Quantum Field Theory

Example Sheet 1
Michelmas Term 2013
Solutions by:
Rachel J. Dowdall [email protected]
Alec A. H. Graham [email protected]
Rahul Jha [email protected]
Emanuel Malek [email protected]
Laurence Perreault Levasseur [email protected]
Note: in the present solutions, we are using the mostly plus metric convention, i.e. = diag(, +, +, +). Also, the
current version has been updated to use the conserved current denition that was used class, j

=
L
(

a
)

a
+ F

Exercise 1
Lagrangian:
L =
_
a
0
dx
_

2
_
y
t
_
2

T
2
_
y
x
_
2
_
(1)
Express y(x, t) as a Fourier series:
y(x, t) =
_
2
a

n=1
q
n
(t) sin
_
nx
a
_
. (2)
The derivative of y(x, t) with respect to x and t is
y(x, t)
t
=
_
2
a

n=1
q
n
(t) sin
_
nx
a
_
(3)
y(x, t)
x
=
_
2
a

n=1
n
a
q
n
(t) cos
_
nx
a
_
, (4)
where we abbreviate q q/t. Substituting Eqs. (3) and (4) in (12) gives
L =
1
a

n,m
_
a
0
dx
_
q
n
(t) q
m
(t) sin
_
nx
a
_
sin
_
mx
a
_
T
nm
2
a
2
q
n
(t)q
m
(t) cos
_
nx
a
_
cos
_
mx
a
_
_
. (5)
Using the orthonormality relations
2
a
_
a
0
dx sin
_
nx
a
_
sin
_
mx
a
_
=
nm
and
2
a
_
a
0
dx cos
_
nx
a
_
cos
_
mx
a
_
=
nm
(6)
we see that the terms with n ,= m vanish. We obtain:
L =

n
_
a
0
dx
_

2
q
2
n
(t)
T
2
_
n
a
_
2
q
2
n
(t)
_
. (7)
The Euler-Lagrange equations are given by
d
dt
L
q
n

L
q
n
= 0, n = 0, 1, 2, . . . . (8)
With
L
q
n
= q
n
and (9)
L
q
n
= T
_
n
a
_
2
q
n
(10)
we obtain the equations of motion:
q
n
(t) +
T

_
n
a
_
2
q
n
(t) = 0, (11)
where n = 0, 1, 2, . . .. These are the equations of motion of harmonic oscillators with frequency
n
=
n
a
_
T

.
1
Exercise 2
We are given the classical Hamiltonian for a string:
H =

n=0
_
1
2
p
2
n
+
1
2

2
n
q
2
n
_
(12)
Upon quantization, the p
n
and q
n
are promoted to Hermitian operators, and we impose the canonical commutation relations
upon them:
[ q
n
, q
m
] = [ p
n
, p
m
] = 0 and [ q
n
, p
m
] = i
nm
.
We then introduce the creation and annihilation operators a

n
and a
n
as a decomposition of p
n
and q
n
:
q
n
=
1

2
n
_
a
n
+ a

n
_
and p
n
= i
_

n
2
_
a

n
a
n
_
(13)
or (inverting these) a
n
a
n
=
_

n
2
q
n
+
i

2
n
p
n
and a

n
a

n
=
_

n
2
q
n

2
n
p
n
Where we have dropped the hats in the last line to simplify the notation. To check that such denitions of a
n
s and a

n
s
indeed satisfy the canonical commutation relations, we compute:
[a
n
, a
m
] =
__

n
2
q
n
+
i

2
n
p
n
,
_

n
2
q
m
+
i

2
m
p
m
_
=

m
2
[q
n
, q
m
] +
i
2
_

n
[p
n
, q
m
] +
i
2
_

m
[q
n
, p
m
]
1
2

m
[p
n
, p
m
]
= +
i
2
(i
nm
) +
i
2
i
nm
0 . (14)
Taking the adjoint of this latter commutator, we nd:
[a

