Lecture Notes for Quantum Physics 3C26E

Andrew Horseld
8th March 2004
2
Contents
1 Formal Aspects of Quantum Theory 5
1.1 The Wave Function . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.2 Principle of Superposition . . . . . . . . . . . . . . . . . . . . . . 7
1.3 Time Dependent Schroedinger Equation . . . . . . . . . . . . . . 8
1.4 Expectation Values . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.5 Hermitian Operators . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.6 Eigenstates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.7 Expansion Postulate and Complete Sets of Eigenfunctions . . . . 17
1.8 Compatible Observables . . . . . . . . . . . . . . . . . . . . . . . 19
1.9 Simultaneous Measurement . . . . . . . . . . . . . . . . . . . . . 19
1.10 Commuting Operators . . . . . . . . . . . . . . . . . . . . . . . . 19
1.11 The Generalised Uncertainty Relations . . . . . . . . . . . . . . . 20
1.12 Dirac Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
1.13 Matrix Representation of States and Operators . . . . . . . . . . 24
1.14 Time Evolution of Operators . . . . . . . . . . . . . . . . . . . . 26
1.15 Step-Operator Approach to Harmonic Oscillator . . . . . . . . . 28
2 Angular Momentum [8] 33
2.1 A Refresher on Commutation Relations . . . . . . . . . . . . . . 33
2.2 Eigenvalues and Eigenfunctions of Orbital Angular Momentum
Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
2.3 Generalized Angular Momentum and Step Operator Techniques
in Angular Momentum Theory . . . . . . . . . . . . . . . . . . . 37
2.4 Spin-1/2 Angular Momentum and Pauli Matrices . . . . . . . . . 39
2.5 Magnetic Moments . . . . . . . . . . . . . . . . . . . . . . . . . . 43
2.6 Combination of Angular Momenta . . . . . . . . . . . . . . . . . 43
3 Approximate Methods [6] 47
3.1 Time-Independent Perturbation Theory . . . . . . . . . . . . . . 47
3.2 Variational Principle . . . . . . . . . . . . . . . . . . . . . . . . . 53
4 Simple Time-dependent systems [3] 55
4.1 Superposition of States of Dierent Energies . . . . . . . . . . . . 55
4.2 Electron in Magnetic Field . . . . . . . . . . . . . . . . . . . . . . 56
4.3 Time Evolution of Entangled States of Two Spin-1/2 Particles
with Total Spin Zero. . . . . . . . . . . . . . . . . . . . . . . . . . 58
3
4 CONTENTS
5 Identical Particles [2] 59
5.1 Systems of Two Identical Particles . . . . . . . . . . . . . . . . . 60
5.2 Independent Particle Model of He Atom . . . . . . . . . . . . . . 62
6 Interpretations of quantum mechanics [4] 65
6.1 Wave-Particle Duality and Indeterminacy . . . . . . . . . . . . . 65
6.2 Double slit experiment . . . . . . . . . . . . . . . . . . . . . . . . 65
6.3 Beam splitter experiment . . . . . . . . . . . . . . . . . . . . . . 66
6.4 Copenhagen Interpretation . . . . . . . . . . . . . . . . . . . . . 67
6.5 Hidden Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
6.6 Non-Locality and the Einstein, Podolsky, Rosen (EPR) Paradox 68
6.7 Bells Inequalities . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
6.8 The Aspect Experiments . . . . . . . . . . . . . . . . . . . . . . . 70
6.9 The Problem of Measurement . . . . . . . . . . . . . . . . . . . . 71
6.10 Schrodingers Cat . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
6.11 Mind and Matter . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
6.12 Indelible Records . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
6.13 Path Integral Approach . . . . . . . . . . . . . . . . . . . . . . . 73
6.14 The Many Universes Interpretation . . . . . . . . . . . . . . . . . 74
6.15 Alternative Histories . . . . . . . . . . . . . . . . . . . . . . . . . 75
6.16 The Ghiradi-Rimini-Weber scheme . . . . . . . . . . . . . . . . . 75
6.17 Gravitationally Induced Reduction . . . . . . . . . . . . . . . . . 76
6.18 Quantum Information . . . . . . . . . . . . . . . . . . . . . . . . 76
Chapter 1
Formal Aspects of Quantum
Theory
1.1 The Wave Function
Consider a set of N particles with positions r
1
r
N
The wave function ( r
1
r
N
, t) is a complex function (it has both a real
and an imaginary part) of the coordinates of the particles (and time). It
contains all the information about the particles at time t.
The fundamental property of the wave function is that it gives the proba-
bility of nding the particles in a given conguration. That is [( r
1
r
N
, t)[
2
d r
1
d r
N
is the probability of nding particle 1 in volume d r
1
centred about r
1
and
particle 2 in volume d r
2
centred about r
2
etc., at a particular time t.
We have used the notation [[
2
=

= Re
2
+ Im
2
dr = dxdydz = r
2
dr sindd
This interpretation in terms of probability leads to the normalisation of
the wave function dened by

[( r
1
r
N
, t)[
2
d r
1
d r
N
= 1. This just
means that the probability of nding the system in one of the allowed
congurations is one.
Note that because it is the modulus squared of the wave function which
denes the physically observable properties, we can multiply the wave
function by a phase factor exp(i) without changing any of the results.
Note that normalisation imposes a condition on allowed wave functions.
One particularly important wave function is not normalisable, namely a
plane wave (exp(i

k r)). We can get around this problem by working

with a wave packet, in which the plane wave is multiplied by another
function which decays rapidly to zero for large distances, or we can use
box normalisation with periodic boundaries.
5
6 CHAPTER 1. FORMAL ASPECTS OF QUANTUM THEORY
Figure 1.1: Innite square well
V
x L
0
Other average properties of the particles are found by taking expectation
values. Consider the position of a particle:
'x
i
` =

[( r
1
r
N
, t)[
2
x
i
dr
1
d r
N
(1.1)
We will see later that in general we can write

( r
1
r
N
, t)

Q( r
1
r
N
, t) d r
1
d r
N
= 'Q` (1.2)
where 'Q` is the average measured quantity, and

Q is the corresponding
operator.
Example
Consider an electron in a box with innitely hard walls:
Allowed wave functions are

n
(x) =

2
L
sin(
nx
L
) 0 x L
0 Otherwise
(1.3)
Normalisation leads to the following

[
n
(x)[
2
dx =
2
L

L
0
sin(
nx
L
) sin(
nx
L
) dx
=
2
L

L
0
1
2
(1 cos(
n2x
L
)) dx
= 1
1.2. PRINCIPLE OF SUPERPOSITION 7
Figure 1.2: Probability distribution
0 0.2 0.4 0.6 0.8 1
x/L
0
0.5
1
1.5
2
P
(
x
)
L
n=1
n=2
n=3
The probability of nding a particle near x depends on the state
P
n
(x) dx =

2
L
sin
2
(
nx
L
) dx 0 x L
0 Otherwise
(1.4)
1.2 Principle of Superposition
In optics in the two slit experiment we have waves emanating from two
separate states that have a constant phase relationship, and thus produce
an interference pattern.
In quantum mechanics we have the same eect. It is possible to have
two coherent sources of particles (for example electrons) that can interfere
with one another.
This interference eect in both cases is produced by the superposition of
two coherent waves.
In general to combine several wave functions for the same particle to pro-
duce a total wave function we must add the wave functions together.
Consider our particle in a box

n
=

2
L
sin(
nx
L
) exp(
E
n
t
i
)
=

2
L
1
2i

exp(
inx
L
) exp(
inx
L
)

exp(
E
n
t
i
)
=
i

2L
exp(
inx
L
) exp(
E
n
t
i
) +
i

2L
exp(
inx
L
) exp(
E
n
t
i
)
=
+
(x)

(x)
8 CHAPTER 1. FORMAL ASPECTS OF QUANTUM THEORY
The wave function for a particle in free space is

free
(x) = N exp(ikx)
When k > 0 the particle is moving to the right, and when k < 0 the
particle is moving to the left.
Thus we can interpret our particle in a box as being in a state that is a
superposition of a left travelling and right travelling states, as you might
expect for a particle bouncing between two walls.
This superposition produces interference eects
[[
2
= [
1
+
2
[
2
= [
1
[
2
+[
2
[
2
+ (

2
+

1
)
If we dene the phase angles
1
and
2
by the equations
1
= [
1
[ exp(i
1
)
and
2
= [
2
[ exp(i
2
) then we get the following interference expression
[[
2
= [
1
[
2
+[
2
[
2
+ 2[
1
[[
2
[ cos(
1

2
)
This produces the standing wave pattern for the particles in the box.
1.3 Time Dependent Schroedinger Equation
From the photoelectric eect we have E = h and p = h/.
De Broglie postulated that these expressions apply to all particles.
If we write k = 2/, = 2 and = h/2 then we can re-express these
as E = and p = k.
We use plane waves (exp(i(kx t)) to describe the propagation of elec-
tromagnetic waves in free space. We would like to identify this with
the wave function of the photon and hence of all free particles ((xt) =
N exp[i(kx t)]).
This function satises the following equations
i

t
= = E
i

x
= k = p
These equations provide the basic structure of quantum mechanics
Observables are represented by operators
E i

t
p
x
i

x
The state of the system is described by the wave function.
1.3. TIME DEPENDENT SCHROEDINGER EQUATION 9
To determine wave functions for more general systems we need to extend
our ideas. The constraints we have are
The equations need to be linear in so that we can make use of
superposition.
They need to be linear in

t
so that we do not need to dene any
derivatives of the wave function as part of our initial conditions (t =
0) since we wish to have the system completely dened by the wave
function.
They should reproduce the results of classical mechanics in the limit
of large systems.
We also have the quantum mechanical postulates. One way to put all this
together is the following
i

t
=

E
op

Here

E
op
is an energy operator motivated by classical mechanics and has
the form

E
op
=

T
op
+

V
op
where

T
op
is the kinetic energy operator and

V
op
is the potential energy operator. But from classical mechanics we know
that T = p
2
/2M, and we already have an expression for the momentum
operator ( p
op
= i

x
). Thus we propose the following equations of
motion for the wave function
i

t
= [

2
2M

2
x
2
+V (x)]
This forms the heart of quantum dynamics. It is the quantum mechanical
equivalent of Newtons second law of otion. It is, of course, Schroedingers
equation.
The more precise, if abstract, form of the equation is

H = i

t
where

H is the Hamiltonian operator. The Hamiltonian is a version of the total

energy borrowed from classical mechanics which has the property that it
is constant throughout the motion of a system in the absence of external
time varying forces. Often it just equals the total energy, which is what
we assume here.
In three dimensions, but still for one particle, the Schroedinger equation
becomes
i
(x, y, z, t)
t
=

2
2M
(

2
x
2
+

2
y
2
+

2
z
2
) +V (x, y, z)

(x, y, z, t)
This can be written in a more compact form using the following notation
i
(r, t)
t
=

2
2M

2
+V (r)

(r, t)
10 CHAPTER 1. FORMAL ASPECTS OF QUANTUM THEORY
When the potential is independent of time we can make the following very
useful separation (r, t) = (r)(t). If we substitute this product into
Schroedingers equation we obtain the following
1
(t)
i
(t)
t
=
1
(r)

2
2M

2
+V (r)

(r)
Since the left-hand side depends only on time, and the right-hand side
depends only on position, both sides must be equal to a constant which
we shall call E. We can then separate our one equation into two equations.

2
2M

2
+V (r)

(r) = E(r)
i
(t)
t
= E(t)
The rst of these equations is the time independent Schroedinger equation.
The second of these equations we can solve immediately (t) = exp(Et/i).
We can now form a general solution to the time dependent Schroedinger
equation by using the principle of superposition.
Let us index the allowed values of E by n (E E
n
). Then we have the
following expansion (r, t) =

n
C
n

n
(r) exp(E
n
t/i). C
n
is a constant
coecient.
The coecients C
n
are determined from the wave function at time t = 0
((r, 0) =

n
C
n

n
(r)). As we will prove later we have the following
condition

m
(r)
n
(r) dr =
nm
where
nm
is the Kroenecker -function
given by

nm
=

0 n = m
1 n = m
Hence we can obtain the coecients

n
C
n

m
(r)
n
(r) dr = C
m
=

m
(r)(r, 0) dr.
Example
In free space we have no potential. Thus the Hamiltonian contains only
the kinetic energy.
Our one-dimensional Schroedinger equation is thus i

t
=

2
2M

x
2
. If
we make a separation of variables we obtain
1
(t)
i
(t)
t
=
1
(x)

2
2M

2
(x)
x
2

=
E. If we dene E = = p
2
/2M, and p = k then we obtain (x, t) =
N exp(i[t kx]), which should now all look familiar.
To produce the general wave function we construct the sum of individ-
ual terms. However in this case we have a continuum of energy levels
allowed, since the wave vector k can have any real value. Therefore
we must replace our sum by an integral. We therefore have (x, t) =

dk c(k) exp(i[(k)t kx]).

