FullNotes PHAS3226

PHAS3226: Quantum Mechanics
David Bowler
(based on notes by D. Tovee)
December 19, 2011
Administration
Ofce hours will be Mondays at 1pm in my ofce (4C2 in the London Centre for Nanotechnology). Attendance sheets
must be lled in. Theyll be given out at the start of a lecture and collated over the weeks.
Problem sheets will be given out through the term, roughly every two weeks. The best three problem sheets will count
for 10% of the nal course mark. N.B. There will be four sheets during term.
Full sets of lecture notes will be made available approximately one week after the lecture. A complete PDF le will
be available at the end of the course.
Prerequisites
To have attended and passed the departments introductory quantum mechanics courses, PHAS2222 or equivalent courses.
PHAS2224, the Atomic and Molecular Physics course is desirable. Frequent reference is made to the material in
PHAS2222.
The following topics should have been covered previously :
The time-independent Schrdinger wave equation and its solution for:
1. quantum wells and quantum barriers/steps
2. The harmonic oscillator (classical and quantum)
3. The hydrogen atom including the radial equation as well as the angular equation and its solution with spherical
harmonics.
An understanding of atomic spectroscopic notation (n, l, m quantum numbers) and their physical basis is assumed.
The expansion postulate; the Born interpretation of the wave function; simple calculations of probabilities and
expectation values.
For a time-independent Hamiltonian, an understanding of the separability of the full Schrdinger equation into a
time-independent wave equation in position space and a time-dependent component is assumed. Familiarity with
applications to eigenstates and/or superpositions thereof is assumed.
A basic understanding of the postulates of quantum theory is assumed.
Aims & Objectives
Aims This course aims to
Introduce the basic concepts of Heisenbergs matrix mechanics. The second year course (PHAS2222) dealt primar-
ily with the Schrdingers matter wave dynamics. In PHAS3226, matrix mechanics is introduced as an alternative
approach to quantum dynamics. It is also shown to provide a complementary approach, enabling the treatment
of systems (such as spin systems) where solutions of the non-relativistic Schrodinger wave equation in position
coordinates is not possible.
Apply matrix mechanics and operator algebra to the Quantum Harmonic Oscillator and its relation to the 2nd year
wave dynamics solutions using Hermite polynomials.
Apply matrix mechanics and operator algebra to quantum angular momentum.
Demonstrate that matrix mechanics predicts and permits solution of spin-1/2 systems using Pauli matrices.
Develop understanding of fundamental concepts using these new methods. The Heisenberg uncertainty principle is
shown to be just one among a family of generalized uncertainty relations, which arise from the basic mathematical
structure of quantum theory, complementing arguments (e.g. Heisenberg microscope) introduced in the second
year.
Explore some concepts of two-particle systems. The addition of two-spins is analysed including fundamental
implications, exemplied by the Einstein-Podolsky-Rosen paradox and Bell inequalities.
Introduce approximate methods (time-independent perturbation theory, variational principle) to extend the PHAS2222
analytical solution of the hydrogen atom to encompass atoms in weak external electric and magnetic elds and two-
electron systems like helium atoms.
Introduce symmetry requirements and the Pauli Principle .
ii
Objectives After completing the module the student should be able to:
Formulate most quantum expressions using abstract Dirac notation and understand that it is not simply shorthand
notation for second-year expressions.
Understand how to formulate and solve simple quantum problems expressing quantum states as vectors and quan-
tum operators as matrices.
Use commutator algebra and creation/annihilation operators to solve for the Quantum Harmonic Oscillator.
Derive generalized uncertainty relations; calculate variances and uncertainties for arbitrary observables. In this
context, have a clear understanding of the relation and difference between operators and observables.
Use commutator algebra and raising/lowering operators to calculate angular momentum observables.
Calculate the states and observables of spin-1/2 systems using Pauli matrices.
Understand and estimate corrections to the energy and properties of low-lying states of helium atoms using approx-
imate methods and symmetry.
Understand time-independent perturbation theory.
Understand the variational principle.
Understand the Pauli Principle and symmetrisation requirements on quantum states including for combinations of
space/spin states.
Textbooks
Those which are closest to the material and level of the course are (in alphabetical order)
B. H. Bransden and C. J. Joachain, Quantum Mechanics, Second Edition (Pearson, 2000)
S. Gasiorowicz, Quantum Physics, Third Edition (Wiley, 2003).
A.Rae, Quantum Mechanics, Fifth edition, (Taylor & Francis 2007).
Syllabus
The approximate allocation of lectures to topics is shown in brackets below.
Formal quantum mechanics [7]
Revision of year 2 concepts using Dirac bracket notation: Introduction to Dirac notation and application to PHAS2222
material including orthonormality of quantum states, scalar products, expansion postulate, Linear Hermitian Operators
and derivations of their properties; simple eigenvalue equations for energy, linear and angular momentum.
Representations of general operators as matrices and states as vectors using a basis of orthonormal states. Basis set
transformations; proof that eigenvalue spectrum is representation independent. Similarity transformations and applica-
tions to calculations of observables such as eigenvalues and expectation values.
The Quantum Harmonic Oscillator; Generalised Uncertainty Relations [7]
Introduction of creation and annihilation operators: main commutator relations, the number operator. Solution of the
QHO eigenstates and spectrum using this method. Relation to wave dynamics and Hermite polynomial solutions obtained
previously. The zero-point energy as a consequence of both commutator algebra as well as wave solutions. Uncertainty
relations for general operators. Solutions of simple examples including Heisenberg Uncertainty Principle.
Generalised Angular Momentum [4]
Commutator algebra and raising/lowering operators. Obtaining angular momentum eigenvalue and eigenstates using
raising and lowering operators. Matrix representation and solution of simple problems.
iii
Spin-1/2 systems [8]
Introduction to spin-1/2 systems. Matrix representations of spin using eigenstates of z-component of spin: spinors and
Pauli matrices. Matrix representations of eigentates of spin operators along arbitrary directions. Basis set and similarity
transformations between different basis sets. Addition of two spins. The EPR paradox and derivation of Bell inequalities.
Approximate methods and many-body systems [7]
Derivation of time independent Perturbation theory. First and second order theory and examples: helium and atoms in
external elds. The variational principle; example with helium.
Symmetry, fermions, bosons and the Pauli principle. Two-particle space-spin symmetry. Slater determinants for many
body systems. Exchange corrections in helium spectrum.
iv
CONTENTS
Contents
1 Formal Quantum Mechanics 3
1.1 Introduction - Review of basic quantum mechanics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.1.1 The Expansion Postulate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.2 Dirac Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.2.1 Dirac notation for state vectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.3 The position (coordinate) representation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.4 Basis states . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.4.1 Change of basis - change of representation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.5 Matrix representation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.5.1 Eigenvalue equation and eigenvalues in matrix representation . . . . . . . . . . . . . . . . . . . 14
1.5.2 Change of basis - an example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
1.5.3 Solving for energy eigenvalues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2 QHO 23
2.1 Classical mechanics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.2 Schrdinger equation for harmonic oscillator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.3 Algebraic operator approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.3.1 Eigenvectors of

N . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.3.2 Eigenvalues of the Hamiltonian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.3.3 Actions of operators a
+
and a
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.3.4 Spatial wavefunctions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.3.5 Matrix representations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.3.6 Minimum uncertainty for the harmonic oscillator . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.3.7 Eigenstates of a
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.4 Compatible observables and commuting operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.5 Heisenberg Uncertainty Relations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.5.1 Uncertainty relation for the harmonic oscillator . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3 Generalised Angular Momentum 37
3.1 Orbital angular momentum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.2 Eigenvalues and eigenfunctions of

L
z
and

L
2
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.3 Central potentials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
3.3.1 Review of hydrogen atom . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
3.4 Generalized angular momentum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3.4.1 Raising and lowering operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3.5 General angular momentum eigenvalue problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.5.1 Actions of the

J
+
and

J
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
4 Spin 1/2 Systems 47
4.1 Useful relations for spin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
4.2 Representation of spin 1/2 operators and eigenfunctions - Pauli matrices . . . . . . . . . . . . . . . . . . 49
4.2.1 Matrix representations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
4.2.2 Commutators for the Pauli matrices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4.2.3 Basis states . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4.2.4 Determination of eigenvalues and eigenvectors . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4.2.5 Spin along an arbitrary direction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
4.3 Space-spin wavefunctions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
4.3.1 Stern-Gerlach experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
4.4 Addition of angular momentum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
1
PHAS3226: Quantum Mechanics CONTENTS
4.4.1 Addition of two spin-1/2 angular momenta . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.4.2 Coupling of spin-1/2 and orbital angular momentum . . . . . . . . . . . . . . . . . . . . . . . . 57
4.4.3 Addition of orbital and spin-1/2 angular momenta . . . . . . . . . . . . . . . . . . . . . . . . . . 58
5 Approximate methods & Many-body systems 61
5.1 Time-independent perturbation theory for a non-degenerate energy level . . . . . . . . . . . . . . . . . . 61
5.1.1 First-order . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
5.1.2 Second-order . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
5.1.3 Observations on the energy corrections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
5.2 The Degenerate case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
5.3 Applications of perturbation theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
5.3.1 First-order examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
5.3.2 Second-order example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
5.3.3 Degenerate example - rst order . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
5.4 Variational method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
5.4.1 Proof of the variational principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
5.4.2 Excited States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
5.4.3 Variational examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
5.5 Systems of identical particles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
5.5.1 Identical particles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
5.5.2 Exclusion principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
5.5.3 N-particle states . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
5.5.4 Helium atom and the exchange force . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
A Problem Sheets 91
2
CHAPTER 1. FORMAL QUANTUM MECHANICS
Chapter 1
Formal Quantum Mechanics
1.1 Introduction - Review of basic quantum mechanics
Quantum mechanics is probabilistic and deterministic. The system of N particles has the same set of dynamical
variables (positions, momenta, kinetic energies, etc.) as in classical mechanics. However it is a fundamental feature of
quantum mechanics that it is impossible to have a precise knowledge of all the dynamical variables at the same time
t. The Heisenberg Uncertainty Principle states that it is impossible to know precisely the position and momentum of a
particle at any given time. In particular, the uncertainty in the x-component of position, x, and momentum, p
x
, of a
particle are related by
xp
1
2
h . (1.1)
In the wave mechanics formulation of quantum mechanics due to Schrdinger, a basic postulate is that for every
dynamical system there exists a wavefunction that is a (single-valued) function of the parameters of the system and
time from which all possible information of the physical system can be obtained. In this formulation the wavefunction for
a single particle is denoted by (r, t), and by (r
1
, r
2
, . . . r
N
, t) for N particles.
The usual interpretation given to is that its square modulus [[
2
[ gives the probability density for nding the particle
at specied positions; [(r)[
2
d
3
r is the probability of nding the particle inside a volume element d
3
r located at vector
position r . (Recall that the square modulus [(r)[
2
= (r)
(r), where
(r) is the complex conjugate of (r).)

On the other hand, quantum mechanics is deterministic in that given the wavefunction at some time t its value at a
later time t
can be found as (r, t) evolves according to the time-dependent Schrdinger equation

ih
(r, t)
t
=

H (r, t) , (1.2)
where

H is the Hamiltonian operator for the system. For a particle of mass m moving in a potential V (x, t) in one-
dimension the Hamiltonian is
H =
h
2
2m
d
2
dx
2
+ V (x, t) . (1.3)
If the potential is independent of time, the time-dependent Schrdinger equation separates with solutions of the form
(x) e
iEt/ h
and the general solution (by the expansion postulate) is
(x, t) =
n
c
n
n
(x) e
iE
n
t/ h
(1.4)
with
n
(x) satisfying the time-independent Schrdinger equation
h
2
2m
d
2
n
(x)
dx
2
+ V (x, t)
n
(x) = E
n
n
(x) (1.5)
with energy eigenvalue E
n
.
It is a basic principle of quantum mechanics that if
1
and
2
are two possible states of a given system then the linear
combination (superposition) = c
1
1
+ c
2
2
is also a possible state of the system where c
1
, c
2
are constants. The
probability density associated with (r) becomes
3
PHAS3226: Quantum Mechanics CHAPTER 1. FORMAL QUANTUM MECHANICS
[ (r)[
2
= [c
1
1
(r) + c
2
2
(r)[
2
(1.6)
= (c
1
(r) + c
2
(r)) (c
1
1
(r) + c
2
2
(r))
P (r) = [c
1
[
2
[
1
(r)[
2
+[c
2
[
2
[
2
(r)[
2
+ c
1
c
2
1
(r)
2
(r) + c
1
c
1
(r)
2
(r) (1.7)
= [c
1
[
2
[
1
(r)[
2
+[c
2
[
2
[
2
(r)[
2
+ 21[c
1
c
2
1
(r)
2
(r)] (1.8)
= [c
1
[
2
P
1
(r) +[c
2
[
2
P
2
(r) + 21[c
1
c
2
1
(r)
2
(r)] (1.9)
where P
1
(r) = [
1
(r)[
2
and P
2
(r) = [
2
(r)[
2
are the probability densities associated with
1
(r) and
2
(r). However
the total probability density is not just the weighted sums of P
1
(r) and P
2
(r) as there is an additional interference term
21[c
1
c
2
1
(r)
2
(r)].
The Hamiltonian operator was mentioned above. It is a postulate of quantum mechanics that every dynamical (vari-
able) observable is represented by a linear Hermitian operator. A linear operator

A is such that

A(c
1
1
+ c
2
2
) =
c
1

A
1
+ c
2

A
2
.
Another postulate is that the only possible results of a measurement of an observable A are the eigenvalues of the
corresponding operator

A. That is
A
n
=
n
n
, (1.10)
where
n
is the eigenvalues for eigenfunction
n
. Immediately after such a measurement the wavefunction of the system
will be the same as the eigenfunction corresponding to the eigenvalue. Examples encountered in the previous course are
p = ih

x
; p = pe
ikx
= ih
x
= hke
ikx
, (1.11)
L
z
= ih

;

L
z
=

L
z
e
im
= mhe
im
,
L
2
Y
m
(, ) = ( + 1) h
2
Y
m
(, )
For an operator

A, the Hermitian conjugate

A
is dened by
_

1
_
A
2
_
d =
_
_
1
_
2
d =
__

2
_
1
_
d
_
(1.12)
for any functions
1
and
2
. If the operator is Hermitian then

A =

A
. and so
_

1
_
A
2
_
d =
_
_
A
1
_
2
d =
__

2
_
A
1
_
d
_
(1.13)
The denition of an Hermitian operator ensures that
1. the eigenvalues are real quantities (a necessity for a measurement).
Since

A
n
= a
n
n
then by Hermiticity
a
n
_

n
d =
_

n

A
n
d =
_
_
A
n
_
n
d = a
n
_

n
d
a
n
= a
n
and so a
n
is real.
2. the eigenfunctions
n
,
m
corresponding to different eigenvalues are orthogonal, i.e.
_

n
(r)
m
(r) d
3
r = 1.
If
k
and
n
are two eigenfunctions corresponding to different eigenvalues a
k
and a
n
respectively then

A
k
=
a
k
k
and

A
n
= a
n
n
.
Hence
_

n

A
k
d = a
k
_

k
d
_
_
A
n
_
k
d = a
k
_

k
d
a
n
_

k
d = a
k
_

k
d
(a
n
a
k
)
_

k
d = 0.
4
As a
n
,= a
k
then
_

k
d = 0 (1.14)
i.e. the eigenfunctions are orthogonal.
{Note: This proof is not satisfying as it stands as sometimes an operator has more than one eigenfunction for one
eigenvalue (differing by more than a multiplicative factor). The eigenvalue is said to be degenerate. Suppose
n
and
m
are two such degenerate eigenfunctions with a
n
= a
m
. Then =
m
+c
n
is also an eigenfunction of

Awith eigenvalue
a
n
(=a
m
);
A =

A
m
+ c

A
n
= a
m
m
+ ca
n
n
= a
m
(
m
+ c
n
) = a
m
.
As c is arbitrary it can be chosen to make orthogonal to
n
, as
_

n
d =
_

n
(
m
+ c
n
) d =
_

m
d + c
_
[
n
[
2
d = 0
if
c =
_

m
d /
_
[
n
[
2
d.
Thus if the eigenvalue is degenerate, the eigenfunctions can always be chosen to be orthogonal.}
Note on Hermitian operators.
If

A,

B are two Hermitian operators then

A

B +

B

A and i(

A

B

B

A) are also Hermitian. From the denition of an
Hermitian operator
_

A
_
d =
_
_
A
_
d
then
_

A

Bd =
_
_
A
_
B
_
d =
_
_
B

A
_
d.
Similarly
_

B

Ad =
_
_
A

B
_
d.
Adding gives
_

(

A

B +

B

A)d =
_
_
(

A

B +

B

A)
_
d
so

A

B +

B

A is Hermitian.
Likewise
_

i

A

Bd =
_
_
A
_
i
_
B
_
d =
_
_
B

A
_
id =
_
_
i

B

A
_
d
_

i

B

Ad =
_
_
i

A

B
_
d
so subtracting gives
_

i(

A

B

B

A)d =
_
_
i(

A

B

B

A)
_
d
and so i(

A

B

B

A) is Hermitian, and i
_
A,

B
_
is Hermitian.
1.1.1 The Expansion Postulate
The expansion postulate is an essential postulate of quantum mechanics. It says that for every dynamical observable that
can be measured its eigenfunctions form a complete set,
n
i.e. that any wavefunction describing a possible state of the
system can be expressed as a linear combination of the eigenstates (in same domain and dimensionality). Hence
=
n
c
n
n
(1.15)
where the constants, c
n
, are called the expansion coefcients. Taking the scalar product of equ(1.15) with
k
gives
_

k
d =
n
c
n
_

n
d =
n
c
n
kn
= c
k
.
5
Thus the expansion coefcients are obtained from
c
n
=
_

n
d. (1.16)
The number of eigenfunctions may be nite or innite. In some cases the eigenfunctions of an observable form a
continuum, in which case the sum in equ(1.15) is replaced by an integral.
As an example, the functions cos (nx/2a) and sin (nx/2a) are eigenfunctions of the Hamiltonian for a particle in
a one-dimensional innite potential well (V (x) = 0, [x[ a, V(x) = 0, [x[ > a) and give rise to the Fourier expansion
of a function,
f (x) = a
0
+
n=1
_
a
n
cos
_
nx
2a
_
+ b
n
sin
_
nx
2a
__
. (1.17)
A further postulate is that if a measurement of observable A is repeated many times, always with the system in the
same initial state , the average value of A, the expectation value of A is given by
A =
_

Ad. (1.18)
Using the expansion postulate the expectation becomes
A =
_

k
(c
k
)

A
n
(c
n
n
) d
=
_

k
n
(c
k
) (c
n
a
n
n
) d
=
n
c
k
c
n
a
n
_

n
d =
n
c
k
c
n
a
n
kn
A =
n
[c
n
[
2
a
n
=
n
p
n
a
n
, (1.19)
where p
n
is the probability of value a
n
, i.e. p
n
= [c
n
[
2
. Furthermore if is normalized
1 =
_

d =
_

k
(c
k
)
n
(c
n
n
) d
=
n
c
k
c
n
_

n
d =
n
c
k
c
n
kn
=
n
[c
n
[
2
and
n
[c
n
[
2
= 1, (1.20)
as is to be expected if [p
n
[
2
is interpreted as the probability of obtaining value a
n
as a result of a measurement.
Many properties of a state are completely independent of how it is represented, e.g.probabilities of obtaining different
measured values of an observable; the expectation value of an observable cannot depend on whether the state was given
as a function of position or some other way. Thus it is desirable to have a way of discussing states which is independent of
the representation. When the concept of intrinsic angular momentum (spin) of a particle was introduced for electrons to
explain the results of a Stern-Gerlach experiment and the ne structure of spectral lines it was noted that the spin operator
cannot be represented in term of spatial coordinates, nor can the associated eigenfunction.
1.2 Dirac Notation
Dirac introduced a compact general notation for quantum states. The idea behind it is based on the observation that the
wavefunctions behave like vectors in many circumstances. The superposition of two wavefunctions
= c
1
1
+ c
2
2
, (1.21)
where c
1
and c
2
are constants is similar to the addition of two vectors B and C:
D = c
1
B+ c
2
C . (1.22)
6
Furthermore if
1
and
2
are eigenfunctions of the same operator

A they are orthonormal, so can be specied by its
two components (c
1
, c
2
) just as the position vector r = xi + yj can be specied by its two components (x, y). For three
eigenfunctions
= c
1
1
+ c
2
2
+ c
3
3
is analogous to
B = B
x
i + B
y
j + B
z
k , (1.23)
where B
x
, B
y
and B
z
are the Cartesian components of B. The full expansion
=
n
c
n
n
(1.24)
just expands the dimensionality of the space.
The wavefunction can be represented in terms of different complete orthonormal sets,
n
as
=
n
c
n
. (1.25)
Similarly, the vector B can be expressed in an alternative set of Cartesian axes, having unit vectors i
, j
, k
as
B = B
x
i
+ B
y
j
+ B
z
k
. (1.26)
For two vectors B = B
x
i + B
y
j + B
z
k and C = C
x
i + C
y
j + C
z
k the scalar product is
B C = B
x
C
x
+ B
y
C
y
+ B
z
C
z
.
If two wavefunctions and are expanded in terms of the orthonormal set of functions
n
as
=
m
b
m
m
=
n
c
n
n
the overlap integral
_

dx =
n
b
m
c
n
_

n
dx =
n
b
m
c
n
mn
=
n
b
n
c
n
(1.27)
is analogous to the scalar product B C = B
x
C
x
+B
y
C
y
+B
z
C
z
, but with the difference that (1) it involved the complex
conjugate for one of the components, (2) it can involve more dimensions than three (even an innite number).
Wavefunctions are acted on by linear operators, which map the wavefunctions onto each other e.g.
=

A . (1.28)
Vectors can also be acted on by operators; a rotation by some chosen angle about a chosen axis; or a reection in a chosen
plane. Under such operations vectors are mapped onto each other.
1.2.1 Dirac notation for state vectors
The wavefunction and its complex conjugate
are denoted by the symbols [ and [, which are called a ket and
a bra respectively so [, and
[. (The reason for these names is that a bra and a ket together make a
bracket!) The kets are members of a linear vector space whose dimensionality depends on the quantum system, and is
often innite. The same is true of the bras. The kets are vectors in the sense that
(i) they can be multiplied by (generally complex) numbers c so if [ is a member of the space then so is c[.
(ii) they can be added so if [ and [ are members of the vector space then so is [ +[.
To each ket, there is a corresponding bra. With [ the bra corresponding to [, the bra corresponding to c[ is
c
[. The bra corresponding to c

1
[ +c
2
[ is c
1
[ +c
2
[. The overlap integral of two wavefunctions and has
the properties of a scalar product,
_

d [
7
in the Dirac notation. The complex conjugate of such a quantity is
__

d
_
=
_

d (1.29)
[
= [ (1.30)
and is a key property of the scalar product of a bra and a ket. The normalization condition becomes [ = 1 and the
orthogonality
_

n
d =
mn
(1.31)
m
[
n
=
mn
. (1.32)
The eigenfunctions of an operator

A are usually characterised by some quantum number or numbers. For example,
the eigenfunctions of a particle in an innite one-dimensional well are characterised by integers, n. The eigenfunctions
for a hydrogen atom are specied by knowing the principal quantum number, n, (related to the energy level), the orbital
angular momentum quantum number, , and the magnetic quantum number, m, so that they are written
nm
(r). In
Dirac notation these are denoted by
n
[
n
[n;
nm
[
nm
[n, , m. The expansion
=
n
c
n
n
becomes
[ =
c
n
[
n
=
n
c
n
[n.
Taking the inner product (= scalar product) with a bra
m
[ (= m[) gives
m
[ = m[ =
n
c
n
m
[
n
=
n
c
n
m[n =
n
c
n
mn
.
Thus
c
n
=
n
[ = n[. (1.33)
The basis states [
m
dene a linear vector space. The terminology is adopted that in one representation
wavefunction
state function
eigenfunction

state vector
eigenvector
vector space
A closure relation can be developed for a complete set of states since
[ =
n
c
n
[n,
c
n
= n[
and then
[ =
n
n[[n =
n
[nn[
so that formally the closure relation
n
[nn[ = 1 (1.34)
behaves as an identity operator when the summation is over all eigenstates [n.
The result of acting with operator

A on ket [ is another ket written as

A[. The scalar product of the ket

A[ with
the bra [ is written as [
A[. The Hermitian conjugate

A
of an operator

A was dened by
_

A
_
d =
_
_
d =
__

d
_
. (1.35)
In Dirac notation, this relationship reads:
[
A[ =
[ = [
. (1.36)
Thus the bra corresponding to the ket

A[ can be written as [
. This means that the quantity [
A[ can be
interpreted in two equivalent ways: (i) it is the scalar product of bra [ with ket

A[; (ii) it is the scalar product of bra
[
A with ket [. This equivalence is an elegant feature of the Dirac notation.