n
, a

m
] 0 . (15)
The remaining commutator is
[a
n
, a

m
] =
__

n
2
q
n
+
i

2
n
p
n
,
_

n
2
q
n

2
n
p
n
_
=
i
2
_

m
[q
n
, p
m
] +
i
2
_

n
[p
n
, q
m
]
=
i
2
_

m
(i
mn
) +
i
2
_

n
(i
nm
)
=
mn
(16)
Now, using (13), and plugging into the expression for the Hamiltonian, we get:
H =

n=1
_
1
2
_
i
_

n
2
_
a

n
a
n
_
__
i
_

n
2
_
a

n
a
n
_
_
+

2
n
2
_
1

2
n
_
a
n
+ a

n
_
__
1

2
n
_
a
n
+ a

n
_
__
(17)
=

n=1
_

n
4
_
a

n
a

n
a
n
a

n
a

n
a
n
+ a
n
a
n
_
+

n
4
_
a
n
a
n
+ a

n
a
n
+ a
n
a

n
+ a

n
a

n
_
_
(18)
=

n=1

n
2
_
a
n
a

n
+ a

n
a
n

(19)
as required.
Suppose now the existence of a ground state [0 such that a
n
[0 = 0 for every n. Using the commutation relation (16), we
can rewrite the Hamiltonian in normal order as follows:
H =

n=1

n
2
_
2 a
n
a

n
+
nn

n=1

n
a
n
a

n
+

n=1

n
2

nn
.
The last, innite term is due to the vacuum energy. Indeed, we nd that the vacuum expectation value is given by:
0[H[0 =

n=1

n
2

nn
. (20)
2
Physical experiments are typically only sensitive to dierences of energy
1
, not absolute magnitudes. This justies us therefore
neglecting the vacuum energy. Performing this subtraction, we recover the normal ordered prescription for the Hamiltonian:
: H :=

n=1
=
n
a

n
a
n
. (21)
From this, we calculate the energy of the states [(n, l) =
_
a

n
_
l
[0 containing l particles in the n-th mode. First, we may
prove by induction that
_
H,
_
a

n
_
l
_
= l
n
_
a

n
_
l
, l 1 .
For l = 1, this is a straightforward application of the commutation relation (16) to obtain that:
_
H, a

m=1

m
_
a

m
a
m
, a

m=0
_
a

m
(
nm
)
_
=
n
a

n
. (22)
Supposing now that (22) is valid for l k, we nd
_
H,
_
a

n
_
k+1
_
=
_
H,
_
a

n
_
k
_
a

n
+
_
a

n
_
k
Ha

_
a

n
_
k+1
H
= k
n
_
a

n
_
k+1
+
_
a

n
_
k
_
H, a

= k
n
_
a

n
_
k+1
+
n
_
a

n
_
k+1
= (k + 1)
n
_
a

n
_
k+1
(23)
as required. Having removed the vacuum energy, we now have H[0 = 0, so the energies of the states [(n, l) are:
H[(n, l) =
__
H,
_
a

n
_
l
_
+
_
a

n
_
l
H
_
[0 = l
n
_
a

n
_
l
[0 = l
n
[(n, l) (24)
Using this, we can now move on to compute the energy of the Fock state of the form
[(l
1
, l
2
, ..., l
N
) =
_
a

1
_
l
1
_
a

2
_
l
2
...
_
a

N
_
l
N
[0 (25)
Acting on these with H to obtain:
H[(l
1
, l
2
, ..., l
N
) =
__
H,
_
a

1
_
l
1
_
+
_
a

1
_
l
1
H
_
[(l
1
, l
2
, ..., l
N
)
= l
1

1
[l
1
, l
2
, ..., l
n
+
_
a

1
_
l
1
(H[(l
1
, l
2
, ..., l
N
)) (26)
This relation can be applied iteratively, to nd
H[(l
1
, l
2
, ..., l
N
) =
_
N

i=1
l
i

i
_
[(l
1
, l
2
, ..., l
N
) +
_
N

i=1
_
a

i
_
l
i
_
H[0 =
_
N

i=1
l
i

i
_
[(l
1
, l
2
, ..., l
N
) (27)
i.e. the energy of the multi-particle state [(l
1
, l
2
, ..., l
N
) is