We obtain the expansion coecients from the equation (x, 0) =

dk c(k) exp(ikx).
1.3. TIME DEPENDENT SCHROEDINGER EQUATION 11
Figure 1.3: Spread of gaussian wavepacket
-10 -5 0 5 10
0
0.2
0.4
0.6
0.8
1
t=0
t=1
t=3
t=10
Example
There is a second solution to the Schroedinger equation for a free particle
which corresponds to what is called a wave packet. To nd this solution
we will write down its form and show that it satises the Schroedinger
equation.
The trial wave function is
(x, t) = N
1

1 +t
f

(x/)
2
1 +t

In order to substitute this into the Schroedinger equation we need the

following derivatives

t
= N

/2

(1 +t)
3
f
(x/)
2

(1 +t)
3
f

x
2
= N

2/
2

(1 +t)
3
f

+
4
2
x
2
/
4

(1 +t)
5
f

Schroedingers equation is satised provided the following conditions hold

true: f(y) = f

(y) = f(y) = exp(y), =

1
4
and = i/2M
2
. If these
conditions are satised then we have
(x, t) = N exp

1
4
(x/)
2
1 + it/2M
2

1 + it/2M
2
Let us introduce a timescale into this problem = 2M
2
/. For an
electron, assuming 10
10
m, this will have a value of about 10
16
s.
12 CHAPTER 1. FORMAL ASPECTS OF QUANTUM THEORY
We can then rewrite our wave function as
(x, t) = N exp

1
4
(x/)
2
1 + it/

1 + it/
From this we are able to calculate the probability density function which
is
[(x, t)[
2
= N
2
exp

1
2
(x/)
2
1 + (t/)
2

1 + (t/)
2
When the time is much greater than our timescale (t >> ) we get
[(x, t)[
2
N
2

t
exp

1
2

x
t/

If we dene the eective width of our wave packet by

(t) = t/, then

we see that the wave packet is spreading with a constant speed of about
10
6
m/s.
1.4 Expectation Values
If you perform an experiment many times and measure some quantity
which we will call Q (for example position or momentum) each time, then
you can nd an average value
'Q` =
1
N
N

i=1
Q
i
'Q` can also be calculated from the quantum mechanical operator

Q ( x,
p
x
= i

x
, etc.) for this observable and the wave function. In this case
it is referred to as an expectation value and is given by:
'Q` =

(r)

Q(r) dr
This will be elaborated upon later.
Note that this is an average value. What you obtain from any one given
experiment will in general dier from this.
Once again consider a particle in a box with innitely hard walls for which
the wave functions are given by:

n
(x) =

2
L
sin(
nx
L
) 0 x L
0 Otherwise
1.5. HERMITIAN OPERATORS 13
The expectation value of position is
'x`
n
=
2
L

L
0
sin(
nx
L
)xsin(
nx
L
) dx
=
2
L

L
0
1
2
[x xcos(
2nx
L
)] dx
=
L
2
On average we will nd a particle in the middle of the box. The expecta-
tion value of the momentum is
'p`
n
=
2
L

L
0
sin(
nx
L
)(i

x
) sin(
nx
L
) dx
=
2
L
(i
n
L
)

L
0
sin(
nx
L
) cos(
nx
L
) dx
= 0
The particle is equally likely to be travelling to the right or to the left, as
it is bouncing between two walls.
Expectation values can behave like classical variables. The Ehrenfest equa-
tions are interesting example
m
d'x`
dt
= 'p`
d'p`
dt
= '
V
x
`
These results will be proven later.
1.5 Hermitian Operators
An operator acts on a function to modify that function:

Of(r) = g(r)
For example, the derivative operator (

O =
d
dx
) acting on the function
f(x) = sinx:

Of(x) =
d
dx
sinx
= cos x
Linear operators obey the following relation

L( +) =

L +

L.
Operators corresponding to physical observables are hermitian.
Now consider integrals of the form

(r)

O(r) dr. We dene the her-
mitian conjugate of

O to be

O

where

(r)

O

(r) dr =

(O(r))

(r) dr
14 CHAPTER 1. FORMAL ASPECTS OF QUANTUM THEORY
Hermitian operator satises

O

=

O. That is

(r)

O

(r) dr =

(O(r))

(r) dr
=

(r)(

O(r)) dr
The operator can act on either wave function and give the same result.
Consider two operators

(r)(

A

B)

(r) dr =

(

A

B(r))

(r) dr
=

(

B(r))

(

A

(r)) dr
=

(r)

(

A

(r)) dr
(

A

B)

=

B

The expectation values of hermitian operators are always real

(

Adr)

dr
=

(

A)

dr
=

dr
=

Adr
If a number equals its complex conjugate, then it must be real.
Example
Consider the position operator x which satises xf(x) = xf(x):

(r) x

(r) dr =

( x(r))

(r) dr
=

(x(r))

(r) dr
=

(r)

x(r) dr
=

(r)

x(r) dr
Now consider the momentum operator i

x
. This also is hermitian:

(x)

(i

x
)(x) dx = i

(x)

(x)

x
(x)

(x) dx

=

i

x
(x)

(x) dx
1.6. EIGENSTATES 15
In the rst line we have used integration by parts. In the second line we
have assumed that at innity the wave functions decay to zero.
1.6 Eigenstates
We can ask the question: is there a state that produces one, and only
one, value for measurements of a particular observable? Mathematically,
we want a state that produces a particular expectation value for that
observable, and for which the standard deviation of the measurements is
zero. That is, 'A` = and 0 =

(

A )
2
dr. For an hermitian
operator this produces the following
0 =

(

A)
2
dr
=

(

A)(

A)dr
=

[(

A)]

(

A)dr
=

[(

A)[
2
dr
Because the integrand is positive denite, it must be the case that

A =
. This is an eigenequation.
We can generalise the above and say that eigenstates correspond to the
value measured in individual experiments (in contrast to expectation val-
ues which are averages over many experiments).
Eigenstates are properties of operators and are dened by the equation

Of

= f

where f

is an eigenfunction, and is a scaler, the eigenvalue.

An operator will in general have many eigenfunctions with corresponding
eigenvalues.
Example
Considered the momentum operator
i

x
f

(x) = f

(x)
f

(x) = Aexp(x/i)
Thus eigenfunctions of the momentum operator are complex exponentials
(or plane waves).
Note that in this example that are no constraints on the value of .
16 CHAPTER 1. FORMAL ASPECTS OF QUANTUM THEORY
Hermitian operators have the very important property that all their eigen-
values are real. This is necessary if they are to correspond to physical
observables.
Let

H be an hermitian operator which satises the following eigenequation

Hf

= f

. We shall now prove that its eigenvalues are real:

Hf

dr =

dr

dr =

dr
The second line is just the complex conjugate of the rst line. Since

H is
hermitian:

dr =

(

Hf

dr =

dr =

Hf

dr
Therefore

= , and therefore the eigenvalues is real.

The eigenfunctions of hermitian operators have the property that two
dierent eigenfunctions are orthogonal to one another:

f

1
f
2
dr = 0. To
prove this considered to dierent eigenstates of the same operator

Hf
1
=
1
f
1

Hf
2
=
2
f
2
This allows us then to write the following integrals

2

Hf
1
dr =
1

2
f
1
dr

1

Hf
2
dr =
2

1
f
2
dr
We now take the complex conjugate of the second equation, while recalling
that the eigenvalues are real

f
1

H

2
dr =
2

2
f
1
dr
If we now make use of the fact that the operator is hermitian we get from
above

2
f
1
dr =

(

Hf
1
)f

2
dr =
1

2
f
1
dr
Hence we get (
1

2
)

2
f
1
dr = 0, which means either the eigenvalues
are the same, which is not the case by construction, or that the eigenfunc-
tions are orthogonal.
If it turns out that two eigenvalues are in fact the same (the states are
degenerate), we can form linear combinations of the corresponding eigen-
functions which will also satisfy the eigenequation. We can choose those
linear combinations such that they are orthogonal to each other. There-
fore, all eigenfunctions are orthogonal to one another.
1.7. EXPANSIONPOSTULATE ANDCOMPLETE SETS OF EIGENFUNCTIONS17
Example
Once again we will consider the Hamiltonian for a particle in a box with
innitely hard walls

H =

2
2m

2
x
2
+V
box
(x)
Inside the box potential is 0, and so we have

2
2m

n
(x)
x
2
= E
n

n
(x)
This has the eigenvalues and eigenfunctions

n
(x) =

2
L
sin(
nx
L
) 0 x L
0 Otherwise
E
n
=

2
2m
(
n
L
)
2
We see that E
n
is real for all n.The wave functions are also orthonormal:
O
nm
=

n
(x)
m
(x)dx
=
2
L

L
0
sin(
nx
L
) sin(
mx
L
) dx
=
1
L

L
0

cos

(n m)x
L

cos

(n +m)x
L

dx
=

0 n = m
1 n = m
1.7 Expansion Postulate and Complete Sets of
Eigenfunctions
Suppose we have two Hamiltonians that are dierent, but they operate in
the same space. For example we could have a particle in a box, and another
particle in a similar box but which has a sloping oor. That is, in both
cases we need to nd a wave function for one particle in one dimension
over the same range of positions. What distinguishes the two cases is just
the shape of the potential. We thus solve very similar equations for the
two cases, but they have dierent eigenfunctions:

H
1

(1)
n
= E
(1)
n

(1)
n

H
2

(2)
n
= E
(2)
n

(2)
n
But because they operate in the same space we can expand one set of
eigenfunctions in terms of the other. This is analogous to choosing a
dierent set of axes to dene coordinates. Thus we can write:

(2)
n
(x) =

m
c
nm

(1)
m
(x)
18 CHAPTER 1. FORMAL ASPECTS OF QUANTUM THEORY
The coecients C
mn
can be determined by making use of the orthonor-
mality of the eigenfunctions:

(1)
p
(x)
(1)
q
(x) dx =
pq
Thus we get:

(1)
p
(x)
(2)
n
(x) dx =

m
c
mn

(1)
p
(x)
(1)
m
(x) dx
=

m
c
nm

pm
= c
np
The expansion is exact in this case if we include all the eigenfunctions.
An alternative way of generating a set of eigenfunctions is to consider the
same system twice, but consider two dierent operators. This reveals more
about the nature of expectation values.
Consider the following

Of
n
=
n
f
n
(x) =

m
c
m
f
m
(x)
c
m
=

m
dx
Where is the wave function that we are interested in. The expectation
value of the operator is then given by
'O` =

Odx
=

mn

c
m
f

m

Oc
n
f
n
dx
=

mn
c
m
c
n

n
f
n
dx
=

mn
c
m
c
n

mn
=

n
[c
n
[
2

n
Thus we can interpret [c
n
[
2
= [

f
n
dx[
2
as the probability of measuring
the physical quantity associated with the operator

O and nding that it
has a value
n
.
There is an important property of orthogonal functions that makes the use
of expansions very robust for practical calculations. Suppose we were only
allowed to use a fraction of the total number of functions to approximate
1.8. COMPATIBLE OBSERVABLES 19
our wave function

N
n=1

n
. We now try to nd the best possible
set of coecients:
R =

n
[
2
dr
R

m
=

m
(

n
) dr = 0

m
dr =
m
= c
m
Thus the best possible coecients for a nite sequence are identical to
those for the full expansion.
1.8 Compatible Observables
An observable corresponds to an eigenvalue of an hermitian operator
(

O

).
Compatible observables share eigenfunctions:

A = a

B = b
1.9 Simultaneous Measurement
Once a wave function collapses due to a measurement, and so enters an eigen-
state, you can extract all the eigenvalues of simultaneous observables without
disturbing the wave function further.
1.10 Commuting Operators
We dene a commutator by the following equation
[

A,

B] =

A

B

B

A
Example
We will consider the commutator between the position and momentum
operators.

A = x

B = p
x
= i

x
[

A,

B] =

A

B

B

A
[ x, p
x
] = x p
x
p
x
x
= x(i

x
)(i

x
)(x)
= i(x

x
+ +x

x
)
20 CHAPTER 1. FORMAL ASPECTS OF QUANTUM THEORY
= i
[ x, p
x
] = i
We now make the connection between commuting operators and simulta-
neous observables.
Suppose we have the following

A
n
= a
n

B
n
= b
n

n
=

n
c
n

n
We then obtain the following result
[

A,

B] = (

A

B

B

A)
=

n
c
n
(

A

B

B

A)
n
=

n
c
n
(

Ab
n

Ba
n
)
n
=

n
c
n
(a
n
b
n
b
n
a
n
)
n
= 0
If [

A,

B] = 0, the operators are said to commute.
Thus we see that simultaneous observables will have quantum mechanical
operators that commute.
Example
Consider a free particle. The Hamiltonian operator in this case is

H =
p
2
/2M.
Now consider the commutator between the Hamiltonian and the momen-
tum operator
[

H, p] =

H p p

H
=
1
2M
(

p
3
p
3
)
= 0
Thus the wave functions for this Hamiltonian are also eigenstates of the
momentum operator. Thus a free particle has a well-dened momentum.
1.11 The Generalised Uncertainty Relations
The uncertainty principle relates to two sets of repeated measurements
on the same system. In one set we measure one quantity (for example
position), while in the other set we measure another quantity (for example
momentum).
1.11. THE GENERALISED UNCERTAINTY RELATIONS 21
For each set we can calculate a mean value and a standard deviation. Let
the mean values be 'A` and 'B` and let the standard deviations be
A
and
B
.
In classical mechanics it is possible in theory to perform measurements for
which
A
and
B
are both arbitrarily small. That is, the system can have
unique values of both A and B.
In quantum mechanics there are pairs of observables (for example position
and momentum) for which this is not true.
Variables for which this is not true, have corresponding quantum operators
that fail to commute. We have already seen that [ x, p
x
] = i.
We can now put this on a proper mathematical footing. Let the two op-
erators corresponding to the two observables be

A and

B, and let them
satisfy [

A,

B] = i. We then have the following denitions of the expecta-
tion values and variances
'A` =

Adx
'B` =

Bdx

2
A
=

(

A'A`)
2
dx =

((

A'A`))

(

A'A`)dx

2
B
=

(

B 'B`)
2
dx =

((

B 'B`))

(

B 'B`)dx
Let us now dene
a(x) = (

A'A`)(x)
b(x) = (

B 'B`)(x)
We then get

2
A
=

[a(x)[
2
dx

2
B
=

[b(x)[
2
dx
Let us now dene a new function (x) = a(x)/
A
e
i
b(x)/
B
.
Clearly this function satises the following inequality

[(x)[
2
dx 0.
We therefore obtain the following
0

(a

(x)/
A
e
i
b

(x)/
B
)(a(x)/
A
e
i
b(x)/
B
) dx

2
A

[a(x)[
2
dx +
1

2
B

[b(x)[
2
dx
1

[a

(x)b(x)e
i
+b

(x)a(x)e
i
] dx
1 + 1
1

[a

(x)b(x)e
i
+b

(x)a(x)e
i
] dx
22 CHAPTER 1. FORMAL ASPECTS OF QUANTUM THEORY

B

1
2

[a

(x)b(x)e
i
+b

(x)a(x)e
i
] dx

1
2
[e
i

(x)(

A'A`)(

B 'B`)(x) dx +
e
i

(x)(

B 'B`)(

A'A`)(x) dx]
If we let e
i
= i, then we obtain

B

i
2

(x)[(

A'A`), (

B 'B`)](x) dx

i
2

(x)[

A

B

A'B` 'A`

B +'A`'B`

B

A+'B`

A+

B'A` 'B`'A`](x) dx

i
2

(x)[

A,

B](x) dx

2
The wave function with the smallest combined uncertainty (
A

B
= /2)
satises the equation (x) = 0. Therefore it satises a(x)/
A
+ib(x)/
B
=
0.
Example
If we choose our operators to be the position and momentum operators
then the equation we have to solve is

x 'x`

x
i'p`

(x) = 0
The solution to this equation is the following
(x) = N exp

ix'p`

x 'x`
2
x

This is a Gaussian wave packet. It forms one possible link with classical
mechanics.
1.12 Dirac Notation
This is a very convenient notation that achieves two things
1. It allows us to represent integrals in a compact fashion
2. It allows us to write quantum mechanics in a more abstract, and
hence general, fashion.
This is achieved by thinking of functions as vectors in an abstract space,
and integrals as dot products.
1.12. DIRAC NOTATION 23
Consider the following integral

dr. In Dirac notation we would

write this as '[`.
The quantity '[ is called a bra, and [` is called a ket. Think bra-c-ket.
Note that these exist in dierent spaces (they are complex conjugates of
one another), so cannot be added to each other.
The absence of explicit coordinates is what allows us to be more abstract.
We immediately have the following result '[` = '[`

