i

2
. We could use the standard procedure for nding conserved currents, or
we could realize that this is just the same as saying e
i
in the Dirac equation, which yeilds
the conserved current J

. Writing this out we obtain

J

=
_

1
i
T
2

2
_
_
0 1
1 0
__
0

0
__

1
i
2

2
_
=
_

1
i
T
2

2
_
_

0
0

__

1
i
2

2
_
=

1
+
T
2

2
=

1
+
T
2

T

2
=

2
To show that this current vanishes, I need the equations of motion.

1
=
L

1
=

= m
T
2

2
=
L

2
=

= m
T
1

2
And their daggers,

1
= m
2

2
= m
2

1
HW 3 DIRAC EQUATION PART II 13
Now, using the equations of motion, Ill show explicitely that the divergence of this current
vanishes.

= (

1
)

1
+

1
(

2
)

2
= m
T
2

1
+ m

2
m
T
1

2
m

1
= m
T
2

1
+ m

2
m
T
2

1
m

2
= 0
The next current, J

, is supposed to be for the majorana eld

S =
_
d
4
x
_

i +
im
2
_

_
_
This would come from a U)1) symmetry of the eld, but we can see that with the mass term,
this is not a real symmetry. Lets calculate the divergence of this current using the equation of
motion = m
2

= m
T

= (

= m
T

2
+ m

We can see that this does not vanish unless m=0, as expected.
Constructing a theory of N majorana elds with O(N) is simple; we just take N copies of the
Majorana equation.
S =
_
d
4
x
_

a
i
a

im
2
_

T
a

a
_
_
Where a runs from one to N and labels the eld number. With O O(N) and a vector of
components
a
, daggered and transposed appropriately,
O

T
(O)
T
=
T
O
1

2

2
O = O
2

O = O
Thus S S
The transformations involving passing O through sigmas and derivatives can be seen more ex-
plicitely component-wise, with being a vector index, , being spinor indices, and a, b being
group indices.
=

O
ab

b
= O
ab

b
= O
4.5. Quantization of the Majorana Field. To Quantize the Majorana eld, we must rst nd
the general solutions so we can express the eld as a sum of modes. The book suggested that we
compare the Majorana eld to the rst two components of the Dirac eld. So we should write out
the left handed components of the Dirac eld.

L
(x) =
_
d
3
p
(2)
3
_
2E
p

s
_
a
p,s
u
s
L
(p)e
ipx
+ b

p,s
v
s
L
(p)e
ipx

with
u
s
L
(p) =

p
s
, v
s
L
(p) =

p
s
and

s
,
s
=
_
1
0
_
,
_
0
1
_
14 PHIL SAAD
We expect that this will work as a solution to the Majorana equation under certain conditions,
probably that
s
= i
2

s
, since if you take the Dirac lagrangian and replace
R
with i
2

L
you get the Majorana lagrangian. Lets test this by plugging our test solution into the Majorana
equation.
i
L
= im
2

L
i
L
=
_
d
3
p
(2)
3
_
2E
p

s
_
a
p,s
p

p
s
e
ipx
b

p,s
p

p
s
e
ipx

Since (p )(p ) = m
2
this reduces to
m
_
d
3
p
(2)
3
_
2E
p

s
_
a
p,s

p
s
e
ipx
b

p,s

p
s
e
ipx

This must be equal to

im
2

L
= im
_
d
3
p
(2)
3
_
2E
p

s
_
a

p,s

2
_
p

s
e
ipx
+ b
p,s

2
_
p

s
e
ipx

So
m
_
d
3
p
(2)
3
_
2E
p

s
__
a
p,s

p
s
ib
p,s

2
_
p

s
_
e
ipx

_
b

p,s

p
s
+ ia

p,s

2
_
p

s
_
e
ipx

= 0
The square root of the conjugate of sigma is giving us trouble, we need to convert this to something
else. Notice that

=
2

2
so
p

=
2
p
2
and by inserting
2
between successive powers of p
(p

)
n
=
2
(p )
n

2
So we can replace

p

with
2

p
2
and
m
_
d
3
p
(2)
3
_
2E
p

s
__
a
p,s

p
s
ib
p,s

p
2

s
_
e
ipx

_
b

p,s

p
s
+ ia

p,s

p
2

s
_
e
ipx

= 0
m
_
d
3
p

p
(2)
2
_
2E
p

s
__
a
p,s

s
ib
p,s

s
_
e
ipx

_
b

p,s

s
+ ia

p,s

s
_
e
ipx

= 0
We can see that this is satsied if a
p,s
= b
p,s
and = i
2

, as suspected
Thus we can write the Majorana eld as
(x) =
_
d
3
p
(2
3
)
_
2E
p

s
_
a
p,s

p
s
e
ipx
ia

p,s

p
2

s
e
ipx
_
Notice that it is its own antiparticle. We must now nd the anticommutation relations of the
creation and annihilation operators given the anticommutation relations of the elds. Furthermore,
the chosen basis spinors,
s
, are real, so we can get rid of the conjugate there.
{
a
(x),