N
i=1
l
i

i
.
Exercise 3
The scalar eld (x) transforms under a Lorentz transformation x

as
(x)

(x) = (y) = (
1
x).
Note that this is an active transformation, i.e. the elds are transformed but the coordinates are left unchanged, and the
new eld equals the old eld at the coordinates transformed backward. Using

=
y

= (
1
)

Now,

(x) + m
2
(x)

(x) + m
2

(x)
=

(
1
)

(
1
)

(y) + m
2
(y)
=

(x) + m
2
(y). (28)
1
Note that this does not hold in systems where gravitational interactions are considered, since in this case the energy-momentum tensor -
enclosing the total energy density - sources the Einstein equations
3
In the last step we used that is a Lorentz transformation and so its inverse
1
preserves the inverse Minkowski metric

,
_

1
_

1
_

.
Renaming y to be x in Eq. (28), this equation demonstrates that if (x) fulls the Klein-Gordon equation

(x) + m
2
(x) = 0 ,
then (
1
x) fulls it as well.
Exercise 4
Lagrangian density:
/ =

m
2

2
(

)
2
. (29)
Euler-Lagrange equation for

:
/

/
(

)
= 0. (30)
With
/

= m
2
(

) and (31)
/
(

)
=

(32)
we obtain the equation of motion for

+ m
2
+ (

) = 0. (33)
Similarly, we can calculate the eld equation for . The result is the complex conjugate of Eq. (33).
Consider the U(1) transformation
(x) e
i
(x), and

(x) e
i

(x), (34)
where [0, 2) is a constant. It is straightforward to check that the Lagrangian (29) is invariant under this transformation.
This is an example of a global symmetry, i.e., the symmetry transformation acts in the same way on the elds at each point
in space and time.
The innitesimal transformation of (34) is:
+ , where = i, and (35)

= i. (36)
We can check explicitly that the Lagrangian density is invariant under this transformation:
/ =

() m
2
(

) (

)(

)
= i

+ i

+ im
2
(

) i(

)(

) = 0. (37)
According to Noethers theorem, the global symmetry implies the existence of a conserved current. If / /+/, where
/ =

, then the conserved current is given by the formula:

=
/
(

a
(x))

a
+

(38)
where a labels all the elds in the theory. In the present case, there are two independent elds and

, and
= i,

= i, / = 0 . (39)
The concerned current is thus:
j

=
/
(

)
(i)
/
(

)
(i

)
= i (

) (

)) . (40)
We can check explicitly that this current is conserved:

= i (

) (

)(

) + (

)(

))
= i (

))
= i
_

_
m
2
+ (

_
m
2

+ (

_
= 0. (41)
In the last step, we used the equation of motion for and

, Eq. (33).
4
Exercise 5
Intuitively we expect the Lagrangian to be invariant under SO(3) rotations of the elds
a
, a = 1, 2, 3, since only the length

a
enters the Lagrangian. To see this explicitly, we consider the transformation of
a

a
under a rotation of the elds
a
by an innitesimal angle

a

a
+
abc
n
b

c
, i.e.
a
=
abc
n
b

c
(42)
where n
a
is a constant unit vector.
a

a
transforms as

a
(
a
+
abc
n
b

c
)(
a
+
ade
n
d

e
)
=
a

a
+
abc
n
b

a
+
ade
n
d

a
+O(
2
)
=
a

a
+O(
2
).
In the last step we used that
abc
is skew-symmetric under exchange of c and a whereas
c

a
is symmetric under this
exchange, and therefore the sum vanishes. The same applies to the other term linear in . Dropping the
2
term we nd
that
a