.
If '[` = 0 then [` and ` are orthogonal.
We have the following relations between bras and kets:
[` = [a` +[b` '[ = 'a[ +'b[
c[a` = [a`c c

'a[ = 'a[c

[` = c
1
[a` +c
2
[b` '[ = c

1
'a[ +c

2
'b[

A[` '[

[a` +[b` = [b` +[a` 'a[ +'b[ = 'b[ +'a[

[a` + ([b` +[c`) = ([a` +[b`) +[c` 'a[ + ('b[ +'c[) = ('a[ +'b[) +'c[
c([a` +[b`) = c[a` +c[b` c

('a[ +'b[) = c

'a[ +c

'b[
'c[([a` +[b`) = 'c[a` +'c[b` ('a[ +'b[)[c` = 'a[c` +'b[c`
What is the meaning of [a`'b[? Now [a`'b[p` is a ket ([a`) multiplied by a
number ('b[p`), which is another ket. Thus the action of [a`'b[ on a ket is
to generate another ket. Thus [a`'b[ is a linear operator.
[a`'b[ also operates on bras: 'p

[ = 'p[a`'b[.
The expectation value of an operator

A is written:

Adr = '[

A[`

(

A)

dr = '[

[`
We can extend our use of the notation to eigenequations

A[
n
` = a
n
[
n
`.
Here [
n
` is an abstract representation of an eigenfunction.
We can write expansions for functions in the following way: [` =

n
c
n
[
n
`.
This can be very convenient. For example
'[

A[` =

nm
(c

n
'
n
[)

A(c
m
[
m
`)
=

nm
c

n
c
m
'
n
[

A[
m
`
=

nm
c

n
c
m
'
n
[a
m
[
m
`
24 CHAPTER 1. FORMAL ASPECTS OF QUANTUM THEORY
=

nm
c

n
c
m
a
m
'
n
[
m
`
=

nm
c

n
c
m
a
m

nm
=

n
[c
n
[
2
a
n
We can use this result to produce a very convenient way to represent oper-
ators. First we notice the following: '
n
[` =

m
c
m
'
n
[
m
` = c
n
. This
allows us to write our expectation value as '[

A[` =

n
'[
n
`a
n
'
n
[`.
This suggests that we can represent our operator by

A =

n
[
n
`a
n
'
n
[
There is an important special case of this, which is the unit operator.
For this operator all the eigenvalues are equal to 1. This use of the unit
operator is called inserting a complete set of states. To see this consider
the following two states:
[` =

n
c
n
[
n
`
[` =

n
d
n
[
n
`
'[` =

mn
c

n
d
m
'
n
[
m
`
=

n
c

n
d
n
But c
n
= '
n
[` and d
n
= '
n
[`. Hence we can write
'[` = '[(

n
[
n
`'
n
[
n
)[`
This is an example of the insertion of a complete set of states.
1.13 Matrix Representation of States and Oper-
ators
As we have seen, we are able to represent wave functions as sums over
complete sets of states. The matrix notation allows us to exploit this fact
more easily.
Consider a wave function and an operator

A. We can dene the function
by =

A.
We can now introduce the complete set of states
n
. We can expand
our previous two functions in this set of states giving
=

n
c
n
1.13. MATRIX REPRESENTATION OF STATES AND OPERATORS 25
=

n
d
n
We can then combine these equations to produce

n

n
d
n
=

A
n
c
n
.
We can now perform some integrals on this equation to produce the fol-
lowing expression:

n

n
dr d
n
=

m

A
n
dr c
n
.
We can of course write this using Dirac notation

n
'
m
[
n
` d
n
=

n
'
m
[

A[
n
` c
n
.
If the set of states is orthonormal then we obtain the following result
d
m
=

n
'
m
[

A[
n
` c
n
.
We can think of '
m
[

A[
n
` as a matrix element, which we can represent
as A
mn
. Thus we have d
m
=

n
A
mn
c
n
, which in matrix notation we can
write as

d = Ac.
In this notation eigenequations become Ac
n
=
n
c
n
. This is the approach
normally used to perform quantum mechanics on a computer.
The time dependent Schrodinger equation can also be written in this form:

H(rt) = i
(rt)
t
(rt) =

n
c
n
(t)
n
(r)

n
c
n
(t)
n
(r) = i

t

n
c
n
(t)
n
(r)

n
c
n
(t)

m
(r)

H
n
(r) dr =

n
i
c
n
(t)
t

m
(r)
n
(r) dr

n
H
mn
c
n
(t) = i
c
m
(t)
t
H
mn
=

m
(r)

H
n
(r) dr
Hc = i
c
t
Matrix elements of the hermitian conjugate of an operator has the follow-
ing property

n

A

m
dr =

(

A
n
)

m
dr
=

m

A
n
dr

nm
= A

mn
Therefore matrix elements of hermitian operators obey the condition A

nm
=
A
nm
= A

mn
.
26 CHAPTER 1. FORMAL ASPECTS OF QUANTUM THEORY
Inner products and overlaps of wave functions are easy to evaluate
=

n
c
n
=

n
d
n

dr =

nm
c

n
d
m

m
dr
=

n
c

n
d
n
= c

d
1.14 Time Evolution of Operators
Schroedingers equation gives us the time evolution of the wave function.
Operators (such as those for position and momentum) have no time de-
pendence. However, their expectation values do
'A(t)` = '(t)[

A[(t)`

H[(t)` = i

t
[(t)`
In the Heisenberg picture we turn this round, and let the operators vary
in time, keeping the wavefunction constant.
The two approaches must yield the same expectation values. Let us look
at the rate of change of this expectation value
d
dt
'A(t)` = '

t
(t)[

A[(t)` +'(t)[

A
t
[(t)` +'(t)[

A[

t
(t)`
=

(t)
t

Adr +

(t)

A
t
(t) dr +

A
(t)
t
dr
=

(
1
i

H(t))

A(t) dr +

(t)

A
t
(t) dr +

(t)

A(
1
i

H(t)) dr
=
1
i

[(

H(t))

A(t) +

(t)

A
t
(t) +

(t)

A

H(t)] dr
=
1
i

(t)

H

A(t) +

(t)

A
t
(t) +

(t)

A

H(t)] dr
=
1
i

(t)[

A,

H](t) dr +

(t)

A
t
(t) dr
Heisenberg suggested that we treat the wave function as constant in time,
and allow the operators to be time varying:
'A(t)` = '(0)[

A(t)[(0)`
1.14. TIME EVOLUTION OF OPERATORS 27
Let us write [(t)` =

U(t)[(0)`. Then 'A(t)` = '(0)[

(t)

A

U(t)[(0)` =
'(0)[

A(t)[(0)`, and hence

A(t) =

U

(t)

A

U(t)
The equation of motion is thus
d

A(t)
dt
=

U

(t)

1
i
[

A,

H] +


A
t

U(t)
=
1
i
[

A(t),

H(t)] +

U

(t)

A
t

U(t)
Examples
1. The Hamiltonian is time invariant

A =

H
d

H(t)
dt
=
1
i
[

H(t),

H(t)] = 0

H(t) =

H(0)
Thus the Schroedinger and Heisenberg Hamiltonians are the same.
2. The time rate of change of position is velocity

A = x

H =
p
2
2m
+

V
d x
dt
=
1
i
[ x,
p
2
2m
+

V ]
=
1
i
[ x,
p
2
2m
] +
1
i
[ x,

V ]
[ x, f( x)] = 0
d x
dt
=
1
i
[ x,
p
2
2m
]
=
1
i
1
2m
( x p
2
p
2
x)
=
1
i
1
2m
([ x, p] p + p[ x, p])
[ x, p] = i
d x
dt
=
p
m
Compare this last result with the classical expression p = mdx/dt.
3. The rate of change of momentum equals the force:

A = p
d p
dt
=
1
i
[ p,
p
2
2m
+

V ]
=
1
i
[ p,
p
2
2m
] +
1
i
[ p,

V ]
=
1
i
[ p,

V ]
28 CHAPTER 1. FORMAL ASPECTS OF QUANTUM THEORY
If we work in a position representation then we have p = i

x
. And
hence we get
d p
dt
= [

x
, V (x)]
= (
V
x
+V

x
V

x
)
=
V
x
Compare this with the classical expression dp/dt = F = dV/dx.
1.15 Step-Operator Approach to Harmonic Os-
cillator
Let us begin by recalling the standard way of nding the eigenstates for
an harmonic oscillator.
[

2
2m

2
x
2
+
1
2
kx
2
]
n
(x) = E
n

n
(x)
E
n
=
n

2
= k/m
= x

2
= m/
d
2

n
d
2
+ (
n

2
)
n
= 0
We now consider the limit of very large displacements
lim

n
= 0 lim

d
2

n
d
2

2

= 0

n
( )
p
exp(
2
/2)
This suggests that we try the following form for the wave functions

n
() = N
n
H
n
() exp(
2
/2)
0 =
d
2
H
n
d
2
2
dH
n
d
+ (
n
1)H
n
The function H
n
() is a polynomial. The full solution is
E
n
= (n +
1
2
)
H
n
() = exp(
2
/2)[
d
d
]
n
exp(
2
/2)
We can now turn to the step operator method of Dirac. We begin by den-
ing three new operators in terms of which we can we write Hamiltonian
1.15. STEP-OPERATOR APPROACH TO HARMONIC OSCILLATOR 29
operator:
a =

m
2
x + i

1
2m
p
a

m
2
x i

1
2m
p

N = a

H =
p
2
2m
+
1
2
m
2
x
2
= (

N +
1
2
)
The proof that we can rewrite the Hamiltonian in this form proceeds as
follows

N = a

a
=

m
2
x i

1
2m
p

m
2
x + i

1
2m
p

=
m
2
x
2
+
1
2m
p
2
+
i
2
[ x, p]
=
1

H
1
2
We are now in a position to go through a series of steps to nd the eigen-
states of the harmonic oscillator using these operators.
1. The eigenstates of

N are also eigenstates of the Hamiltonian.

N[n` = n[n`

H[n` = (n +
1
2
)[n`
2. Note that the number operator

N is hermitian:

N

= ( a

a)

= a

( a

=
a

a =

N.
3. The eigenvalues of the number operator are non-negative: n = 'n[

N[n` =
'n[ a

a[n` = '[` 0, where [` = a[n`.

4. The operators a and a

do not commute:
[ a, a

] = [

m
2
x + i

1
2m
p,

m
2
x i

1
2m
p]
=
[ x, p]
i
= 1
5. These operators do not commute with the number operator either:
[

N, a] = a

a a a a

a
30 CHAPTER 1. FORMAL ASPECTS OF QUANTUM THEORY
= ( a a

1 a a

) a
= a
[

N, a

] = a

a a

a
= a

(1 + a

)
= a

6. We can use our operators to generate new eigenstates of the number oper-
ator:

N a[n` = a(

N 1)[n` = (n1) a[n`. Therefore if [n` is an eigenstate
of the number operator with eigenvalue n, then a[n` is also an eigenstate
of the number operator but with the new eigenvalue n1. By considering
normalisation we obtain the following:
a[n` = [n 1`
'n 1[n 1` = 'n[n` = 1

2
= 'n[ a

a[n`
= 'n[

N[n`
= n
a[n` =

n[n 1`
7. Thus if n is an eigenvalue of

N then so is n 1, and therefore so are
n2, n3, . . .. But we have the condition n 0. Therefore the eigenvalues
must be integers so that this sequence terminates: a[0` = 0.
8. We are also able to generate eigenstates using the conjugate operator:

N a

[n` = a

(

N + 1)[n` = (n + 1) a

[n`. Thus if n is an eigenvalue of

N,
so is n + 1. Thus the eigenvalues of the number operator are elements of
the set of non-negative integers. Once again by considering normalisation
we obtain
a

[n` = [n + 1`
'n + 1[n + 1` = 'n[n` = 1

2
= 'n[ a a

[n`
= 'n[1 +

N[n`
= n + 1
a

[n` =

n + 1[n + 1`
9. Thus if we know [0` then we can generate all other eigenstates
[1` = a

[0`
[2` =

1
2
a

[1` =

1
2
( a

)
2
[0`
[n` =

1
n!
( a

)
n
[0`
10. We have thus found all the eigenvalues of the Hamiltonian (E
n
= (n +
1
2
)), and know how to generate all the eigenvectors once we have at
1.15. STEP-OPERATOR APPROACH TO HARMONIC OSCILLATOR 31
least one of them. What we now need is a procedure to generate one
eigenvector. We do this as follows:
a[0` = 0

m
2
x + i

1
2m

i

x

0
(x) = 0

0
x
=
mx

0
(x) = Aexp(
mx
2
2
)

0
dx = 1
A =

1
4
32 CHAPTER 1. FORMAL ASPECTS OF QUANTUM THEORY
Chapter 2
Angular Momentum [8]
2.1 A Refresher on Commutation Relations
1. The classical expression for orbital angular momentum is

L = r p. In
quantum mechanics it is dened by

L =

x
y
z

p
x
p
y
p
z

L
x

L
y

L
z

y p
z
z p
y
z p
x
x p
z
x p
y
y p
x

2. We have the commutation relations for position and momentum:

[ x, p
x
] = [ y, p
y
] = [ z, p
z
] = i
[ x, p
y
] = [ x, p
z
] = [ y, p
x
] = [ y, p
z
] = [ z, p
x
] = [ z, p
y
] = 0
3. The angular momentum operator is hermitian:

x
= ( y p
z
z p
y
)

= ( y p
z
)

( z p
y
)

= p

z
y

y
z

= p
z
y p
y
z
= y p
z
z p
y
=

L
x
etc.
4. The commutators between angular momentum and position are
[

L
x
, x] = [ y p
z
z p
y
, x] = 0
[

L
x
, y] = [ y p
z
z p
y
, y] = z[ p
y
, y] = i z
[

L
x
, z] = [ y p
z
z p
y
, z] = y[ p
z
, z] = i y
[

L
y
, x] = i z
33
34 CHAPTER 2. ANGULAR MOMENTUM [8]
[

L
y
, y] = 0
[

L
y
, z] = i x
[

L
z
, x] = i y
[

L
z
, y] = i x
[

L
z
, z] = 0
This can be written compactly as [n

L, r] = i r n were n is a unit
vector, and r is the position operator.
5. The commutators between angular momentum and momentum operators
are:
[