b
(y)} =
ab
(x y)
HW 3 DIRAC EQUATION PART II 15
Ill expand out the expression for the anticommutator and postulate the anticommutations for the
creation and annihilation operators. If this gives the correct result for the anticommutator of the
elds, than the postulated anticommutation relations for the creation and annihilation operators
must be correct. With a, b, c, d, e, and f being spinor indices,
{
a
(x),

b
(y)} =
_
d
3
p
(2)
3
_
2E
p
d
3
q
(2)
3
_
2E
q

s,r
_
a
p,s

p
ac

s
c
e
ipx
ia

p,s

p
ac

2
ce

s
e
e
ipx
__
a

q,r

r
d

q
db
e
iqy
+ ia
q,r

r
f

2
fd

q
db
e
iqy
_
_
a

q,r

r
d

q
db
e
iqy
+ ia
q,r

r
f

2
fd

q
db
e
iqy
__
a
p,s

p
ac

s
c
e
ipx
ia

p,s

p
ac

2
ce

s
e
e
ipx
_
Multiplying out and combining terms with the same creation and annihilation operators we get
=
_
d
3
p
(2)
3
_
2E
p
d
3
q
(2)
3
_
2E
q

s,r
_
{a
p,s
, a

q,r
}e
i(pxqy)

p
ac

s
c

r
d

q
db
+i{a
p,s
, a
q,r
}e
i(px+qy)

p
ac

s
c

r
f

2
fd

q
db
i{a

p,s
, a

q,r
}e
i(px+qy)

p
ac

2
ce

s
e

r
d

q
db
+{a

p,s
, a
q,r
}e
i(pxqy)

p
ac

2
ce

s
e

r
f

2
fd

q
db
_
Well this is just awful. Fortunately I have wizard powers and can forsee what the anticommuta-
tion relations of the creation and annihilation operators must be to make this work. Voila!
{a
p,s
, a

q,r
} = (2)
3

sr
(p q), {a
p,s
, a
q,r
} = {a

p,s
, a

q,r
} = 0
Lets impose these relations and watch our eld anticommutation relations emerge. Our anti-
commutator of the elds is now
_
d
3
p
(2)
3
_
2E
p
d
3
q
(2)
3
_
2E
q

s,r
_
(2)
3

sr
(p q)e
i(pxqy)

p
ac

s
c

r
d

q
db
+(2)
3

sr
(p q)e
i(pxqy)

p
ac

2
ce

s
e

r
f

2
fd

q
db
_
Now we apply our Kroneker delta to the spin terms and integrate over q to get rid of our Dirac
deltas.
=
_
d
3
p
(2)
3
2E
p

s
_
e
ip(xy)

p
ac

s
c

s
d

p
db
+ e
ip(xy)

p
ac

2
ce

s
e

s
f

2
fd

p
db
_
Much more manageable, but these spinor terms look pretty nasty. Fortunately, the sum over s lets
us simplify this a lot more, since

s
a

s
b
=
ab
Our eld anticommutator becomes
_
d
3
p
(2)
3
2E
p
_
e
ip(xy)

p
ac

cd

p
db
+ e
ip(xy)

p
ac

2
ce

ef

2
fd

p
db
_
=
_
d
3
p
(2)
3
2E
p
_
e
ip(xy)

p
ac

p
cb
+ e
ip(xy)

p
ac

2
ce

2
ed

p
db
_
16 PHIL SAAD
And since
2
squares to one,
=
_
d
3
p
(2)
3
2E
p
_
e
ip(xy)

p
ac

p
cb
+ e
ip(xy)

p
ac

cd

p
db
_
=
_
d
3
p
(2)
3
2E
p
_
e
ip(xy)
(p )
ab
+ e
ip(xy)
(p )
ab
_
(p )
ab
= E
p

ab
p
i

i
ab
, so we get
=
_
d
3
p
2(2)
3

ab
_
e
ip(xy)
+ e
ip(xy)
_

_
d
3
p
(2)
3
2E
p
p
i

i
ab
_
e
ip(xy)
+ e
ip(xy)
_
The second integral over the p
i
is odd, so it vanishes. We are left with
_
d
3
p
2(2)
3