a
is indeed invariant under (42). Similarly one can also show that

a
is invariant under (42). Therefore / is
invariant under (42).
To obtain the associated Noether current, we refer to the formula (38) above. In our case = 0 and
a
=
abc
n
b

c
,
so that
j

= (

a
)
abc
n
b

c
.
The Noether current j

implies a conserved charge

abc
n
b

c
.
This is conserved (

Q = 0) for any unit vector n. If we choose n
b
=
bd
for d = 1, 2, 3 (i.e. n is any of the standard basis
vector of R
3
), we obtain three linearly independent charges
Q
d
=
_
d
3
x
adc

c
=
_
d
3
x
dac

c
.
Let us nally show explicitly that these charges are conserved using the equations of motion. These are given by

a
+ m
2

a
=

a
+ m
2

a
= 0.
Then we get
dQ
a
dt
=
d
dt
_
d
3
x
abc

c
=
_
d
3
x
abc
(

c
+

c
)
=
_
d
3
x
abc
(

c
)
=
_
d
3
x
abc
__

b
_

c
m
2

=
_
d
3
x
abc

c
= 0.
Here we used the antisymmetry of the tensor several times and integrated by parts to go to the last line. We assume that
the eld falls o suciently fast so that we can neglect the boundary term of the partial integration.
Exercise 6
Under a Lorentz transformation, the vector x

transforms as:
x

, (43)
and such transformations preserve the metric, i.e. we know
g

= g

. (44)
But replacing the x

s by their denition in terms of

,
g

= g

)
_

_
=
_
g

_
x

=
_
g

_
x

. (45)
5
In the last step, we have interchanged the indices and with and . This is allowed since they are contracted, and
therefore this is just a renaming. Now, equating (44) with the last step in (45), it follows straightforwardly that
g

= g

. (46)
Now, let us consider an innitesimal Lorentz transformation of the form

, [[ 1 . (47)
Using (46), we nd that, keeping only the terms up to rst order in
g

=
_

_
g

_
= g

+O
_

2
_
, (48)
which requires that

is an antisymmetric tensor for the above transformation to indeed be a Lorentz transformation.

= g

(t)

(t)

. (49)
Looking innitesimally corresponds to dierentiating at the identity, i.e. consider
d
dt
_
g

(t)

(t)

t=0
= g

(0)

(0)

+ g

(0)

(0)

= g

(0)

+ g

(0)

.
However, according to (49), this is precisely
d
dt
_
g

(t)

(t)

t=0
=
d
dt
g

t=0
= 0 .
Therefore, if we dene a matrix , by

(0)

,
we have in all that
0 = g

+ g

.
To make the connection with our earlier innitesimal form of the Lorentz transformation

, we can write down the Taylor

= (0)

+ t

(0)

+O(t
2
)
=

+ t

+O(t
2
) .
Besides being rigorous, the benet of this approach is that innitesimal is characterised by a parameter t being small,
while

is any skew-symmetric matrix and not an innitesimal object. One-parameter families of transformations, and in
particular one-parameter subgroups (i.e. smooth homomorphisms : R G), are precisely how one denes the Lie algebra
of a group and its exponential map in a more general setting - i.e. when the group isnt given as being embedded in GL(n, R).
The form of the innitesimal rotations and boosts are given by:
Rotation:
A generic rotation by and angle about the x
3
-axis is given by:
_
_
_
_
1 0 0 0
0 cos sin 0
0 sin cos 0
0 0 0 1
_
_
_
_
. (50)
Taylor expanding the trigonometric functions up to rst order in , we nd that an innitesimal rotation around the
x
3
-axis is given by:

+
_
_
_
_
0 0 0 0
0 0 0
0 0 0
0 0 0 0
_
_
_
_
. (51)
6
As a check on the above, exponentiate the matrix
=
_
_
_
_
0 0 0 0
0 0 1 0
0 1 0 0
0 0 0 0
_
_
_
_
.
In order to do this, note that

2
=
2
_
_
_
_
0 0 0 0
0 1 0 0
0 0 1 0
0 0 0 0
_
_
_
_
,
so that

2k
= (1)
k

2k
_
_
_
_
0 0 0 0
0 1 0 0
0 0 1 0
0 0 0 0
_
_
_
_
, k 1 and
2k+1
= (1)
k

2k+1
_
_
_
_
0 0 0 0
0 0 1 0
0 1 0 0
0 0 0 0
_
_
_
_
, k 0 .
Now the exponentiation can be calculated from
exp () = I +

k=1
1
2k!