L
x
, p
x
] = [ y p
z
z p
y
, p
x
] = 0
[

L
x
, p
y
] = [ y p
z
z p
y
, p
y
] = p
z
[ y, p
y
] = i p
z
etc.
This can be written compactly as [n

L, p] = i p n
6. The commutators between angular momentum operators are:
[

L
x
,

L
x
] = 0
[

L
x
,

L
y
] = [ y p
z
z p
y
, z p
x
x p
z
] = i

L
z
etc.
This can be written compactly as

L

L = i

L.
7. The commutators between the squares of the angular momentum opera-
tors and the angular momentum operators is given by the following:
[

L
2
x
,

L
x
] = 0
[

L
2
y
,

L
x
] = i(

L
z

L
y
+

L
y

L
z
)
[

L
2
z
,

L
x
] = i(

L
z

L
y
+

L
y

L
z
)
[

L
2
,

L
x
] = [

L
2
x
+

L
2
y
+

L
2
z
,

L
x
] = 0
More generally we have [

L
2
,

L] = 0.
2.2 Eigenvalues and Eigenfunctions of Orbital An-
gular Momentum Operators
1. Consider [

L
2
,

L
z
] = 0. This implies that the eigenstates of

L
z
are also
eigenstates of

L
2
. Let us dene one of these eigenstates by [` where

L
2
[` = [` and

L
z
[` = [`.
2. We now show that the eigenvalues are not negative.
'[

L
2
[` =

i
'[

L
i
[` =

L
i
[` = [i`
=

i
'i[i` 0
2.2. EIGENVALUES ANDEIGENFUNCTIONS OF ORBITAL ANGULAR MOMENTUMOPERATORS35
In the light of this, we shall write =
2
l(l + 1), = m, and the
eigenstates as [lm`.
3. Note that [

L
x
,

L
z
] = 0 and [

L
y
,

L
z
] = 0. Therefore the eigenstates of

L
x
and

L
y
are not [lm` . However these operators commute with

L
2
. These
apparently contradictory statements can be satised if the eigenstates of
these operators are linear combinations of [lm`, where all eigenstates in
the sum have the same value of l.
4. We shall now write the operators and eigenfunctions for orbital angular
momentum in terms of specic coordinates. This will allow us to produce
functions that we can visualise. It is conventional to use spherical coor-
dinates, and this is what we shall use here. The formulae that connect
rectangular coordinates to spherical coordinates are the following

x
y
z

r sin cos
r sin sin
r cos

r
x

x
r
y

y
r
z

r
x

x
r
y

y
r
z

sin cos
cos cos
r
sin
r sin
sin sin
cos sin
r
cos
r sin
cos
sin
r
0

L =

L
x

L
y

L
z

y p
z
z p
y
z p
x
x p
z
x p
y
y p
x

= i

y

z
z

y
z

x
x

z
x

y
y

x

= i

sin

cot cos

cos

cot sin

5. We get immediately

L
2
=
2

cot

+

2

2
+
1
sin
2

6. We want states that are simultaneous eigenstates of

L
z
and

L
2
. We give
the symbol Y
lm
(, ) to the eigenstates. Thus we have:

L
z
Y
lm
(, ) = mY
lm
(, )
i

Y
lm
(, ) = mY
lm
(, )

Y
lm
(, ) = imY
lm
(, )
Y
lm
(, ) = e
im
f
lm
()
36 CHAPTER 2. ANGULAR MOMENTUM [8]
7. For a single valued wave function we must have Y
lm
(, +2) = Y
lm
(, ).
Therefore m must be an integer.
8. To determine the function f
lm
() we make use of the fact that our state
is also an eigenstate of

L
2
.

L
2
e
im
f
lm
() =
2
l(l + 1)e
im
f
lm
()

cot

+

2

2
+
1
sin
2

e
im
f
lm
() =
2
l(l + 1)e
im
f
lm
()

cot

+

2

2

m
2
sin
2

f
lm
() = l(l + 1)f
lm
()
9. We can eliminate all the trigonometric functions by making the following
substitution = cos . If we dene F
lm
() = f
lm
() we get
df
lm
()
d
=
d
d
dF
lm
()
d
= sin
dF
lm
()
d

d
2
f
lm
()
d
2
=

sin
d
d

2
F
lm
()
=

cos
d
d
+ sin
2
d
2
d
2

F
lm
()
l(l + 1)F
lm
() =

2
d
d
(1
2
)
d
2
d
2
+
m
2
1
2

F
lm
()
10. The factor of 1/(1
2
) makes nding a solution problematic. So we
introduce another substitution to remove it F
lm
() = [1
2
]
|m|/2
G
lm
().
Hence
dF
lm
()
d
= [m[[1
2
]
|m|/21
G
lm
() + [1
2
]
|m|/2
dG
lm
()
d
d
2
F
lm
()
d
2
= [1
2
]
|m|/2

[m[
1
2
+
2
2
[m[([m[/2 1)
(1
2
)
2

2[m[
1
2
d
d
+
d
2
d
2

G
lm
()
0 = [1
2
]
|m|/2

l(l + 1) [m[([m[ + 1) 2([m[ + 1)

d
d
+ (1
2
)
d
2
d
2

G
lm
()
11. We now make a polynomial expansion for G
lm
() =

n=0
C
(n)
lm

n
. This
gives:
C
(n+2)
lm
=

n(n 1) + 2n([m[ + 1) +[m[([m[ + 1) l(l + 1)

(n + 1)(n + 2)

C
(n)
lm
12. To obtain a convergent series, there must be an N such that C
(N+2)
lm
= 0.
Hence
0 = N(N 1) + 2N([m[ + 1) +[m[([m[ + 1) l(l + 1)
N = ([m[ +
1
2
) (l +
1
2
)
N = l [m[ as N 0
l = integer
2.3. GENERALIZEDANGULAR MOMENTUMANDSTEP OPERATOR TECHNIQUES INANGULAR MOMENTUMTHEORY37
13. Thus, putting all this together we have
Y
lm
(, ) = N
lm
e
im
(sin)
|m|
l|m|

n=0
C
(n)
lm
(cos )
n
= (1)
m

(2l + 1)(l m)!

4(l +m)!
P
m
l
(cos )e
im
14. P
m
l
() is an associated Legendre polynomial given by P
m
l
() = (1

2
)
|m|/2 d
m
d
m
P
l
(), and P
l
() is a Legendre polynomial, where
P
0
() = 1
P
1
() =
0 = (l + 1)P
l+1
() (2l + 1)P
l
() +lP
l1
()
15. The eigenstates are orthonormal

2
0
d

0
sind Y

lm
(, )Y
l

m
(, ) =
ll

mm

16. Some of the functions are

Table 2.1: Spherical harmonics
Y
00
(, ) =
1

4
Y
10
(, ) =

3
4
cos Y
11
(, ) =

3
8
sine
i
2.3 Generalized Angular Momentum and Step Op-
erator Techniques in Angular Momentum The-
ory
We now write down formal expressions for angular momenta that apply
to orbital angular momentum, but also more generally to things like spin.
Generalised angular momenta are dened by the commutation relations
between the operators.
1. We dene the raising and lowering operators

L

=

L
x
i

L
y
. Note that

=

L
+
.
2. We now consider the commutation relations for the raising and lowering
operators

L
+

= (

L
x
+ i

L
y
)(

L
x
i

L
y
)
=

L
2
x
+

L
2
y
+ i(

L
y

L
x

L
x

L
y
)
=

L
2

L
2
z
+

L
z
38 CHAPTER 2. ANGULAR MOMENTUM [8]

L
+
=

L
2

L
2
z

L
z
[

L
+
,

L

] = 2

L
z
[

L
z
,

L

] =

L

3. Equipped with these commutation relations we are now in a position to

determine the eigenvalues and eigenvectors of the angular momenta oper-
ators. First we determine the range for the magnetic quantum number.
If we dene

L

[lm` = [`, then we have 'lm[

L
+

[lm` = '[` 0.
Similarly we have 'lm[

L
+
[lm` 0. Therefore we can write
'lm[

L
+
[lm` = 'lm[

L
2

L
2
z
+

L
z
[lm`
=
2
[l(l + 1) m
2
+m]
0
(l +
1
2
)
2
(m
1
2
)
2
l m
'lm[

L
+

[lm` = 'lm[

L
2

L
2
z

L
z
[lm`
=
2
[l(l + 1) m
2
m]
0
(l +
1
2
)
2
(m+
1
2
)
2
l m
l m l
4. Now we build the eigenstates using the step operators

L
z

[lm` =

L

L
z
)[lm`
=

L

(m )[lm`
= (m1)

[lm`

[lm` = [lm1`
'lm[

L
+

[lm` =
2
'lm1[lm1`
But'lm[

L
+

[lm` =
2
[l(l + 1) m
2
m]

2
=
2
[l(l + 1) m
2
m]

[lm` =

[l(l + 1) m(m1)][lm1`
5. By a similar argument we can show that

L
+
[lm` =

[l(l + 1) m(m+ 1)][lm+ 1`

6. From the above we see that if m is an eigenstate of

L
z
, then so are
(m+ 1), (m+ 2), etc. But we have the condition that m l, therefore
for some m l the coecient

[l(l + 1) m(m+ 1)] must vanish in order

for the series of eigenstates to terminate. Clearly this occurs for m = l.
Thus m l, l 1, l 2 . . ..
2.4. SPIN-1/2 ANGULAR MOMENTUM AND PAULI MATRICES 39
7. We also have m l. The smallest value of msatises

[l(l + 1) m(m1)] =
0. This occurs for m = l. Thus m l, l 1, l 2 . . . , l +1, l. There-
fore l = l +N where N is a non-negative integer. Thus
l =
N
2
=

integer Neven
integer +
1
2
Nodd
2.4 Spin-1/2 Angular Momentum and Pauli Ma-
trices
Electrons have an internal degree of freedom (spin), as seen in the Stern-
Gerlach experiment.
It is a form of angular momentum with magnitude
1
2
.
The operator for spin is s. It obeys the rules of angular momentum oper-
ators:
s s = i s
[ s
2
, s] = 0
s
2
[
1
2
m
s
` =
1
2
(
1
2
+ 1)
2
[
1
2
m
s
`
=
3
4

2
[
1
2
m
s
`
s
z
[
1
2
m
s
` = m
s
[
1
2
m
s
`
m
s

1
2
,
1
2

The convention for the state vectors (Dirac notation) is

Table 2.2: Spin states
[` = [
1
2
1
2
` [` = [
1
2

1
2
`
s
2
[` =
3
4

2
[` s
2
[` =
3
4

2
[`
s
z
[` =
1
2
[` s
z
[` =
1
2
[`
'[` = '[` = 1 '[` = '[` = 0
Now consider s
x
and s
y
:
40 CHAPTER 2. ANGULAR MOMENTUM [8]
Table 2.3: s
x
and s
y
s
+
= s
x
+ i s
y
s

= s
x
i s
y
s
+
[` = 0 s

[` = [`
s
+
[` = [` s

[` = 0
s
x
=
1
2
( s
+
+ s

) s
y
=
1
2i
( s
+
s

)
s
x
[` =
1
2
[` s
x
[` =
1
2
[`
s
y
[` = i
1
2
[` s
y
[` = i
1
2
[`
s
z
[` =
1
2
[` s
z
[` =
1
2
[`
We now derive the Pauli spin matrices. A general spin state can be written
as [` = a[` + b[`. A general spin operator o acting on [` produces
a new vector [` = o[`. But we can expand this state vector in our
original basis set: [` = c[`+d[`. Combining these equations we obtain
(c[` +d[`) = o(a[` +b[`). Evaluating overlap integrals gives us
c = '[ o[`a +'[ o[`b
d = '[ o[`a +'[ o[`b
In matrix notation we can write this as

c
d

'[ o[` '[ o[`

'[ o[` '[ o[`

a
b

Therefore we can represent the operator o by the matrix

o =

'[ o[` '[ o[`

'[ o[` '[ o[`

The spin matrices are thus given by

Table 2.4: Spin matrices
s
2
=

3
4

2
0
0
3
4

s
x
=

0
1
2

1
2
0

s
y
=

0 i
1
2

i
1
2
0

s
z
=

1
2
0
0
1
2

Pauli introduced the matrices

i
=
2

s
i
Table 2.5: Pauli spin matrices

x
=

0 1
1 0

y
=

0 i
i 0

z
=

1 0
0 1

I =

1 0
0 1

2.4. SPIN-1/2 ANGULAR MOMENTUM AND PAULI MATRICES 41

The Pauli spin matrices obey the following relations
[
x
,
y
] = 2i
z
[
y
,
z
] = 2i
x
[
z
,
x
] = 2i
y

2
x
=
2
y
=
2
z
= 1
Tr
x
= Tr
y
= Tr
z
= 0
det
x
= det
y
= det
z
= 1
[
i
,
j
]
+
= 2I
ij
where [A, B]
+
= AB +BA

z
=

z
=

2
= 3

2
= 3
Example - spin about an arbitrary axis
Consider the operator for a spin aligned along an arbitrary direction n
( s n) where
n =

sin cos
sin sin
cos

Figure 2.1: Axes

42 CHAPTER 2. ANGULAR MOMENTUM [8]
The matrix representing the operator equals
s
n
= s n =

2
n
=

2

sin cos
sin sin
cos

=

2

cos sin e
i
sin e
i
cos

The eigenvalues of this operator are found from

det [s
n
I[ = 0

cos
2

sin e
i
sin e
i
cos
2

= 0

2
=

2
4
=

2
The eigenvalues are independent of direction. The eigenvectors (X

( n))
satisfy
s
n
X

( n) = X

( n)

cos sin e
i
sin e
i
cos

x
1
x
2

x
1
x
2

X
+
( n) =

cos(/2)
sin(/2) e
i

( n) =

sin(/2)
cos(/2) e
i

We have normalised the vectors to 1. In Dirac notation

[X
+
( n)` = cos(/2)[` + sin(/2) e
i
[`
[X

( n)` = sin(/2)[` + cos(/2) e

i
[`
Consider the special cases of the axes
[X
+
( x)` =
1

2
[[` +[`]
[X

( x)` =
1

2
[[` +[`]
[X
+
( y)` =
1

2
[[` + i[`]
[X

( y)` =
1

2
[[` + i[`]
[X
+
( z)` = [`
[X

( z)` = [`
2.5. MAGNETIC MOMENTS 43
If the system starts in the up eigenstate of s n, then the probability of a
measurement of s
z
giving up or down are cos
2
(/2) and sin
2
(/2) respec-
tively. If the system started in the down eigenstate they are sin
2
(/2) and
cos
2
(/2) respectively.
2.5 Magnetic Moments
The electron has an intrinsic magnetic moment associated with its spin
m
s
= g
s

B
s/ = g
s

B
/2, where g
s
is the gyromagnetic ratio. g
s
2.