8
If

A is Hermitian,

A =

A
and
[
A[ = [
A[
. (1.37)
Hence in Dirac notation the proof that the eigenvalues are real becomes, if [n is an eigenstate of operator

A then
n[
A[n = n[
A[n
An[n
,
a
n
n[n = a
n
n[n
so a
n
is real.
The expectation value of observable A is
A = [
A[ =
n
c
m
c
n
a
n
m[n =
n
[c
n
[
2
a
n
. (1.38)
The normalization is
1 = [ =
n
m[c
m
c
n
[n =
n
c
m
c
n
m[n,
1 =
n
[c
n
[
2
.
Hence from the expression for the expectation value it follows that [c
n
[
2
is probability that eigenvalue a
n
is obtained
as a measurement of the observable A. After measurement the eigenfunction is the eigenstate [n, i.e. the wavefunction
"collapses".
In Dirac notation an operator

A transforms a ket [ into another ket [ as
[ =

A[.
Taking the scalar product with the bra [ gives
[ = [
A[.
The quantity [
A[ is called the matrix element of the operator

A between the states [ and [.
1.3 The position (coordinate) representation
Since a particles position along the x-axis can be measured there is an Hermitian operator x for this observable. The
result of the measurement can be any real number, < x < . If the eigenvalues of operator x are x
they form a
continuum. The corresponding eigenvector [x
satises
x[x
= x
[x
. (1.39)
The expansion in discrete states [n, [ =
n
n[[n has to become an integral
[ =
_
x/[[x
dx
. (1.40)
The components x
[ constitute a complex-valued function of the real variable x
. This function is identied with the

wavefunction as
(x
) = x
[
so establishing the connection between the (abstract) state vector [ and the wavefunction (x
). Thus (x
) is a
component of the state vector in the innitely dimensional abstract vector space. Thus (x
) is merely just one of many

possible ways of representing the state vector. This representation is called the coordinate or position representation.
Equally the momentum eigenvectors of the momentum operator p[p
= p
[p
and the momentum wavefunction (p
) =
p
[ would be obtained in the momentum representation.

Note, by taking the scalar product of [ with [x gives
x[ =
_
x[x
[dx
so the integral version of the closure relation is

_
[xx[dx = 1. (1.41)
9
This implies that normalization of the states [x
has to be
x[x
= (x x
) (1.42)
where is the Dirac -function.
As an advanced application, begin with the basic commutator
[ x, p
x
] = ih. (1.43)
Its matrix element is
x
[ x p
x
p
x
x[ x
= ihx
[x
= ih (x
)
(x
) x
[ p
x
[x
= ih (x
)
x
[ p
x
[x
=
ih (x
)
(x
)
. (1.44)
Making use of the property of the derivative of the Dirac delta-function, x
(x) = (x) then

x
[ p
x
[x
= ih

x
(x
) . (1.45)
Consider p
x
operating on a state [ =
_
[x
[dx
as
p
x
[ =
_
p
x
[x
[dx
and so its expectation value is

[ p
x
[ =
_ _
[x
dx
[ p
x
[x
dx
[,
=
_ _
[x
dx
_
ih

x
(x
)
_
dx
[,
=
_ _

(x
) dx
_
ih

x
(x
)
_
dx
(x
) . (1.46)
Evaluating the (x
) dx
-function over x
, (i.e. at x
= x
) gives
[ p
x
[ =
_

(x
)
_
ih

x
_
dx
(x
) (1.47)
as the expectation value of p
x
in the state [. Hence the representation of the momentum operator p
x
in the coordinate
representation is
p
x
= ih

x
(1.48)
as plausibly suggested in an earlier quantum mechanics course!
1.4 Basis states
The state of a quantumsystemcan be specied by writing the wavefunction as a function of position, (r); (r
1
, r
2
. . . r
N
).
However there are other ways of specifying the state. For example, can be expressed as a linear combination of the
complete set of eigenfunctions
n
of an operator

A, as =
n
c
n
n
. In this representation the wavefunction is
specied by the list of coefcients c
n
instead of being given as an explicit function of position. {If is given as a function
of position, essentially this is a representation in terms of the eigenstates of the position operator x.} For example,
(x) =
n
A
n
1
a
sin
_
nx
2a
_
=
n
A
n
n
(x) (1.49)
as
n
(x) =
1
a
sin
_
nx
2a
_
are eigenfunctions of the Hamiltonian

H for the innite potential well of width 2a.
The wave function could be specied in terms of the eigenfunctions of the momentum operator
p(x) = ih

x
(x) = p(x) ,
(p) = e
ipx/ h
.
10
Then
(x) =
_

(p) e
ipx/ h
dp (1.50)
(a Fourier transform) and the state of the system is specied by the function

(p). For discrete momentum values
(x) =
m
B
m
1
2a
e
ik
m
x
=
m
B
m
m
(x) .
If the energy is measured (x) will reduce to an energy eigenstate
n
(x) with probability [A[
2
; if momentum is
measured the value p = hk
m
is obtained with a probability [B
m
[
2
. In the above case, conversion from one representation
is straight forward since sin k
m
x =
1
2i
_
e
ik
m
x
e
ik
m
x
_
. Other cases are less straight forward and require formal
methods.
Many properties of a state are completely independent of how it is represented, e.g.probabilities of obtaining different
measured values of an observable cannot depend on whether the state was given as a function of position or some other
way. Thus it is desirable to have a way of discussing states which is independent of the representation. When the concept
of intrinsic angular momentum (spin) of a particle was introduced for electrons to explain the results of the Stern-Gerlach
experiment and ne structure of spectral lines it was noted that the spin operator cannot be represented in terms of spatial
coordinates, or their derivatives, and nor can the associated eigenfunctions.
1.4.1 Change of basis - change of representation
Transformation of states
Cold atom physicists can easily adjust the shape of the trapping potential which connes the atoms. Typically this is an
harmonic potential V
T
(x). The Hamiltonian is
H (x, p) =
p
2
2m
+ V
T
(x) , (1.51)
with V
T
(x) =
1
2
m
2
x
2
. Solutions of the time-independent Schrdinger equation are
n
(x) = N
n
H
n
(x) e
x
2
/2
(1.52)
where H
n
(x) is a Hermite polynomial of degree n and = (m/h)
1/2
. The physicist can increase (and thus ) by
increasing the strength of the magnetic elds which conne the atom. A tight trap has large and , and hence widely
spaced energy levels E
n
=
_
n +
1
2
_
h and rapid decay of the tail of the probability distribution [ (x)[
2
showing the
atoms are tightly conned; conversely for a weak potential - small and .
Suppose experimenter A prepares atoms in a well dened state by using dissipative laser cooling processes until all
the atoms are in the ground state. The cooling lasers are turned off and the atoms are in state
0
(x) of a potential
V

0
T
(x) =
1
2
m
2
0
x
2
. The conning elds are abruptly weakened and the trap centre is displaced to x = x
1
. The state
function is still
0
(x) but it is no longer an eigenfunction of the new Hamiltonian of the system

H =
p
2
2m
+V

T
(x x
1
).
A theorist B represents the state in the new basis as
[
0
=
n
c
n
[
n
(1.53)
with
x[
n

n
(x) = N
n
H
n
(x x
1
) e
(xx
1
)
2
/2
(1.54)
with = (m/h)
1/2
. The coefcients c
n
are calculated by
c
n
=
n
[
0
(1.55)
to get probabilities [c
n
[
2
and expectation values. However suppose the settings in the apparatus are wrong and the trap
frequency is really
, = (m
/h)
1/2
with eigenstates [
m
,
x[
m

m
(x) = N
m
H
m
(x x
1
) e
(xx
1
)
2
/2
. (1.56)
The theorists expansion should have been
[
0
=
m
d
m
[
m
. (1.57)
11
The eigenstates [
n
and [
m
both form a complete set of basis functions and mathematically both expansions are
equally valid. But only for eq(1.57) can the time evolution be written as
[
0
(t) =
m
d
m
[
m
e
iE
m
t/ h
(1.58)
as these states (and energies) correspond to the real apparatus. It is often advantageous to change representations, even in
mid-calculation. In the experiment different potentials V
T
(x) may exist for different times.
The basis states can be expanded in terms of each other, as each forms a complete set. Explicitly
[
n
=
m
[
n
[
m
=
m
S
mn
[
m
(1.59)
and conversely
[
m
=
n
[
m
[
n
. (1.60)
Hence from eq(1.57) and (1.53) the expansion coefcients are given by
d
m
=
m
[
0
=
m
[
n
c
n
[
n
=
m
[
n
c
n
(1.61)
d
m
=
n
S
mn
c
n
. (1.62)
This is an example of matrix multiplication of column vectors d and c and matrix S with S
mn
=
m
[
n
,
d = Sc,
_
_
_
_
_
d
1
d
2
.
.
.
d
N
_
_
_
_
_
=
_
_
_
_
_
1
[
1

1
[
2

1
[
N
2
[
1

2
[
2

2
[
N
.
.
.
.
.
.
.
.
.
N
[
1

N
[
2

N
[
N
_
_
_
_
_
_
_
_
_
_
c
1
c
2
.
.
.
c
N
_
_
_
_
_
. (1.63)
S is the matrix which transforms between the two representations of [
0
. Noting that c
n
=
n
[
0
and using the
expansion [
0
=
m
d
m
[
m
gives the inverse transformation
_
_
_
_
_
c
1
c
2
.
.
.
c
N
_
_
_
_
_
=
_
_
_
_
_
1
[
1

1
[
2

1
[
N
2
[
1

2
[
2

2
[
N
.
.
.
.
.
.
.
.
.
N
[
1

N
[
2

N
[
N
_
_
_
_
_
_
_
_
_
_
d
1
d
2
.
.
.
d
N
_
_
_
_
_
(1.64)
c = S
d (1.65)
where S
has elements S
nm
(transpose and complex conjugation of S).
From
d = Sc
S
1
d = S
1
Sc = c = S
d
S
1
= S
and so S is unitary, i.e.

SS
= I
n
. (1.66)
The same results can easily be found using the closure property of a complete set of states. Recall that [ =
n
c
n
[
n
and c
n
=
n
[
0
so that
[
0
=
n
[
0
[
n
=
n
[
n

n
[
0
(1.67)
so that formally
n
[
n
n
[ = 1. (1.68)
Then since
d
m
=
m
[ , (1.69)
12
using the closure relation
d
m
=
m
[
n
n
[
0
=
n
S
mn
c
n
(1.70)
with
S
mn
=
m
[
n
(1.71)
and in matrix form
d = Sc. (1.72)
1.5 Matrix representation
A state [ can be expanded in terms of a complete set of basis states
n
as
[ =
n
c
n
[
n
. (1.73)
The constants c
n
can be considered as the elements of a columns vector
c =
_
_
_
_
_
c
1
c
2
.
.
.
c
n
_
_
_
_
_
(1.74)
with c
n
=
n
[.
The action of an operator

A on the wavefunction is to generate another wavefunction [ =

A[. The state [ can
also be expanded in terms of the basis states
n
as
[ =
m
d
m
[
n
(1.75)
with d
m
=
m
[. Thus
d
m
=
m
[ =
m
[
A[ =
m
[
A[
n
c
n
[
n
=
m
[
A[
n
c
n
. (1.76)
The quantities A
mn
=
m
[
A[
n
are called the matrix elements of the operator

Ain the basis
n
. The above equation
becomes
d
m
=
n
A
mn
c
n
and d
m
are elements of a column vector
d =
_
_
_
_
_
d
1
d
2
.
.
.
d
n
_
_
_
_
_
(1.77)
with the matrix equation
d = Ac (1.78)
_
_
_
_
_
d
1
d
2
.
.
.
d
n
_
_
_
_
_
=
_
_
_
_
_
A
11
A
12
. . . A
1n
A
21
A
22
. . . A
2n
.
.
.
.
.
.
.
.
.
.
.
.
A
n1
A
n2
. . . A
nn
_
_
_
_
_
_
_
_
_
_
c
1
c
2
.
.
.
c
n
_
_
_
_
_
(1.79)
In the coordinate representation
A
mn
=
m
[
A[
n
=
_

m
(x)

A
n
(x) d. (1.80)
The scalar product
[ =
n
d
m
c
n
m
[
n
=
n
d
n
c
n
as
m
[
n
=
mn
so
[ =
n
d
n
c
n
= d
c =(d
1
, d
2
, . . . d
n
)
_
_
_
_
_
c
1
c
2
.
.
.
c
n
_
_
_
_
_
(1.81)
These ideas lead to a matrix representation of quantum systems - a formulation put forward by Heisenberg.
13
1.5.1 Eigenvalue equation and eigenvalues in matrix representation
Suppose

A[ = a[. In matrix representation this is written

Ac = ac. Considering the i-th component,
n
A
in
c
n1
= ac
i1
= a
in
c
n1
(A
in
in
a) c
n1
= 0.
This is equivalent to
_
_
_
_
_
A
11
a A
12
A
13
. . .
A
21
A
22
a A
23
. . .
A
31
A
32
A
33
a . . .
.
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
_
_
_
c
1
c
2
c
3
.
.
.
_
_
_
_
_
= 0. (1.82)
This equation has a non-trivial solution for the cs if
det [A
in
in
a[ = 0 = det [AaI[ . (1.83)
This is a good way of nding the eigenvalues for operators represented by nite matrices - not so simple for innite
matrices!
Transformation of operators
Recall that the matrix representation of an operator equation such as
=

A (1.84)
is
b = Ac (1.85)
where
[ =
n
b
n
[n; [ =
n
c
n
[n (1.86)
in the same basis [n and the operator is represented by the matrix
A
kj
=
_
k[
A[j
_
(1.87)
whence
b
k
=
j
A
kj
c
j
. (1.88)
The operator

A in the basis [
n
is represented by the matrix
_
i
[
A[
n
_
and by
_
A[
m
_
in the basis [
n
.
Using the closure relation twice
_
A[
m
_
=
j
_
[
i
i

A[
j
j
[
m
_
(1.89)
_
_
m
=
j
S
li

A
ij
S
jm
(1.90)
= S

A
= S

A
S
1
, (1.91)
and the converse
= S
1

A
S. (1.92)
Eq(1.91) and (1.92) are called similarity transformations of the matrices representing the operator

A.
Let
A
m
=
_
A[
m
_
and
A
in
=
_
i
[
A[
n
_
But
[
m
=
n
[
m
[
n
(1.93)
14
and
[
[ =
[
i

i
[ (1.94)
so that
_
A[
m
_
=
[
i

i
[
n
[
n

n
[
m
(1.95)
=
[
i

i
[
A[
n

n
[
m
(1.96)
=
n
S
i
A
in
S
nm
(1.97)
A = SA
= SA
S
1
, (1.98)
A
= SA
S
1
. (1.99)
Conversely
AS = SA
S
1
S = SA
(1.100)
A
= S
1
AS. (1.101)
Eq(1.98) and (1.101) are called similarity transformations of the matrices representing the operator

A.
Transforming operators, especially the Hamiltonian between bases can be an efcient way to solve the Schrdinger
equation. As an example: represent

H with the basis [
n
(of harmonic oscillator states) with
H[
n
= E
n
[ =
_
n +
1
2
_
h[, (1.102)
so that
H
nn
=
_
n
[
H[
n

_
= E
n
nn
(1.103)
and H is a diagonal matrix
H =
_
_
_
_
_
_
E
1
0
0 E
2
0
.
.
. 0 E
3

.
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
. (1.104)
{The matrix of an operator in a basis of its own eigenstates is diagonal, i.e. if

A[n =
n
[n, then
_
n[
A[n
_
=
n
nn
.}
However the matrix H
of

H in the basis [
n
is not diagonal in this example as [
n
are not the eigenfunctions;
the [
m
are. However the eigenvalues of H
are still
_
n +
1
2
_
h, i.e. the (energy) spectrum is independent of the
representation.
Proof that H and H
have the same spectrum To nd the eigenvalues of H
; H
[n =
n
[n, the characteristic
equation
det (H
I) = 0
must be solved. But H
= S
1
HS and I = S
1
S so and
H
I = S
1
HS S
1
S = S
1
(H I) S
and
det (H
I) = det
_
S
1
(H I) S
.
But det (AB) = det Adet B so
det (H
I) = det
_
S
1
_
det (H I) det S
= det
_
S
1
_
det S det (H I)
= det (H I) ,
since det
_
S
1
_
det S = det
_
S
1
S
_
= det (I). The eigenvalues of H are known and hence so are those of H
. Also
both the eigenvalues as well as the trace of the matrix of any operator

A are independent of the (basis) representation.
15
Demonstration that the expectation value is independent of the (basis) representation
The expectation value of an operator

A can be evaluated from
_
A
_
=
_
[
A[
_
. (1.105)
If

A is represented in a basis [
n
which is not formed from its eigenfunctions, the matrix representing

A is non-
diagonal, with elements
A
nn
=
_
n
[
A[
n
_
. (1.106)
Expanding [ in the same basis gives
[ =
n
c
n
[
n
(1.107)
and
_
[
A[
_
=
n
[c
n

A
c
n
[
n
=
n
A
nn
c
n
(1.108)
=
_
c
T
_
c (1.109)
Using the closure relation
_
[
A[
_
=
_
[
n
n
[
A[
n

n
[
_
, (1.110)
=
n

A
nn
c
n
, (1.111)
= c
c.
If [
j
is an eigenbasis of

A then
A[
j
=
j
[
j
(1.112)
and
j
[A[
i
=
j
ji
(1.113)
i.e. is diagonal. It could also have been used to expand [ as
[ =
j
d
j
[
j
(1.114)
then
_
[
A[
_
=
j
[d
j

A
j
d
i
[
i
=
j
d
j
[
A[
i
d
i
, (1.115)
=
_
d
T
_
d = d
d. (1.116)
_
[
A[
_
= (d
1
, d
2
, . . .)
_
_
_
1
0
0
2

.
.
.
.
.
.
_
_
_
_
_
_
d
1
d
2
.
.
.
_
_
_. (1.117)
=
d
2
j
j
. (1.118)
Eq(1.109) and (1.118) give the same result. This follows from SS
1
= 1; d = Sc; d
= c
and A
= SA
S
1
, so
d
d = c
Sc = c
S
1
A
Sc,
= c
c, (1.119)
and thus
A =
_
[
A[
_
= c
c = d
d. (1.120)
16
Transforming one state vector to another
Here we ask the question, what operator

A will transform a given state [ into [ ?
Suppose operator

A transforms state [ to state [ so that
[ =

A[. (1.121)
In the basis [n
[ =
m
d
m
[m; [ =
n
c
n
[n (1.122)
and
d = Ac (1.123)
m
d
m
[m =

A
n
c
n
[n, (1.124)
d
m
=
n
c
n
m[
A[n =
n
A
mn
c
n
(1.125)
But
n
[c
n
[
2
= 1 so
d
m
= d
m
n
[c
n
[
2
=
n
m[c
n
c
n
d
m
=
n
m[[nc
n
(1.126)
Comparing the expressions eq(1.125) and (1.126) for d
m
gives
A
mn
= m[[n (1.127)
or
A = [[. (1.128)
as the required projection operator.
Alternatively, since
[ =

A[
then
[[ =

A[[ (1.129)
=

A
m
c
m
[m
n
c
n
n[ =

A
n
c
m
c
n
n[m =

A
n
c
m
c
mn
(1.130)
=

A
n
[c
n
[
2
=

A (1.131)
since for the expansion coefcients, normalization of [ requires
n
[c
n
[
2
= 1.
Note that for any basis [n which is a complete set of states, a state [ has the expansion
[ =
n
c
n
[n (1.132)
and
c
n
= n[. (1.133)
Thus
[ =
n
n[[n =
n
[nn[. (1.134)
Thus there is the identity
1 =
n
[nn[ (1.135)
for a complete set of states.
17
1.5.2 Change of basis - an example
N.B. The example below uses ideas from later in the course, but is included for completeness Spin is associated with
a magnetic dipole moment carried by a spin-
1
2
particle, e.g. an electron where
= g
s
B
h
s (1.136)
with
B
=
e h
2m
e
the Bohr magneton, g
s
the Land g-factor for the electron ( 2). In an external magnetic eld B the
energy is
H = E = B =g
s
B
h
s B =
e
s B (1.137)
with
e
= g
s
B
/h. In MRI/NMR it is the nuclear protons spin that is involved with
N
= eh/2m
p
instead of the Bohr
magneton.
Consider a spin-
1
2
particle (or system) in a magnetic eld B = B
x
+ B
z
k. The Hamiltonian is
H =
e
s B
=
e
_
B
x

S
x
+ B
z

S
z
_
=
e
h
2
(B
x

x
+ B
z

z
)
H = a
z
+ b
x
. (1.138)
Suppose a/h = 4MHz and b/h = 3MHz . The matrix representation of

H in the basis functions of
z
is
H
(z)
= a
_
1 0
0 1
_
+ b
_
0 1
1 0
_
=
_
a b
b a
_
. (1.139)
The matrix representation of

H in the basis functions of
x
is
H
(x)
=
_
x
+[
H[+
x x
+[
H[
x
x
[
H[+
x x
[
H[
x
_
. (1.140)
But
[+
x
=
1
2
([ +[) =
1
2
_
1
1
_
and
[
x
=
1
2
([ [) =
1
2
_
1
1
_
and hence
x
+[
H[+
x
=
1
2
([ +[) (a
z
+ b
x
) ([ +[) .
But
z
[ = [;
z
[ = [;
x
[ = [;
x
[ = [ so
x
+[
H[+
x
=
1
2
([ +[) (a[ a[ + b[ + b[)
=
1
2
(a + b a + b) = b. (1.141)
Using the Pauli matrices
x
+[
H[+
x
=
1
2
(1, 1)
_
a b
b a
__
1
1
_
=
1
2
(1, 1)
_
a + b
b a
_
= b,
x
[
H[+
x
=
1
2
(1, 1)
_
a b
b a
__
1
1
_
=
1
2
(1, 1)
_
a + b
b a
_
= a,
x
[
H[
x
=
1
2
(1, 1)
_
a b
b a
__
1
1
_
=
1
2
(1, 1)
_
a b
b + a
_
= b.
Hence
H
(x)
=
_
b a
a b
_
(1.142)
18
H
(x)
and

H
(z)
are the same Hamiltonian but represented in different bases. It is easy to verify that

H
(x)
and

H
(z)
have
the same spectrum (of eigenvalues). For

H
(z)
one has
a b
b a
= 0
giving (a ) (a + ) b
2
= 0,
2
a
2
b
2
= 0 and =
a
2
+ b
2
= 5. For

H
(x)
one has
b a
a b
= 0
so (b ) (b + ) a
2
= 0,
2
a
2
b
2
= 0 and =
a
2
+ b
2
= 5.
Suppose that the system is in the ground state [
0
of

H with eigenvalue
a
2
+ b
2
= 5. What is this state in
terms of the basis of
z
? (If a measurement of a sample of these spins along the z-axis is measured the magnetization is
M
z
=
e
N S
z
.) If the eigenvalue is and [
0
= p[ + q[ then
H[
0
= [
0
(a
z
+ b
x
) (p[ + q[) = (p[ + q[)
or in terms of matrices
_
a b
b a
__
p
q
_
=
_
p
q
_
_
4 3
3 4
__
p
q
_
= 5
_
p
q
_
4p + 3q = 5p
3p 4q = 5q
so q = 3p. Normalization requires [p[
2
+[q[
2
= 1 so p =
1
10
and q =
3
10
and
[
0
=
1
10
[
3
10
[. (1.143)
For = +5 the eigenstate is
[
1
=
3
10
[ +
1
10
[. (1.144)
The expectation value
_
S
z
_
=
h
2

0
[
z
[
0
=
h
2
1
10
(1, 3)
_
1 0
0 1
__
1
3
_
1
10
_
S
z
_
=
h
2
1
10
(8) =
4
5
_
h
2
_
,
which is just the probability of "up"
_
1
10
_
h
2
__
and "down"
_
9
10
_
h
2
__
.
What is the probability of the spin pointing along the +xaxis? The ground state [
0
=
1
10
[
3
10
[
c
1
[ + c
2
[ needs to be expanded in terms of the eigenstates of

S
x
as
[
0
= d
1
[+
x
+ d
2
[
x
= c
1
[ + c
2
[
Taking the scalar product with [+
x
gives
d
1x
+[+
x
+ d
2x
+[
x
= c
1x
+[ + c
2x
+[
But [+
x
and [
x
are orthogonal so
d
1x
=
x
+[c
1
+
x
+[c
2
.
Similarly using [
x
d
2x
=
x
[c
1
+
x
[c
2
.
19
These equations can be represented by the matrix equation
d = Sc (1.145)
_
d
1
d
2
_
=
_
x
+[
x
+[
x
[
x
[
__
c
1
c
2
_
. (1.146)
But [+
x
=
1
2
([ +[) and [
x
=
1
2
([ [) so
x
+[ =
1
2
etc. so that
S =
_
1
2
1
2
1
2

1
2
_
=
1
2
_
1 1
1 1
_
and
S
=
1
2
_
1 1
1 1
_
. (1.147)
Thus S is the matrix that transforms representations in the z-basis to those of the x-basis. S
does the reverse. Note that

SS
=
1
2
_
1 1
1 1
__
1 1
1 1
_
=
1
2
_
2 0
0 2
_
= I
2
as it should!
{If working out S for the transformation between the z-basis and the y-basis care must be taken as there are is when
taking transposes and complex conjugations.}
In this case
_
d
1
d
2
_
=
1
2
_
1 1
1 1
_
1
10
_
1
3
_
=
1
5
_
1
2
_
and so
[
0
=
1
5
[+
x
+
2
5
[
x
. (1.148)
The probability of measuring spin along the +x-axis is [d
1
[
2
=
1
5
and the probability of measuring spin along the x-axis
is [d
2
[
2
=
4
5
and the magnetization
P
+
P
=
0
[
x
[
0
=
3
5
. (1.149)
The expectation (and magnetization) are measurable quantities and should be independent of the representation. This is
easily veried as in the z-representation
0
[
x
[
0
=
1
10
(1, 3)
_
0 1
1 0
_
1
10
_
1
3
_
=
3
5
. (1.150)
Transformation of the operators between representations
The Hamiltonian operators

H
(x)
and

H
(z)
are related by a similarity transform
H
(x)
= S

H
(z)
S
1
= S

H
(z)
S
(1.151)
Explicitly
_
b a
a b
_
=
1
2
_
1 1
1 1
__
a b
b a
_
1
2
_
1 1
1 1
_
(1.152)
For the simple 2 2 matrix and spin-
1
2
problems most quantities can be calculated using the transformation of the basis,
e.g.
[+
x
=
1
2
([ +[) ; [
x
=
1
2
([ [) (1.153)
and their inverses
[ =
1
2
([+
x
+[
x
) ; [ =
1
2
([+
x
[
x
) (1.154)
But the similarity transforms are best for complex systems of many coupled spins where the matrices can be quite large.
Also if the magnet eld is constantly changing as in

H (t) =
e

S
z
B
z
(t) +
e

S
x
B
x
(t). Note that the matrix S is
independent of B
x
(t) and B
z
(t) so one can switch back and forth thousands of times to calculate the magnetizations
M
x
, M
y
, M
z
at different times.
20
1.5.3 Solving for energy eigenvalues
Nowadays most solutions of the time-independent Schrdinger equation are achieved by diagonalizing a matrix such as
H
rather than solving second-order partial differential equations.