ab
_
e
ip(xy)
+ e
ip(xy)
_
=
ab
(x y)
So our postulated anticommutation relations work! Big surprise, provided x
0
= y
0
(I really should
have imposed this earlier but I dont want to go back and change it)
Now we need to nd the Hamiltonian for the Majorana eld. The Lagrangian density is given
by
L =

i +
im
2
_

_
=

i
0

i
+
im
2
_

_
=
L

0

= i

H =
0
L =

i
i

im
2
_

_
H =
_
d
3
xH =
_
d
3
x
_

i
i

i

im
2
_

_
_
Now for the fun part....plugging the eld into the Hamiltonian and simplifying the gory mess that
results. Ill do this term by term to avoid clutter. First lets look at
_
d
3
x

i
i

Keeping in mind that these elds are at t = 0 and letting px be the 3-vector dot product, we get

i
=
_
d
3
p
(2)
3
_
2E
p

s
_
ip
i
a
p,s

p
s
e
ipx
p
i
a

p,s

p
2

s
e
ipx
_
So our rst term becomes
_
d
3
x
d
3
p
(2)
3
_
2E
p
d
3
q
(2)
3
_
2E
q

s,r
_
a

q,r
e
iqx

q + ia
q,r
e
iqx

q
_
i
i
_
ip
i
a
p,s

p
s
e
ipx
p
i
a

p,s

p
2

s
e
ipx
_
=
_
d
3
x
d
3
p
(2)
3
_
2E
p
d
3
q
(2)
3
_
2E
q

s,r
_
a

q,r
a
p,s
e
ix(qp)

q
i
p
i

p
s
+ia

q,r
a

p,s
e
ix(p+q)

q
i
p
i

p
2

s
+ ia
q,r
a
p,s
e
ix(p+q)

q
i
p
i

p
s
a
q,r
a

p,s
e
ix(qp)

q
i
p
i

p
2

s
_
HW 3 DIRAC EQUATION PART II 17
I can integrate over x to simplify this a bit.
_
d
3
x

i
i

i
=
_
d
3
pd
3
q
2(2)
3
_
E
p
E
q

s,r
_
a

q,r
a
p,s
(p q)
r

q
i
p
i

p
s
+ia

q,r
a

p,s
(p + q)
r

q
i
p
i

p
2

s
+ ia
q,r
a
p,s
(p + q)
r

q
i
p
i

p
s
a
q,r
a

p,s
(p q)
r

q
i
p
i

p
2

s
_
Then integrating over q gives us this, and when I write p i mean (E
p
, p)
_
d
3
p
2(2)
3
E
p

s,r
_
a

p,r
a
p,s

p
i
p
i

p
s
+ia

-p,r
a

p,s

p
i
p
i

p
2

s
+ ia
-p,r
a
p,s

p
i
p
i

p
s
a
p,r
a

p,s

p
i
p
i

p
2

s
_
It doesnt seem like theres much else I can do here right now, so lets move on to the next part
_
d
3
x
T

=
_
d
3
x
d
3
p
(2)
3
_
2E
p
d
3
q
(2)
3
_
2E
q

s,r
_
a
q,r
e
iqx

Tr
_
q
T
ia

q,r
e
iqx

2T
_
q
T
_

2
_
a
p,s
e
ipx

p
s
ia

p,s
e
ipx

p
2

s
_
=
_
d
3
x
d
3
p
(2)
3
_
2E
p
d
3
q
(2)
3
_
2E
q

s,r
_
a
q,r
a
p,s
e
ix(p+q)

Tr
_
q
T

p
s
ia
q,r
a

p,s
e
ix(qp)

Tr
_
q
T

p
2

s
ia

q,r
a
p,s
e
ix(qp)

2T
_
q
T

p
s
a

q,r
a

p,s
e
i(p+q)

2T
_
q
T

p
2

s
_
We can simplify this a bit rst by recalling that

= ,
2T
=
2
, and
T
=
2

2
. Then this
term in our Hamiltonian is equal to
=
_
d
3
x
d
3
p
(2)
3
_
2E
p
d
3
q
(2)
3
_
2E
q

s,r
_
a
q,r
a
p,s
e
ix(p+q)

Tr

p
s
ia
q,r
a

p,s
e
ix(qp)