B
Is the Bohr magneton.
Protons also have a spin of
1
2
, and a magnetic moment. However the
constants dier from those for the electron: m
s
= +g
P

N
/2,
N
=
e/2M
P
and g
P
5.5883.
The magnetic moment couples to a magnetic eld. The contribution to
the Hamiltonian from this coupling is

H = m
s

B
2.6 Combination of Angular Momenta
We will deal with this as an abstract problem rst of all, and then apply
it to specic cases.
Consider the two general angular momenta which belong to two indepen-
dent subsystems

J
1
and

J
2
. For example they might be orbital angular
momentum and spin. Because they belong to independent subsystems
they must commute [

J
1
,

J
2
] = 0.
These operators obey the usual angular momentum eigenequations

J
2
1
[j
1
m
1
` = j
1
(j
1
+ 1)
2
[j
1
m
1
`

J
2
2
[j
2
m
2
` = j
2
(j
2
+ 1)
2
[j
2
m2
1
`

J
1,z
[j
1
m
1
` = m
1
[j
1
m
1
`

J
2,z
[j
2
m
2
` = m
2
[j
2
m
2
`
If we work with wave functions, then the combined wave function for two
angular momenta is just the product of the individual wave functions:
(j
1
m
1
j
2
m
2
) =
1
(j
1
m
1
)
2
(j
2
m
2
). We represent this by the following
state vector [j
1
j
2
m
1
m
2
`.
This joined vector satises the following relations

J
2
1
[j
1
j
2
m
1
m
2
` = j
1
(j
1
+ 1)
2
[j
1
j
2
m
1
m
2
`

J
2
2
[j
1
j
2
m
1
m
2
` = j
2
(j
2
+ 1)
2
[j
1
j
2
m
1
m
2
`

J
1,z
[j
1
j
2
m
1
m
2
` = m
1
[j
1
j
2
m
1
m
2
`

J
2,z
[j
1
j
2
m
1
m
2
` = m
2
[j
1
j
2
m
1
m
2
`

J
z
=

J
1,z
+

J
2,z

J
z
[j
1
j
2
m
1
m
2
` = (m
1
+m
2
)[j
1
j
2
m
1
m
2
`
44 CHAPTER 2. ANGULAR MOMENTUM [8]
We also have the following commutation relations

J
2
= (

J
1
+

J
2
)
2
=

J
2
1
+

J
2
2
+ 2

J
1

J
2
[

J
2
,

J
2
1
] = [

J
2
,

J
2
2
] = 0
[

J
2
,

J
1,z
] = 0
[

J
2
,

J
2,z
] = 0
Therefore the eigenfunctions of

J
2
and

J
z
are also eigenfunctions of

J
2
1
and

J
2
2
, but not of

J
1,z
and

J
2,z
.
We therefore have two complete, but distinct, descriptions of our system:
1. Eigenfunctions of

J
2
1
,

J
2
2
,

J
1,z
and

J
2,z
([j
1
j
2
m
1
m
2
`).
2. Eigenfunctions of

J
2
1
,

J
2
2
,

J
2
and

J
z
([j
1
j
2
jm`).
The two sets of eigenstates are related by [j
1
j
2
jm` =

m
1
m
2
[j
1
j
2
m
1
m
2
`'j
1
j
2
m
1
m
2
[j
1
j
2
jm`.
'j
1
j
2
m
1
m
2
[j
1
j
2
jm` is a Clebsch-Gordan coecient.
We can immediately dene one condition for the Clebsch-Gordan coe-
cients

J
z
[j
1
j
2
jm` = m[j
1
j
2
jm`
=

m
1
m
2
(m
1
+m
2
)[j
1
j
2
m
1
m
2
`'j
1
j
2
m
1
m
2
[j
1
j
2
jm`
'j
1
j
2
m
1
m
2
[j
1
j
2
jm` = 0 if m
1
+m
2
= m
We can impose upper limits on the values of the eigenvalues of the total
angular momentum operators as follows
max m
1
= j
1
max m
2
= j
2
max m = j
1
+j
2
m j, j + 1, . . . , j 1, j
max j = j
1
+j
2
We can impose a lower bound on the eigenvalues of the total angular
momentum as follows
There are (2j
1
+ 1)(2j
2
+ 1) possible states which we break up into
multiplets corresponding to dierent values of the total angular mo-
mentum.
For j = j
1
+j
2
, we have a multiplet of 2(j
1
+j
2
) + 1 states.
For j = j
1
+j
2
1, we have a multiplet of 2(j
1
+j
2
1) + 1 states.
We run out of states when j = [j
1
j
2
[. Therefore [j
1
j
2
[ j
j
1
+ j
2
. This is the relation obeyed by the third side of a triangle
given the rst two.
2.6. COMBINATION OF ANGULAR MOMENTA 45
We now begin the process of building the set of Clebsch-Gordan coe-
cients.
For m = j = j
1
+j
2
there is only one term in the expansion [j
1
j
2
j
1
+
j
2
j
1
+ j
2
` = [j
1
j
2
j
1
j
2
`'j
1
j
2
j
1
j
2
[j
1
j
2
j
1
+ j
2
j
1
+ j
2
`. Normalisation
then gives ['j
1
j
2
j
1
j
2
[j
1
j
2
j
1
+j
2
j
1
+j
2
`[ = 1.
We now introduce the raising and lowering operators

J

=

J
1
+

J
2
.
Matrix elements of these operators give
'j
1
j
2
m
1
m
2
[

[j
1
j
2
jm` =

j(j + 1) m(m1)'j
1
j
2
m
1
m
2
[j
1
j
2
jm1`
= 'j
1
j
2
m
1
m
2
[

J
1
+

J
2
[j
1
j
2
jm`
=

j
1
(j
1
+ 1) m
1
(m
1
1)'j
1
j
2
m
1
1m
2
[j
1
j
2
jm` +

j
2
(j
2
+ 1) m
2
(m
2
1)'j
1
j
2
m
1
m
2
1[j
1
j
2
jm`
These expressions allow us to build all the Clebsch-Gordan coe-
cients for a given multiplet starting from the one at the extreme
value of the total angular momentum.
To x the absolute values of the Clebsch-Gordan coecients within
a given multiplet we need the following normalisation condition
1 = 'j
1
j
2
jm[j
1
j
2
jm`
=

m
1
m
2
'j
1
j
2
jm[j
1
j
2
m
1
m
2
`'j
1
j
2
m
1
m
2
[j
1
j
2
jm`
=

m
1
m
2
['j
1
j
2
jm[j
1
j
2
m
1
m
2
`[
2
Example - the addition of two spins
The total spin operator is given by s = s
1
+ s
2
. The maximum and
minimum values of the total spin eigenvalues are 1 and 0 respectively.
For the multiplet of maximum total spin (the triplet state) the largest
magnetic quantum number that can be obtained is also 1. The state is
given by [sm` = [11` = [`.
If we apply the lowering operator to the state we obtain the following
s

[11` =

2[10`
= ( s
1
+ s
2
)[`
= ([` +[`)
[10` =
1

2
([` +[`)
s

[10` =

2[1 1`
= ( s
1
+ s
2
)
1

2
([` +[`)
=

2[`
[1 1` = [`
46 CHAPTER 2. ANGULAR MOMENTUM [8]
The other multiplet with s = 0 contains only one state (the singlet state).
This must be normalised to 1 and be orthogonal to all the states in
the above multiplet. Since its magnetic quantum number is 0 it must
be formed from a linear combination of [` and [`. If we solve the
two equations '00[00` = 1 and '00[10` = 0 we obtain the result [00` =
1

2
([` [`).
Chapter 3
Approximate Methods [6]
3.1 Time-Independent Perturbation Theory
Suppose we have solved the time independent Schroedinger equation for
some problem (for example the particle in a box). Now we want the
solution for a problem in which a very small additional potential is applied
to our original system.
We wish to do this in a way that allows us to use all the work we have
already done for the original problem.
For the original problem we might have

H[
n
` =
n
[
n
`. For our new
problem we would have (

H+

V )[
n
` = E
n
[
n
`, where

V is the additional
potential, and is a scalar that we associate with the perturbing potential
which allows us to perform systematic expansions as we shall now see.
We make the following expansions in the parameter
E
n
() =

r=0
E
n,r

r
[
n
()` =

r=0
[
n,r
`
r
If we substitute our expansions into the original eigenequation we obtain
the following (

H +

V )

r=0
[
n,r
`
r
=

s=0
E
n,s

r=0
[
n,r
`
r
.
Since our parameter can have any value, for the above equation to always
be true the coecients preceding each power of the parameter must be
the same on the left and on the right of the equation.

0
:

H[
n,0
` = E
n,0
[
n,0
`

1
:

H[
n,1
` +

V [
n,0
` = E
n,0
[
n,1
` +E
n,1
[
n,0
`

2
:

H[
n,2
` +

V [
n,1
` = E
n,0
[
n,2
` +E
n,1
[
n,1
` +E
n,2
[
n,0
`
47
48 CHAPTER 3. APPROXIMATE METHODS [6]
From these equations we can obtain the following results. From the 0th
power of the parameter (
0
) we just obtain the eigenequation for the
unperturbed system. Therefore we can write
[
n,0
` = [
n
`
E
n,0
=
n
From the rst power of the parameter (
1
)
(

H
n
)[
n,1
` = (E
n,1

V )[
n
`
[
n,1
` =

p
C
(n,1)
p
[
p
`
(

H
n
)

p
C
(n,1)
p
[
p
` = (E
n,1

V )[
n
`

p
C
(n,1)
p
(
p

n
)[
p
` = (E
n,1

V )[
n
`
C
(n,1)
p
(
p

n
) = E
n,1

pn
'
p
[

V [
n
`
p = n : E
n,1
= '
n
[

V [
n
`
p = n : C
(n,1)
p
=
'
p
[

V [
n
`

p
[
n,1
` =

p(=n)
[
p
`
'
p
[

V [
n
`

p
+C
(n,1)
n
[
n
`
C
(n,1)
n
is found by normalizing the wavefunction [
n
`.
From the square of the parameter (
2
) we get the following results
E
n,2
[
n
` = (

H
n
)[
n,2
` + (

V E
n,1
)[
n,1
`
E
n,2
= '
n
[(

H
n
)[
n,2
` +'
n
[(

V E
n,1
)[
n,1
`
= '
n
[(
n

n
)[
n,2
` +'
n
[

V [
n,1
` E
n,1
'
n
[
n,1
`
= '
n
[

V [
n,1
` '
n
[

V [
n
`'
n
[
n,1
`
=

p(=n)
'
n
[

V [
p
`
'
p
[

V [
n
`

p
=

p(=n)
['
p
[

V [
n
`[
2

p
Therefore to second order the perturbed energy is
E
n

n
+'
n
[

V [
n
` +

p(=n)
['
p
[

V [
n
`[
2

p
The above theory breaks down when we have degenerates states (
n

p
=
0). This is because the terms in the perturbation expansion involve ratios
of matrix elements divided by
n

p
. These terms are innite unless
3.1. TIME-INDEPENDENT PERTURBATION THEORY 49
the matrix elements are zero. Degenerates states have the property that
any normalised linear combination of them produces another state of the
same energy. Therefore the procedure for handling degenerates states in
perturbation theory is to take linear combinations that produce matrix
elements of the perturbing potential that are zero. That is
[

` =

p{degenerate states}
C
,p
[
p
`
'

V [

` = '

V [

The rst order term in the perturbation expansion then obeys

C
(,1)

) = E
,1

'

V [

`
0 = E
,1

'

V [

E
,1
= '

V [

`
C
(,1)
p
(
p

) = E
,1

p
'
p
[

V [

`
C
(,1)
p
=
'
p
[

V [

p
[
,1
` =

p/ {degenerate states}
[
p
`
'
p
[

V [

p
We nd the orthogonal states by diagonalising the matrix V
pq
= '
p
[

V [
q
`:

q
V
pq
C
,q
=

C
,p

pq
C

,p
V
pq
C
,q
=

'

V [

` =

Example - charged harmonic oscillator

Consider a charged harmonic oscillator in an electric eld. The electric
eld results in an additional force on the harmonic oscillator which can be
represented by the additional potential

V = QE x where Q is the charge
on the oscillator, E is the strength of the electric eld and x is the operator
representing the displacement of the oscillator.
The eigenstates of the harmonic oscillator are not degenerate in one di-
mension. Therefore we can use ordinary perturbation theory. For this we
need the following matrix elements 'p[

V [n` = QE'p[ x[n`, where [p` is an

eigenstate of the unperturbed harmonic oscillator.
We can use the Dirac raising and lowering operators to represent the
displacement since x =

2M
( a + a

). The matrix elements are thus

'p[

V [n` = QE

2M
'p[( a + a

)[n`
= QE

2M

n
p,n1
+

n + 1
p,n+1

50 CHAPTER 3. APPROXIMATE METHODS [6]

The perturbed energy to second order is thus
E
n
=
n
+'
n
[

V [
n
` +

p(=n)
['
p
[

V [
n
`[
2

p
= (n +
1
2
) + 0 +
Q
2
E
2

2M

n + 1

= (n +
1
2
)
Q
2
E
2
2M
2
Thus each level receives the same constant shift downwards.
Example - Stark eect
The Stark eect in atomic spectra is observed when an electric eld is
applied to an atom. It can be understood from perturbation theory in
which the perturbing potential is once again an electric eld interacting
with a charged particle (this time an electron).
We will study this eect using the n = 2 levels of the hydrogen atom in a
uniform electric eld pointing in the z direction.
There are four degenerate states in the n = 2 shell of hydrogen, namely
[S
0
`, [P
1
`, [P
0
` and [P
1
`. The states are labelled by the orbital angular
momentum. They satisfy the following equations

L
z
[S
0
` = 0

L
z
[P
1
` = [P
1
`

L
z
[P
0
` = 0

L
z
[P
1
` = [P
1
`
The perturbing potential we can write as

V = eE z. Therefore we need
to consider matrix elements of the form eE'lm[ z[l

`.
We now recall that [ z,

L
z
] = 0. If we take matrix elements of this expres-
sion we obtain
0 = 'lm[[ z,

L
z
][l

`
= 'lm[ z

L
z

L
z
z[l

`
= (m

m)'lm[ z[l

`
if m

= mthen0 = 'lm[ z[l

`
Now consider 'lm[ z[lm`. These are all zero because the hydrogen wave
functions have the following property ['r[nlm`[
2
= ['r[nlm`[
2
(they are
in variant under the parity operation).
The full matrix of matrix elements of the perturbation potential is there-
fore
V =

0 v 0 0
v

0 0 0
0 0 0 0
0 0 0 0

3.1. TIME-INDEPENDENT PERTURBATION THEORY 51

Where v = 'S
0
[ eE z[P
0
` = 3eEa
0
. Thus the electric eld only causes
two of the states to mix, namely [S
0
` and [P
0
`. Therefore we only need to
diagonalize a 2 2 matrix.