Suppose that the true eigenstates [ are not known, but that [ is a known basis. Then
H
n
= E
n
[
n
=

H
j
c
(n)
j
[
j
(1.155)
but also the coefcients c
(n)
j
are unknown (if they were known then so would the [ be known).
Procedure:
1. The effect of an operator on a state is given by the effect of its matrix on the states column vector
2. Represent

H in the basis [ by H
3. Matrix H
has the correct energy spectrum, thus

H
c
(n)
=
_
_
_
H
11
H
12

H
21
H
22

.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
c
(n)
1
c
(n)
2
.
.
.
_
_
_
_
= E
n
_
_
_
_
c
(n)
1
c
(n)
2
.
.
.
_
_
_
_
(1.156)
4. Solve det (H
I) = 0 to get the eigenvalues E

n
5. Each eigenvector c
(n)
contains the coefcients for eq(1.155) so solve
H
c
(n)
= E
n
c
(n)
(1.157)
Hence now all the eigenvalues and eigenstates are known without solving the wave equation.
21
22
CHAPTER 2. QHO
Chapter 2
Quantum Harmonic Oscillator; Generalised
Uncertainty Relations
2.1 Classical mechanics
The harmonic oscillator is an important system to study as many physical systems are approximated by it. Many systems
have an interaction potential of this general form, with a quadratic potential.
Expanding the potential function about the minimum at x = x
0
as a Taylor series gives
V (x) = V (x
0
) + (x x
0
)
_
dV
dx
_
x
0
+
1
2
(x x
0
)
2
_
d
2
V
dx
2
_
x
0
+
1
6
(x x
0
)
3
_
d
3
V
dx
3
_
x
0
+ (2.1)
At the minimum
_
dV
dx
_
x
0
= 0. The force of interaction and equation of motion of a particle of mass m are
F =
dV
dx
x = m
d
2
x
dx
2
x
m
d
2
x
dx
2
= (x x
0
)
_
d
2
V
dx
2
_
x
0
1
2
(x x
0
)
2
_
d
3
V
dx
3
_
x
0
.
For small displacements from the equilibrium position x = x
0
the leading term is
d
2
x
dx
2
=
1
m
_
d
2
V
dx
2
_
x
0
(x x
0
) (2.2)
which is the equation of simple harmonic motion about x = x
0
.
Dening the potentials with respect to the minimum V (x
0
) and displacements relative to x
0
gives
V (x) =
1
2
kx
2
(2.3)
with
k =
_
d
2
V
dx
2
_
x
0
(2.4)
and
d
2
x
dx
2
=
k
m
x =
2
x, (2.5)
with angular frequency =
_
k/m and solution x(t) = Acos (t) + sin (t). The turning points are at x =
_
2E/m
2
_
1/2
.
2.2 Schrdinger equation for harmonic oscillator
Expressing the potential as V (x) =
1
2
m
2
x
2
, the time-independent Schrdinger equation is
h
2
2m
d
2
(x)
dx
2
+
1
2
m
2
x
2
(x) = E (x) . (2.6)
23
PHAS3226: Quantum Mechanics CHAPTER 2. QHO
By introducing the dimensionless variable = 2E/h and = x where = (m/h)
1/2
the Schrdinger equation
becomes
_
d
dx
=
d
d
_
d
2
()
d
2
+
_

2
_
() = 0. (2.7)
This is a second-order differential equation which can be solved by a power series approach. The quantum harmonic
oscillator eigenfunctions are
n
() = N
n
H
n
() e
2
/2
, (2.8)
where H
n
() is Hermite polynomial of degree n, parity (1)
n
and N
n
is a normalisation constant
N
n
=
_

2
n
n!
_
1/2
. (2.9)
The eigenvalues
n
= 2n + 1 give energy eigenvalues
E
n
=
_
n +
1
2
_
h, (2.10)
with n = 1, 2, . . ..
In unscaled variables
(x) = N
n
() H
n
(x) e
2
x
2
/2
. (2.11)
Thus large implies a deep potential well, large and a short tail (rapid fall off) of the wavefunction (and vice versa).
The eigenfunctions are orthonormal i.e.
_

m
(x)
n
(x) dx =
N
m
N
n
2
H
m
() H
n
() d =
mn
, (2.12)
and
_

2
H
m
() H
n
() d =
2
n
n!
mn
. (2.13)
Recurrence relations exist for the Hermite polynomials
H
n+1
(x) 2xH
n
(x) + 2nH
n1
(x) = 0
dH
n
(x)
dx
= 2nH
n1
(x) .
The recurrence relations enable evaluation of the matrix elements, e.g.
x
kn
=
_

m
(x) x
n
(x) dx =
k
[x[
n
=
_
h
2m
_
1/2 _
k + 1
k+1,n
+
k
k1,n
_
. (2.14)
2.3 Algebraic operator approach
In many physical systems involving harmonic oscillators the interest is in the energy eigenvalues or the matrix elements
and the explicit form of the eigenfunctions is not needed (or is even hard to nd). This can be achieved by an elegant
operator approach.
The Hamiltonian can be written as
H =
p
2
2m
+
1
2
m
2
x
2
(2.15)
with the momentum and position operators satisfying
[ x, p
x
] = ih. (2.16)
Two new operators a
+
and a
are dened by
a
+
=
_
1
2 hm
_
1/2
(m x i p) =
1
2
__
_
m
h
_
1/2
x
i p
(mh)
1/2
__
(2.17)
a
+
=
1
2
_
x
i
h
p
_
, (2.18)
a
=
_
1
2 hm
_
1/2
(m x + i p) =
1
2
_
x +
i
h
p
_
. (2.19)
24
The operators a
+
and a
are called raising and lowering operators (also creation and annihilation operators) respectively
for reasons that will be clearer later. Since x and p are Hermitian operators, a
+
and a
are Hermitian conjugates of each

other,
a
+
= a
; a
= a
+
(2.20)
These operators satisfy the commutator
[ a
, a
+
] = 1. (2.21)
Explicitly
[ a
, a
+
] =
1
2 hm
([m x + i p, m x i p])
1
2 hm
im ([ x, p] + [ p, x])
since [ x, x] = [ p, p] = 0. But [ x, p
x
] = ih so
[ a
, a
+
] =
i
2 h
(ih ih) = 1. (2.22)
Introducing the Hermitian operator
N = a
+
a
(2.23)
then
a
+
a
=
1
2hm
(m x i p) (m x + i p)
=
1
2hm
_
m
2
2
x
2
+ p
2
+ im ( x p p x)
_
=
1
2hm
_
m
2
2
x
2
+ p
2
+ imih
_
=
1
h
_
1
2
m
2
x
2
+
p
2
2m
_
1
2
=
1
h
H
1
2
.
Hence the Hamiltonian can be written as
H = h
_
a
+
a
+
1
2
_
= h
_

N +
1
2
_
. (2.24)
The operator

N = a
+
a
is called the number operator for reasons that will be more obvious later! It follows that
N = a
+
a
commutes with the Hamiltonian

H and thus they have a common set of eigenvectors.
{Using the form a
+
=
1
2
_
x
i
h
p
[ a
, a
+
] =
1
2
__
x +
i
h
p
_
,
_
x
i
h
p
__
Expanding using commutator algebra and noting that x and p commute with themselves, so
[ a
, a
+
] =
1
2
_
x,
i
h
p
_
+
_
i
h
p, x
_
=
1
2
__
i
h
_
[ x, p] +
i
h
[ p, x]
_
=
1
2
__
i
h
_
ih +
i
h
(ih)
_
= 1.
Also
a
+
a
=
1
2
_
x
i
h
p
_
1
2
_
x +
i
h
p
_
(2.25)
=
1
2
_
2
x
2
+
p
2
h
2
2
+ x
i
h
p
i
h
p x
_
=
1
2
_
2
x
2
+
p
2
h
2
2
+
i
h
[ x, p]
_
=
1
2
_
2
x
2
+
p
2
h
2
2
+
i
h
ih
_
=
1
2
_
2
x
2
+
p
2
h
2
2
1
_
=
1
2
_
m
h
_
x
2
+
p
2
2hm

1
2
.
=
1
h
_
1
2
m
2
x
2
+
p
2
2m
_
1
2
=
1
h
H
1
2
, (2.26)
25
and so
H =
_
a
+
a
+
1
2
_
h.
as before.}
Useful commutators are _
H, a
+
_
= h a
+
;
_
H, a
_
= h a
. (2.27)
Proof:
_
H, a
+
_
= h
__
a
+
a
+
1
2
_
, a
+
_
= h [ a
+
a
, a
+
] = h a
+
[ a
, a
+
] + [ a
+
, a
+
] a
= h a
+
_
H, a
_
= h
__
a
+
a
+
1
2
_
, a
_
= h [ a
+
a
, a
] = h a
+
[ a
, a
] + [ a
+
, a
] a
= h a
.
2.3.1 Eigenvectors of

N
Suppose [ is an eigenvector of

N with eigenvalue , then
N[ = [
and
= [ =
[ a
+
a
[ =
a
[ a
=
since a
+
= a
. Consider the state [
= a
[, then from above

_
_
= .
But by necessity
0, and hence 0.
Consider
N[
= ( a
+
a
) a
[ = ( a
a
+
1) a
[
= a
( a
+
a
) [ a
[ = a

N[ a
[
= a
[ a
N ( a
[) = ( 1) ( a
[) (2.28)
Thus a
[ is also an eigenvector of

N but with eigenvalue 1. Hence the reason that a
is called a lowering operator.

Note that a
+
has the opposite effect, i.e.

N ( a
+
[) = ( + 1) ( a
+
[) and hence is a raising operator and obviously
a
+
[
= a
+
a
[ = [.
Thus from the state [ a sequence of eigenstates can be formed by repeated application of a
i.e. a[, ( a
)
2
[,
( a
)
3
[, . . . ( a
)
n
[ having eigenvalues 1, 2, 3, . . . n. Hence can only have integer values (positive
or zero) since 0. If positive non-integer values of existed, the sequence , 1, 2, 3, . . . n would be
unending with some values which are negative contrary to the requirement that they are all greater or equal to zero. If
is an integer, then n can be zero and its associated eigenstate [
0
satises
a
[
0
= 0[
0
= 0,
a
+
a
[
0
= 0[
0
N[
0
= 0[
0
, (2.29)
so the eigenvalues of

N are positive integers and zero. The corresponding eigenstate is denoted by [ = [n, such that
N[n = n[n (2.30)

and
N ( a
[n) = (n 1) ( a
[n) ,
N ( a
+
[n) = (n + 1) ( a
+
[n) .
Since

N[n = n[n and E
n
=
_
n +
1
2
_
h the state [n can be interpreted as containing n quanta each of energy h.
The state a
+
[n contains n + 1 quanta and so a
+
creates a quantum of energy and is referred to as a creation operator.
Likewise a
absorbs a quantum of energy and is referred to as the annihilation operator. The operator

N is called the
occupation number operator.
26
2.3.2 Eigenvalues of the Hamiltonian
Since

H = h
_

N +
1
2
_
then
H[ = h
_

N +
1
2
_
[ =
_
n +
1
2
_
h[ = E
n
[, (2.31)
so the energy eigenvalues are E
n
=
_
n +
1
2
_
h with n = 0, 1, 2, . . . Note that the lowest energy is E
0
=
1
2
h, the
so-called "zero-point" energy. In this approach it arises as a consequence of the non-commutability of the position and
momentim operators, and their assocated uncertainty xp
1
2
h.
Note also that the deternination of the energy eigenvalues has been done without solving the second-order differential
equation for the harmonic oscillator in the coordinate representation.
If [
E
is a state of energy E then
H[
E
= E[
E
and since
_
H, a
_
= h a
then
_
H a

H
_
= h a
and
H a
[
E
=
_
a

H h a
_
[
E
= ( a
E h a
) [
E
= (E h) a
[
E
.
Thus a
[
E
is an eigenstate of

H with eigenvalue E h, so a
+
raise the value of E by h and a
lowers E by h.
2.3.3 Actions of operators a
+
and a
Assume that the eigenstates [n of

N, are normalised so n[n = 1. Then
N[n = n[n;

N[n + 1 = (n + 1) [n + 1 (2.32)
But a
+
[n is also an eigenstate of

N with eigenvalue n + 1, hence
N ( a
+
[n) = (n + 1) ( a
+
[n) (2.33)
but
N[n + 1 = (n + 1) [n + 1 (2.34)
hence ( a
+
[n) must be some multiple of [n + 1, such as
a
+
[n = c
n
[n + 1. (2.35)
Normalisation gives, since a
+
= a
,
n[ a
a
+
[n = [c
n
[
2
n + 1[n + 1 = [c
n
[
2
(2.36)
But [ a
, a
+
] = 1, so a
a
+
= a
+
a
+ 1 and
n[ a
+
a
+ 1[n = [c
n
[
2
_
n[
N + 1[n
_
= [c
n
[
2
n + 1 = [c
n
[
2
(2.37)
so
c
n
=
n + 1
and
a
+
[n =
n + 1[n + 1. (2.38)
Similarly for a
where a
[n is an eigenstate of

N with eigenvalue n 1. Hence a
[n is a multiple of [n 1, as
a
[n = c
n
[n 1.
27
Then
a
+
a
+
[n = c
n
a
+
[n 1 = c
n
n[n
n[ a
+
a
+
[n = c
n
nn[n = c
n
n
n[
N[n = n = c
n
n
so c
n
=
n and
a
[n =
n[n 1. (2.39)
If [0 is the lowest energy eigenstate then
a
+
[0 = [1,
( a
+
)
2
[0 = a
+
[1 =
2[2,
( a
+
)
3
[0 = a
+
2[2 =
3[3,
( a
+
)
4
[0 = a
+
3[3 =
4[4
and by repeated action of a
+
( a
+
)
n
[0 =
n![n (2.40)
and hence the state
[n =
1
n!
( a
+
)
n
[0 (2.41)
generates all the eigenstates of

N.
2.3.4 Spatial wavefunctions
The eigenfunctions in the coordinate representation can be generated in a similar manner. The lowest eigenstate satises
a
[0 = 0,
or equivalently
_
1
2 hm
_
1/2
(m x + i p) [0 = 0.
In the coordinate representation this is
_
m x + i
_
ih
d
dx
__
0
(x) = 0 (2.42)
d
0
(x)
dx
+
_
m
h
_
x
0
(x) = 0. (2.43)
This equation has solution
0
(x) = Ae
(m/2 h)x
2
(2.44)
satisfying the requirements that
0
(x) 0 as x . Normalisation requires
_

[A[
2
e
(m/ h)x
2
dx = 1
and gives A = (m/h)
1/4
and
0
(x) =
_
m
h
_
1/4
e
(m/2 h)x
2
= [0. (2.45)
Then
1
(x) is obtained from
0
(x) using [1 = a
+
[0, i.e.
1
(x) =
_
1
2 hm
_
1/2
(m x i p)
0
(x) (2.46)
=
_
1
2 hm
_
1/2
_
m x h
d
dx
_
0
(x)
and similarly for
2
etc. Hence from [n =
1
n!
( a
+
)
n
[0 comes
n
(x) =
1
n!
_
_
1
2hm
_
1/2
_
m x h
d
dx
_
_
n
0
(x) . (2.47)
28
2.3.5 Matrix representations
Since a
+
[n =
n + 1[n + 1 then taking the scalar product with another state (ket) [k gives
k[ a
+
[n =
n + 1k[n + 1 =
n + 1
k,n+1
(2.48)
and from a
[n =
n[n 1 comes
k[ a
[n =
n
k,n1
. (2.49)
Thus the matrix representation of a
+
is
a
+
=
_
_
_
_
_
_
_
0 0 0 0
1 0 0 0
0
2 0 0
0 0
3 0
.
.
.
.
.
. 0
4
.
.
.
_
_
_
_
_
_
_
.
From the denitions of a
+
and a
,
a
+
=
_
1
2 hm
_
1/2
(m x i p) , (2.50)
a
=
_
1
2 hm
_
1/2
(m x + i p) . (2.51)
adding gives the position operator in terms of a
+
and a
as
x =
_
h
2m
_
1/2
( a
+
+ a
) =
1
2
( a
+
+ a
) , (2.52)
and by subtracting, the momentum operator is
p = i
_
mh
2
_
1/2
( a
+
a
) =
ih
2
( a
+
a
) . (2.53)
The matrix elements of x are
k[ x[n =
_
h
2m
_
1/2
_
n + 1
k,n+1
+
n
k,n1
_
(2.54)
for k, n = 1, 2, . . .. Note that there are no diagonal elements, whereas in the basis states [nthe Hamiltonian

H is
diagonal with
H =
_
_
_
_
_
_
0[
H[0
_ _
0[
H[1
_

_
1[
H[0
_ _
1[
H[1
_
.
.
.
.
.
.
.
.
.
_
_
_
_
_
= h
_
_
_
1
2
0
0
3
2

.
.
.
.
.
.
.
.
.
_
_
_. (2.55)
To obtain matrices for say x
2
there are several ways to proceed; by repeated action of x; by expanding x
2
in terms of a
+
,
and a
, i.e.
x[n =
_
h
2m
_
1/2
( a
+
+ a
) [n =
_
h
2m
_
1/2
_
n + 1[n + 1 +
n[n 1
_
,
x x[n =
_
h
2m
_
1/2
( a
+
+ a
)
_
h
2m
_
1/2
_
n + 1[n + 1 +
n[n 1
_
,
x
2
[n =
_
h
2m
_
_
a
+
_
n + 1[n + 1 +
n[n 1
_
+ a
_
n + 1[n + 1 +
n[n 1
_
,
=
_
h
2m
_
_
n + 1
n + 2[n + 2 +
n[n +
n + 1
n + 1[n +
n 1[n 2
,
x
2
[n =
_
h
2m
_
_
_
(n + 2) (n + 1)[n + 2 + (2n + 1) [n +
_
n(n 1)[n 2
_
, (2.56)
k[ x
2
[n =
_
h
2m
_
_
_
(n + 2) (n + 1)
k,n+2
+ (2n + 1)
kn
+
_
n(n 1)
k,n2
_
. (2.57)
29
or using [ a
, a
+
] = 1 = a
a
+
a
+
a
,
x =
_
h
2m
_
1/2
( a
+
+ a
) ,
x
2
=
_
h
2m
_
1/2
( a
+
+ a
)
_
h
2m
_
1/2
( a
+
+ a
) ,
=
_
h
2m
_
( a
+
+ a
) ( a
+
+ a
) =
_
h
2m
_
_
a
2
+
+ a
+
a
+ a
a
+
+ a
2
_
,
x
2
=
_
h
2m
_
_
a
2
+
+ a
+
a
+ a
+
a
+ 1 + a
2
_
=
_
h
2m
_
_
a
2
+
+ 2

N + 1 + a
2
_
, (2.58)
k[ x
2
[n =
_
h
2m
_
_
_
(n + 2) (n + 1)
k,n+2
+ (2n + 1)
kn
+
_
n(n 1)
k,n2
_
. (2.59)
2.3.6 Minimum uncertainty for the harmonic oscillator
The uncertainty in the x-position x is given by
(x)
2
=
_
( x x)
2
_
=
x
2
_
x
2
.
In terms of the raising and lowering operators a
+
, a
, the x operator is
x =
_
h
2m
_
1/2
( a
+
+ a
) (2.60)
so that
x[n =
_
h
2m
_
1/2
( a
+
+ a
) [n
=
_
h
2m
_
1/2
_
n + 1[n + 1 +
n[n 1
_
and hence
x = n[ x[n = 0. (2.61)
Since x is an Hermitian operator and a
+
= a
and a
= a
+
then
n[ x x[n =
_
h
2m
_
_
n + 1[
n + 1 +n 1[
n
_ _
n + 1[n + 1 +
n[n 1
_
=
_
h
2m
_
(n + 1) n + 1[n + 1 + nn 1[n 1 =
_
(2n + 1) h
2m
_
,
and
(x)
2
=
_
n +
1
2
_
h
m
. (2.62)
In terms of a
+
, a
, the momentum operator p is

p = i
_
mh
2
_
1/2
( a
+
a
) (2.63)
and hence
p[n = i
_
mh
2
_
1/2
( a
+
a
) [n
= i
_
mh
2
_
1/2
_
n + 1[n + 1
n[n 1
_
and hence
p = n[ p[n = 0. (2.64)
30
Also
p
2
_
= n[ p p[n =
_
mh
2
_
_
n + 1[
n + 1 n 1[
n
_ _
n + 1[n + 1
n[n 1
_
=
_
mh
2
_
(n + 1 + n) = mh
_
n +
1
2
_
,
and
(p)
2
=
_
n +
1
2
_
mh. (2.65)
Thus the product
(x)
2
(p)
2
=
_
n +
1
2
_
h
m
_
n +
1
2
_
mh
=
_
n +
1
2
_
2
h
2
(2.66)
and
(x) (p) =
_
n +
1
2
_
h. (2.67)
Thus in the lowest state n = 0, the product of uncertainties is (x) (p) =
1
2
h. The quantum oscillator satises the
lower bound of the generalised uncertainty relation.
2.3.7 Eigenstates of a
Suppose that [ is an eigenstate of a
with eigenvalue , then

a
[ = [. (2.68)
Expanding [ in terms of the complete set of states [n as
[ =
n
c
n
[n.
Then
a
[ =
n
c
n
a
[n =
n
c
n
n[n 1 =
n
c
n
[n.
Taking the scalar product with the state [k gives
k[ a
[ =
n
c
n
nk[n 1 =
n
c
n
k[n
=
n
c
n
n
k,n1
=
n
c
n
kn
. (2.69)
Thus for k = 0, 1, 2, . . . in turn,
c
1
= c
0
;
2c
2
= c
1
;
3c
3
= c
2
; . . . (2.70)
Hence c
1
= c
0
, c
2
=

2
2
c
0
; c
3
=

3
2
c
0
, etc. and in general
c
n
=

n
n!
c
0
(2.71)
so that
[ = c
0
n=0
n!
[n. (2.72)
Normalisation requires [ = 1 so
1 = [c
0
[
2
n=0
(
n
)
n!
m=0
m!
n[m
= [c
0
[
2
n=0
[[
2n
n!
= [c
0
[
2
e
||
2
.
31
Hence
[c
0
[ = e
||
2
/2
giving
[ = e
||
2
/2
n=0
n!
[n. (2.73)
The state [n contains n quanta each of energy h. Thus the probability that state [ contains k quanta is (= [c
k
[
2
)
P
k
= [k[[
2
=
e
||
2
/2
n=0
n!
k[n
2
P
k
=
e
||
2
/2

k
k!
=
_
[[
2
_
k
k!
e
||
2
. (2.74)
This is a Poisson distribution with a mean value [[
2
. The average number of quanta in state [ can be found from the
expectation value of the occupation number operator

N in the state [, i.e.
[
N[ = [ a
+
a
[ = [[
2
. (2.75)
Alternatively using the probability P
k
above, the average number of quanta is
n =
k=0
kP
k
=
k=0
k
_
[[
2
_
k
k!
e
||
2
=
k=1
e
||
2
_
[[
2
_
k
(k 1)!
=
k=1
[[
2
e
||
2
_
[[
2
_
k1
(k 1)!
= [[
2
e
||
2
k=0
_
[[
2
_
k
(k)!
= [[
2
e
||
2
e
||
2
= [[
2
.
2.4 Compatible observables and commuting operators
Two physical observables A, B are said to be compatible if a measurement of A, giving a value a, followed immediately
by a measurement of B, giving value b, leaves the system in a state where an immediate remeasurement of A or B gives
the same values a or b again. {Statement requires slight modication for a degenerate system.} The requirement that the
two measurements and remeasurement be carried out immediately ensures that between measurements the values of the
observables have not altered owing to the state of the system evolving according to the equation of motion.
Since a denite value a
n
can be assigned to a system only if it is in an eigenstate
n
of the operator

A, and likewise a
value b
n
for operator

B then
n
is an eigenstate of both

A and

B, i.e.

A
n
= a
n
n
and

B
n
= b
n
n
. The observables
A and B have simultaneous eigenfunctions. The two observables are compatible, otherwise they are incompatible. It
follows that
B

A
n
=

Ba
n
n
= b
n
a
n
n
= a
n
b
n
n
=

A

B
n
,
_
A

B

B

A
_
n
= 0. (2.76)
Since any function can be expanded in terms of the complete set
n
then
_
A

B

B

A
_
= 0 and the commutator
_
A,

B
_
= 0, (2.77)
i.e. the two operators commute with each other. Thus compatibility implies commuting operators.
The converse is true. If A and B are two commuting observables then a common set of eigenfunctions exits. If
n
is
an eigenfunction of

A with eigenvalue a
n
then

A
n
= a
n
n
. Thus

B

A
n
= a
n

B
n
But

B

A =

A

B so

A
_
B
n
_
=
a
n
_
B
n
_
and hence

B
n
is an eigenfunction of

A with eigenvalue a
n
. If a
n
is a non-degenerate value there is only one
eigenfunction having eigenvalue a
n
, hence

B
n
must be a multiple of
n
, i.e.