Tr

p
2

s
+ ia

q,r
a
p,s
e
ix(qp)

p
s
+a

q,r
a

p,s
e
i(p+q)

p
2

s
_
Then we integrate over x to get some delta functions
=
_
d
3
pd
3
q
2(2)
3
_
E
p
E
q

s,r
_
a
q,r
a
p,s
(p + q)
Tr

p
s
ia
q,r
a

p,s
(p q)
Tr

p
2

s
+ ia

q,r
a
p,s
(p q)
r

p
s
+a

q,r
a

p,s
(p + q)
r

p
2

s
_
And we integrate over q to get rid of those delta functions. Again, when I write p I mean (E
p
, p).
18 PHIL SAAD
_
d
3
p
2(2)
3
E
p

s,r
_
a
-p,r
a
p,s

Tr

p
s
ia
p,r
a

p,s

Tr

p
2

s
+ ia

p,r
a
p,s

p
s
+a

-p,r
a

p,s

p
2

s
_
Now lets look at the last part of the Hamiltonian
_
d
3
x

=
_
d
3
x
d
3
p
(2)
3
_
2E
p
d
3
q
(2)
3
_
2E
q

s,r
_
a

q,r
e
iqx

q + ia
q,r
e
iqx

q
_

2
_
a

p,s
e
ipx
_
p

s
ia
p,s
e
+ipx
_
p

s
_
=
_
d
3
x
d
3
p
(2)
3
_
2E
p
d
3
q
(2)
3
_
2E
q

s,r
_
a

q,r
a

p,s
e
ix(p+q)

q
2
_
p

s
ia

q,r
a
p,s
e
ix(qp)

q
2
_
p

s
+ ia
q,r
a

p,s
e
ix(qp)

q
2
_
p

s
a
q,r
a
p,s
e
i(p+q)

q
2
_
p

s
_
Doing our usual tricks with
2
and such, then integrating over x and q, and keeping in mind what
I mean by p, we get that this is equal to
=
_
d
3
p
2(2)
3
E
p

s,r
_
a

-p,r
a

p,s

p
2

s
ia

p,r
a
p,s

p
s
+ ia
p,r
a

p,s

q
2

p
2

s
a
-p,r
a
p,s

p
s
_
Now weve got all the parts to our Hamiltonian, so lets put them together. Using the fact that
p = p and p = p
H =
_
d
3
p
2(2)
3
E
p

s,r
_
a

p,r
a
p,s

r
_

p
i
p
i

p +
m
2

p +
m
2

p
_

s
+a
p,r
a

p,s

r
_

p
i
p
i

p
2

m
2

p
m
2

p
_

s
+a

-p,r
a

p,s

r
_
i

p
i
p
i

p
2

im
2
p
2
+
im
2
p
2
_

s
+a
-p,r
a
p,s

r
_
i
2

p
i
p
i

p
im
2

2
p +
im
2

2
p
_

s
First, lets try to get rid of the terms with 2 creation or 2 annihilation operators. Notice that
p p = 2p
i

i
. We can also express
i
p
i
as
1
2
(p p ), so

p
i
p
i

p =
1
2
_
p (p )
3
2
(p )
3
2

p
_
=
m
2
_
p p
_
= m
i
p
i
HW 3 DIRAC EQUATION PART II 19
Thus the coecients of the double creation or double annihillation operators become (ignoring the
s and
2
s)
im
i
p
i
im
i
p
i
= 0
Now we must deal with the other parts of the Hamiltonian. First, we use the fact that

p =

p = m. Then we do the same trick with

i
p
i

p
i
p
i

p =
1
2
_
(p )
2
m
2
)
(p )
2
= (E
p
p
i

i
)
2
= E
2
p
2E
p

i
p
i
+ p
i
p
j

j
= E
2
p
2E
p

i
p
i
+
p
i
p
j
2
{
i

j
}
= E
2
p
2E
p

i
p
i
+ p
2
= 2E
2
p
m
2
2E
p

i
p
i
so

p
i
p
i

p = E
2
p
m
2
E
p

i
p
i
And the coecient for the combinations of creation and annihilation operators becomes

r
_
E
2
p
m
2
E
p

i
p
i
+ m
2
_

s
=
s,r
_
E
2
p
E
p

i
p
i
_
Our Hamiltionian then becomes
H =
_
d
3
p
2(2)
3
E
p

s
_
E
2
p
E
p

i
p
i
__
a

p,s
a
p,s
a
p,s
a

p,s
_
The integral over E
p

i
p
i
is odd in p so we can get rid of this term.
H =
_
d
3
p
2(2)
3
E
p

s
_
a

p,s
a
p,s
a
p,s
a

p,s
_
Using our anticommutation relations on the second term and chucking the delta function, we arrive
at our nal Hamiltionian,
H =
_
d
3
p
(2)
3
E
p

s
a

p,s
a
p,s