0 v
v 0

C
s
C
p

C
s
C
p

vC
p
= C
s
vC
s
= C
p
v
2
C
s
C
p
=
2
C
s
C
p
= 3eEa
0

C
s
C
p

=
1

1
1

Example - spin-orbit coupling

Our next example is the spin-orbit correction in hydrogen. The perturbing
potential has the form

V =
1
2m
2
c
2
Ze
2
4
0
1
r
3

L

S
= (r)

L

S
As previously we will try to make use of as much symmetry as possible.
We note the following commutation relation [

L
2
,

L

S] = [

L
2
,

L]

S = 0.
Therefore [

L
2
,

V ] = 0, and so the total angular momentum remains a good
quantum number, and the perturbation can only mix states which have
the same value of l.
For a given principal quantum number all the states are degenerate. There-
fore we must use degenerate perturbation theory. The basis states for hy-
drogen we can represent by [nlmm
s
`, where m
s
is the magnetic quantum
number associated with spin.
If we dene

J =

L +

S, then we can obtain the following

J
2
=

L
2
+

S
2
+ 2

L

S

V = (r)
1
2
(

J
2

L
2

S
2
)
0 = [

L
2
,

V ]
= [S
2
,

V ]
= [

J
2
,

V ]
= [

J
Z
,

V ]
Thus the eigenstates of the perturbing Hamiltonian will also be eigen-
states of the total angular momentum, the spin, and the orbital angular
momentum. To obtain eigenstates of the total angular momentum start-
ing from our basis states we need to use the normal rules for coupling
angular momentum. That is [nljm
j
` =

mm
s
[nlmm
s
`'l
1
2
mm
s
[l
1
2
jm
j
`,
where 'l
1
2
mm
s
[l
1
2
jm
j
` is a Clebsch-Gordan coecient.
52 CHAPTER 3. APPROXIMATE METHODS [6]
Using these new basis states we obtain the rst order energy shifts as
E
nljm
j
= 'nljm
j
[

V [nljm
j
`
= 'nljm
j
[(r)
1
2
(

J
2

L
2

S
2
)[nljm
j
`
=

2
2
[j(j + 1) l(l + 1)
3
4
]'nljm
j
[(r)[nljm
j
`
=

2
2
[j(j + 1) l(l + 1)
3
4
]
nl

nl
= 'nljm
j
[(r)[nljm
j
`
=
1
2m
2
c
2
Ze
2
4
0
Z
3
a
2
0
n
3
l(l + 1)(l +
1
2
)
If l = 0 then the energy shift is 0. Otherwise we have j = l
1
2
, and the
energy shifts are
E
nljm
j
=

nl

2
2
l j = l +
1
2

nl

2
2
(l + 1) j = l
1
2
Example - positronium
Positron (positive electron) and electron orbit about each other.
The energy levels are the same as for hydrogen, except that they are
multiplied by a factor of one half because the reduced mass in positronium
is one half of that for hydrogen.
The magnetic moments on the electron and positron interact to give a
spin-spin interaction.

H =

H
0
+A s
e
s
p
Both the spin operators commute with the Hamiltonian

H
0
, thus there
are four degenerate eigenstates corresponding to the four possible spin
states (one singlet state and three triplets states), provided we ignore the
spin-spin interaction.
The perturbation is already diagonal in the total spin states
'SM
s
[ s
e
s
p
[S

s
` =
1
2
'SM
s
[

S
2
s
2
e
s
2
p
[S

s
`
=

2
2
(S(S + 1)
3
2
)
SS

M
S
M

S
=

3
2
4
S = 0

2
4
S = 1
Thus the rst order degenerate perturbation shift in energy is given by
E =

3
2
4
A S = 0 (singlet)

2
4
A S = 1 (triplet)
3.2. VARIATIONAL PRINCIPLE 53
Positronium decays when the positron and electron annihilate each other.
To conserve energy, linear momentum, and angular momentum, the singlet
state must decay into two photons, and the triplet state must decay into
three photons. The photon carries a spin angular momentum of one.
3.2 Variational Principle
Consider a system described by a Hamiltonian

H. Consider also a guess
for the wave function for the system [`. We dene the energy of the
system by E

= '[

H[`/'[`. We now expand this wave function

in terms of the eigenstates of the Hamiltonian [` =

n
C
n
[
n
` where

H[
n
` = E
n
[
n
`. Therefore we get
E

nm
C

n
C
m
'
n
[

H[
m
`

nm
C

n
C
m
'
n
[
m
`
=

n
[C
n
[
2
E
n

n
[C
n
[
2
=

n
[C
n
[
2
(E
n
E
0
+E
0
)

n
[C
n
[
2
= E
0
+

n
[C
n
[
2
(E
n
E
0
)

n
[C
n
[
2
E
0
This tells us that for any guess for the wave function, our estimate for
the energy will be no less than the ground state energy. This means that
we can optimise our guess by varying it in such a way as to minimise the
estimated energy E

.
We can extend this idea to excited states as well. Here we consider just
the rst excited state. This will be orthogonal to the ground state, so
we need to consider a trial wave function that obeys '
0
[` = 0. We
can always achieve this by writing the trial wave function in the following
way [` = [

` + [
0
`. Imposing the orthogonality condition gives
0 = '
0
[` = '
0
[

` + . From this we get [` = (1 [

0
`'
0
[)[

`.
If we expand the function [

` in terms of the eigenstates we have [

` =

n=0
C
n
[
n
`, and hence
[` =

n=0
C
n
(1 [
0
`'
0
[)[
n
`
=

n=0
C
n
(1 [
0
`'
0
[)[
n
`
=

n=1
C
n
[
n
`
where in the last line we make use of the fact that the term (1[
0
`'
0
[)
is 1 unless n = 0, in which case it is 0. This allows us to change the range
54 CHAPTER 3. APPROXIMATE METHODS [6]
of the sum from 0 to 1 . If we now use this to evaluate E

, we
obtain the result E

E
1
. Therefore, provided our trial wave function is
orthogonal to the ground state we can use exactly the same procedure for
the rst excited state as we use for the ground state.
Example - helium atom
We now perform a variational calculation for the helium atom. The Hamil-
tonian is given by

H =
p
2
1
2m
+
p
2
2
2m

Ze
2
r
1

Ze
2
r
2
+
e
2
[r
1
r
2
[
We now assume that the wave function has the form (r
1
, r
2
) =
1s
(r
1
)
1s
(r
2
),
where
1s
(r) =
1

Q
a
0

3/2
exp(Qr/a
0
). Note that

dr
1
dr
2
[(r
1
, r
2
)[
2
=
1.
The estimate for the energy is then given by
E

dr
1
dr
2

1s
(r
1
)
1s
(r
2
)

p
2
1
2m
+
p
2
2
2m

Ze
2
r
1

Ze
2
r
2
+
e
2
[r
1
r
2
[

1s
(r
1
)
1s
(r
2
)
=

dr
1

1s
(r
1
)

p
2
1
2m

Ze
2
r
1

1s
(r
1
) +

dr
2

1s
(r
2
)

p
2
2
2m

Ze
2
r
2

1s
(r
2
)
+

dr
1
dr
2

1s
(r
1
)
1s
(r
2
)
e
2
[r
1
r
2
[

1s
(r
1
)
1s
(r
2
)
= 2

Q
2
e
2
2a
0

ZQe
2
a
0

. .. .
Hydrogen like
+
5Qe
2
8a
0
. .. .
e e
To optimise our wave function we now minimise the energy with respect
to our parameter Q. This gives
0 =
E

Q
=
e
2
a
0

5
8
+ 2Q2Z

Q = Z
5
16
Thus we see that the eective charge appearing in the exponent of the wave
function is reduced relative to the true nuclear charge by the amount 5/16.
This is a result of the repulsion between the electrons which screens the
nucleus. The minimum energy is E = (Z
5
16
)
2 e
2
a
0
= 7620 kJ/mol. The
experimental value is 7477 kJ/mol. So we see that our calculated value is
indeed larger than the true value. The error is a result of our very simple
form for the wave function.
Chapter 4
Simple Time-dependent
systems [3]
4.1 Superposition of States of Dierent Energies
As we saw earlier the solution of the time-dependent Schroedinger equation
for a system of xed energy E
n
is
n
(xt) = e
E
n
t/i

n
(x).
A general wave function can be constructed as a linear combination of
these functions (xt) =

n
C
n
e
E
n
t/i

n
(x), where C
n
=

n
(x)(x0) dx.
Consider the expectation value of an operator

O with respect to this wave
function
'O` = '[

O[`
=

nm
C

n
C
m
e
(E
m
E
n
)t/i
'
n
[

O[
m
`
Suppose the operator is the Hamiltonian (

O =

H). Then we have

H[
n
` =
E
n
[
n
`, and hence
'H` =

nm
C

n
C
m
e
(E
m
E
n
)t/i
'
n
[

H[
m
`
=

nm
C

n
C
m
e
(E
m
E
n
)t/i
'
n
[E
m
[
m
`
=

nm
[C
n
[
2
E
n
This is independent of time. In fact in general it is straightforward to
show that any operator that commutes with the Hamiltonian will have an
expectation value that is independent of time
d'O`
dt
=
1
i
'[[

O,

H][`
= 0 since [

O,

H] = 0
55
56 CHAPTER 4. SIMPLE TIME-DEPENDENT SYSTEMS [3]
The expectation value of all other operators varies with time. Consider a
wave function that is a linear superposition of just two states:
[` = C
1
e
E
1
t/i
[
1
` +C
2
e
E
2
t/i
[
2
`
'O` = (C

1
e
E
1
t/i
'
1
[ +C

2
e
E
2
t/i
'
2
[)

O(C
1
e
E
1
t/i
[
1
` +C
2
e
E
2
t/i
[
2
`)
= [C
1
[
2
O
11
+[C
2
[
2
O
22
+C

1
C
2
e
(E
2
E
1
)t/i
O
12
+C

2
C
1
e
(E
1
E
2
)t/i
O
21
O
nm
= '
n
[

O[
m
`
If the operator

O is hermitian then we have O
12
= O

21
. In this case we
obtain
'O` = [C
1
[
2
O
11
+[C
2
[
2
O
22
+

1
C
2
e
(E
2
E
1
)t/i
O
12

1
C
2
e
(E
2
E
1
)t/i
O
12

If we dene = E
2
E
1
, and further we assume that C
1
, C
2
and O
12
are all real, then we obtain
'O` = [C
1
[
2
O
11
+[C
2
[
2
O
22
+ 2C
1
C
2
cos(t)O
12
=

O + cos(t)

O = [C
1
[
2
O
11
+[C
2
[
2
O
22
= 2C
1
C
2
O
12
If either C
1
or C
2
is zero then the expectation value is time independent
because we are now in an eigenstate.
4.2 Electron in Magnetic Field
We now consider a system that contains one electron. We now place this
system in a magnetic eld, and allow the electronic spin to couple to the
magnetic eld. We will neglect the interaction between the orbital motion
of the electron and eld (which is a much smaller eect). The Hamiltonian
that describes this system is

H =

H
0

g
s

S

B
We now assume that the magnetic eld points in the z direction and is
everywhere uniform

B = Be
z
. For this case the Hamiltonian simplies to

H =

H
0

g
s

B
B

S
z
=

H
0

S
z
.
Provided the unperturbed Hamiltonian does not depend on spin then we
have [

H
0
,

H] = 0, and thus the eigenstates of the unperturbed Hamiltonian
are also eigenstates of the perturbed Hamiltonian. Let the eigenstates be
represented by [nm
s
`, where m
s
is the spin magnetic quantum number.
The eigenvalues of our Hamiltonian are then
E
nm
s
= 'nm
s
[

H[nm
s
`
= 'nm
s
[

H
0
[nm
s
` +'nm
s
[

S
z
[nm
s
`
= E
(0)
n
m
s
4.2. ELECTRON IN MAGNETIC FIELD 57
Suppose we can ignore all eigenstates of the unperturbed Hamiltonian
except the ground state. Then the general time-dependent way function
for our system in a magnetic eld becomes
[(t)` = C
1
e
(E
0
+
1
2
)t/i
[0
1
2
` +C
2
e
(E
0

1
2
)t/i
[0
1
2
`
If the system starts in an eigenstate of the operator

S
z
, then one or other
of the constant coecients must be 0. Either way the system remains in
that eigenstate for all time.
Consider the expectation value of the z component of the spin operator
'S
z
` = [C
1
[
2
(
1
2
) +[C
2
[
2
(
1
2
)
=
1
2
([C
2
[
2
[C
1
[
2
)
= constant
Now considered the expectation value of the x component of the spin
operator

S
x
=
1
2
(

S
+
+

S

S
+
[
1
2
` = 0

S
+
[
1
2
` = [
1
2
`

[
1
2
` = [
1
2
`

[
1
2
` = 0
'S
x
` = 0 + 0 + 2Re

1
C
2
e
t/i
'0
1
2
[
1
2
(

S
+
+

S

)[0
1
2
`

= Re

1
C
2
e
it

Similarly the expectation value of the y component of the spin operator is

'S
y
` = Im

1
C
2
e
it

Now consider the special case C

1
= C
2
=
1

2
'S
z
` = 0
'S
y
` =

2
sint
'S
x
` =

2
cos t
For this special case the spin precesses perpendicular to the magnetic eld.
58 CHAPTER 4. SIMPLE TIME-DEPENDENT SYSTEMS [3]
4.3 Time Evolution of Entangled States of Two
Spin-1/2 Particles with Total Spin Zero.
If two particles are in an entangled state then any measurement made on
one of the particles will inuence the state of the other particle. Math-
ematically this means that the wave function for the combined system
cannot be expressed as a single product of the wave functions for the two
individual particles.
If we have two electrons in a state of zero spin then we know that the wave
function for the spins can be written as [S = 0, m
s
= 0` =
1

2
([`[`).
This is clearly an entangled state.
[THIS HAS YET TO BE FINISHED]
Chapter 5
Identical Particles [2]
If you have many particles, you have a set of operators for each particle
( p
i
, r
i
,

S
i
,

L
i
. . .).
The Hamiltonian for a system of many particles is characterised by the
following equations

H( p
i
, q
i
) =

i
p
2
i
2M
i
+

V ( q
i
)

H( p
i
, q
i
)(q
i
, t) = i

t
(q
i
, t)
Other system wide operators include

L =

L
i

P =

P
i

S =

S
i
All operators for particle i commute with those for particle j: [

P
i
,

P
j
] =
[

P
i
, r
j
] = . . . = 0.
Identical particles cannot be distinguished by any intrinsic property (for
example all electrons are equivalent to one another). Further, they cannot
be distinguished by their trajectories in quantum mechanics. This opens
the way for a new system wide operator which interchanges identical par-
ticles

P
ij
f(q
1
. . . q
i
. . . q
j
. . . q
N
) = f(q
1
. . . q
j
. . . q
i
. . . q
N
)
Clearly

P
2
ij
f = f. Therefore the eigenvalues of

P
ij
must be 1. If

P
ij
f =
+f, then the function f is a symmetric eigenfunction of

P
ij
, while if

P
ij
f = f, then the function f is an anti symmetric eigenfunction.
59
60 CHAPTER 5. IDENTICAL PARTICLES [2]
For identical particles

P
ij

Hf =

H

P
ij
f for all pairs of particles. Therefore
the eigenfunctions of the Hamiltonian must also be eigenfunctions of all
permutation operators

P
ij
.
However [

P
ij
,

P
rs
] = 0. To accommodate this fact the wave function
must either be completely symmetric under interchange of particles, or
completely anti symmetric. Under this condition we nd that

P
ij

P
rs
f =

P
rs

P
ij
f.
Therefore wave functions for identical particles must be fully symmetric
or fully anti symmetric under exchange of particles
Bosons (particles with integer spin such as photons) have fully sym-
metric wave functions
Fermions (particles with half integer spin, such as electrons) have
fully anti symmetric wave functions.
5.1 Systems of Two Identical Particles
5.1.1 Two fermions
We now consider two electrons. The wave function for one electron can
be written as
n
=
n
(r)

, where
n
(r) is the spatial part of the wave
function and

is the spin part.