B
n
= b
n
n
and
n
is an eigenfunction
of

B with eigenvalue b
n
. Thus every eigenfunction of

A is an eigenfunction of

B.
{The argument requires modication if the eigenvalue a
n
is degenerate. Following a measurement of observable A
the system will have an wavefunction corresponding to one of the set of -degenerate eigenfunctions
ns
, s = 1, 2, . . . ,
for eigenvalue a
n
. In general this need not be an eigenfunction of observable B and the exact result of a measurement
of B is not predictable even though A and B commute. Nevertheless a subsequent measurement of B will result in a
32
wavefunction being equivalent to one of the set of common eigenfunctions so that the results of further measurements of
A or B are completely predictable. If a
n
is -fold degenerate,
nk
, k = 1, 2, . . . , are orthogonal eigenfunctions of
A with eigenvalue a
n
. Any linear combination
n
=
k=1
c
k
nk
is also an eigenfunction. If this is an eigenfunction of
B then

B
n
= b
n
n
and
k=1
c
k
nk
= b
n
k=1
c
k
nk
.
Taking the scalar product with
nm
, m = 1, 2, . . . , gives
k=1
c
k
_

nm

B
nk
d = b
n
k=1
c
k
_

nm
nk
d.
Writing
_

nm

B
nk
= B
mk
gives
k=1
(B
mk
b
n
mk
) c
k
= 0. (2.78)
This set of homogeneous linear equations in the unknowns c
1
, c
2
, . . . , c
possess a non-zero (non-trivial) solution if

det [B
mk
b
n
mk
[ = 0. This is a polynomial equation of degree in b
n
having roots. Corresponding to each
root b
(p)
n
where p = 1, 2, . . . , there is a solution for c
kp
giving the corresponding eigenfunction of

B which is also an
eigenfunction of

A. }
The theorem can be extended to more than two observables. If A
1
, A
2
,. . . A
n
are commuting observables whose
operators satisfy
_
A
r
,

A
s
_
= 0 r, s = 1, 2, . . . , n, (2.79)
they possess a complete set of simultaneous eigenfunctions (and vice-versa). This is a powerful result as it is often easier
to evaluate commutators, even in complicated situations, when there is no hope of obtaining explicit eigenfunctions.
To exploit this result some properties of commutators are be needed,
[A, B] = [B, A] (2.80)
[A, B + C] = [A, B] + [A, C] (2.81)
[A, BC] = [A, B] C + B[A, C] (2.82)
[AB, C] = A[B, C] + [A, C] B (2.83)
[A, [B, C]] + [B, [C, A]] + [C, [A, B]] = 0. (2.84)
The order of the factors must be preserved. From [A, BC] = [A, B] C + B[A, C] it follows that

A commutes with

A
2
,
and more generally with

A
n
and so if f
n
_
A
_
=
n
c
n

A
n
then
_
A, f
n
_
A
__
= 0. (2.85)
2.5 Heisenberg Uncertainty Relations
The most famous pair of non-commuting observables are position and momentum in the same dimension, i.e.
[ x, p
x
] = ih. (2.86)
Recall that in the coordinate representation p
x
= i

x
, so
[ x, p
x
] (x) = ihx
x
+ ih
(x)
x
= ih.
For each component
[ x, p
x
] = [ y, p
y
] = [ z, p
z
] = ih (2.87)
but different Cartesian components commute, e.g.
[ x, p
y
] = [ x, p
z
] = 0. (2.88)
The product of the uncertainties (x) (p
x
) forms the Heisenberg Uncertainty Principle
(x) (p
x
)
1
2
h. (2.89)
33
This is an example of a general principle that the product of the uncertainties of two observables is related to the expecta-
tion value of their commutator.
Suppose the state of the quantum system is . In this state measurement of the observable A results in obtaining one
of its eigenvalues of operator

A with probability [c
n
[
2
where =
n
c
n
n
. The average value is A and the observed
values are scattered about this value. The observable AA will be scattered about zero, so a measure of the uncertainty
is
(A)
2
=
_
_
A
_
A
__
2
_
, (2.90)
where
_
A
_
=
_

Ad. (This is just the standard deviation of A.) Similarly for another observable B,
(B)
2
=
_
_
B
_
B
__
2
_
. (2.91)
Note that
(A)
2
=
_
_
A
_
A
__
2
_
=
__
A
_
A
___
A
_
A
___
=
_
A
2
2

A
_
A
_
+
_
A
_
2
_
=
_
A
2
_
2
_
A
__
A
_
+
_
A
_
2
(A)
2
=
_
A
2
_
A
_
2
. (2.92)
Dene Hermitian operators =

A
_
A
_
,

=

B
_
B
_
and their linear combination (which is non-Hermitian)
O = + i
where is a real parameter. Clearly

_

2
d 0
and is real and non-negative. Then
_
_
O
_
O
_
d 0,
_
__
+ i
__
+ i
_
d 0.
From the Hermiticity of and

,
_

_
i
__
+ i
_
d 0.
But from denitions of and

,
_
i
__
+ i
_
=
2
+
2
2
+ i
_

_
=
2
+
2
2
+ i
_
,

_
.
Also _
,

_
=
_
A
_
A
_
,

B
_
B
__
=
_
A,

B
_
since
_
A
_
and
_
B
_
are just numbers. Hence
_

_

2
+
2
2
+ i
_
A,

B
__
d 0.
Thus the function
f () =

2
_
+
2
_
2
_
+ i
__
A,

B
__
0,
and implies that
__
A,

B
__
is purely imaginary. This function has a minimum value for
0
=
i
__
A,

B
__
2
_
2
_ .
34
The value of f () at this minimum is
f (
0
) =

2
_
__
A,

B
__
2
4
_
2
_ +
__
A,

B
__
2
2
_
2
_ =

2
_
+
__
A,

B
__
2
4
_
2
_ 0.
Thus

2
_
_
2
_

1
4
__
A,

B
__
2
.
But
__
A,

B
__
is purely imaginary, say
__
A,

B
__
= iC
where C is real. Then
_
_
A
_
A
__
2
__
_
B
_
B
__
2
_

1
4
(iC)
2
=
1
4
C
2
,
(A)
2
(B)
2
1
4
C
2
(A) (B)
1
2
C
i.e.
(A) (B)
1
2
__
A,

B
__
. (2.93)
This is the Generalised Uncertainty Principle.
For the position-momentum commutator [ x, p
x
] = ih so that (x) (p
x
)
1
2
h.
35
2.5.1 Uncertainty relation for the harmonic oscillator
The uncertainties are (x)
2
=
_
( x x)
2
_
=
x
2
_
x
2
and (p)
2
=
_
( p p)
2
_
. In the ground state the
eigenfunction is
0
(x) =
_

_
1/2
e
2
x
2
/2
(2.94)
with = (m/h)
1/2
. Then clearly
x =
_

__

xe
2
x
2
dx = 0
as the integrand is an odd function of x. Then
x
2
_
=
_

__

x
2
e
2
x
2
dx =
_

_
1
2
2
_

2
=
1
2
2
,
using the standard integral
_
x
2
exp
_
ax
2
_
dx =
1
2a
_
a
, hence
(x) =
1
2
. (2.95)
Note comparing the probability distribution P (x) = [
0
[
2
=

exp
_
2
x
2
_
with the normal (Gaussian) distribution
1
2
exp(x
2
/2
2
) the variance is
1
2
2
= or =
1
2
.
For the momentum operator,
p
x
= ih

2
x
2
/2
_
d
dx
_
_
e
2
x
2
/2
_
dx = ih

2
xe
2
x
2
dx = 0,
as integrand is an odd function of x. To evaluate
p
2
x
_
use integration by parts as
p
2
x
_
= h
2

2
x
2
/2
_
d
2
dx
2
_
_
e
2
x
2
/2
_
dx
=
h
2
_
_
e
2
x
2
/2
_
d
dx
_
_
e
2
x
2
/2
_
_
__
d
dx
_
_
e
2
x
2
/2
_
_
2
dx
_
=
h
2
2
xe
2
x
2
/2
_
2
dx =
h
2
4
x
2
e
2
x
2
dx =

5
h
2
1
2
2
_

2
=
h
2
2
2
.
and
(p
x
) =
h
2
. (2.96)
Two useful sandard integrals are
_

e
ax
2
dx =
_
a
;
_

x
2
e
ax
2
dx =
1
2a
_
a
. (2.97)
The product of the uncertainties
(x) (p
x
) =
h
2
. (2.98)
Thus the ground state of the harmonic oscillator minimises the uncertainty principle. In general for a harmonic oscillator
in energy state E
n
=
_
n +
1
2
_
h, the product of the uncertainties is
(x) (p
x
) =
_
n +
1
2
_
h. (2.99)
36
CHAPTER 3. GENERALISED ANGULAR MOMENTUM
Chapter 3
Generalised Angular Momentum
3.1 Orbital angular momentum
One set of non-commuting (and incompatible) operators which have already been met are the components of orbital
momentum. In general the components of orbital angular momentum cannot be assigned denite values simultaneously
(except for L
x
= L
y
= L
z
= 0). A review of some of the properties of orbital angular momentum is given before
moving on to its generalizations.
In classical mechanics the principle of conservation of angular momentum is a powerful tool in the solution of
rotational problems, e.g. motion of satellites, planets, gyroscopes, etc. The role of angular momentum in quantum
mechanics is probably even more important. This is due to the fact that the angular momentum properties of a system are
largely independent of the details of the forces acting. (The angular momentum properties are related to the rotational
symmetry properties.) For example, for a particle moving in a central potential V (r) a part of the wavefunction is fully
determined by its angular momentum properties, independent of the exact form of the potential. As a consequence, the
angular momentum and hence qualitative features of the energy level schemes and spectroscopy of many-electron atoms
can be understood without formally solving the Schrdinger equation.
Classically the orbital angular momentum L with respect to the origin of a particle at position r moving with momen-
tum p is
L = r p. (3.1)
This has Cartesian components
L
x
= yp
z
zp
y
(3.2)
L
y
= zp
x
xp
z
(3.3)
L
z
= xp
y
yp
x
. (3.4)
and square of the magnitude of the orbital angular momentum
L
2
= L
2
x
+ L
2
y
+ L
2
z
. (3.5)
Using the fundamental position-momentum commutator relations
[ x, p
x
] = [ y, p
y
] = [ z, p
z
] = ih
and commutator algebra leads to
_
L
x
,

L
y
_
= ih
L
z
;
_
L
y
,

L
z
_
= ih
L
x
;
_
L
z
,

L
x
_
= ih
L
y
. (3.6)
Proof:
_
L
x
,

L
y
_
= [ y p
z
z p
y
, z p
x
x p
z
]
= [ y p
z
, z p
x
] [ z p
y
, z p
x
] [ y p
z
, x p
z
] + [ z p
y
, x p
z
]
= y p
x
[ p
z
, z] 0 0 + p
y
x[ z, p
z
] = y p
x
(ih) + p
y
x(ih)
= ih( x p
y
y p
x
) = ih
L
z
.
Relations for the other components may be obtained by cyclic permutation of the x, y, z as there can be nothing special
about the choice of x, y, z. These commutation relations mean that two components of the angular momentum can not be
measure simultaneously. This is very different from classical mechanics. The commutator relations can be expressed as
L = ih
L. (3.7)
37
PHAS3226: Quantum Mechanics CHAPTER 3. GENERALISED ANGULAR MOMENTUM
The generalized uncertainty relation
(A) (B)
1
2
[
__
A,

B
__
[
gives
_
L
x
__
L
y
_

1
2
[
__
L
x
,

L
y
__
=
1
2
_
ih
L
z
_
L
x
__
L
y
_

1
2
h
_
L
z
_
. (3.8)
So in general all three components can not be known except for L
x
= L
y
= L
z
= 0.
The components of L may not commute among themselves but each commutes with

L
2
, as for example
_
L
x
,
L
2
_
=
_
L
x
,

L
2
x
_
+
_
L
x
,

L
2
y
_
+
_
L
x
,

L
2
z
_
= 0 +
_
L
x
,

L
y
_

L
y
+

L
y
_
L
x
,

L
y
_
+
_
L
x
,

L
z
_

L
z
+

L
z
_
L
x
,

L
z
_
= ih
L
z
L
y
+ ih
L
y
L
z
ih
L
y
L
z
ih
L
z
L
y
_
L
x
,
L
2
_
= 0. (3.9)
Similarly for the other components so that
_
L
x
,
L
2
_
=
_
L
y
,
L
2
_
=
_
L
z
,
L
2
_
= 0. (3.10)
Thus it is possible to construct simultaneous eigenfunctions of

L
2
and one component of L. This is by convention chosen
to be the z-component

L
z
(owing to its simple representation in spherical polar coordinates). Thus
L
2
= a, (3.11)
L
z
= b. (3.12)
Since L
2
= L
2
x
+ L
2
y
+ L
2
z
then
L
2
_
L
2
z
_
which implies that a b
2
.
The differential operators for the components of

L are obtained from

L =r p using p = ihas

L = ihr .
In Cartesian form they are
L
x
= ih
_
y

z
z

y
_
, (3.13)
L
y
= ih
_
z

x
x

z
_
, (3.14)
L
y
= ih
_
z

x
x

z
_
. (3.15)
However as orbital angular momentum is intimately related to rotations it is more convenient to express the operators in
spherical polar coordinates (r, , ) related to the Cartesian coordinates (x, y, z) by
x = r sin cos , (3.16)
y = r sin sin , (3.17)
z = r cos (3.18)
with 0 r , 0 , 0 2. A tedious bit of algebra relating the partial derivatives with respect to
(x, y, z) to those with respect to (r, , ) gives
L
x
= ih
_
sin

+ cot cos

_
, (3.19)
L
y
= ih
_
cos

+ cot sin

_
, (3.20)
L
z
= ih

, (3.21)
and
L
2
= h
2
_
1
sin
_
sin

_
+
1
sin
2
_
. (3.22)
38
{Note as the interest is only in

L
z
and

L
2
the following procedure can be used. In spherical polar coordinates
=r

r
+
r sin
then

L = ihr becomes
L = ih
_
sin
_
.
Noting that z = cos r sin
then L
z
= L z = ih

. For
L
2
= h
2
(r ) (r )
= h
2
r [(r )]
= h
2
r
_
sin
_
using the general relation a (c d) = c (d a). Evaluating the curl
r sin
_
=
1
r
2
sin
r r
r sin
0
r
sin
r sin

.
The interest is only in the r component so
L
2
= h
2
r
1
r
2
sin
_

_
r sin

_
r
sin
__
L
2
= h
2
_
1
sin
_
sin

_
+
1
sin
2
2
_
. (3.23)
3.2 Eigenvalues and eigenfunctions of

L
z
and

L
2
The operators

L
z
and

L
2
have simultaneous eigenfunctions (, ) satisfying
L
2
(, ) = a (, ) ,
L
2
z
(, ) = b (, ) .
In polar coordinates
ih
(, )
= b (, ) .
This clearly has solution (, ) = () () = () e
ib/ h
. It is usual to impose the condition that must be a
single-valued function. In this case that ( + 2) = (). This requires e
ib(+2)/ h
= e
ib/ h
i.e. e
i2b/ h
= e
i2m
and so b = mh where m = 0, 1, 2, . . ..
{The single-valued requirement looks reasonable in view of the probability interpretation of [ (, )[
2
. However this
would only seem to require that [ (, )[
2
is single valued. This would admit both integer and half-integer values of m.
However in the present context the half-integer values of m can be ruled out because they lead to solutions for () for
which the probability ux has a physically impossible behaviour (innite at angles = 0 and ). Half-integer values of
m do arise in connection with eigenstates used to describe "spin" of elementary particles.}
Normalization is achieved by
_
2
0
m
()
m
() d = 1
so that
m
() =
1
2
e
im
. (3.24)
The full wavefunction (, ) = () () is such that it satises
L
2
(, ) = h
2
_
1
sin
_
sin

_
+
1
sin
2
2
_
()
1
2
e
im
= a()
1
2
e
im
39
and carrying out the differentiation with respect to gives
1
sin
_
sin
d()
d
_
+
_
a
h
2

m
2
sin
2
_
() = 0. (3.25)
Putting x = cos ,
d
d
= sin
d
dx
=
_
1 x
2
_
1/2
d
dx
, P (cos ) = () gives
d
dx
_
_
1 x
2
_
dP
|m|
(x)
dx
_
+
_
a
h
2

m
2
(1 x
2
)
_
P
|m|
(x) = 0 (3.26)
which is the associated Legendre differential equation. Physically acceptable solutions which remain nite over the range
0 exist only if
a
h
2
= ( + 1)
where = 0, 1, 2, . . ., and m = , + 1, . . . 1, . That is is a positive integer or zero. The solutions are called
the associated Legendre polynomials denoted by P
|m|
(x). The functions () are normalized so that

_

0
m
()
m
() sin d =
and are given in terms of the Legendre polynomials P

|m|
(x) as
m
() = (1)
m
_
(2 + 1) ( m)!
2 ( + m)!
_
1/2
P
|m|
(x) , m 0, (3.27)
= (1)
|m|

|m|
() m < 0. (3.28)
The simultaneous eigenfunctions (, ) = () () are called the spherical harmonics Y
m
(, ) given by
Y
m
(, ) = (1)
m
_
(2 + 1) ( m)!
4 ( + m)!
_
1/2
P
|m|
(x) e
im
m 0, (3.29)
Y
,m
(, ) = (1)
m
Y

m
(, ) m < 0. (3.30)
The spherical harmonics satisfy the orthonormality relation
_
Y

m
(, ) Y
m
(, ) d =
_
2
0
d
_

0
sin Y

m
(, ) Y
m
(, ) d =
m
. (3.31)
The rst few spherical harmonics are
Y
00
=
1
4,
(3.32)
Y
1,1
=
_
3
8
sin e
im
, (3.33)
Y
10
=
_
3
4
cos =
_
3
4
_
z
r
_
, (3.34)
Y
11
+ Y
11
= 2
_
3
8
sin cos = 2
_
3
8
_
x
r
_
. (3.35)
The basic eigenvalue equations for

L
2
and

L
z
are
L
2
Y
m
= ( + 1) h
2
Y
m
, (3.36)
L
z
= mhY
m
, (3.37)
with = 0, 1, 2, . . ., and m = , +1, . . . 1, . Thus the ( + 1) h
2
eigenvalue is (2 + 1)-fold degenerate. The
quantities , and m
are called the orbital angular momentum quantum number and the magnetic quantum number respec-
tively. A particle is referred to to as being in a state with angular momentum - not as one of magnitude h
_
( + 1)!
For historical reasons the states with = 0, 1, 2, 3 are called s, p, d, and f states. The higher values of (= 4, 5, . . .) are
referred to in alphabetical order as g, h, etc. states.
40
3.3 Central potentials
The Hamiltonian for a particle moving in a potential V (r) is
H =
h
2
2m
2
+ V (r) .
For a central potential V depends only on the magnitude of r and is spherically symmetric giving
H =
h
2
2m
2
+ V (r) .
In polar coordinates
H =
h
2
2m
_
1
r
2
r
_
r
2

r
_
+
1
r
2
sin
_
sin

_
+
1
r
2
sin
2
2
_
+ V (r) , (3.38)
=
h
2
2m
_
1
r
2
r
_
r
2

r
_
L
2
h
2
r
2
_
+ V (r) ,
=
h
2
2m
1
r
2
r
_
r
2

r
_
+ V (r) +
L
2
2mr
2
. (3.39)
Since

L
2
and

L
z
operate only on the angular coordinates (, ), then
_
H,

L
2
_
=
_
H,

L
z
_
= 0, (3.40)
and hence there are simultaneous eigenfunctions of the total energy E, orbital angular momentum and one component,
Em
(r, , ) = R
E
(r) Y
m
(, ) . (3.41)
The energy level corresponding to the same value of are (2 + 1)-fold degenerate.
3.3.1 Review of hydrogen atom
The time-independent Schrdinger equation is
_
h
2
2m
e
e
2
4
0
r
_
(r, , ) = E (r, , ) .
In atomic units where e = m
e
= h = 1/4
0
= 1 so
H =
1
2
1
r
,
= f (r) +
L
2
2r
2
and thus
H =
_
1
2r
2
d
dr
_
r
2
d
dr
_
+
L
2
2r
2

1
r
_
= E.
In general since V (r) is a central potential
_
H,

L
2
_
=
_
H,

L
z
_
= 0
and hence there are simultaneous eigenfunctions of

H, the orbital angular momentum

L
2
and one component. The
wavefunction can be written as
Em
(r, , ) = R
E
(r) Y
m
(, )
and the energy eigenvalues
E
nm
E
n
=
1
2n
2
, n = 1, 2, 3, . . . .
The radial wavefunctions have the form
R
n
(r) = P
n
(r) e
r/n
where P
n
(r) is a polynomial of degree n 1. Note that 0 n 1, and m = , + 1, . . . 1, . For a
given there are 2 + 1 values of m all giving the same energy. Without a spin-orbit interaction all states of given n have
the same energy regardless of the value, i.e. are degenerate. The spin-orbit interaction removes some of this degeneracy.
The spectroscopic notation is
= 0 1 2 3 4 5
notation s p d f g h
.
41
3.4 Generalized angular momentum
An orbiting electron constitutes a current loop which by electromagnetic theory (Biot-Savart law for example) has asso-
ciated with it a magnetic moment
L
proportional to the electrons angular momentum L. Thus a hydrogen atom in its
ground state ( = 0) would not be expected to have a magnetic moment. However when a beam of such atoms is passed
through an inhomogeneous magnetic eld, the beam splits into two components (Stern-Gerlach type experiment). This
suggests that the H-atom has indeed a magnetic moment which can have two quantized orientations to the direction of
the magnetic eld. Furthermore this magnetic moment
S
is ascribed to an intrinsic property of the electron. Then by
analogy with orbital magnetic moment
L
one deduces that the electron possesses an intrinsic angular momentum s in
its own rest frame proportional to its magnetic moment
S
. This angular momentum is called spin. This spin has two
projections on the z-axis and this m =
1
2
or
1
2
as 2 + 1 = 2 gives =
1
2
. {The ne structure of spectral lines is also
explained in terms of the interaction between the magnetic moments associated with spin and the orbital angular momenta
of the electron, via the spin-orbit coupling s .}
Clearly spin is not an angular momentum of the familiar r p kind. It has no classical analogue. To discuss it a
generalized denition of angular momentum is needed. Also clearly the eigenfunctions associated with the spin variable
are not ordinary functions of position, x, and momentum, p, or time of the system. Thus one has to generalize the idea of
angular momentum in quantum mechanics. The Dirac notation offers a good way to do this.
Orbital angular momentum satises the commutator relations
_
L
x
,

L
y
_
= ih
L
z
and with cyclic permutations of the components. The will be the starting position for a general denition of angular mo-
mentum. As a denition J is an angular momentum if its components are Hermitian operators satisfying the commutation
relations
_
J
x
,

J
y
_
= ih

J
z
(3.42)
and with its cyclic permutations., i.e.

J
J = ih
J.
3.4.1 Raising and lowering operators
Dene two operators

J
+
and

J
by
J
+
=

J
x
+ i

J
y
, (3.43)
=

J
x
i

J
y
. (3.44)
These operators are not Hermitian (but they are self-adjoint, or Hermitian conjugates) nor do they represent any physically
measurable quantity, but they are very useful in the mathematical developments. Note that
_
[
J
+
_
=
_
[
_
=
_
[
. (3.45)
Commutator relations for

J
+
and

J
Clearly
_
J
2
,

J
_
= 0 (3.46)
as

J
2
commutes with

J
x
and

J
y
. The product
J
+

J
=
_
J
x
+ i

J
y
__
J
x
i

J
y
_
=

J
2
x
+

J
2
y
i
_
J
x

J
y

J
y

J
x
_
J
+

J
=

J
2

J
2
z
+ h

J
z
(3.47)
Similarly

J
+
=

J
2

J
2
z
h

J
z
. (3.48)
Subtracting gives
_
J
+

J

J
+
_
=
_
J
+
,

J
_
= 2 h

J
z
.
and in
_
,

J
_
= 2h

J
z
. (3.49)
42
Finally
_
J
z
,

J
_
=
_
J
z
,

J
x
i

J
y
_
=
_
J
z
,

J
x
_
i
_
J
z
,

J
y
_
= ih

J
y
i
_
ih

J
x
_
= ih

J
y
h

J
x
= h
_
J
x
i

J
y
_
_
J
z
,

J
_
= h

J
. (3.50)
3.5 General angular momentum eigenvalue problem
Let the eigenvalues of

J
2
and

J
z
be and respectively and the corresponding simultaneous eigenfunction be . Thus
J
2
[ = [, (3.51)
J
z
[ = [. (3.52)
The eigenvalues of

J
2
and

J
z
are real as

J
2
and

J
z
are Hermitian operators. Moreover the eigenvalues of

J
2
must be
positive or zero because the expectation value of the square of an Hermitian operator must be positive or zero.
{Note
_
J
2
x

_
0. Choose [ =

J
x
[ then
_
J
x

_
=
_
J
x
= [
= [ 0, as

J
x
is
Hermitian. Therefore
_
J
2
x
_
,
_
J
2
y
_
,
_
J
2
z
_
are all 0. But
_
J
2
_
=
_
J
2
x
_
+
_
J
2
y
_
+
_
J
2
z
_
0 and so
_
J
2
_

_
J
2
z
_
and hence
2
.
Operating with

J
+
J
+

J
z
[ =

J
+
[.
Using
_
J
z
,

J
+
_
= h

J
+
i.e.