2k
+

k=1
1
(2k + 1)!

2k+1
=
_
_
_
_
1 0 0 0
0 1 0 0
0 0 1 0
0 0 0 1
_
_
_
_
+
_

k=1
1
2k!
(1)
k

2k
_
_
_
_
_
0 0 0 0
0 1 0 0
0 0 1 0
0 0 0 0
_
_
_
_
+
_

k=1
1
(2k + 1)!
(1)
k

2k+1
_
_
_
_
_
0 0 0 0
0 0 1 0
0 1 0 0
0 0 0 0
_
_
_
_
=
_
_
_
_
1 0 0 0
0 0 0 0
0 0 0 0
0 0 0 1
_
_
_
_
+ cos
_
_
_
_
0 0 0 0
0 1 0 0
0 0 1 0
0 0 0 0
_
_
_
_
+ sin
_
_
_
_
0 0 0 0
0 0 1 0
0 1 0 0
0 0 0 0
_
_
_
_
=
_
_
_
_
1 0 0 0
0 cos sin 0
0 sin cos 0
0 0 0 1
_
_
_
_
.
Boost:
A generic boost along the x
1
-axis is given by:
_
_
_
_
v 0 0
v 0 0
0 0 1 0
0 0 0 1
_
_
_
_
. (52)
Recall that the rapidity is dened via tanh = v, so that the above boosts are parameterised in analogy to the
rotations by
_
_
_
_
cosh sinh 0 0
sinh cosh 0 0
0 0 1 0
0 0 0 1
_
_
_
_
. (53)
Taylor expanding to rst order about = 0 (i.e. v = 0), the innitesimal boost by v is given by:

+
_
_
_
_
0 1 0 0
1 0 0 0
0 0 0 0
0 0 0 0
_
_
_
_
. (54)
7
We will again check that the above innitesimal transformation gives the correct nite transformation when we
exponentiate the matrix
=
_
_
_
_
0 1 0 0
1 0 0 0
0 0 0 0
0 0 0 0
_
_
_
_
.
Calculate the square and notice the pattern emerging

2
=
2
_
_
_
_
1 0 0 0
0 1 0 0
0 0 0 0
0 0 0 0
_
_
_
_
,
so that the even and odd powers are

2k
=
2k
_
_
_
_
1 0 0 0
0 1 0 0
0 0 0 0
0 0 0 0
_
_
_
_
, k 1
2k+1
=
2k+1
_
_
_
_
0 1 0 0
1 0 0 0
0 0 0 0
0 0 0 0
_
_
_
_
, k 0 .
Now the exponentiation can be calculated from
exp () = I +

k=1
1
2k!

2k
+

k=1
1
(2k + 1)!

2k+1
=
_
_
_
_
1 0 0 0
0 1 0 0
0 0 1 0
0 0 0 1
_
_
_
_
+
_

k=1
1
2k!

2k
_
_
_
_
_
1 0 0 0
0 1 0 0
0 0 0 0
0 0 0 0
_
_
_
_
+
_

k=1
1
(2k + 1)!