If we combine these for two electrons the natural starting point is to form
a product (1, 2) =
n
(1)
n

(2). But this has the wrong symmetry

properties under exchange of particles 1 and 2. So consider

A
(1, 2) =
1

2
(
n
(1)
n

(2)
n
(2)
n

(1)) =
A
(2, 1)
This wave function now has the correct symmetry. This is the simplest
case of a Slater determinant.
Consider the special case n = n

. We can now rewrite the wave function

as

A
(1, 2) =
n
(1)
n
(2)
1

2
(

(1)

(2)

(2)

(1))
If we now also set =

then we nd that the wave function vanishes.

Thus we cannot have both electrons in the same single particle state. This
is the Pauli exclusion principle for fermions.
If we now set

= and

= , then we get

A
(1, 2) =
n
(1)
n
(2)
1

2
((1)(2) (2)(1))
But this new spin state is just the S = 0, m
s
= 0 state.
5.1. SYSTEMS OF TWO IDENTICAL PARTICLES 61
Now consider the alternative case in which n = n

. We are able in this

case to obey the Pauli exclusion principle while having =

. There are
two possible solutions

A
(1, 2) =
1

2
(
n
(1)
n
(2)
n
(2)
n
(1))(1)(2)

A
(1, 2) =
1

2
(
n
(1)
n
(2)
n
(2)
n
(1))(1)(2)
The spin parts of the wave function correspond to the states S = 1,
m
s
= 1. This suggests that there should be a third possibility with
S = 1, m
s
= 0. This wave function would have the form

A
(1, 2) =
1

2
(
n
(1)
n
(2)
n
(2)
n
(1))
1

2
((1)(2) +(2)(1))
This wave function has the necessary property that
A
(1, 2) =
A
(2, 1).
Thus we see that we have two sets of solutions
The singlet set with symmetric spatial part of the wave function but
anti symmetric spin part (S = 0).
The triplet set with anti symmetric spatial part of the wave function
but symmetric spin part (S = 1).
That are also mixed solutions for which n = n

and =

. But these
are not good wave functions because they are not eigenstates of the spin
operator. Provided [

H,

S] = 0, then the eigenstates of the Hamiltonian
should also be eigenstates of the spin operator.
5.1.2 Two bosons
Consider two photons. We can go through the same procedure as above,
except that now we must produce a symmetric wave function.

S
(1, 2) =
1

2
(
n
(1)
n

(2) +
n
(2)
n

(1)) =
S
(2, 1)
We can see the most important dierence between bosons and fermions
by considering the special case of n = n

and =

. In the case of the

fermions this caused the wave function to vanish. But in this case we get

S
(1, 2) =
n
(1)
n
(2)
Notice that we have dropped a factor of

2 to obtain the correct normal-
isation. From this we see that more than one boson is able to occupy a
particular single particle state. One consequence of this is that you can
have many photons in one place with the same energy and travelling in
the same direction.
62 CHAPTER 5. IDENTICAL PARTICLES [2]
5.2 Independent Particle Model of He Atom
As we saw earlier when looking at the variational principle, the Hamilto-
nian for the helium atom is

H =
p
2
1
2m
+
p
2
2
2m

Ze
2
r
1

Ze
2
r
2
+
e
2
[r
1
r
2
[
Note that [

L
2
,

H] = [

L
z
,

H] = 0. Therefore the eigenstates of the Hamil-
tonian will also be angular momentum eigenstates.
If we have normalised wave functions then the energy of the atom is given
by E

= '[

H[`.
For the singlet states we have
E
S
=

dr
1
dr
2
(r
1
)(r
2
)

H(r
1
)(r
2
)'s = 0m
s
= 0[s = 0m
s
= 0`
=

dr
1
dr
2
(r
1
)(r
2
)

H(r
1
)(r
2
)
=

dr
1
(r
1
)

p
2
1
2m

Ze
2
r
1

(r
1
)

dr
2
[(r
2
)[
2
+

dr
2
(r
2
)

p
2
2
2m

Ze
2
r
2

(r
2
)

dr
1
[(r
1
)[
2
+

dr
1
dr
2
e
2
[(r
1
)[
2
[(r
2
)[
2
[r
1
r
2
[
= 2

dr(r)

p
2
2m

Ze
2
r
+
1
2

dx
e
2
[(x)[
2
[r x[

(r)
Thus the energy looks like that from two independent electrons each of
which is moving in an eective potential
V (r) =
Ze
2
r
+
1
2

dx
e
2
[(x)[
2
[r x[
This potential is that generated by the nucleus plus a term produced by
the average interaction between the electrons which partially screens the
nuclear potential. This result is the single particle picture result.
To determine the wave function we use the variational principle. We vary
the wave function until we have minimised the energy E
s
.
We now consider the triplet state
E
T
=
1
2

dr
1
dr
2
(
1
(r
1
)
2
(r
2
)
1
(r
2
)
2
(r
1
))

H(
1
(r
1
)
2
(r
2
)
1
(r
2
)
2
(r
1
))'1m
s
[1m
s
`
=
1
2

dr
1
dr
2
(
1
(r
1
)
2
(r
2
)
1
(r
2
)
2
(r
1
))

H(
1
(r
1
)
2
(r
2
)
1
(r
2
)
2
(r
1
))
5.2. INDEPENDENT PARTICLE MODEL OF HE ATOM 63
Consider the one body operators. We will use the operator
p
2
1
2m
as an
example:

p
2
1
2m

=
1
2

dr
1
dr
2
(
1
(r
1
)
2
(r
2
)
1
(r
2
)
2
(r
1
))
p
2
1
2m
(
1
(r
1
)
2
(r
2
)
1
(r
2
)
2
(r
1
))
=
1
2

dr
1

1
(r
1
)
p
2
1
2m

1
(r
1
)

dr
2

2
(r
2
)
2
(r
2
)
+
1
2

dr
1

2
(r
1
)
p
2
1
2m

2
(r
1
)

dr
2

1
(r
2
)
1
(r
2
)

1
2

dr
1

1
(r
1
)
p
2
1
2m

2
(r
1
)

dr
2

2
(r
2
)
1
(r
2
)

1
2

dr
1

2
(r
1
)
p
2
1
2m

1
(r
1
)

dr
2

1
(r
2
)
2
(r
2
)
=
1
2

dr
1

1
(r
1
)
p
2
1
2m

1
(r
1
) +
1
2

dr
1

2
(r
1
)
p
2
1
2m

2
(r
1
)
Now consider the two body operator
e
2
|r
1
r
2
|

e
2
[r
1
r
2
[

=
1
2

dr
1
dr
2
(
1
(r
1
)
2
(r
2
)
1
(r
2
)
2
(r
1
))
e
2
[r
1
r
2
[
(
1
(r
1
)
2
(r
2
)
1
(r
2
)
2
(r
1
))
=
1
2

dr
1
dr
2
e
2
[
1
(r
1
)[
2
[
2
(r
2
)[
2
[r
1
r
2
[
+
1
2

dr
1
dr
2
e
2
[
1
(r
2
)[
2
[
2
(r
1
)[
2
[r
1
r
2
[

dr
1
dr
2

1
(r
1
)
2
(r
2
)
e
2
[r
1
r
2
[

1
(r
2
)
2
(r
1
)
Combining all the terms we obtain
E
T
=

dr
1
(r)

p
2
2m

Ze
2
r
+
1
2

dx
e
2
[
2
(x)[
2
[r x[

1
(r)
+

dr
2
(r)

p
2
2m

Ze
2
r
+
1
2

dx
e
2
[
1
(x)[
2
[r x[

2
(r)

drdx
1
(r)
2
(x)
e
2
[r x[

1
(x)
2
(r)
The rst two terms are clearly completely analogous to the single term in
the singlet state.
However the third term is new. It is called the exchange energy, and is a
direct consequence of the anti symmetric form of the spatial part of the
wave function.
The exchange energy lowers the energy of states with parallel spins. This is
because the symmetry of the wave function keeps these electrons spatially
separated from each other. That is, it is a result of correlation in the
motion of the electrons.
64 CHAPTER 5. IDENTICAL PARTICLES [2]
Chapter 6
Interpretations of quantum
mechanics [4]
6.1 Wave-Particle Duality and Indeterminacy
We wish to clarify the key concepts of quantum mechanics, which involves some
strange concepts. We will illustrate these concepts by reference to experiments.
The concepts are
Waveparticle duality: consider one component of the material world (for
example, electrons). Some experiments are best understood by thinking
of it as a wave (for example, diraction eects), while others are best un-
derstood by thinking of it as a particle (for example, momentum transfer).
Indeterminacy: a wave function does not tell you what you will mea-
sure. Instead it gives you probabilities. But you in fact measure denite
values. Heisenbergs uncertainty principle quanties this: if you make
many measurements of non simultaneous observables, then the product
of the standard deviations of their measured values has a lower bound
(AB /2).
6.2 Double slit experiment
If light from a point source passes through a pair of slits, then you get a
diraction pattern (Youngs slits experiment). This is easily explained in
terms of waves. It is possible to perform the experiment one photon at a
time, producing one dot at a time on the recording screen. This is also
particle like.
If one slit is closed, a single slit diraction pattern is produced
If a slit changes from open to closed once the photon is in-ight, the
pattern is still correct for the single slit.
If we monitor the slits we can observe whether the photon goes to one or
the other.
65
66 CHAPTER 6. INTERPRETATIONS OF QUANTUM MECHANICS [4]
We can make this quantitative
[` = [A` +[B`
where [A` is the wave from slit A, and [B` is the wave from slit B. Thus
['r[`[
2
= [[
2
['r[A`[
2
+ [[
2
['r[B`[
2
+

'B[r`'r[A` +

'A[r`'r[B`,
which gives interference.
We introduce a two state monitor with state vectors [M` (registered that
a particle is present) and [

M` (did not register that a particle is present).
The motion of a particle after passing through a slit is dened by its mo-
mentum vector

k. Therefore the amplitudes that tell us what the proba-
bilities are that the particle is travelling with a prticular momentum after
passing through a particular slit are
'

kM[` = '

kM[A` +'

kM[B`
'

k

M[` = '

k

M[A` +'

k

M[B`
If '

kM[B` = '

k

M[A` = 0 then we know which slit the particle went
through.
6.3 Beam splitter experiment
Figure 6.1: Beam splitter experiment
[A`
1

2
([B` + i[C`)

2
(i[D` [E`)

2
(i[
1

2
([G` + i[F`)] [
1

2
([F` + i[G`)]) = [F`
6.4. COPENHAGEN INTERPRETATION 67
A measurement at M or N gives a 50% chance of detection.
If an obstruction is placed at N then
[A`
1

2
([B` + i[C`)

2
([B` [E`)

2
([B`
1

2
([F` + i[G`))
Thus the photon strikes D1 and D2 with equal probability.
If we remove Y, then the nal state is
1

2
(i[D` [E`). Therefore D1 and
D2 respond with equal probability.
Note that we can insert or remove Y after the photon has left X and we
get the same results.
6.4 Copenhagen Interpretation
This is a scheme for providing outcomes of measurements, and not for
explaining mechanisms.
Collapse of the wave function: the system enters an eigenstate during a
measurement.
Principle of complementarity: mutually exclusive descriptions can be ap-
plied to a quantum system, but not simultaneously (for example, waves
and particles).
This interpretation has several problems, despite being the most com-
monly used one:
Measurements are macroscopic and deterministic. This interpreta-
tion says the microscopic world is probabilistic. Where is the bound-
ary?
A measurement involves interactions between the quantum system
and measuring device. If this obeyed Schroedingers equation it
would not lead to the collapse of the wave function. Thus we have
an inconsistency.
Complementarity is vague and may not even be true.
It appears to reject objective reality.
6.5 Hidden Variables
Statistics is about averaging over many measurements. This averaging
could be hiding the eects of variables we are not treating explicitly. For
example, the kinetic theory of gases allows us to talk about pressure,
temperature etc by averaging over the atomic positions and momenta.
68 CHAPTER 6. INTERPRETATIONS OF QUANTUM MECHANICS [4]
The main contenders for hidden variable theory are
de Broglie pilot wave theory
David Bohms (from Birkbeck College) quantum force theory
6.6 Non-Locality and the Einstein, Podolsky, Rosen
(EPR) Paradox
The EPR thought experiment proceeds as follows
A single stationery particle explodes into two identical fragments
Momentum conservation requires that the measured momentum of
the two particles be equal and opposite
Symmetry requires that each particle has travelled the same distance
Does measurement of position (momentum) of A reveals the position
(momentum) of B.
This measurement gives B a xed position (momentum) instanta-
neously.
David Bohm modied the experiment to use spin.
Two spin
1
2
particles, with zero total spin, move apart. The spin part
of the wave function is 1/