J
z

J
+

J
+

J
z
= h

J
+
_
J
z

J
+
h

J
+
_
[ =

J
+
[
J
z
_
J
+
[
_
= ( + h)
_
J
+
[
_
. (3.53)
Similarly operating with

J
gives
J
z
_
[
_
= ( h)
_
[
_
. (3.54)
Thus
_
J
+
[
_

J
z
with eigenvalue ( + h) and
_
[
_

J
z
with eigenvalue
( h). Thus

J
+
and

J
raise or lower respectively the eigenvalues of

J
z
(They are called step-up, step-down or raising,
lowering, or ladder operators.)
Also from

J
2
[ = [ and using
_
J
2
,

J
_
= 0

J
2
[ =

J
J
2
_
[
_
=
_
[
_
(3.55)
so

J
[ are also eigenfunctions of

J
2
with eigenvalue .
Hence by repeatedly operating with

J
+
a sequence of eigenstates of

J
z
can be constructed with eigenvalues + h,
+ 2h, . . . + ph, each eigenstate being an eigenfunction of

J
2
with eigenvalue . Similarly repeated application of
produces a sequence of eigenfunctions of

J
z
with eigenvalues h, 2h, . . . qh. Since ( + ph)
2
, and
( qh)
2
these sequences must terminate. There will be a maximum value of + ph, say m
T
with eigenfunction
[
T
such that
J
+
[
T
= 0 (3.56)
and
J
z
[
T
= m
T
[
T
. (3.57)
Similarly there is a minimum value of qh = m
B
with eigenfunction [
B
such that
[
B
= 0, (3.58)
J
z
[
B
= m
B
[
T
(3.59)
Furthermore as

J
+
and

J
raise and lower the eigenvalues in steps of h, then

m
T
m
B
= nh (3.60)
43
where n = 0, 1, 2, . . ..
Applying

J
to

J
+
[
T
= 0 then

J

J
+
[
T
= 0. But

J

J
+
=

J
2

J
2
z
h

J
z
so
_
J
2

J
2
z
h

J
z
_
[
T
= 0. (3.61)
Thus evaluating using equ(3.51, 3.59)
_
m
2
T
m
T
h
_
[
T
= 0
and
= m
T
(m
T
+ h) (3.62)
Similarly from

J
+

J
[
B
= 0, but

J
+

J
=

J
2

J
2
z
+ h

J
z
and so
_
m
2
B
+ m
B
h
_
[
T
= 0
and
= m
B
(m
B
h) (3.63)
Equations (3.62, 3.63) give
m
T
(m
T
+ h) = m
B
(m
B
h) ,
m
2
T
+ (m
T
+ m
B
) h m
2
B
= 0.
This equation has two solutions
m
T
= m
B
,
m
T
= m
B
h.
The second one violates m
T
m
B
thus the only valid solution is m
T
= m
B
and so m
T
0. But m
T
m
B
= nh so
m
T
= n
h
2
, n = 0, 1, 2, . . . . (3.64)
Putting n/2 = j means m
T
= m
B
= jh and
= j (j + 1) h
2
(3.65)
with j = 0,
1
2
, 1,
3
2
, . . . , i.e. j is an integer or half-integer and
= mh
with m ranging between j and j in integer steps., i.e. m has 2j + 1 values for a given j value.
The general denition of angular momentum based on the commutator relations has led to a positive integer (or zero)
and half-integer values for j (and for m). The half odd-integer values for j are excluded for orbital angular momentum for
which j = 0, 1, 2, . . .. There is overwhelming evidence that electrons, protons, neutrons and many more particles possess
an intrinsic angular momentum, called spin, of
1
2
. For pions s = 0, for photons s = 1, for the
particle s =
3
2
. Note
properties of electron spin can be derived from Diracs relativistic formulation of quantum mechanics.
3.5.1 Actions of the

J
+
and

J
The state [j, m is an eigenstate of

J
z
with eigenvalue mh so
J
z
[j, m = mh[j, m
so
J
+

J
z
[j, m = mh

J
+
[j, m
_
J
z

J
+
h

J
+
_
[j, m = mh
_
J
+
[j, m
_
J
z
_
J
+
[j, m
_
= (m + 1) h
_
J
+
[j, m
_
, (3.66)
so the state
_
J
+
[j, m
_
is an eigenstate of

J
z
with eigenvalue (m + 1) h. But also
J
z
[j, m + 1 = (m + 1) h[j, m + 1 (3.67)
44
and so the state

J
+
[j, m must be a multiple of state [j, m + 1. Hence
J
+
[j, m = C
+
(j, m) [j, m + 1. (3.68)
Similarly for the lowering operator
[j, m = C
(j, m) [j, m1. (3.69)

Applying

J
to equ(3.68) and taking the scalar product with state [j, m,

j, m

J
+
[j, m = C
+
(j, m) j, m[
[j, m + 1. (3.70)
But

J
+
and

J
are Hermitian adjoint operators such that

[
J
+
=
[ = [
and so
j, m[
[j, m + 1 = j, m + 1[
J
+
[j, m
= j, m + 1[C
+
(j, m) [j, m + 1
= C
+
(j, m) .
Equ(3.70) reads
j, m[

J
+
[j, m = [C
+
(j, m)[
2
. (3.71)
Alternatively noting that since

J
+
and

J
are Hermitian adjoints of each other then as
J
+
[j, m = C
+
(j, m) [j, m + 1
then
j, m[
+

J
+
[j, m = j, m + 1[C
+
(j, m) C
+
(j, m) [j, m + 1
j, m[

J
+
[j, m = [C
+
(j, m)[
2
But

J

J
+
=

J
2

J
2
z
h

J
z
so
j, m[
J
2

J
2
z
h

J
z
[j, m = [C
+
(j, m)[
2
_
j (j + 1) m
2
m
h
2
= [C
+
(j, m)[
2
and
C
+
(j, m) = [j (j + 1) m(m + 1)]
1/2
h,
the usual convention being to take C
+
(j, m) as positive so that
J
+
[j, m = [j (j + 1) m(m + 1)]
1/2
h[j, m + 1, (3.72)
and similarly
[j, m = [j (j + 1) m(m1)]
1/2
h[j, m1. (3.73)
45
46
CHAPTER 4. SPIN 1/2 SYSTEMS
Chapter 4
Spin 1/2 Systems
The theory of the last section provides the formalism for a discussion of the spin angular momentum of a particle. The
usual convention is that s denotes the spin angular momentum number,

S the spin angular momentum operator. The
corresponding quantities for orbital angular momentum are and

L. The discussion will be conned to spin-
1
2
particles -
an important case as electrons and protons have spin-
1
2
. For such particles the only possible values of m (denoted now by
m
s
to distinguish it from the orbital magnetic quantum number) are
1
2
h and
1
2
h. These are often referred to as "spin-up"
and "spin-down" states (with respect to quantization axis). Since [s[ =
_
s (s + 1)h =
_
3
4
h and s
z
=
1
2
h the classical
vector for s makes an angle of 54.7
to the z-axis!
The eigenstates of S
z
are denoted by [
1
2
,
1
2
[ for m
s
=
1
2
h, and [
1
2
,
1
2
[ for m
s
=
1
2
h. Then
S
2
[ = s (s + 1) h
2
[ =
3
4
h
2
[, (4.1)
S
2
[ =
3
4
h
2
[, (4.2)
S
z
[ =
1
2
h[, (4.3)
S
z
[ =
1
2
h[. (4.4)
Since [ and [ belong to different eigenvalues of S
z
they are orthogonal, [ = [ = 0 and are assumed to be
normalized, [ = [ = 1.
The angular momentum commutation relations are
_
S
x
,

S
y
_
= ih
S
z
(4.5)
with cyclic permutations. The action of the raising and lowering operators

S
+
=

S
x
+ i
S
y
and

S
=

S
x
i
S
y
are, in
general,
[s, m
s
= [s (s + 1) m
s
(m
s
1)]
1/2
h[s, m
s
1, (4.6)
[
1
2
, m
s
=
_
3
4
m
s
(m
s
1)
_
1/2
h[
1
2
, m
s
1. (4.7)
In particular
S
+
[ = 0;

S
+
[ = h[, (4.8)
[ = h[;

S
[ = 0. (4.9)
Since
S
x
=
1
2
_
S
+
+

S
_
;

S
y
=
1
2i
_
S
+
_
(4.10)
this leads to the set
S
x
[ =
h
2
[

S
x
[ =
h
2
[
S
y
[ = i
h
2
[

S
y
[ = i
h
2
[
S
z
[ =
h
2
[

S
z
[ =
h
2
[
. (4.11)
An arbitrary spin-
1
2
state [ can be written as a linear superposition of states [ and [ as
[ = a[ + b[. (4.12)
47
PHAS3226: Quantum Mechanics CHAPTER 4. SPIN 1/2 SYSTEMS
Taking the scalar product with [ and [ in turn gives
a = [; b = [,
so
[ = [[ +[[ (4.13)
which is just an example of the general expansion in eigenstates. If [ is normalized as [ = 1 then
1 = [[ +[[
1 = [a[
2
+[b[
2
. (4.14)
The expectation value of

S
z
in the state [ from
S
z
[ =

S
z
[[ +

S
z
[[
=
h
2
[[
h
2
[[
[
S
z
[ =
h
2
[[[
2
h
2
[[[
2
h
2
[a[
2
h
2
[b[
2
, (4.15)
as would be expected since [a[
2
is the probability of nding the particle in state [ with spin-up and [b[
2
the probability
of spin-down.
4.1 Useful relations for spin
The spin-
1
2
operators have some other useful properties. One is that

S
2
is a purely numerical operator since
S
2
[ = s (s + 1) h
2
[ =
3
4
h
2
[ (4.16)
for any state [, so
S
2
3
4
h
2
.
For spin-
1
2
only:
S
2
x
[ =
h
2
S
x
[ =
h
2
4
[,
S
2
x
[ =
h
2
S
x
[ =
h
2
4
[
so
S
2
x
[ =
h
2
4
[. (4.17)
Similar results hold for

S
2
y
and

S
2
z
, hence
S
2
x
=

S
2
y
=

S
2
z
=
h
2
4
. (4.18)
Since
S
+
[ = 0;

S
+
[ = h[,
S
2
+
[ = 0;

S
2
+
[ = 0
then for an arbitrary state [
S
2
+
[ = 0. (4.19)
Similarly for

S
S
2
[ = 0 (4.20)
and thus
0 =
_
S
x
i
S
y
__
S
x
i
S
y
_
,
0 =

S
2
x

S
2
y
i
_
S
x

S
y
+

S
y

S
x
_
. (4.21)
48
It follows that
S
x

S
y
+

S
y

S
x
= 0 (4.22)
and that

S
x
and

S
y
are said to anticommute,
_
S
x
,

S
y
_
=

S
x

S
y
+

S
y

S = 0. (4.23)
Similar anticommutation relations exist between the other components so,
_
S
x
,

S
y
_
=
_
S
y
,

S
z
_
=
_
S
z
,

S
x
_
= 0. (4.24)
Combining the anticommutation relations above with the commutation relations, e.g.
S
x

S
y
+

S
y

S
x
= 0
and
S
x

S
y

S
y

S
x
= ih
S
z
gives
S
x

S
y
= i
h
2
S
z
. (4.25)
A similar relation holds for a cyclic permutation of the indices x, y, z so that in general
S
x

S
y
= i
h
2
S
z
(4.26)
with cyclic permutation of (x, y, z).
This relation is useful because when combined with

S
2
x
=

S
2
y
=

S
2
z
=
h
2
4
it shows that any arbitrary product of spin-
1
2
operators can be reduced to a spin independent term or a term linear in

S
x
,

S
y
,

S
z
, i.e. the most general spin-
1
2
operator
is
A =

A
0
+

B
S =

A
0
+

B
x

S
x
+

B
y

S
y
+

B
z

S
z
. (4.27)
4.2 Representation of spin 1/2 operators and eigenfunctions - Pauli matrices
So far the discussion of spin has made use solely of the spin operators and their eigenstates [ and [. For some
calculations it is useful to have an explicit representation for the operators and eigenfunctions. It should be obvious from
the earlier discussion that they cannot be represented as functions of the particles position and momentum as in the case
of orbital angular momentum. They can, however, be very conveniently handled by representing the spin operators by
2 2 matrices and the spin states by column vectors.
4.2.1 Matrix representations
Since

S
2
[s, m
s
= s (s + 1) h
2
[s, m
s
then s
, m
s
[
S
2
[s, m
s
= s (s + 1) h
2
s
m
s
and the matrix representation of
S
2
is diagonal with
S
2
=
3
4
h
2
_
1 0
0 1
_
. (4.28)
Since

S
z
[s, m
s
= m
s
h[s, m
s
then s, m
s
[
S
z
[s, m
s
= m
s
h
m
s
m
s
and
S
z
=
1
2
h
_
1 0
0 1
_
=
1
2
h
z
(4.29)
where
z
is a Pauli matrix.
The matrix representations of

S
x
and

S
y
can be found using
S
x
=
1
2
_
S
+
+

S
_
;

S
y
=
1
2i
_
S
+
_
and
[s, m
s
= [s (s + 1) m
s
(m
s
1)]
1/2
h[s, m
s
1.
Hence
S
x
[s, m
s
=
1
2
[s (s + 1) m
s
(m
s
+ 1)]
1/2
h[s, m
s
+ 1 + [s (s + 1) m
s
(m
s
1)]
1/2
h[s, m
s
1. (4.30)
49
Explicity

S
x
[
1
2
,
1
2
=
h
2
[
1
2
,
1
2
and

S
x
[
1
2
,
1
2
=
h
2
[
1
2
,
1
2
and so
S
x

_

1
2
,
1
2
[
S
x
[
1
2
,
1
2

1
2
,
1
2
[
S
x
[
1
2
,
1
2
1
2
,
1
2
[
S
x
[
1
2
,
1
2

1
2
,
1
2
[
S
x
[
1
2
,
1
2
_
=
_
[
S
x
[ [
S
x
[
[
S
x
[ [
S
x
[
_
, (4.31)
S
x
=
h
2
_
0 1
1 0
_
=
h
2
x
. (4.32)
By similar steps
S
y
=
h
2
_
0 i
i 0
_
=
h
2
y
. (4.33)
The three Pauli matrices
x
=
_
0 1
1 0
_
;
y
=
_
0 i
i 0
_
;
z
=
_
1 0
0 1
_
; (4.34)
and the identity
I =
_
1 0
0 1
_
(4.35)
form a complete set in terms of which any 2 2 matrix may be expanded, possibly with complex coefcients.
4.2.2 Commutators for the Pauli matrices
Since Pauli matrices are dened by
S =
h
2
(4.36)
and
_
S
x
,

S
y
_
= ih
S
z
then
[
x
,
y
] = 2i
z
(4.37)
together with the cyclic permutations of the indices.
Also since

S
x

S
y
= i
h
2
S
z
then

x

y
= i
z
. (4.38)
Note also that the traces of all the Pauli matrices vanish,
Tr
x
= Tr
y
= Tr
z
= 0 (4.39)
and
det
x
= det
y
= det
z
= 1. (4.40)
4.2.3 Basis states
The eigenfunctions [ and [ are represented by the column vectors
[ =
_
1
0
_
; [ =
_
0
1
_
. (4.41)
Thus the operator equation
S
z
[ =
1
2
h[
becomes
1
2
h
_
1 0
0 1
__
1
0
_
=
1
2
h
_
1
0
_
. (4.42)
The arbitrary spin state [ = a[ + b[ becomes
[ = a
_
1
0
_
+ b
_
0
1
_
=
_
a
b
_
. (4.43)
If [ = c[ + d[ then
[ = c
a + d
b. (4.44)
50
This expression can be obtained if the bra [ is represented by a two-component row vector
[ = (c
, d
) (4.45)
and so
[ = (c
, d
)
_
a
b
_
= c
a + d
b
by the rules of matrix multiplication.
As an example consider the expectation value of

S
z
in the state [;
[
S
s
[ =
h
2
(a
, b
)
_
1 0
0 1
__
a
b
_
=
h
2
(a
a b
b) =
h
2
_
[a[
2
[b[
2
_
(4.46)
as before.
4.2.4 Determination of eigenvalues and eigenvectors
To solve, say,

S
x
[ = a[ in the matrix representation. Write a =
h
2
and

S
x
=
h
2

x
and express
[ =
_
u
v
_
, (4.47)
so that
h
2
_
0 1
1 0
__
u
v
_
=
h
2
_
u
v
_
,
then
_
1
1
__
u
v
_
= 0. (4.48)
This equation has a non-trivial solution if
det
1
1
= 0 (4.49)
i.e.
2
1 = 0
and hence = 1.
For = +1,
h
2
_
0 1
1 0
__
u
v
_
=
h
2
_
u
v
_
gives
v = u
u = v.
Normalization of the eigenvector [ as 1 = [ = [u[
2
+ [v[
2
gives u = v =
_
1
2
. The eigenvalue is thus
h
2
and the
eigenvector is
[
+
=
1
2
_
1
1
_
=
1
2
([ +[) . (4.50)
For = 1 one gets u = v and eigenvector
[
=
1
2
_
1
1
_
=
1
2
([ [) . (4.51)
In a similar way the eigenvalues and eigenvectors of

S
y
and

S
z
can be found. These are summarized below:
eigenvalue eigenstate matrix form
S
x
1
2
h [
+
= [+
x
=
1
2
([ +[)
1
2
_
1
1
_
1
2
h [
= [
x
=
1
2
([ [)
1
2
_
1
1
_
S
y
1
2
h [+
y
=
1
2
([ + i[)
1
2
_
1
i
_
1
2
h [
y
=
1
2
([ i[)
1
2
_
1
i
_
S
z
1
2
h [+
z
= [
_
1
0
_
1
2
h [
z
= [
_
0
1
_
(4.52)
51
The eigenvectors can also be found via the basis states [ and [ Express [ = a[ + b[ then
S
x
[ = a[ =
1
2
h[
1
2
_
S
+
+

S
_
(a[ + b[) =
1
2
h[
b
h
2
[ + a
h
2
[ =
1
2
h(a[ + b[) (4.53)
Taking the scalar product with [ and [, or noting that [ and [ are linearly independent so that coefcients on each
side may be equated, gives
b
h
2
=
1
2
ha
a
h
2
=
1
2
hb.
Thus
2
= 1, a
2
= b
2
, = 1. For = +1, a = b and normalization gives a = b =
1
2
as before.
4.2.5 Spin along an arbitrary direction
The component of spin S along a direction n is

S
n
=

S n. Taking the direction
n =(cos sin , sin sin , cos )
then
S
n
=
h
2
(
x
cos sin +
y
sin sin +
z
cos )
=
h
2
__
0 1
1 0
_
cos sin +
_
0 i
i 0
_
sin sin +
_
1 0
0 1
_
cos
_
=
h
2
_
cos cos sin i sin sin
cos sin + i sin sin cos
_
S
n
=
h
2
_
cos sin e
i
sin e
i
cos
_
. (4.54)
The eigenvalue equation is
S
n
[ =
h
2
[
with [ = a[ + b[, or in matrix form
_
cos sin e
i
sin e
i
cos
__
a
b
_
=
_
a
b
_
. (4.55)
Hence for a non-trivial solution

cos sin e
i
sin e
i
cos
= 0
(cos ) (cos + ) sin
2
= 0
2
= 1,
= 1. (4.56)
Thus the eigenvalues are h/2 for a spin-
1
2
system regardless of the axis chosen. To nd the eigenvectors, for = +1
a cos + b sin e
i
= a
b2 sin

2
cos

2
e
i
= a (1 cos ) = 2a sin
2

2
so
b
a
=
sin

2
e
i
cos

2
. (4.57)
Choose for = +1,
[+
n
=
_
cos

2
sin

2
e
i
_
,
52
and for = 1,
[
n
=
_
sin

2
cos

2
e
i
_
; or [
n
=
_
sin

2
cos

2
e
i
_
. (4.58)
In state [+
n
the probability of "spin-up" (along + z) is P
+
= cos
2
2
; probability of "spin-down" (along - z) is P
=
sin
2
2
.
4.3 Space-spin wavefunctions
The wavefunction of a spin-
1
2
particle also has a spatial dependence. The complete wavefunction is
=
m
s
=1/2
m
s
=1/2
c
m
s
m
s
(r)
m
s
, (4.59)
which for a spin-
1
2
particle can be written a
(r, s) = c1
2
1
2
(r) [ + c
1
2
1
2
(r) [, (4.60)
(r, s) =
_
c1
2
1
2
(r)
c
1
2
1
2
(r)
_
. (4.61)
Thus a spin-
1
2
particle is described by a two-component wavefunction (a spinor). The probability that the particle will be
found in a volume element d with "spin up"
c1
2
1
2
(r)
2
, and the probability that it is found anywhere with "spin-up" is
c1
2
2
provided 1
2
(r) is normalized, as
_

1
2
(r)
2
d = 1. The expectation value
_
[
S
z
[
_
=
h
2
_
c1
2
1
2
2
_
(4.62)
Suppose the electron has spin fully aligned along the z-axis, i.e. its spin state is [ and it is moving along the x-axis in
a particle free region. Then

H =
h
2
2m
d
2
dx
2
and the wavefunction is
(x) = Ae
ikx
[ = Ae
ikx
_
1
0
_
=
_

+
(x)
(x)
_
(4.63)
with
+
(x) = Ae
ikx
and
(x) = 0.
4.3.1 Stern-Gerlach experiment
This is one of the famous experiments that demonstrated the existence of electron spin (1922, Nobel prize 1943). Orig-
inally performed with Ag atoms which have an unpaired electron in the outer shell. The beam of Ag atoms is directed
along the x-axis and passes between the poles of a magnet with a very non-uniform eld along the z-axis. The potential
energy is
V = B =
_
g
s
B
h
s B
_
= g
s
B
h
s B. (4.64)
and the force is
F
z
=
dV
dz
= g
s
B
h
s
dB
dz
k =g
s
B
h
S
z
dB
dz
. (4.65)
Depending on the sign of S
z
(+h/2 or h/2) the force is upwards or downwards. Hence two "spots" are seen on a screen.
Classically s would be orientated at random, so all points on the screen between the two spot limits would be lled in.
Suppose that the atoms are initially polarized with m
s
= + h/2, i.e.
z
= +1. The incoming wavefunction is
in
(r, t) = f (r vt) e
ikx
[ (4.66)
where f is an "envelope" function describing the shape of the wave packet - the spatial part of the wave packet at some
time t. Particles are located within the envelope whose maximum advances at speed v = hk/m. When the atoms emerge
from the magnet they will have been deected by an amount r = z
k and the outgoing wavefunction is
out
(r, t) = f
_
r vt + z
k
_
e
ikx
[. (4.67)
53
Similarly the "spin-down" atoms have
z
= 1 and
in
(r, t) = f (r vt) e
ikx
[, (4.68)
out
(r, t) = f
_
r vt z
k
_
e
ikx
[. (4.69)
Suppose that initially the atoms were fully polarized along the +x-axis, so then
[+
x
=
1
2
([ +[) .
Then before the magnet
_

in
(r, t)
in
(r, t)
_
=
1
2
e
ikx
_
f (r vt)
f (r vt)
_
. (4.70)
After the magnet induce deections
_

out
(r, t)
out
(r, t)
_
=
1
2
e
ikx
_
_
f
_
r vt + z
k
_
f
_
r vt z
k
_
_
_
. (4.71)
The intensity of each beam will be equal. If the lower beam (state [) is taken through a second magnet with B = B
x
(x)
the atoms will be found with a 50:50 mix along the +x-axis and the x-axis since the lower z-beam was in a state
[ =
1
2
([+
x
+[
x
).
4.4 Addition of angular momentum
For an atom in free space, its total angular momentum is conserved (i.e. it commutes with the Hamiltonian). This means
that the stationary states (energy eigenstates) of the atom can be chosen so that the square of total angular momentum and
the z-component of angular momentum are precisely dened. But the total angular momentum of an atom is generally
the sum of a number of individual angular momenta. Each electron in the atom has an orbital angular momentum as well
as an intrinsic spin angular momentum. The vector sum of all these angular momenta is the total angular momentum of
the electrons. In addition, the atomic nucleus generally has an intrinsic angular momentum, which must also be included
to obtain the total angular momentum of the atom. So in order to understand atoms, we need to know how to add angular
momenta. The question is: if two (quantised) angular momenta are add them together, what are the possible values of
the j and m quantum numbers of the resulting total angular momentum, and how are the eigenvectors of total angular
momentum expressed in terms of the eigenvectors of the individual angular momenta? The general theory of addition
of angular momenta is rather complicated, and is presented in the 4th-year quantum mechanics course. Here, only two
important special cases are studied: the addition of two spin-
1
2
angular momenta, and the addition of a spin-
1
2
angular
momentum and an orbital angular momentum. The rst case is relevant to the addition of the spin of two electrons in
an atom (e.g. the He atom), and the second case is relevant to the addition of the orbital and spin angular momenta of an
electron.
4.4.1 Addition of two spin-1/2 angular momenta
The simplest, but important example, is the addition of the spins of two spin-
1
2
particles, e.g. two electrons. The vector
operators representing the spin angular momenta of these two particles are denoted by

S
1
and

S
2
. Then the total spin
angular momentum

S is the vector sum of

S
1
and

S
2
:
S =

S
1
+

S
2
. (4.72)
Since

S is an angular momentum, it is quantized: its square can only have values S(S + 1)h
2
, with S an integer or a
half-odd-integer; its z-component

S
z
can only have the values m
s
h, with m
s
= S, S + 1, . . . S 1, S.
Let the two eigenvectors of s
1z
be called
1
and
1
then
S
1z
1
=
1
2
h
1
,

S
1z
1
=
1
2
h
1
. (4.73)
These are also eigenstates of

S
2
1
:
S
2
1
1
=
3
4
h
2
1
,

S
2
1
1
=
3
4
h
2
1
. (4.74)
54
Similarly, for particle 2, denote the eigenvectors of

S
2z
by
2
,
2
:
S
2z
2
=
1
2
h
2
,

S
2z
2
=
1
2
h
2
S
2
2
2
=
3
4
h
2
2
,

S
2
2
2
=
3
4
h
2
2
. (4.75)
Note that the state vectors
1
and
1
on the one hand, and
2
and
2
on the other hand, inhabit completely different
spaces. Linear combinations of
1
and
1
, e.g.
1
2
(
1
+
1
), can be constructed, describing different quantum states of
particle 1. Similarly, linear combinations of
2
and
2
describe states of particle 2. But
1
+
2
has no meaning.
All possible states of the system of two spin-
1
2
particles can be represented as linear combinations of the following
four basic states:
2
,
1
2
,
1
2
,
1
2
. (4.76)
These are eigenstates of both