2k+1
_
_
_
_
_
0 1 0 0
1 0 0 0
0 0 0 0
0 0 0 0
_
_
_
_
=
_
_
_
_
0 0 0 0
0 0 0 0
0 0 1 0
0 0 0 1
_
_
_
_
+ cosh
_
_
_
_
1 0 0 0
0 1 0 0
0 0 0 0
0 0 0 0
_
_
_
_
sinh
_
_
_
_
0 1 0 0
1 0 0 0
0 0 0 0
0 0 0 0
_
_
_
_
=
_
_
_
_
cosh sinh 0 0
sinh cosh 0 0
0 0 1 0
0 0 0 1
_
_
_
_
.
Food for thought: There is some subtlety involved in the above operations:
1. If one takes a group, is it true that exponentiation (a representation of) its algebra gives an element of (a representation
of) the original group? Seen from a dierent perspective, can two groups have the same algebra?
2. If the answer to the previous question is positive, can one obtain all elements in this way?
Unfortunately, the answer to both of these questions is negative. Fortunately, the answer is almost as nice as one would
wish, and is certainly much more interesting than a bland armative.
Aspects of the relevant theory may be covered in the course Symemtries and Particles and is covered in the course Lie
Algebras and Their Representations though, needless to say, in great abstraction. Otherwise, see Warner, Foundations of
Dierentiable Manifolds for general background of Fulton & Harris, Representation Theory for full details.
Exercise 7
From the above derivation, under an innitesimal Lorentz transformation, a vector x

transforms as
x

= x

. (55)
Similarly, a scalar eld transforms according to
(x

(x

) =
_
(
1
)

_
= (x

) . (56)
8
Where in the last step we needed to know what
1
looks like innitesimally. To gure this out, we look at the denition
that it preserves the metric
g

,
and we nd that the inverse Lorentz transformation is simply given by
_

1
_

= g

.
Note that if the metric were Euclidean, this would simply be the statement
T
=
1
, i.e. is an orthogonal matrix. We
now nd the inverse of the innitesimal transformation (neglecting terms of O(
2
))
_

1
_

g
mu
=

.
In the second line we use the skew-symmetry of

.
Now, Taylor expanding to rst order in

we obtain
(x

(x

) (x

(x

) . (57)
The Lagrangian density, /, is a Lorentz scalar and will therefore transform in the way just found under an innitesimal
Lorentz transformation
/(x

) /

(x

) = /(x

) t

/(x

) , (58)
so that / =

/. If you are unconvinced by this reasoning, check explicitly that the Lagrangian for the scalar eld
theory, given the transformation for the scalar eld , does indeed change in this way.
Now look at

/) =

) / +

/
=

(x

) / +

/
=

/ +

/
=

/, (59)
where in the second step we have used that

is a constant tensor, and in the last step we made use of the skew-symmetry
of

to infer that

= 0. Therefore, we obtain that the variation of the Lagrangian density id a total derivative
/ =

/) . (60)
From Noethers theorem, there must be a conserved current associated with this symmetry, given by
j

=
/
(

)
+ F

, (61)
with F dened via / =

. Inserting the expressions derived above for (x

(x

) and F

, we obtain:
j

=
_

_
_

/
(

/
_
=

. (62)
Here, T refers to the energy-momentum tensor. The total charge Q is given by
Q
_
d
3
x(j
0
) (63)
=
_
d
3
x

T
0

. (64)
For pure spatial rotation we have
i
j
,= 0 and
0
i
=
i
0
= 0 where i, j = 1, . . . 3 are spatial indices. In this case
Q =
_
d
3
x
j
k
x
k
T
0
j
=
_
d
3
x
jk
x
k
T
0j
=
1
2
_
d
3
x
jk
_
x
j
T
0k
x
k
T
0j
_
=

jk
2
_
d
3
x
_
x
j
T
0k
x
k
T
0j
_
, (65)
9
where in the third line the skew-symmetry was used to discard the symmetric part of x
k
T
0j
.