2(
1

2
).
Measurement of one spin causes collapse of the wave function to
1

2
or
1

2
, so if particle 1 is spin up, then particle 2 is spin down, and
vice versa.
Thus if the z component of the spin of particles 1 and 2 are measured
then they will have opposite values. But if the z component of the
spin of particle 1 is measured, and then the x component of the spin
of particle 2, there will be a 50% chance of getting either of the two
possible values.
If the x component of the spin had been measured for particles 1 and
2, then particle 2 would have the opposite value to particle 1.
Special relativity can introduce a further complication. Depending on
your state of motion, the order of two measurements can change:
6.7. BELLS INEQUALITIES 69
Figure 6.2: Relativistic measurement
If A and B ashlight when they are making measurements, O
L
and O
R
might disagree about the order in which pulses reached them.
6.7 Bells Inequalities
These are a test of the constraints of hidden variable theories
Bell provided inequalities that must hold between joint probabilities of
spin measurements made on two particle systems and that are a necessary
consequence of their being separate entities.
Consider spins in a singlet state [` =
1

2
([` [`)
Consider the correlation between measurements

K = (

S
1
a)(

S
2

b)
The expectation value of the correlation is E( a,

b) = 'K` = '[(

S
1
a)(

S
2

b)[` = a

b = cos
Suppose that is a hidden variable that allows the system to be com-
pletely dened. Each spin 0 system has a value for this variable. Let the
probability distribution of be beP(), with

dP() = 1.
Let measurement of spin 1 be /2A( a, ), and of spin 2 be /2B(

b, ).
A, B 1. Since S = 0, A( a, ) = B( a, ).
Correlation coecient
( a,

b) =

P()A( a, )B(

b, ) d
( a,

b) ( a, c) =

P()[A( a, )B(

b, ) A( a, )B( c, )] d
70 CHAPTER 6. INTERPRETATIONS OF QUANTUM MECHANICS [4]
=

P()A( a, )A(

b, )[1 +A(

b, )B( c, )] d
[( a,

b) ( a, c)[

[P()A( a, )A(

b, )[1 +A(

b, )B( c, )][ d
P() 0
1 +A(

b, )B( c, ) 0
[A( a, )A(

b, )[ = 1
[( a,

b) ( a, c)[

P()[1 +A(

b, )B( c, )] d
[( a,

b) ( a, c)[ 1 +(

b, c)
[( a,

b) ( a, c)[ (

b, c) 1
If a,

b and c lie in a plane with

b bisecting the angle between a and

b, then
E( a,

b) = cos
E( a, c) = cos 2
E(

b, c) = cos
[ cos + cos 2[ + cos 1
The inequality is violated for 0 /2.
6.8 The Aspect Experiments
Measurements were made on the polarisation of a pair of photons emitted
by a calcium atom.
The atom was put into an excited state by two lasers
It then relaxes in two stages releasing two photons
These photons then meet polarisers: they either pass through or are re-
ected. The detectors record +1 if the photon is found to be linearly
polarised parallel to the polariser, or -1 if normal to it.
For an emitted photon pair [` =
1

2
([ ` +[ `)
We dened coincidence rates by N
++
( a,

b) = the rate for recording +1 in

both detectors etc. The correlation function is then
E( a,

b) =

N
++
+N

N
+
N
+
N
++
+N

+N
+
+N
+

We now dene X = ( a,

b) +( a,

) ( a

b) +( a

)
For a local real theory we have the condition 2 X 2
Aspect chose a

b = a

b = a

= cos , and a

= cos 3. The inequality

was violated.
6.9. THE PROBLEM OF MEASUREMENT 71
This shows that the hidden variable theory must be nonlocal.
Figure 6.3: Calcium energy levels
6.9 The Problem of Measurement
Copenhagen interpretation
During measurement the wave function collapses to an eigenstate
Only then do we have knowledge of the quantum system
But
When does measurement take place?
How do we separate system, apparatus and observer?
When does the collapse of the wave function take place?
The von Neuman chain
Supposed polarised light is incident at 45
0
to a polariser.
The photon is in a superposition of two states: one parallel and one
perpendicular to the polariser.
72 CHAPTER 6. INTERPRETATIONS OF QUANTUM MECHANICS [4]
The measuring apparatus will thus have two states, and prior to
measurement will be in a superposition.
The measurement apparatus only makes a measurement when a mea-
surement is made on it.
Etc.
Consider the following example
Suppose we have an electron that can be in one of two spin states
[` and [`, and that we have a measuring device with two lights on
it: one for up ([u`) and one for down ([d`) .
Before any interaction the system is in the state [B` = (C

[` +
C

[`) (a
u
[u` +a
d
[d`)
After the interaction it becomes [A` = C

[` [u` +C

[` [d`
Which light is shining? It is not dened.
To collapse the wave function we need a second measurement. Etc.
6.10 Schrodingers Cat
This famous thought experiment highlights the nature of the measurement
problem.
A cat in a sealed box with a Geiger counter, some radioactive material, a
ask of hydrogen cyanide, and a mechanism for breaking the ask.
After one hour this radioactive material will have undergone either zero
or one decay events. If there was a decay the Geiger counter will register
this, and cause the ask of hydrogen cyanide to break, killing the cat.
From quantum mechanics, after one hour the state of the system will be
[` = C
a
[No decay; Cat alive` +C
d
[Decay; Cat dead`.
Before opening the box the cat is in a superposition of states. After the box
is opened, and a measurement is made, the cats wave function collapses
into one state or the other.
At which point is the fate of the cat determined?
6.11 Mind and Matter
It is possible that consciousness leads to collapse of the wave function
But how complex a being is needed for this?
What happened before life appeared in the universe?
6.12. INDELIBLE RECORDS 73
6.12 Indelible Records
An isolated quantum system evolves reversibly according to Schroedingers
equation.
If a measurement is made then an irreversible process takes place that
cannot be undone.
If the system could be described by a single wave function before the
measurement, this is no longer true afterwards. The system is in a mixed
state.
Bohr spoke of a process of irreversible amplication. Quantum eects are
amplied to the classical level.
6.13 Path Integral Approach
This was developed by Richard Feynman
It calculates the probability that a particle starting at position a at time
t
a
arrives at b at time t
b
.
This approach does not work with wave functions, but instead takes the
view that the particle travels every single trajectory between a and b.
The probability is given by the square of the following term
K(b, a) =

all paths ab
e
S(a,b)/i
S(a, b) is the action associated with a path and is given by
S(a, b) =

t
b
t
a
dt (K.E. P.E.)
The classical path is the one with the minimum value of the action.
The nal probability is a result of interference between neighbouring paths.
74 CHAPTER 6. INTERPRETATIONS OF QUANTUM MECHANICS [4]
Figure 6.4: Path integral
6.14 The Many Universes Interpretation
This was rst proposed by Hugh Everett
When a measurement is made the universe splits into a number of copies in
each of which one of the outcomes is realised. There is no communication
between the universes.
For example, if we measure an electron spin, the universe branches into
two copies, one in which the electron has spin up, and the other in which
it has spin down.
Proponents point out its simplicity: only one assumption is needed.
Opponents protest that invoking innitely many universes to solve the
problem of wave function collapse is excessive.
It also does not explain what a measurement is.
David Deutsch has reformulated the theory to work with a pre-existing
nite number of universes. When quantum alternatives are presented, the
universes partitioned themselves into groups, in each of which a dierent
outcome happens. The universes exist in parallel and change in content
completely in accordance with the principle of increase of entropy.
6.15. ALTERNATIVE HISTORIES 75
6.15 Alternative Histories
A history is a sequence in time of events.
There is no distinction between microscopic and macroscopic, no split
between system, apparatus and environment.
Focuses on probabilities for dierent histories.
6.15.1 Fine-grained and coarse-grained histories
This approach was proposed by Gell-Mann and Hartle.
It does not assign probabilities to histories, but instead a decoherence
functional D(A, B) to a pair of histories A and B.
If (Aor B) is the combined history of A or B then
D(Aor B, Aor B) = D(A, A) +D(B, B) +D
i
(A, B)
D
i
(A, B) = D(A, B) +D(B, A)
Where D
i
is the interference term. If D
i
(A, B) = 0, then A and B cannot
be assigned probabilities as they interfere with each other. If D
i
(A, B) =
0 then D(A, A) and D(B, B) can be interpreted as the probabilities of
histories A and B respectively.
Fine-grained histories give as complete a description of the universe as is
possible. Interference between two ne-grained histories does not generally
vanish.
In coarse-grained histories only certain things are followed, and with a
limited amount of detail. The rest of the universe is not followed.
6.15.2 Decoherence
Two histories decohere if their interference term is zero.
Decoherence occurs because of the entanglement of what is followed with
what is not. It gives rise to branching of histories into distinct alternatives
with real probabilities, and thus can explain the transition to the classical
domain.
Consider Schroedingers cat. [cat` =

l
a
l
[l` +

d
b
d
[d`. There are many
scenarios involving the living cat, and many involving the dead cat. These
interfere, and after the decoherence time they cancel out. Thus there is
no interference between live and dead.
6.16 The Ghiradi-Rimini-Weber scheme
Schroedingers equation is modied by the introduction of random hits.
76 CHAPTER 6. INTERPRETATIONS OF QUANTUM MECHANICS [4]
At random times the wave function is multiplied by a narrow Gaussian
(about once every 10
8
years per particle). This causes collapse of the wave
function and suppresses linear superpositions.
However, it is ad hoc, not based on known physical laws, and involves
small violations of energy conservation.
6.17 Gravitationally Induced Reduction
Roger Penrose believes that incorporation of gravity into quantum me-
chanics could provide an explanation of wave function reduction as a real
physical process.
Suppose a system evolves by Schroedingers equation into a linear su-
perpositions of two states that occupy signicantly dierent locations in
space. The evolved state must then involve a superpositions of gravita-
tional elds with dierent space-time geometries. Reduction of the wave
function could occur if the geometries of the superposed states became so
dierent that they could not coexist.
6.18 Quantum Information
The unit of information is the bit (binary digit), equal to 0 or 1.
A classical system can reside in one of the two states.
A quantum system can exist in a superposition of both states at once:
[` = C
0
[0` +C
1
[1`.
These qubits can be represented by electron spins, photon polarisations,
atomic energy levels, states of trapped ions...
6.18.1 Quantum computers
For a qubit, the probability of measuring 0 is [C
0
[
2
, and of measuring 1 is
[C
1
[
2
.
A classical register composed of L bits can store one of 2
L
possible num-
bers.
A quantum register can store all 2
L
possible numbers simultaneously in
superposition. But we can only see one of them after a measurement.
Thus this is not useful for bulk storage.
Operations can be performed simultaneously on all the numbers, so that
2
L
parallel operations can be performed.
This can produce important savings in time and memory for some algo-
rithms
searches
6.18. QUANTUM INFORMATION 77
factorisation (the factorisation of a 100 bit number takes about 10
44
s
on a classical computer, but only a few seconds on a quantum com-
puter. The age of the universe is 10
17
s).
Logical gates have already been demonstrated.
A major obstacle is decoherence (the destruction of the coherent state).
6.18.2 Quantum cryptography
Cryptography is the art of concealing the content of messages.
We know how to generate an unbreakable cipher. You need a genuinely
random string of bits with which to modify your message (for example,
using exclusive or). The encoded message can now be sent by a public
channel (for example electronic mail).
The problem is to transmit this random series of bits (the key) between
two people (Alice and Bob) without others being able to intercept it.
Quantum mechanics can be used to solve the problem of communicating
the key.
One scheme is the following
Alice has a source of photons and two polarisers. One polariser lin-
early polarises the light vertically (0) and the other at 45
0
(1).
Bob has two analysers. One measures photons linearly polarised at
-45
0
(0) and the other horizontally (1).
Alice chooses a polariser at random and sends a photon to Bob.
Bob chooses an analyser at random and records whether he detects
a signal or not.
For example, if Alice sends a vertical photon and Bob uses a -45
0
analyser, then Bob has a 50% chance of detecting a photon. If he
does detect a photon both Bob and Alice selected the same value for
the bit. Bob tells Alice if he detected the photon or not.
Bob and Alice only retain bits for which a photon was detected and
use this string as the key.
A second scheme uses entangled photons
a supply of pairs of entangled photons is available to Alice and Bob.
From each pair, one goes to Alice and one to Bob.
They each make a series of plane polarisation measurements, half the
time distinguishing between perpendicular directions, and the other half
between 45
0
rotated directions.
When the measurements are nished they exchange information about
what measurement was made on each photon so that they know on which
occasions they made the same measurement. Then by the EPR eect they
know that common measurements give the same result.
78 CHAPTER 6. INTERPRETATIONS OF QUANTUM MECHANICS [4]
This produces a string of random bits.
If there is an eavesdropper (Eve), then Alice and Bob will not always get
the same result, which can be detected by comparing a fraction of their
results.
6.18.3 Quantum Teleportation
We can make an exact replica of a photon at a remote location.
Consider the following four Bell states
[
+
` =
1

2
([10` +[01`)
[

` =
1

2
([10` [01`)
[
+
` =
1

2
([11` +[00`)
[

` =
1

2
([11` [00`)
We can switch between these Bell states by means of the following oper-
ations
phase shift (change of sign)
bit ip ([0` [1` or [1` [0`)
combined phase shift and bit ip
identity operation
The states can all be generated and operated on using a combination of
non linear crystals, polarisers and mirrors.
Alice wishes to teleport a photon (A) in state [T` = u[1` +v[0` to Bob.
A pair of entangled photons are sent to Alice and Bob (B and C).
Alice performs a joint measurement on photons A and B, and obtains
one of the Bell states with probability
1
4
. This collapses photon C
into a well-dened state uniquely related to the state of A.
Alice then transmits the result of her measurement to Bob over a
public channel.
Bob can then perform the required unitary transforms on C to pro-
duce a photon in state [T`.
Example
the ancillary photons are prepared in state [

`. The three photon state

of A, B and C is
[` =
1

2
[u[110` +v[010` u[101` v[001`]
6.18. QUANTUM INFORMATION 79
This can be regrouped in terms of Bell states
[` =
1
2
[u[
+
0`v[
+
1`+u[

0`+v[

1`u[
+
1`+v[
+
0`u[

1`v[

0`]
Suppose Alice, in her Bell states measurement, nds [

`. Then C is
projected into the state [c` = u[1` +v[0` = [T`.
Alice tells Bob the result of her measurement, and Bob knows he need do
nothing.