S
1z
and

S
2z
, for example,
S
1z
2
=
1
2
h
1
2
,

S
2z
2
=
1
2
h
1
2
, (4.77)
so that in the state
1
2
, S
1z
has the precisely dened value
1
2
h, and S
2z
also has the precisely dened value
1
2
h. Similarly
in the state
1
2
, S
1z
and S
2z
have the values
1
2
h and
1
2
h respectively; in state
1
2
they have values
1
2
h and
1
2
h; and
in state
1
2
, they have values
1
2
h and
1
2
h. (Note that
1
2
does not mean
1
multiplied by
2
in the normal sense
of the word multiplied. It is just a notation for the state of the system in which S
1z
and S
2z
have the values
1
2
h and
1
2
h.)
The total spin is
S =

S
1
+

S
2
(4.78)
and
S
z
=

S
1z
+

S
2z
. (4.79)
Denote the general eigenstate by [S, M. There are four possible product combinations
1
2
,
1
2
,
1
2
and
1
2
which are simultaneous eigenvectors of

S
z
with eigenvalues (M = m
s
1
+ m
s
2
) of 1, 0, 0, 1 respectively. The state
2
with M = 1 must belong to the state S = 1 as
S
z
[1, 1 =
_
S
1z
+

S
2z
_
2
=
h
2
2
+
h
2
2
= h
1
2
. (4.80)
There are two M = 0 states,
1
2
,
1
2
. From these can be formed two linearly independent combinations. One
corresponds to the state [1, 0 and the other to state [0, 0. These can be found by using the lowering operators

S
S
1
+ S
2
and
[1, 1 =
_
S
1
+ S
2
_
2
= h
1
2
+ h
1
2
= h(
1
2
+
1
2
) . (4.81)
But in general

S
[s, m
s
= [s (s + 1) m
s
(m
s
1)]
1/2
h[s, m
s
1 so
[1, 1 =
2 h[1, 0
and hence
[1, 0 =
1
2
(
1
2
+
1
2
) (4.82)
Note that

S
is symmetric in the label 1, 2 of the two electrons. Hence operating on

1
2
which is also symmetric under
the interchange of labels 1 and 2 must produce a symmetric state. Note also that [1, 0 is already normalized. The state
with S = 1 and M = 1 is clearly
1
2
. It could be found by operating with

S
on [1, 0.
The state [0, 0 with S = 0, M = 0 must also be a linear combination of
1
2
and
1
2
. It is orthogonal to [1, 0. As
[1, 0 is symmetric then [0, 0 is antisymmetric, so
[0, 0 =
1
2
(
1
2
) . (4.83)
Formally one can write [0, 0 = a
1
2
+ b
1
2
then

S
+
or

S
applied to it must give zero, i.e.
S
+
(a
1
2
+ b
1
2
) =
_
S
1+
+

S
2+
_
(a
1
2
+ b
1
2
) = 0 (4.84)
= bh
1
2
+ ah
1
2
= 0 (4.85)
so a = b. Normalization requires [a[
2
+[b[
2
= 1 so a =
1
2
.
55
The triplet states with S = 1 are
[1, 1 =
1
2
; [1, 0 =
1
2
(
1
2
+
1
2
) ; [1, 1 =
1
2
, (4.86)
and the singlet S = 0 is
[0, 0 =
1
2
(
1
2
) . (4.87)
An alternative approach can be followed that shows that these eigenstates are also eigenstates of
S
2
=
_
S
1
+

S
2
_
S
1
+

S
2
_
(4.88)
=

S
2
1
+

S
2
2
+ 2
S
1
S
2
(4.89)
=

S
2
1
+

S
2
2
+ 2
_
S
1x

S
2x
+

S
1y

S
2y
+

S
1z

S
2z
_
(4.90)
Since

S
x
=
1
2
_
S
+
+

S
_
and

S
y
=
1
2i
_
S
+
_
then
S
1x

S
2x
+

S
1y

S
2y
=
1
2
_
S
1+

S
2
+

S
1

S
2+
_
(4.91)
and
S
2
=

S
2
1
+

S
2
2
+ 2
S
1z

S
2z
+
_
S
1+

S
2
+

S
1

S
2+
_
. (4.92)
Express [1, 0 = a
1
2
+ b
2
1
then
S
2
[1, 0 = 1 (1 + 1) h
2
[1, 0 = 2a
1
2
+ 2b
2
1
(4.93)
and _
S
2
1
+

S
2
2
+ 2
S
1z

S
2z
+
_
S
1+

S
2
+

S
1

S
2+
__
(a
1
2
+ b
2
1
) =
+a
_
3
4
h
2
2
+
3
4
h
2
2
+ 2
_
h
2
__
h
2
_
2
+ h
2
2
_
+b
_
3
4
h
2
1
+
3
4
h
2
1
+ 2
_
h
2
__
h
2
_
1
+ h
2
1
_
=
1
2
_
3
2
h
2
a
1
2
h
2
a + h
2
b
_
+
2
1
_
3
2
h
2
b
1
2
h
2
b + h
2
a
_
=
1
2
(a + b) h
2
+
1
2
(a + b) h
2
. (4.94)
Thus from eq(4.93,4.94) comparing coefcients of
1
2
and
1
2
gives
2a = a + b; 2b = a + b
a = b.
Normalization requires [a[
2
+[b[
2
= 1 so a = b =
1
2
and
[1, 0 =
1
2
(
1
2
+
2
1
) . (4.95)
If a = b then the eigenvalue is 0 and
1
2
(
1
1
) corresponds to the state [0, 0.
Applying the same method to
1
2
:
S
2
2
=
_
S
2
1
+

S
2
2
+ 2
S
1z

S
2z
+
_
S
1+

S
2
+

S
1

S
2+
__
2
=
_
3
4
h
2
+
3
4
h
2
+ 2
h
2
h
2
+) + 0 + 0
_
2
= 2 h
2
2
= S (S + 1) h
2
2
and so S = 1. For the z-component
S
z
2
=
_
S
1z
+

S
2z
_
2
=
h
2
2
+
h
2
2
so M = 1 and the state
1
2
= [1, 1.
56
4.4.2 Coupling of spin-1/2 and orbital angular momentum
If a spin-
1
2
particle is in a state such that its orbital angular momentum quantum number is , what are the possible values
of total angular momentum quantum number j, and how are eigenstates of

J
2
and

J
z
, where

J is total angular momentum
constructed?
All possible states of the system having orbital angular momentum quantum number can be represented as linear
combinations of the state vectors Y
m
(, ) and Y
m
(, ) , where and represent spin-up and spin-down, as before.
(For a complete specication of the state, a radial function R(r) is needed, but this is of no interest here, since it does
not affect the angular momentum.) Denote the magnetic quantum number for orbital angular momentum by m
, and
the magnetic quantum number for total angular momentum by m
j
. Recalling the restriction [m
[ , there are 2 + 1
possible values of m, so that the total number of state vectors Y
lm
and Y
lm
is 2(2 + 1).
The total angular momentum

J is:
J =

L +

S , (4.96)
where

L is the orbital angular momentum and

S is the spin angular momentum. The states Y
m
(, ) and Y
m
(, )
are already eigenstates of

J
z
:
J
z
Y
m
(, ) =

L
z
Y
m
(, ) +

S
z
Y
m
(, ) = h
_
m +
1
2
_
Y
m
(, ) (4.97)
J
z
Y
m
(, ) =

L
z
Y
m
(, ) +

S
z
Y
m
(, ) = h
_
m
1
2
_
Y
m
(, ) . (4.98)
From this, it is clear that the highest value of total magnetic quantum number m
j
is +
1
2
, and the lowest possible value
is
1
2
. This means that the highest value of J is +
1
2
, and the values of m
j
are
1
2
, +
1
2
. . ., +
1
2
. There are
therefore 2
_
+
1
2
_
+1 = 2 +2 states having j = +
1
2
. Since the total number of states is 2(2 +1), the other 2 states
must be associated with J =
1
2
. This is conirmed by the fact that there is only a single state for which m
j
= +
1
2
,
and a single state for which m
j
=
1
2
, but there are two states for all the other values of m
j
.
Note that most of the states Y
m
and Y
m
, although they are eigenvectors of

J
z
, are not eigenvectors of

J
2
. The
only exceptions to this statement are the states Y
and Y
. One way to construct all the eigenstates of

J
2
having
J = +
1
2
is to start with Y
and act repeatedly with the total lowering operator

J
=

L
+

S
. The eigenstates of
J
2
for which J =
1
2
can then be obtained by nding the states that are eigenvectors of

J
z
but are orthogonal to the
eigenvectors of

J
z
having J = +
1
2
.
As an explicit example, take the case of j = +
1
2
and m
J
= +
1
2
i.e. the maximal projection state Y

[j, j. Applying

J
to it generates the state of j = +

1
2
with m
J
=
1
2
. For any angular momenta,

J
[j, m =
_
j (j + 1) m(m1)h[j, m1, so
[j, j =
_
2jh[j, j 1;

J
[ +
1
2
, +
1
2
=
2
_
+
1
2
_
h[ +
1
2
,
1
2
.
_
+

S
_
Y
=
_
+

S
_
[, [ =
2h[, 1[ +[, h[
so

2
_
+
1
2
_
h[ +
1
2
,
1
2
=
2h[, 1[ +[, h[, (4.99)

and the state
[ +
1
2
,
1
2
=
_
2
2 + 1
[, 1[ +
_
1
2 + 1
[, h[. (4.100)
The state [
1
2
,
1
2
= a[, 1[ + b[, [ is orthogonal to [ +
1
2
,
1
2
so
+
1
2
,
1
2
[
1
2
,
1
2
= 0, (4.101)
_
_
2
2 + 1
, 1[[ +
_
1
2 + 1
, [[
_
(a[, 1[ + b[, [) = 0, (4.102)
a
_
2
2 + 1
+ b
_
1
2 + 1
= 0,
b = a
2. (4.103)
57
Since normalization requires [a[
2
+[b[
2
= 1, then a =
_
1
2+1
and b =
_
2
2+1
, and
[
1
2
,
1
2
=
_
1
2 + 1
[, 1[
_
2
2 + 1
[, [. (4.104)
In summary the total angular momentum

J is:
J =

L +

S , (4.105)
where

L is the orbital angular momentum and

S is the spin angular momentum and
L
2
Y
m
= ( + 1) h
2
Y
m
, (4.106)
S
2
= s (s + 1) h
2
, (4.107)
J
2
= J (J + 1) h
2
, (4.108)
also
J
z
=

L
z
+

S
z
. (4.109)
Four good quantum numbers are J, M
J
, , s where
M
J
= m
+ m
s
(4.110)
J
z
= M
J
h, (4.111)
and
J
2
[J, M
J
, , s = J (J + 1) h
2
[J, M
J
, , s
with [ s[ J + s and
J
z
[J, M
J
, , s = M
J
[J, M
J
, , s
=
_
m
1
2
_
h[J, M
J
, , s.
4.4.3 Addition of orbital and spin-1/2 angular momenta
{This section can be ignored} If only orbital and one spin angular momentum are to be added the method outlined below
can be followed:
The linear combination
[j, m +
1
2
= a[l, m + b[l, m + 1 (4.112)
is constructed which is an eigenfunction of

J
z
with eigenvalue
_
m +
1
2
_
h. The constants a and b have to be found such
that this state is also an eigenstate of

J
2
. Since
J =

L +

S,
J
2
=

L
2
+

S
2
+ 2
J
2
=

L
2
+

S
2
+ 2L
z
S
z
+

L
+

S
+

L

S
+
(4.113)
Then using
[j, m = [j (j + 1) m(m1)]
1/2
h[j, m1,
= [(j m) (j m + 1)]
1/2
h[j, m1
_
L
2
+

S
2
+ 2L
z
S
z
+

L
+

S
+

L

S
+
_
(a[l, m + b[l, m + 1)
= ah
2
_
( + 1) +
3
4
+ 2m
_
1
2
__
[, m
+bh
2
_
( + 1)
3
4
+ 2 (m + 1)
_
1
2
__
[, m + 1
+ah
2
_
[( m) ( + m + 1)]
1
2
_
[, m + 1
+bh
2
_
[( + m + 1) ( (m + 1) + 1)]
1
2
_
[, m.
58
This will be equivalent to
J
2
[j, m +
1
2
= j (j + 1) h
2
(a[l, m + b[l, m + 1)
provided (by comparing coefcients of [l, m and [l, m + 1) that
ah
2
_
( + 1) +
3
4
+ 2m
_
1
2
__
+ bh
2
_
[( + m + 1) ( (m + 1) + 1)]
1
2
_
= j (j + 1) h
2
a
ah
2
_
[( m) ( + m + 1)]
1
2
_
+ bh
2
_
( + 1)
3
4
+ 2 (m + 1)
_
1
2
__
= j (j + 1) h
2
b
i.e.
_
( + 1) +
3
4
+ mj (j + 1)
_
a + [( m) ( + m + 1)]
1/2
b = 0
[( m) ( + m + 1)]
1/2
a +
_
( + 1) +
3
4
m1 j (j + 1)
_
b = 0
These are two simultaneous equations for a and b which have a non-trivial solution if the determinant is zero,
_
( + 1) +
3
4
+ mj (j + 1)
_ _
( + 1) +
3
4
m1 j (j + 1)
_
( m) ( + m + 1) = 0
Putting A = j (j + 1) ( + 1)
3
4
, gives
(Am) (A + m + 1) [ ( + 1) m(m + 1)] = 0,
and
A =
1
2

_
+
1
2
_
2
so A = or A = 1, and j = +
1
2
or j =
1
2
. To solve for a and b the matrix equation
_
( + 1) +
3
4
+ mj (j + 1) [( m) ( + m + 1)]
1/2
[( m) ( + m + 1)]
1/2
( + 1) +
3
4
m1 j (j + 1)
_
_
a
b
_
= 0
gives for j = +
1
2
(m) a + [( m) ( + m + 1)]
1/2
b = 0
a
b
=
_
+ m + 1
m
_
1/2
.
Normalization requires [a[
2
+[b[
2
= 1, yielding
b =
_
m
2 + 1
; a =
_
+ m + 1
2 + 1
.
Hence the state
[j, m +
1
2
[ +
1
2
, m +
1
2
=
_
+ m + 1
2 + 1
[l, m +
_
m
2 + 1
[l, m + 1. (4.114)
Choosing j =
1
2
gives
[j, m +
1
2
[
1
2
, m +
1
2
=
_
m
2 + 1
[l, m
_
+ m + 1
2 + 1
[l, m + 1. (4.115)
59
60
CHAPTER 5. APPROXIMATE METHODS & MANY-BODY SYSTEMS
Chapter 5
Approximate methods & Many-body systems
As in classical mechanics there are relatively few interesting problems in quantum mechanics that can be solved ex-
actly. Approximate methods are necessary to obtain eigenvalues and eigenfunctions for potentials in the Schrdinger
equation. Such methods conveniently divide into two groups according to whether the Hamiltonian of the system is
time-independent or time-dependent.
In the rst instance the approximate determination of the discrete eigenvalues and eigenfunctions for stationary states
of a time-indpendent Hamiltonian will be discussed.
5.1 Time-independent perturbation theory for a non-degenerate energy level
Suppose that the time-independent Hamiltonian

H of a system can be expressed as
H =

H
0
+ H
(5.1)
where

H
0
is the unperturbed Hamiltonian whose corresponding time-independent Schrdinger equation can be solved
exactly, i.e.
H
0
[
(0)
n
= E
(0)
n
[
(0)
n
(5.2)
where E
(0)
n
is the energy eigenvalue (known) with eigenfunction [
(0)
n
(also known). These eigenfunctions form a com-
plete orthonormal set, with
(0)
k
[
(0)
n
=
kn
. The additional term H
is in some sense a "small" change to

H
0
- a
perturbation. In perturbation theory the solution is expanded as a power series in the perturbation. For the theory to be
valid the series must be convergent in a mathematical sense. For the theory to be useful, however, the series must be
rapidly convergent such that only the rst few terms are important. (The convergence of the series will not be discussed.
In some cases it can even be shown that the series cannot converge and yet the rst few terms do satisfactorily describe
the physical system.) To keep track of the order of the perturbation a dimensionless parameter is introduced into the
Hamiltonian via
H =

H
0
+ H
(5.3)
such that = 0 gives the unperturbed system and = 1 gives the actual system to be solved. It is assumed that

H has a
discrete set of eigenvalues which are non-degenerate and satisfy
H[
n
= E
n
[
n
, (5.4)
_
H
0
+ H
_
[
n
= E
n
[
n
. (5.5)
Considering a particular unperturbed energy E
(0)
n
it is assumed that H
is small enough that E

n
is closer to E
(0)
n
than
any other unperturbed level. Thus
lim
0
E
n
E
(0)
n
and lim
0
[
n
[
(0)
n
. (5.6)
The basis idea of perturbation theory is to assume that the eigenvalues and eigenfunctions can be expanded in a power
series of the perturbation parameter as
E
n
=
j=0
j
E
(j)
n
= E
(0)
n
+ E
(1)
n
+
2
E
(2)
n
+ . . . (5.7)
and
[
n
=
j=0
j
[
(j)
n
= [
(0)
n
+ [
(1)
n
+
2
[
(2)
n
+ . . . (5.8)
61
PHAS3226: Quantum Mechanics CHAPTER 5. APPROXIMATE METHODS & MANY-BODY SYSTEMS
where j, the power of , is the order of the perturbation. Substituting these expressions into equ(5.5) gives
_
H
0
+ H
__
[
(0)
n
+ [
(1)
n
+
2
[
(2)
n
+ . . .
_
(5.9)
=
_
E
(0)
n
+ E
(1)
n
+
2
E
(2)
n
+ . . .
__
[
(0)
n
+ [
(1)
n
+
2
[
(2)
n
+ . . .
_
j=0
j

H
0
[
(j)
n
+
j=1
j

H[
(j1)
n
=
_
_
j=0
j
E
(j)
n
_
_
_

k=0
k
[
(k)
n

_
=
j,k
j+k
E
(j)
n
[
(k)
n
=
j=0
j
k=0
j
E
(k)
n
[
(jk)
n
.
As equ(5.9) is to be true for all values of , coefcients of equal powers of must be the same on both sides. Thus the
various orders give
0
;

H
0
[
(0)
n
= E
(0)
n
[
(0)
n
, (5.10)
1
;

H
0
[
(1)
n
+

H
[
(0)
n
= E
(0)
n
[
(1)
n
+ E
(1)
n
[
(0)
n
, (5.11)
2
;

H
0
[
(2)
n
+

H
[
(1)
n
= E
(0)
n
[
(2)
n
+ E
(1)
n
[
(1)
n
+ E
(2)
n
[
(0)
n
. (5.12)
etc.
H
0
[
(k)
n
+

H
[
(k1)
n
=
k
r=0
E
(r)
n
[
(kr)
n
.
5.1.1 First-order
To obtain the rst-order correction to the energy, E
(1)
n
then equ(5.11) is
_
H
0
E
(0)
n
_
[
(1)
n
=
_
E
(1)
n

H
_
[
(0)
n
(5.13)
and taking the scalar product with [
(0)
n
gives
(0)
n
[
_
H
0
E
(0)
n
_
[
(1)
n
=
(0)
n
[
_
E
(1)
n

H
_
[
(0)
n
= E
(1)
n

(0)
n
[
[
(0)
n

as [
(0)
n
are normalized. The L.H.S. is zero as

H
0
is Hermitian. To see this explicitly,
(0)
n
[
H
0
[
(1)
n
=
(1)
n
[
H
0
[
(0)
n

=
(1)
n
[E
(0
n
[
(0)
n

= E
(0
n

(0)
n
[
(1)
n

as E
(0
n
is real. Thus rst-order correction to the energy, E
(1)
n
is given by
E
(1)
n
=
(0)
n
[
[
(0)
n
. (5.14)
This is one of the most useful equations in quantum mechanics. The rst-order correction to the energy, E
(1)
n
is just the
expectation value of the perturbation in the state [
(0)
n
. Hence to rst-order
E
n
E
(0)
n
+ E
(1)
n
. (5.15)
5.1.2 Second-order
Taking the scalar product of equ(5.12) with [
(0)
n
gives
(0)
n
[
H
0
[
(2)
n
+
(0)
n
[
[
(1)
n
= E
(0)
n

(0)
n
[
(2)
n
+ E
(1)
n

(0)
n
[
(1)
n
+ E
(2)
n

(0)
n
[
(0)
n

(0)
n
[
H
0
E
(0)
n
[
(2)
n
= E
(2)
n
+
(0)
n
[E
(1)
n

H
[
(1)
n
.
The L.H.S. is zero for the same reason as in the rst-order case, so that
E
(2)
n
=
(0)
n
[
E
(1)
n
[
(1)
n
. (5.16)
62
Hence to nd the second-order correction E
(2)
n
it is necessary to nd the rst-order correction, [
(1)
n
, to the eigenfunction
[
(0)
n
. In fact in general [
(0)
n
yields E
(0)
n
, E
(1)
n
; [
(1)
n
yields E
(2)
n
, E
(3)
n
etc. To nd [
(1)
n
it can be expanded in terms
of the complete set of states
_
[
(0)
n

_
as
[
(1)
n
=
k
a
(1)
nk
[
(0)
k
. (5.17)
Substituting this into eq(5.13)
_
H
0
E
(0)
n
_
[
(1)
n
=
_
E
(1)
n

H
_
[
(0)
n

leads to
_
H
0
E
(0)
n
_
k
a
(1)
nk
[
(0)
k
=
_
E
(1)
n

H
_
[
(0)
n

and taking the scalar product with [
(0)
gives
k
a
(1)
nk
(0)
[
_
H
0
E
(0)
n
_
[
(0)
k
=
(0)
[
_
E
(1)
n

H
_
[
(0)
n
(5.18)
k
a
(1)
nk
_
E
(0)
k
E
(0)
n
_
(0)
[
(0)
k
= E
(1)
n

(0)
[
(0)
n

(0)
[
(0)
n
(5.19)
But
(0)
[
(0)
k
=
k
and by introducing the matrix element of the perturbation
n
=
(0)
[
(0)
n
, (5.20)
then
k
a
(1)
nk
_
E
(0)
k
E
(0)
n
_
k
= E
(1)
n

n
n
. (5.21)
For = n the rst-oder result is obtained E
(1)
n
=

H
nn
. For ,= n
a
(1)
n
_
E
(0)
E
(0)
n
_
=
n
,
a
(1)
n
=
n
_
E
(0)
n
E
(0)
_
and so
[
(1)
n
=
k=n
kn
_
E
(0)
n
E
(0)
k
_[
(0)
k
. (5.22)
Note that a sufcient condition for this method is that
kn
_
E
(0)
n
E
(0)
k
_
1.
Note also that the coefcient a
(1)
nn
, the component of [
(1)
n
along [
(0)
n
cannot be found. Returning to
E
(2)
n
=
(0)
n
[
E
(1)
n
[
(1)
n

=
(0)
n
[
E
(1)
n
[
k
a
(1)
nk
[
(0)
k

E
(2)
n
=
(0)
n
[
E
(1)
n
[
_
_
a
(1)
nn
[
(0)
n
+
k=n
a
(1)
nk
[
(0)
k

_
_
, (5.23)
63
but
(0)
n
[
E
(1)
n
[
(0)
n
= 0 (the rst-order correction formula), so
E
(2)
n
=
(0)
n
[
E
(1)
n
[
k=n
a
(1)
nk
[
(0)
k
(5.24)
=
k=n
a
(1)
nk
(0)
n
[
E
(1)
n
[
(0)
k

k=n
a
(1)
nk
_
nk
E
(1)
n

nk
_
=
k=n
a
(1)
nk
nk
E
(2)
n
=
k=n
kn
_
E
(0)
n
E
(0)
k
_

H
nk
, (5.25)
E
(2)
n
=
k=n
kn
2
_
E
(0)
n
E
(0)
k
_ =
k=n
(0)
k
[
[
(0)
n

2
_
E
(0)
n
E
(0)
k
_ . (5.26)
and
[
n
= [
(0)
n
+
k=n
kn
_
E
(0)
n
E
(0)
k
_[
(0)
k
(5.27)
to rst-order in the eigenfunction.
Note that if the term with k = n is included then
[
n
= (1 + a
(1)
nn
[
(0)
n
+
k=n
kn
2
_
E
(0)
n
E
(0)
k
_[
(0)
k
. (5.28)
The coefcients a
(1)
nn
are not determined by these methods. It can be shown that coefcients of this type can not be found
in any order by the method above. Thus one in tempted to conclude that the choice of these quantities has no physical
signicance, thus could set a
(1)
nn
= 0!
{One could attempt to nd a
(1)
nn
=
(0)
n
[
(1)
n
by using the normalization condition
1 =
n
[
n
=
_
(0)
n
[ +
(1)
n
[ +
2
(2)
n
[ + . . .
__
[
(0)
n
+ [
(1)
n
+
2
[
(2)
n
+ . . .
_
.
This could be done to any order. For example, to rst-order in
1 =
_
(0)
n
[ +
(0)
n
[
__
[
(0)
n
+ [
(1)
n

_
= 1 +
(0)
n
[
(1)
n
+
(0)
n
[
(1)
n

1 = 1 + a
(1)
nn
+ a
(1)
nn
shows that the real parts of a
(1)
nn
vanish, but gives no information on the imaginary parts, so a
(1)
nn
= i say, with real. If
normalization to second order is carried out one gets
a
(2)
nn
+ a
(2)
nn
+
a
(1)
nk
2
= 0
which again is only an equation for the real parts of a
(2)
nn
. This feature is found for any order. The imaginary parts of
a
(j)
nn
cannot be found from the normalization condition. This remaining arbitrariness corresponds to the fact that one can
multiply a wave function [
n
by an arbitrary phase factor e
i
( real) without changing the normalization. Thus if a
(j)
nn
is imaginary then
_
1 + a
(j)
nn
_
[
(0)
n
Ae
i
[
(0)
n
. Therefore without loss of generality one can set the imaginary parts
of the coefcients a
(j)
nn
to zero. }
5.1.3 Observations on the energy corrections
1. Unlike the rst-order energy shift E
(1)
n
which only involves the state [
(0)
n
, the second-order correction depends
on the complete set of unperturbed states and, in general, leads to formidable calculations.
64
2. If n is the ground state, E
(0)
n
E
(0)
k
< 0 (always) and thus E
(2)
n
< 0.
3. The rst-order energy shift E
(1)
n
often vanishes exactly on symmetry grounds thus necessitating evaluation of the
second-order correction.
4. The states [
(0)
k
over which the summation is performed are often called intermediate states. Each term in E
(2)
n
is viewed as a succession of two rst-order "transitions" n k; k n weighted by the energy denominator
E
(0)
n
E
(0)
k
. System appears to leave state [
(0)
n
, propagate to intermediate state [
(0)
k
then fall back again to
state [
(0)
n
.
5. If all matrix elements of