jk
is just a constant times a parameterisation of our choice of rotation and so we will aim to remove it from the
expression of the charge. Begin by realising that by its skew-symmetry one can write

jk
=
jki
v
i
,
for some vector v
i
which is normal to the rotation. This can be thought of as dening an angular momentum vector
as in classical dynamics. Then
Q =
1
2
v
i

ijk
_
d
3
x
_
x
j
T
0k
x
k
T
0j
_

1
2
Q
i
v
i
,
and we can identify Q
i
as the three linearly independent charges
Q
i
=
ijk
_
d
3
x
_
x
j
T
0k
x
k
T
0j
_
, (66)
which have the interpretation of angular momenta.
For a Lorentz boost we have only
0
i
,= 0 and
i
j
= 0. A simple calculation also shows that
w
k
0
=
k0
=
0k
=
0
k
.
Therefore in this case
Q =
_
d
3
x
_

0
i
x
i
T
0
0
+
i
0
x
0
T
0
i
_
=
0
i
_
d
3
x
_
x
i
T
0
0
x
0
T
i
0
_
=
0i
_
d
3
x
_
x
0
T
0i
x
i
T
00
_
,
where for the 0 raised and that lowered we each get a 1 so there is no overall sign change. Since there is such a
conserved quantity for each choice of
0i
and there are three such independent choices we nd the 3-vector of conserved
charges
Q
i
=
_
d
3
x
_
x
0
T
0i
x
i
T
00
_
.
Since Q
i
is conserved, taking its derivative with respect to time gives zero
d
dt
Q
i
= 0 0 =
d
dt
__
d
3
x x
0
T
i0
_

d
dt
__
d
3
x x
i
T
00
_
=
_
d
3
x T
i0
+ t
_
d
3
x
d
dt
_
T
i0
_

d
dt
__
d
3
x x
i
T
00
_
,

d
dt
__
d
3
x x
i
T
00
_
=
_
d
3
x T
i0
+ t
_
d
3
x
d
dt
_
T
i0
_
.
Let us investigate the terms in this equation separately. T
00
is the energy density of the quanum eld while T
0k
is
its linear momentum density. Because the theory is invariant under translations, the total linear momentum must be
conserved thus
_
d
3
x T
i0
= const. Similarly, this means that the second term on the right-hand side vanishes. So we
simply get
d
dt
__
d
3
x x
i
T
00
_
=
_
d
3
x T
i0
= const.
This has a simple interpretation: the integral on the LHS gives the centre of energy of the eld and so the equation is
telling us that it moves at a constant velocity.
Exercise 8
Let us begin in the greatest generality, i.e. work in (n+1)-dimensional Minkowski space-time. We are given that the group
(R
+
, ) of positive reals under multiplication acts on the space of eld congurations via the action
( )(x)
D
(
1
x)
10
for some D, the scaling dimension of . In other (perhaps less technical) words, we have a transformation
(x)
D
(
1
x) , > 0.
We show that this is a symmetry of the action
S[] =
_
d
n+1
x/((x), (x)) ,
where the Lagrangian density is given as
/((x), (x)) =
1
2

(x)

(x)
1
2
m
2
(x)
2
g(x)
p
,
for constants m, p and g. We assume p ,= 2, else the third term simply serves to modify the mass.
Recall that a group action is a symmetry i we have
S[ ] = S[] , R
+
,
i.e. if the above transformation leaves S unchanged. So we must show that
S[ ] =
_
d
n+1
x
_

1
2

2D

_
(
1
x)

_
(
1
x)

1
2
m
2

2D
(
1
x)
2
g
pD
(
1
x)
_
does not depend on . Making the subtitution x

=
1
x, then as > 0,
d
n+1
x = [
n+1
[ d
n+1
x

=
n+1
d
n+1
x

=
1

x

and, abbreviating

,
we nd
S[ ] =
_
d
n+1
x

n+1
_

1
2

2D2

(x

(x

)
1
2
m
2

2D
(x

)
2
g
pD
(x

)
_
,
which, since we can replace x

with x as a dummy variable in the integral, is the original value of the action i each term in
the integrand,