H
are of the same order of magnitude - a reasonable rst guess - then the "nearby" levels
have a bigger effect on E
(2)
n
than "distant" ones.
6. If n is not the ground state and if an important level k - lying nearby or having

H
kn
large - lies above level n then
E
(2)
n
is <0; if level k lies below the shift is upwards. One speaks of a tendency of levels to repel each other!
7. For perturbation theory to be useful it must give small corrections so that calculations to low order sufce. For
rst-order changes to [
(0)
n
to be small one needs
nk
E
(0)
n
E
(0)
k
for k ,= n. In general the diagonal

matrix elements

H
nn
will be of the same order of magnitude as the non-diagonal elements

H
nk
. Thus the condition
implies
nn
E
(0)
n
E
(0)
k
and therefore
nn
min
E
(0)
n
E
(0)
k
. That is the rst-order shift E

(1)
n
must be small compared to the level spacing min
E
(0)
n
E
(0)
k
between E
(0)
n
and the nearest lying energy level.
These conditions break down if the level is degenerate. In this case there are some states [
(0)
k
, k ,= n with
the same energy E
(0)
k
= E
(0)
n
. Thus some denominators E
(0)
n
E
(0)
k
vanish making the expressions for a
(1)
nk
meaningless. One can, however, relax the condition that all the unperturbed energy levels are non-degenerate as
one only needs that the level whose energy shift is being calculated is non-degenerate. The difculties only arise
from the degeneracy of level n, the other levels with E
(0)
k
,= E
(0)
n
can be degenerate.
5.2 The Degenerate case
As well as the difculty referred to above, i.e. E
(0)
n
E
(0)
k
being zero for [
(0)
k
, = [
(0)
n
for a degenerate level
there is another difculty. Suppose level E
(0)
n
is -fold degenerate. There are unperturbed wavefunctions [
(0)
n
,
( = 1, 2, . . . ), corresponding to the level and it is not known a priori to which of these functions, or linear combinations,
the perturbed eigenfunction tends as 0.
The -unperturbed eigenfunctions [
(0)
n
, ( = 1, 2, . . . ), for level E
(0)
n
are orthogonal to those [
(0)
k
corresponding
to any other level E
(0)
k
,= E
(0)
n
. Although not necessarily orthogonal amongst themselves it is always possible to construct
linear combinations of them that are orthonormal, i.e.
_
(0)
n
[
(0)
n
_
=
. It will be assumed that this has already been

done so that the set
_
(0)
n
_
are orthonormal.
The "correct" unperturbed eigenfunctions [
(0)
n
that [
n
tend to as 0 are some linear combination of the
[
(0)
n
, so that
lim
0
[
n
= [
(0)
n
=
=1
c
[
(0)
nv
. (5.29)
To avoid too many summations, consideration will be given explicitly to a doubly degenerate level ( = 2) as the nal
result is easily generalized to an -fold degenerate level. In this case the degenerate unperturbed eigenfunctions are [
(0)
n1
,
[
(0)
n2
. The correct zeroth-order eigenfunctions are
[
(0)
n1
= c
11
[
(0)
n1
+ c
12
[
(0)
n2
, (5.30)
[
(0)
n2
= c
21
[
(0)
n1
+ c
22
[
(0)
n2
(5.31)
Corresponding to eqs(5.7, 5.8) of the non-degenerate case is
[
n
= [
(0)
n
+ [
(1)
n
+
2
[
(2)
n
+ . . . , (5.32)
E
n
= E
(0)
n
+ E
(1)
n
+
2
E
(2)
n
+ . . . (5.33)
65
for = 1, 2. Thus for rst-order eq(5.13) becomes
_
H
0
E
(0)
n
_
[
(1)
n
=
_
E
(1)
n

H
_
[
(0)
n
. (5.34)
For a doubly degenerate case this is
_
H
0
E
(0)
n
_
[
(1)
n1
=
_
E
(1)
n1

H
_
[
(0)
n1
(5.35)
=
_
E
(1)
n1

H
__
c
11
[
(0)
n1
+ c
12
[
(0)
n2
_
(5.36)
and
_
H
0
E
(0)
n
_
[
(1)
n2
=
_
E
(1)
n2

H
_
[
(0)
n2
(5.37)
=
_
E
(1)
n2

H
__
c
21
[
(0)
n1
+ c
22
[
(0)
n2
_
(5.38)
Taking the scalar product of eq(5.36)with [
(0)
n1
and [
(0)
n2
in turn gives
(0)
n1
[
_
H
0
E
(0)
n
_
[
(1)
n1
=
(0)
n1
[
_
E
(1)
n1

H
__
c
11
[
(0)
n1
+ c
12
[
(0)
n2
_
(5.39)
(0)
n2
[
_
H
0
E
(0)
n
_
[
(1)
n1
=
(0)
n2
[
_
E
(1)
n1

H
__
c
11
[
(0)
n1
+ c
12
[
(0)
n2
_
(5.40)
The L.H.S is zero just as in the non-degenerate case. Using the orthonormality of [
(0)
n1
and [
(0)
n2
gives
c
11
_
E
(1)
n1

(0)
n1
[
[
(0)
n1
_
c
12
(0)
n1
[
[
(0)
n2
= 0, (5.41)
c
11
(0)
n2
[
[
(0)
n1
+ c
12
_
E
(1)
n1

(0)
n2
[
[
(0)
n2
_
= 0. (5.42)
Writing

H
ij
=
(0)
ni
[
[
(0)
nj
these equations are
_
11
E
(1)
n1
_
c
11
+

H
12
c
12
= 0,
21
c
11
+
_
22
E
(1)
n1
_
c
12
= 0
and have a matrix form
_

H
11
E
(1)
n1
12
21

H
22
E
(1)
n1
_
_
c
11
c
12
_
= 0. (5.43)
These simultaneous equations for c
11
and c
12
have a non-trivial solution if and only if the determinant is zero,
11
E
(1)
n1
12
21

H
22
E
(1)
n1
= 0 (5.44)
This is a quadratic equation in E
(1)
n1
leading to two possible values of E
(1)
n1
. One solution gives E
(1)
n1
, the other gives
E
(1)
n2
. Substituting these back into eq(5.43) leads to two sets of values for the coefcients, One set for c
11
and c
12
and
the other set corresponds to c
21
and c
22
. Thus two zero-order eigenfunctions [
(0)
n1
and [
(0)
n2
are found that are linear
combinations of [
(0)
n1
and [
(0)
n2
, and which, in general, correspond to different rst-order corrections to the energy
E
(1)
n1
and E
(1)
n2
. These two linear combinations are the correct zeroth-order eigenfunctions and can thus be used along
with the unperturbed eigenfunctions of the other states to evaluate higher order corrections in a similar way as in the
non-degenerate case.
The generalization to the -fold degenerate case is straightforward. Since [
(0)
n
=
=1
c
[
(0)
nv
then the determi-
nantal equation becomes
11
E
(1)
n1
12

H
13
. . .

H
21

H
22
E
(1)
n1
23
. . .

H
31

H
32

H
33
E
(1)
n1
. . .

H
3
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
1

H
2

H
3
. . .

H
E
(1)
n1
= 0. (5.45)
66
which has -roots for E
(1)
n1
. If all the roots are different the degeneracy is completely removed by the perturbation. If
some of them are the same the degeneracy is only partially removed. The residual degeneracy may be removed in a higher
order of perturbation. In general the determination of the coefcients c
is a lengthy process.
The determination of the eigenfunctions [
(0)
n
amounts to nding the correct orthogonal linear combination of the
original zero-order degenerate eigenfunctions [
(0)
nv
such that the matrix

H
=
_
(0)
n

(0)
n
_
is diagonal with respect
to the indices , .
A few special cases are worth noting:
1. The matrix of

H
is already diagonal, as in
11
E
(1)
n1
0
0

H
22
E
(1)
n1
= 0 (5.46)
then E
(1)
n1
=

H
11
and E
(1)
n2
=

H
22
. Thus the two states [
(0)
n1
and [
(0)
n2
are not connected in rst-order. The
degeneracy is completely removed and the energies are E
n1
= E
(0)
n
+

H
11
, and E
n2
= E
(0)
n
+

H
22
. This
happens when the perturbation

H
commutes with the operator whose eigenvalues the label represents. Suppose
an Hermitian operator

A commutes with

H
0
and

H
, i.e.
_
H
0
,

A
_
= 0, and
_
,

A
_
= 0. Thus

H
0
and

A possess
a complete set of mutual eigenstates. The eigenstates [
(0)
n
, = 1, . . . , belonging to eigenvalue E
(0)
n
of

H
0
are
also eigenfunctions of

A, with
A[
(0)
n
= a
n
[
(0)
n
. (5.47)
Then
_
(0)
n
,

A
_

(0)
n
_
=
_
(0)
n

A

A

H

(0)
n
_
= 0, (5.48)
=
_
a
n
a
n
_
_
(0)
n

(0)
n
_
= 0. (5.49)
and

H
= 0 if a
n
,= a
n
(note: the eigenvalues are real). Hence if all the -eigenvalues a
n
are different

H
is
already diagonal and the [
(0)
n
are already the correct zero-order states.
In the hydrogen atom a degeneracy is associated with the eigenvalues of

L
z
in that all 2 + 1 m values have the
same energy. Suppose that
_
,

L
z
_
= 0,
then choose [
(0)
n
to be eigenfunctions of

L
z
i.e. m, so that
L
z
[
(0)
nm
= mh[
(0)
nm
. (5.50)
Then
_
(0)
nm
,

L
z
_

(0)
nm
_
=
_
(0)
nm
[
L
z
[
(0)
nm
_
(0)
nm
[
L
z

H
[
(0)
nm
_
(5.51)
= mh
_
(0)
nm
[
[
(0)
nm
_
m
h
_
(0)
nm
[
[
(0)
nm
_
= (mm
) h

H
n
. (5.52)
But the m values are real and thus if m
,= m then

H
n
= 0, i.e.

H
n
is a diagonal matrix.
This example illustrates a general result: a degeneracy due to a symmetry of

H
0
is reduced or removed altogether
by a perturbation of a lower symmetry. In this example,

H
0
is spherically symmetric but

H
possesses only axial

symmetry.
The coefcients c
11
, c
12
are easily found from
_

H
11
E
(1)
n1
0
0

H
22
E
(1)
n1
_
_
c
11
c
12
_
= 0. (5.53)
For E
(1)
n1
=

H
11
, c
12
= 0 so can choose c
11
= 1 and [
(0)
n1
= [
(0)
n1
. For E
(1)
n2
=

H
22
, c
21
= 0 so can choose
c
22
= 1 and [
(0)
n2
= [
(0)
n2
.
67
2. The other interesting case occurs for

H
11
=

H
22
= 0. Then the determinant is
E
(1)
n1
12
21
E
(1)
n1
= 0 (5.54)
and
_
E
(1)
n1
_
2
=

H
12

H
21
=
12
2
(5.55)
since
12
=
(0)
n1
[
[
(0)
n2
=
(0)
n2
[
[
(0)
n1
=
_
21
_
. (5.56)
Hence the rst-order energy correction is
E
(1)
n1
=
12
. (5.57)
The coefcients satisfy
c
11
E
(1)
n1
+ c
12

H
12
= 0
so that
c
11
c
12
=
12
E
(1)
n1
. (5.58)
For E
(1)
n1
= +
12
,
c
11
c
12
= 1 depending on the sign of

H
12
. For E
(1)
n1
=
12
,
c
11
c
12
= 1. The original
eigenstates [
(0)
n1
and [
(0)
n2
are said to be fully mixed with the correct zeroth-order eigenfunctions being
[
(0)
n1
=
1
2
_
[
(0)
n1
[
(0)
n2
_
, (5.59)
[
(0)
n2
=
1
2
_
[
(0)
n1
[
(0)
n2
_
. (5.60)
5.3 Applications of perturbation theory
5.3.1 First-order examples
The anharmonic oscillator To calculate the energy eigenvalues of an anharmonic oscillator represented by the Hamil-
tonian
H =
p
2
2m
+
1
2
k x
2
+ x
4
. (5.61)
The presence of the anharmonic term x
4
makes this problem difcult to solve exactly, but if the term x
4
is small (in a
sense to be claried below), rst-order perturbation theory can be used. The unperturbed Hamiltonian

H
0
is the harmonic
oscillator,
H
0
=
p
2
2m
+
1
2
k x
2
, (5.62)
and the perturbation Hamiltonian is:
= x
4
. (5.63)
Consider the change of energy of the ground state. The unperturbed ground-state energy is E
(0)
0
=
1
2
h, where
is the frequency of the harmonic oscillator = (k/m)
1/2
. According to eqn (5.14), the change of ground-state energy
caused by the term x
4
is:
E
(1)
0
=
0
[
[
0
, (5.64)
where [
0
is the ground-state eigenvector. This quantity could be evaluated using the operator methods for the harmonic
oscillator but here the normal wavefunction methods are used .
The normalized ground-state wave-function of the harmonic oscillator is
0
(x) =
_
m
h
_
1/4
exp(mx
2
/2h) . (5.65)
Hence
E
(1)
0
=
_
m
h
_
1/2
_

x
4
exp(mx
2
/h)dx . (5.66)
68
In order to calculate this integral note that
_

e
x
2
dx =
1/2
1/2
. (5.67)
Differentiating this with respect to once gives
d
d
_

e
x
2
dx =
_

x
2
e
x
2
dx =
1
2
1/2
3/2
(5.68)
and differentiating again
d
d
_

x
2
e
x
2
dx =
_

x
4
e
x
2
dx =
3
4
1/2
5/2
. (5.69)
It follows that
_

x
4
exp(mx
2
/h)dx =
3
4
1/2
(m/h)
5/2
(5.70)
and hence that
E
(1)
0
=
3
4
_
m
h
_
1/2
_
h
m
_
1/2
_
h
m
_
2
=
3 h
2
4m
2
2
. (5.71)
Thus the ground-state energy is raised by an amount proportional to , as should be expected.
It is useful to consider how the total ground-state energy E
0
E
(0)
0
+E
(1)
0
deviates from the unperturbed ground-state
energy E
(0)
0
;
E
0
/E
(0)
0

_
E
(0)
0
+ E
(1)
0
_
/E
(0)
0
= 1 + E
(1)
0
/(
1
2
h) = 1 +
3 h
2m
2
3
. (5.72)
The strength of the perturbation can therefore be characterized by the dimensionless quantity g = 2 h/m
2
3
. The table
shows how the approximation for E
0
/E
(0)
0
from rst-order perturbation theory compares with essentially exact numerical
predictions. This shows that the energy shift is given very accurately for small values of g, but becomes less accurate as g
increases.
g = 2 h/m
2
3
E
0
/E
(0)
0
(pert. th.) E
0
/E
(0)
0
(exact)
0.01 1.00750 1.00735
0.10 1.075 1.065
0.2 1.15 1.12
Table 5.1: The ground-state energy E
0
of the anharmonic oscillator divided by the ground-state energy of the unperturbed
harmonic oscillator E
(0)
0
for different values of . The size of is specied by the dimensionless quantity g = 2 h/m
3
.
Second and third columns give E
0
/E
(0)
0
from rst-order perturbation theory and from almost exact numerical solution of
the Schrdinger equation.
Harmonic oscillator As an example to demonstrate the validity of the perturbation approach consider a harmonic
oscillator where the perturbation is itself harmonic. The unperturbed Hamiltonian
H
0
=
p
2
2m
+
1
2
k x
2
(5.73)
has energy eigenvalues
E
(0)
n
=
_
n +
1
2
_
h
0
,
where
0
=
_
k/m. Suppose that the force constant k is changed slightly, k k
such that k
k = with 0 < k.
The Hamiltonian is now
H =
p
2
2m
+
1
2
k x
2
+
1
2
x
2
=

H
0
+
1
2
x
2
(5.74)
=
p
2
2m
+
1
2
(k + ) x
2
69
and the new force constant is k + so the new frequency of oscillation
=
_
k +
m
_
1/2
=
_
k
m
_
1/2 _
1 +

k
_
1/2
=
0
_
1 +

k
_
1/2
(5.75)
and the exact energy levels are
E
n
=
_
n +
1
2
_
h
0
_
1 +

k
_
1/2
E
n
=
_
n +
1
2
_
h
0
_
1 +
1
2
k

1
8
_
k
_
2
+ . . .
_
. (5.76)
The same problem will now be solved using time-independent perturbation theory. The perturbation is

H
=
1
2
x
2
.
The energy levels are non-degenerate, so the rst-order correction to E
(0)
n
=
_
n +
1
2
_
h
0
is
E
(1)
n
=
(0)
n
[
[
(0)
n
=
1
2
(0)
n
[ x
2
[
(0)
n
=
1
2
_
x
2
_
nn
.
The matrix elements may be evaluated several ways; using the eigenfunctions for
(0)
n
; or using the operator approach.
The latter is easier. The matrix elements
_
x
2
_
nn
were calculated earlier when considering the generalized uncertainty
relation for the harmonic oscillator, i.e.
n[ x
2
[n =
2n + 1
2m
0
h =
_
n +
1
2
_
h
m
. (5.77)
For the second-order correction matrix elements between different states will be needed. In anticipation of this recall that
x =
_
h
2m
0
_
( a
+
+ a
) and
x[n =
_
h
2m
0
_
1/2
( a
+
+ a
) [n =
_
h
2m
0
_
1/2
_
n + 1[n + 1 +
n[n 1
_
so
n[ x x[k =
_
h
2m
0
_
_
n + 1n + 1[ +
nn 1[
_
_
k + 1[k + 1 +
k[k 1
_
,
=
_
h
2m
0
__ _
(n + 1) (k + 1)n + 1[k + 1 +
_
(n + 1) kn + 1[k 1
+
_
n(k + 1)n 1[k + 1 +
nkn 1[k 1
_
,
=
_
h
2m
0
_
_
_
(n + 1) (k + 1)
nk
+
_
(n + 1) k
n,k2
+
_
n(k + 1)
n,k+2
+
nk
nk
_
,
n[ x
2
[k =
_
h
2m
0
_
_
(2n + 1)
nk
+
_
(n + 1) (n + 2)
n,k2
+
_
n(n 1)
n,k+2
_
. (5.78)
The rst-order correction is
E
(1)
n
=
1
2
_
x
2
_
nn
=
1
2
_
n +
1
2
_
h
m
0
. (5.79)
This is the same as the rst term in the expansion equ(5.76) since
_
n +
1
2
_
h
0
1
2
k
=
_
n +
1
2
_
h
0
1
2
m
2
0
=
_
n +
1
2
_
h
2m
0
. (5.80)
The second-order correction is obtained from
E
(2)
n
=
k=n
k[
[n
2
_
E
(0)
n
E
(0)
k
_ =
k=n
k[
1
2
x
2
[n
2
__
n +
1
2
_
h
0
_
k +
1
2
_
h
0
_
=
k=n
_
1
2
_
2

k[ x
2
[n
2
(n k) h
0
=
1
4
2
h
0
k=n
k[ x
2
[n
2
(n k)
. (5.81)
70
But from equ(5.78) the only non-zero matrix elements are for k = n 2; k = n already being excluded from the
summation. Hence
E
(2)
n
=
1
4
2
h
0
_
n + 2[ x
2
[n
2
(n (n + 2))
+
n 2[ x
2
[n
2
(n (n 2))
_
, (5.82)
=
1
4
2
h
0
_
_
_
h
2m
0
_
2
_
(n + 1) (n + 2)
2
2
+
_
h
2m
0
_
2
_
n(n 1)
2
2
_
_,
=
1
4
2
h
0
_
h
2m
0
_
2
_
[(n + 1) (n + 2)[
2
+
[n(n 1)[
2
_
,
E
(2)
n
=
_
h
2
32m
2
3
0
_
(4n 2) =
_
h
2
8m
2
3
0
__
n +
1
2
_
. (5.83)
The perturbation result is in agreement with the exact result equ(5.76), since from the expansion the second-order correc-
tion is
_
n +
1
2
_
h
0
_
1
8
_
k
_
2
_
=
1
8
_
n +
1
2
_
h
2
m
2
3
0
.
Ground state of helium atom The neutral He atom consists of two electrons bound to a nucleus having charge Z[e[,
with Z = 2. Because of the electrostatic repulsion between the electrons, the Schrdinger equation for this system cannot
be solved exactly. One way of tackling it is to treat the inter-electron repulsion by perturbation theory.
The Hamiltonian for the two-electron system is:
H (r
1
, r
2
) =
h
2
2m
2
1
h
2
2m
2
2
Ze
2
4
0
_
1
r
1
+
1
r
2
_
+
e
2
4
0
1
r
12
. (5.84)
Here, r
1
and r
2
are the positions of the two electrons, and r
12
= [r
1
r
2
[ is the distance between the electrons;
2
1
and
2
2
are the Laplacian operators for the two electron positions. Since the electrons have spin-
1
2
the full space-spin wave-
function should describe both the spin state and the space state. The electrons are also fermions so the total wave-function
must be antisymmetric with respect to interchange of electrons. However, the ground state is a spin singlet so that the spin
part of the two-electron wave-function is (
1
2

1
2
)/
2. This is already antisymmetric, so that the spatial part of

the wave-function (r
1
, r
2
) is symmetric. Since the Hamiltonian does not depend on spin, Schrdingers equation is:
_
h
2
2m
2
1
h
2
2m
2
2
Ze
2
4
0
_
1
r
1
+
1
r
2
_
+
e
2
4
0
1
r
12
_
(r
1
, r
2
) = E(r
1
, r
2
) . (5.85)
The inter-electron repulsion e
2
/(4
0
r
12
) makes it impossible to solve Schrdingers equation exactly, and perturba-
tion theory offers one way forward. The unperturbed Hamiltonian

H
0
omits the repulsion
H
0
=
h
2
2m
2
1
h
2
2m
2
2
Ze
2
4
0
_
1
r
1
+
1
r
2
_
, (5.86)
and is a separable Hamiltonian with
H
0
(r
1
, r
2
) =

H
0
(r
1
) +

H
0
(r
1
) . (5.87)
In atomic units
H
0
(r) =
1
2
Z
r
. (5.88)
The unperturbed Schrdinger equation,
H
0
(r
1
, r
2
)
(0)
0
(r
1
, r
2
) = E
0
(0)
0
(r
1
, r
2
) , (5.89)
is easy to solve, because it represents two independent electrons. The ground state wavefunction of

H
0
is just the product
of the ground-state wavefunctions of the individual electrons,
(0)
0
(r
1
, r
2
) =
100
(r
1
)
100
(r
2
) , (5.90)
where
100
(r) is the ground-state wavefunction for a single electron bound to a nucleus of charge Z[e[. This can be
obtained from the usual hydrogenic wave-function by replacing e
2
by (Ze)
2
and the energy for each electron is
E
(0)
n
=
Z
2
2n
2
, (5.91)
71
with
100
(r) = Ae
Zr/a
0
, (5.92)
and with a
0
= 4
0
h
2
/me
2
0.529 the usual Bohr radius. The ground-state energy E
(0)
0
of the He atom in this ap-
proximation is twice the ground-state energy of a single electron bound to a nucleus of charge Z[e[, which is Z
2
/2 a.u, so
that E
(0)
0
= Z
2
= 4 a.u. ("Hartrees"). Not surprisingly, this is a poor approximation compared with the experimental
value of the He ground-state energy, which is 2.905 a.u. .
The electron-electron mutual repulsion is taken as the perturbation part
=
e
2
4
0
r
12
1
r
12
, (5.93)
so that by rst-order perturbation theory, the total energy is shifted by:
E
(1)
0
=
_
(0)
0
e
2
4
0
r
12
(0)
0
_

_
(0)
0
1
r
12
(0)
0
_
. (5.94)
This can be evaluated exactly, but it is long and tedious as
1
r
12
=
1
|r
1
r
2
|
depends on the angular coordinates. The result is
E
(1)
0
= 1.25 a.u., so that the corrected ground-state energy is 4+1.25 = 2.75 a.u. This is still not in perfect agreement
with the experimental value of 2.905 a.u, but it is much better than the value obtained by omitting the repulsion between
electrons.
5.3.2 Second-order example
Stark effect in hydrogen To illustrate the application of perturbation theory to a real physical problem consider the
effect of a time-independent, spatially uniform external electric eld on the energy levels of an atom - the so-called
Stark effect. The Stark effect is the shift of the optical adsorption and emission frequencies of atoms induced by the
application of an external electric eld. This shift arises from the shift of the atomic energy levels. To keep the discussion
as manageable as possible it will be restricted to hydrogen-like atoms, i.e. to one electron atoms.
The unperturbed Hamiltonian

H
0
is, as usual,
H
0
=
h
2
2m
e
2
4
0
r
, (5.95)
whose zero-order eigenfunctions are the hydrogenic wavefunctions. It is useful to recall some of their properties:
(1) The eigenfunctions are
nm
(r, , ) = R
n
(r) Y
m
(, ), with R
n
(r) the radial wavefunction and Y
m
(, )
the spherical harmonic,
(2) the parity of
nm
(r, , ) is (1)
,
(3) the eigenfunctions are orthonormal,
_
0
_
0
_
2
0

m

nm
r
2
dr sin dd =
n
m
(4) the unperturbed energy levels are E
n
=
m
2 h
2
Ze
2
4
0
1
n
2
=
mc
2
2
_
Z
n
_
2
where n = 1, 2, 3, . . . and =
e
2
4
0
hc