2D2+n+1

, m
2

2D+n+1

2
, g
pD+n+1

p
,
does not depend on . Invariance of the rst term entails
D =
1
2
(n 1).
Having found D, invariance of the second term occurs i m = 0. Finally, the third term is invariant either when g = 0, else
if g ,= 0, then
p = 2
n + 1
n 1
.
In the case n = 3, therefore D = 1 and p = 4.
Having established that this is a symmetry, we nd the conserved current associated to the innitesimal transformation in
the case n = 3, i.e. we look at the rst order changes in the elds and Lagrangian. The change in to rst order is, by the
chain rule,
(x) =
d
d
( )(x)

=1
=
d
d
_

1
(
1
x)
_

=1
= (x) x

(x).
The Lagrangian transforms as
/(( )(x), [( )] (x)) =
4
/
_
(
1
x), (
1
x)
_
11
Since / on the right depends on only through the space-time dependence of the elds, we nd the innitesimal change in
/ to be, again by the chain rule,
/ =
d
d
/(( )(x), [( )] (x))

=1
= 4/((x), (x)) x

[/((x), (x))] .
As was to be expected from general principles, this is precisely the divergence
/ =

, where

= x

/.
The conserved current provided by Noethers theorem is,
j

=
/

.
Plugging in the expressions found above, we eventually obtain the conserved current associated to our scaling symmetry
j

( + x

) x

/.
Once the eld equations have been found, it is straightforward to check that this is indeed conserved.
Exercise 9
In the Heisenberg picture the elds are time-dependent. This is related to the Schrodinger picture by

H
(t, x) = e
iHt

S
(x)e
iHt
We omit the subscripts and use the four-vector notation (x) to distinguish the between the two pictures. The Hamiltonian
for the system is given by
H =
1
2
_
dy
3
_

2
(y) +[(y)[
2
+ m
2

2
(y)
_
This system is quantised by imposing equal time commutation relations on (y) and (y) in the Heisenberg picture as,
[(t, x), (t, y)] = [(t, x), (t, y)] = 0 , [(t, x), (t, y)] = i
(3)
(x y)
Now it follows that

(x) = iH(e
iHt
(x)e
iHt
) i(e
iHt
(x)e
iHt
)H = i[H, (x)]
In terms of the mode decompositions, the only non-vanishing piece is the commutator of and .
i[H, (x)] = i
_
1
2
_
dy
3
(
2
(y) +[(y)[
2
+ m
2

2
(y)), (x)
_
=
i
2
_
dy
3
[
2
(y), (x)] = (x) (67)
where we have used the equal time commutation relations and the fact that in (y) the derivative acts on y and hence can
be pulled out of the commutator [(y), (x)] which then equals zero. Simillarly, for the other case we have,
(x) = i[H, (x)] =
i
2
_
dy
3
_

2
(y) +[(y)[
2
+ m
2

2
(y), (x)

= i
2
_
dy
3
_
(y).
(3)
(y x) + m
2
(y)
(3)
(y x)
_
=
2
(x) m
2
(x) (68)
It now follows from Eq.(67) and Eq.(68) that (x) satises the Klein-Gordon equation

(x) +
2
(x) = m
2

2
(x)
Exercise 10
The relativistically normalised one particle states are [p =
_
2E
p
a

p
[0. Using
(x) =
_
d
3
q
(2)
3
1
_
2E
q
(a
q
e
iqx
+ a

q
e
iqx
)
12
we nd
0[(x)[p =
_
d
3
q
(2)
3

2E
p
2E
q
0[(a
q
e
iqx
+ a

q
e
iqx
)a

p
[0
=
_
d
3
q
(2)
3

2E
p
2E
q
e
iqx
0[a
q
a

p
[0
=
_
d
3
q
(2)
3

2E
p
2E
q
e
iqx
(2)
3

(3)
( p q)0[0
= e
ipx
Where we have used 0[a

q
= 0 , [a
q
, a

p
] = (2)
3

(3)
( p q) and 0[0 = 1. Comparing this with the relation x[ p = e
i p.x
from quantum mechanics, we interpret (x)[0 as a particle at x.
13