1
137
is the ne structure constant
(5) for a given , m has 2 + 1 values, , + 1, . . . 1,
(6) for a given n, 0 n 1.
(7) each level is (without considering spin)
n1
=0
(2 + 1) = n
2
degenerate; including spin doubles this value.
The applied electric eld c is uniform over the atomic dimensions and assumed to be directed along the the positive
z-axis, so that the additional potential energy of the electron is
V (r) = ecz = p c (5.96)
where p = ez is the electric dipole moment of the electron. This interaction energy must be added to the Coulomb
potential energy of the electron in the electric eld of the nucleus.
Consider rst the ground state (n = 1, = 0, m = 0). The rst-order shift E
(1)
100
is
E
(1)
100
=
100
[V [
100
=
_
dr
100
(r)V (r)
100
(r) . (5.97)
The ground-state wave-function
100
(r) is:
100
(r) =
1
1/2
a
3/2
0
e
r/a
0
e
r
, (5.98)
72
so that
E
(1)
100
= ec
_
[
100
(r)[
2
zdr . (5.99)
This is exactly zero, since [
100
(r)[
2
is an even function under parity (r r) whereas the perturbation is an odd
function as z z and hence overall the integrand is odd. {Positive values of the integrand cancel negative values. The
reason for this is that the value of the integrand at any point r is exactly equal and opposite to its value at r}. This is
the result of symmetry; the charge distribution in the unperturbed atom is spherically symmetric, whereas the perturbation
Hamiltonian V changes sign when r r. Thus the rst-order shift is zero. Hence for the ground state there is no
energy shift linear in the electric eld strength c. Classically a system with electric dipole moment p will experience an
energy shift p c hence it is concluded that the atom has, in the ground state, no permanent dipole moment.
Since the rst-order shift is zero a second-order calculation must be done. For the ground state the second-order
correction is given by
E
(2)
100
=
n>1
n1
=0
m=
m=
(0)
nm
[
[
(0)
100
2
_
E
(0)
100
E
(0)
nm
_ ,
= e
2
c
2
n>1
n1
=0
m=
m=
(0)
nm
[z[
(0)
100
2
_
E
(0)
100
E
(0)
nm
_ .
Since E
(0)
100
E
(0)
nm
< 0, then E
(2)
n
< 0 i.e. the ground state energy is lowered by the interaction. To evaluate the matrix
elements put z = r cos then
(0)
nm
[z[
(0)
100
=
_ _
R
n
(r) Y

m
(, ) r cos R
10
Y
00
r
2
dr d. (5.100)
But Y
00
= 1/
4 and cos =
_
4
3
Y
10
so that
(0)
nm
[z[
(0)
100
=
_
R
n
(r) R
10
r
3
dr
_
Y

m
(, )
1
3
Y
10
d.
From the orthonormality of the spherical harmonics this integral vanishes unless = 1 and m = 0. Thus only the states
[
(0)
n10
contribute to the integral and
E
(2)
100
= e
2
c
2
n=2
(0)
n10
[z[
(0)
100
2
_
E
(0)
100
E
(0)
n10
_ . (5.101)
Explicit evaluation of
(0)
n10
[z[
(0)
100
2
is very difcult but clearly it is proportional to some length squared and depends
on n. Thus one can write
(0)
n10
[z[
(0)
100
2
=
_
a
0
Z
_
2
f (n) , (5.102)
with a
0
the Bohr radius, {Actually
(0)
n10
[z[
(0)
100
2
=
1
3
2
8
n
7
(n1)
2n5
(n+1)
2n+5
a
2
0
!} so that
E
(2)
100
= e
2
c
2
_
a
0
Z
_
2

n=2
f (n)
mc
2
2
(Z)
2
_
1
1
n
2
_ =
2e
2
c
2
a
2
0
2
Z
4
mc
2
n=2
n
2
f (n)
(n
2
1)
. (5.103)
Hence E
(2)
100
e
2
c
2
as is to be expected from electromagnetic theory as it is due to an induced dipole moment p = ec
which has an energy
1
2
pc c
2
.
5.3.3 Degenerate example - rst order
Stark effect in hydrogen - excited states The rst excited state has n = 2. If ne-structure (i.e. spin-orbit) effects are
neglected this level is 4-fold degenerate. The degenerate unperturbed eigenfunctions are for n = 2 , = 0, [
200
, and
n = 2, = 1, m = 1, 0, 1, [
211
, [
210
, [
211
. Consequently to nd even the rst-order correction for the excited
states, degenerate perturbation theory needs to be used. At rst sight it appears that one needs to calculate the 16 matrix
elements

m
=
2
m
[ecz[
2m
(in a slight deviation from the usual notation to emphasise the states involved)
in the table below
73
, m 0, 0 1, 0 1, 1 1, 1
0, 0
00

H
00
E
(1)
2
00

H
10
00

H
11
00

H
11
1, 0
10

H
00
10

H
10
E
(1)
2
10

H
11
10

H
11
1, 1
11

H
00
11

H
10
11

H
11
E
(1)
2
11

H
11
1, 1
11

H
00
11

H
10
11

H
11
11

H
11
E
(1)
2
and evaluate a 44 determinant. However an examination of the form of these elements can result in a dramatic reduction
in the number (and effort) of them to evaluate explicitly.
1. The perturbation H
= ecz has odd parity and the eigenstates [

2m
have parity (1)
. Thus
_

2m
_
= 0 if
= . (5.104)
Thus all the diagonal elements of

H
m
vanish, i.e.
00

H
00
=
10

H
10
=
11

H
11
0 =
11

H
11
= 0. In addition
the elements
10

H
11
,
10

H
11
,
11

H
10
,
11

H
11
,
11

H
10
, and
11

H
11
are also zero - resulting in a reduction of 10
elements that do no need to be evaluated explicitly, leaving
, m 0, 0 1, 0 1, 1 1, 1
0, 0 E
(1)
2
00

H
10
00

H
11
00

H
11
1, 0
10

H
00
E
(1)
2
0 0
1, 1
11

H
00
0 E
(1)
2
0
1, 1
11

H
00
0 0 E
(1)
2
.
2. The perturbation H
= ecz commutes with

L
z
whose eigenvalues result in the degeneracy index m, i.e.
_
L
z
,

H
_
=
ec
_
L
z
, z
_
= 0. Hence
0 =
_
L
z
,

H

2m
_
=
_
L
z

H
L
z
_

2m
_
,
0 = h(m
m)
_

2m
_
. (5.105)
Thus if m
,= m then
_

2m
_
= 0. Hence elements
00

H
11
,
00

H
11
,
10

H
11
,
10

H
11
,
11

H
00
,
11

H
10
,
11

H
11
,
11

H
00
,
11

H
10
,
11

H
11
are all zero, giving
, m 0, 0 1, 0 1, 1 1, 1
0, 0 E
(1)
2
00

H
10
0 0
1, 0
10

H
00
E
(1)
2
0 0
1, 1 0 0 E
(1)
2
0
1, 1 0 0 0 E
(1)
2
.
Hence only two non-zero matrix elements exist,
00

H
10
=
_
200

210
_
and
10

H
00
=
_
210

200
_
. But
as

H
is Hermitian then
00

H
10
=
10

H
00
so only one matrix element need be calculated! {As a general policy it is
always advisable to examine the symmetries and origin of the degeneracy before using brute force to evaluate all
the matrix elements}
The 4 4 matrix
_
_
_
_
_
00

H
00
E
(1)
2
00

H
10
00

H
11
00

H
11
10

H
00
10

H
10
E
(1)
2
10

H
11
10

H
11
11

H
00
11

H
10
11

H
11
E
(1)
2
11

H
11
11

H
00
11

H
10
11

H
11
11

H
11
E
(1)
2
_
_
_
_
_
_
_
_
_
c
1
c
2
c
3
c
4
_
_
_
_
= 0 (5.106)
reduces to
_
_
_
_
_
E
(1)
2
00

H
10
0 0
10

H
00
E
(1)
2
0 0
0 0 E
(1)
2
0
0 0 0 E
(1)
2
_
_
_
_
_
_
_
_
_
c
1
c
2
c
3
c
4
_
_
_
_
= 0 (5.107)
74
and only the secular equation
E
(1)
2
00

H
10
10

H
00
E
(1)
2
= 0 (5.108)
needs to be solved. This has two roots E
(1)
2
=
00

H
10
. Explicit evaluation
00

H
10
=
200
[ecz[
210
= ec
200
[z[
210
= ec
_ _
R
20
Y

00
(r cos ) R
21
Y
10
r
2
dr d. (5.109)
of
00

H
10
is now needed. Using expressions for the hydrogenic wavefunctions,
R
20
(r) = 2
_
Z
2a
0
_
3/2
_
1
Zr
2a
0
_
e
Zr/2a
0
, (5.110)
R
21
(r) =
1
3
_
Z
2a
0
_
3/2
Zr
a
0
e
Zr/2a
0
with Z = 1, and z = r cos
00

H
10
= ec
_
2
_
Z
2a
0
_
3/2
_
1
Zr
2a
0
_
e
Zr/2a
0
r
3
1
3
_
Z
2a
0
_
3/2
Zr
a
0
e
Zr/2a
0
dr
_
Y

00
(cos ) Y
10
d. (5.111)
Evaluating the angular integral rst (always a wise move!), and noting that Y
00
=
_
1
4
_
1/2
and Y
10
=
_
3
4
_
1/2
cos ,
_
Y

00
cos Y
10
d =
_
1
4
_
4
3
Y
10
Y
10
d =
1
3
(5.112)
using the orthonormality of the spherical harmonics. Hence
00

H
10
=
2
3
ec
_
Z
2a
0
_
3
_

0
r
3
_
Zr
a
0
__
1
Zr
2a
0
_
e
Zr/a
0
dr
=
2
3
ec
_
Z
2a
0
_
3 _
a
0
Z
_
4
_

0
4
_
1

2
_
e
d
setting = Zr/a
0
. Then
00

H
10
=
1
12
ec
_
a
0
Z
_
__

0
4
e
d
1
2
_

0
5
e
d
_
=
1
12
ec
_
a
0
Z
_
_
4!
1
2
5!
_
= 3ec
a
0
Z
, (5.113)
using the standard integral,
_
0
x
n
e
ax
dx = n!/a
n+1
. Hence the rst-order energy correction is
E
(1)
2
=
00

H
10
= 3ec
a
0
Z
, for = 1, 2, (5.114)
= 0 for = 3, 4. (5.115)
Thus the energy shift is proportional to c, i.e. a linear Stark effect. The linear Stark effect is only non-zero because the
n = 2 state is degenerate.
The correct linear combination of the unperturbed eigenfunctions [
2m
can be found by solving for the coefcients
c,
_
E
(1)
2
00

H
10
10

H
00
E
(1)
2
_
_
c
1
c
2
_
= 0
so
c
1
c
2
=
00

H
10
E
(1)
2
(5.116)
Since
00

H
10
= 3ec
a
0
Z
then for E
(1)
21
= +
00

H
10
= 3ec
a
0
Z
,
c
1
c
2
= 1 and
[
(0)
21
= c
1
[
200
+ c
2
[
210
=
1
2
([
200
[
210
) .
75
For E
(1)
22
=
00

H
10
= 3ec
a
0
Z
,
c
1
c
2
= +1 and
[
(0)
22
= c
1
[
200
+ c
2
[
210
=
1
2
([
200
+[
210
) .
For completeness, E
(1)
23
= 0 and [
(0)
23
= [
211
and E
(1)
24
= 0 and [
(0)
24
= [
211
.
The states in the presence of the electric eld are no longer eigenstates of

L
2
or parity as [
(0)
21
and [
(0)
22
mix these
states and so neither nor parity are "good" quantum numbers. The magnetic quantum number m is still a good quantum
number (as

H
commutes with

L
z
) and the system is invariant under rotations about the z-axis. Physically the external
electric eld denes a preferred direction in space so the system cannot be invariant under under arbitrary rotations (i.e.
L
2
) but still invariant for rotations about this preferred direction (i.e.

L
z
).
The degeneracy of the states [
211
and [
211
is not removed in rst-order. The 4-fold degenerate n = 2 level splits
symmetrically into three sub-levels.
This means that the hydrogen atom (in its unperturbed rst-excited state) behaves as if it had a permanent electric
dipole moment of magnitude 3ea
0
which can be orientated in three different ways,
1. antiparallel to electric eld c for [
(0)
21
=
1
2
([
200
[
210
) (Note energy in electric eld E of a dipole is
W = E.)
2. parallel to electric eld c for [
(0)
22
=
1
2
([
200
+[
210
)
3. two states with zero component along electric eld c for [
(0)
23
= [
211
and [
(0)
24
= [
211
.
5.4 Variational method
There are many stationary state problems which cannot be solved exactly and for which the perturbation approach is un-
satisfactory because the rst-order is not sufciently accurate and higher orders involve enormous (complex) calculations.
Basically the perturbation expansion converges too slowly. The variational method - also known as the Rayleigh-Ritz
method - does not presuppose a knowledge of the simpler problem and is, therefore, very versatile. The method is partic-
ularly suitable for calculating the ground state energies of a system.
The quantum ground state of a system has a special importance. For systems of electrons, in particular, one is often
interested only in the ground state. For example, when quantum mechanics is used to calculate the energies of chemical
reactions, or the properties of materials, excited states are usually irrelevant. Because of this, methods for calculating the
ground-state energy of quantum systems are important. In developing such methods, the variational principle of quantum
mechanics usually plays a key role. This principle states that
The expectation value of the Hamiltonian, evaluated in an arbitrary state, is always greater than or equal
to the ground-state energy.
This principle forms the basis for practical approximate methods of calculating the ground-state energy of a complex
system. The idea is to guess an approximate ground-state wave-function, called the trial wave-function, which is
used to calculate the expectation value of the Hamiltonian. To improve the estimate of the ground-state energy, the
trial wave-function contains variable parameters. The expectation value of the Hamiltonian is minimized with respect to
these parameters, and the minimum energy provides the best estimate of the true ground-state energy, within the given
parameterized family of wavefunctions.
5.4.1 Proof of the variational principle
The Hamiltonian

H possesses a complete set of orthonormal eigenstates
n
,
0
,
1
,
2
, . . . ,
n
[
m
=
nm
and
corresponding energies E
0
, E
1
, E
2
, . . . with the ground state energy E
0
E
1
E
2
. . . .and
H[
n
= E
n
[
n
. (5.117)
Let [ be an approximate wavefunction which satises the correct boundary conditions. Expand [ in terms of the
complete set
n
as
[ =
n
c
n
[
n
(5.118)
76
then the expectation value of the energy in the state [ is
E [] = E

[
,
=
m
c
m
c
n
m
[
H[
n
m
c
m
c
n
E
n
m
[
n
,
E
n
[c
n
[
2
E
n
n
[c
n
[
2
.
and
E
E
0
=
n
[c
n
[
2
E
n
n
[c
n
[
2
E
0
=
n
[c
n
[
2
(E
n
E
0
)
n
[c
n
[
2
. (5.119)
Since every term on the R.H.S. of equ(5.119) is positive then it follows that
E
E
0
,

[
E
0
. (5.120)
and gives an upper bound on the ground state energy. Equality will occur if the trial function [ should be the same as
the correct ground state eigenfunction [
0
in which case [c
0
[ = 1 and c
n
= 0 for all n ,= 0. If
[
H[
|
= E
0
but c
n
,= 0
for more than one n then the ground state is degenerate.
In practice a trial function is chosen which depends on some parameters
1

2
, . . .
s
, i.e. [
T
= [
T
(
1
,
2
, . . .
s
)
and one calculates
E
T
[
T
= E (
1
,
2
, . . .
s
) . (5.121)
This energy is minimized with respect to the variational parameters
1
,
2
, . . .
s
by solving the equations
E (
1
,
2
, . . .
s
)
i
= 0, i = 1, 2, . . . , s. (5.122)
The resulting minimum value of E (
1
,
2
, . . .
s
) represents the best estimate of the ground state energy with that
particular trial function.
Clearly the success of the variational method depends on choosing a trial function that incorporates the correct features
of the ground state. Symmetry and other physical properties of the system are useful guides. The use of powerful com-
puters allows trial functions of great complexity and hence great exibility leading to very accurate results. One should
bear in mind that the variational method optimizes the energy. The trial function is not necessarily a good approximation
to the true wavefunction and may give poor results when used in calculations of quantities other than energy.
If [ is almost the true eigenfunction for the ground state [
0
but with a bit of [
1
mixed in so that [ = [
0
+
a
1
[
1
then the approximation to E
0
is as good as
E
=
E
0
+[a[
2
E
1
1 +[a[
2
= E
0
_
1 +[a[
2
E
1
/E
0
1 +[a[
2
_
(5.123)
E
E
0
= [a[
2
(E
1
E
0
)
1 +[a[
2
. (5.124)
and the error E is quadratic in the coefcients. This means that accurate values of the ground-state energy can often be
obtained with rather simple guesses for the approximate ground-state wavefunction.
5.4.2 Excited States
The variational method can be used to obtain an upper bound for the energy of an excited state provided the trial function
is made orthogonal to all the energy eigenfunctions corresponding to states having an energy lower than the energy level
being considered. Unfortunately often the lower eigenfunctions are not known exactly and one has only approximations
(obtained perhaps from a variational calculation) for these functions. In this case the orthogonality conditions cannot be
achieved exactly. Consequently E
1
may not necessarily provide an upper bound on the energy E
1
. These difculties
do not arise if one looks for an excited energy level whose symmetry properties differ from those of lower-lying levels.
Choosing a trial function with the correct symmetry automatically ensures orthogonality to the lower-lying states and the
variational method gives an upper bound to the energy of the selected excited state.
77
5.4.3 Variational examples
Ground state of hydrogen - actually an unrealistic example as it can be solved exactly! However it is useful to see
how to choose a trial function and perform the integrals and the variation.
The ground state is an s-state with = 0. Thus [ depends only on r with [ (0) ,= 0. At large distances a bound
state wavefunction must vanish. Thus take the trial function as (r) = Ce
r
. Normalization requires
[ = 4C
2
_

0
e
2r
r
2
dr = 1. (5.125)
Using the standard integral
_
0
r
n
e
ar
dr = n!/a
n+1
,
1 = 4C
2
2!
(2)
3
and
C =
_
_
1/2
and
(r) =
_
_
1/2
e
r
. (5.126)
The Hamiltonian is

H =
h
2
2m
e
2
4
0
r
. The expectation value of the potential energy is
[
e
2
4
0
r
[ =
_
_
e
2
4
0
4
_

0
e
2r
rdr, (5.127)
=
_
_
e
2
4
0
4
1
(2)
2
,
=
e
2
4
0
. (5.128)
The following "trick" frequently simplies the kinetic energy calculation. For any function (r) a vector identity is
_
V
(
) d =
_

_
d +
_

() d,
and using Gauss theorem on the L.H.S.
_
V
(
) d =
_
S
(
) ndS =
_

_
d +
_
[()[
2
d.
As 0 sufciently rapidly as r the surface integral vanishes (certainly the case for the exponential function) so
that
_

_
d =
_
[()[
2
d. (5.129)
Using this result on the expectation value of the kinetic energy term
[
h
2
2m
2
[ =
h
2
2m
_
[()[
2
d =
h
2
2m
_
[()[
2
4r
2
dr.
But for the trial function
[()[
2
=
2
=
2
[[
2
.
Hence
[
h
2
2m
2
[ =
h
2
2
2m
_
[[
2
4r
2
dr =
h
2
2
2m
(5.130)
Hence the expectation value of the energy is
E () =
h
2
2
2m

e
2
4
0
(5.131)
and
E ()
=
h
2
m

e
2
4
0
78
and
=
e
2
m
4
0
h
=
1
a
0
, (5.132)
where a
0
is the Bohr radius. Consequently the minimum energy is
E
_
1
a
0
_
=
1
2
e
2
4
0
a
0
=
1
2
a.u. (5.133)
Harmonic oscillator As a simple illustration of how the variational principle works consider the harmonic oscillator,
with the usual Hamiltonian:
H =
h
2
2m
d
2
dx
2
+
1
2
kx
2
. (5.134)
For this system the exact ground-state energy E
0
=
1
2
h is known, where is the frequency = (k/m)
1/2
; also the
exact ground-state wavefunction which is
0
(x) = (m/h)
1
4
exp(mx
2
/2 h). However, to show that the variational
principle really works suppose that this wave function is not known and a trial function is constructed. Because the
potential is symmetric about x = 0 and the wavefunction must satisfy (x) 0 as x then take as a trial
wavefunction as
T
(x) = (2/)
1/4
e
x
2
, (5.135)
with (positive) variational parameter. To demonstrate that the expectation value of

H evaluated with this wavefunction
is always greater than or equal to the exact ground-state energy.
First, verify that the
T
(x) as given by equ (5.135) is already correctly normalized,
_

[
T
(x)[
2
dx =
_
2
_
1/2
_

e
2x
2
dx . (5.136)
For any positive number ,
_

e
x
2
dx =
_
_
1/2
; and
_

x
2
e
x
2
dx =
1
2
_
_
1/2
. (5.137)
In the present case, = 2, so that
_
exp(2x
2
) dx = (/2)
1/2
. This conrms that
_

[
T
(x)[
2
dx = 1 . (5.138)
Now to evaluate the expectation value of H, work out separately the expectation values of the kinetic energy T and
the potential energy V ,
T =
T
[
T[
T
=
_

T
(x)
h
2
2m
d
2
dx
2
_
T
(x) dx (5.139)
In general by integrating by parts
_

T
(x)
T
(x)dx = [
T
(x)
T
(x)]
T
(x)
T
(x)dx
=
_
[
T
(x)[
2
dx (5.140)
Thus
T =
T
[
T[
T
=
h
2
2m
_

_
d
dx
_
_
2
_
1/4
e
x
2
__
2
dx,
=
h
2
2m
_
2
_
1/2
_
1/4
4
2
x
2
e
x
2
dx, (5.141)
=
h
2
2m
_
2
_
1/2
4
2
1
2 (2)
_

2
_
1/2
, (5.142)
T =
h
2
2m
. (5.143)
79
The expectation value of the potential energy is
V =
T
[
V [
T
=
1
2
k
_
2
_
1/2
_

x
2
e
2x
2
dx, (5.144)
=
1
2
k
_
2
_
1/2
1
2 (2)
_

2
_
1/2
, (5.145)
V =
k
8
. (5.146)
Adding T and V the expectation value of the Hamiltonian is
H =
h
2
2m
+
k
8
. (5.147)
Now note that as 0, H k/8 +, and as , H h
2
/2m . Clearly, H must have a
minimum for some value of . To nd this minimum, differentiate with respect to ,
d
d
H =
h
2
2m

k
8
2
= 0, (5.148)
so that
2
=
mk
4h
2
=
m
2
2
4 h
2
. (5.149)
This means that for > 0 there is only a single stationary point, which must be a minimum. The value of that gives
this minimum is
min
= m/2 h. (5.150)
Hence the trial wave-function that gives the minimum value of H is:
min
(x) = (m/h)
1/4
exp(mx
2
/2 h) . (5.151)
This is exactly the correct ground-state wave-function.
From equ (5.147), the value of H given by
min
(x) is:
H
min
=
h
2
2m

m
2h
+
k
8

2 h
m
=
1
4
h +
1
4
h =
1
2
h, (5.152)
which (not surprisingly) is exactly the correct result.
In this case, the variational principle gives exactly the correct result, because the trial wavefunction used is capable of
giving the correct result. If some other kind of trial wavefunction had been taken, this would not have happened.
Ground state of helium atom The Schrdinger equation for the helium atom is impossible to solve exactly because of
the electrostatic repulsion between the two electrons. The Hamiltonian is:
H =
h
2
2m
2
1
h
2
2m
2
2
Ze
2
4
0
_
1
r
1
+
1
r
2
_
+
e
2
4
0
1
r
12
, (5.153)
=
1
2
2
1
1
2
2
2
Z
r
1
Z
r
2
+
1
r
12
. (5.154)
Here, Z is the charge on the nucleus in units of e, which for the helium atom is Z = 2.
This was treated earlier as an example of perturbation theory, where

H was separated into an unperturbed part

H
0
,
in which electron-electron repulsion is ignored, and

H
= e
2
/4
0
r
12
, which is the electron-electron repulsion. A
reasonably good estimate of the ground-state energy can be obtained by using rst-order perturbation theory to calculate
the energy shift due to

H
. The zero-th order wavefunctions used in the perturbation calculation assume that each electron
moves in the Coulomb eld of the nucleus, with Z = 2. In fact the effect of the mutual repulsion of the electrons is to
reduce the effect of the nuclear led experienced by each electron. This screening can be accounted for by assuming an
effective nuclear charge (1 < 2). The ground-state wavefunction of a single electron bound to a nucleus of charge
Z is
100
(r) =
1
_
Z
a
0
_
3/2
e
Zr/a
0
Z
3/2
e
Zr
, (5.155)
80
and the single-electron ground-state energy is
E
1
=
Z
2
e
2
8
0
a
0
=
Z
2
h
2
2ma
2
0
=
Z
2
2
a.u. (5.156)
The two-electron ground-state wave-function is
0
(r
1
, r
2
) =
100
(r
1
)
100
(r
2
) =
1
_
Z
a
0
_
3
e
Z(r
1
+r
2
)/a
0
Z
3
e
Z(r
1
+r
2
)
(5.157)
i.e. the product of the two single-electron wave-functions.
This form of the approximate wave-function immediately suggests how to do a variational calculation. This suggests
that Z is replaced in
0
(r
1
, r
2
) by a variable parameter , an effective charge, so that the trial wave-function is:
T
(r
1
, r
2
) =
1
_

a
0
_
3
e
(r
1
+r
2
)/a
0
, (5.158)
e
(r
1
+r
2
)
. (5.159)
This trial function is already normalized since
100
(r
1
) and
100
(r
2
) were normalized.
The Hamiltonian can be written as
H =
1
2
2
1
1
2
2
2
Z
r
1
Z
r
2
+
1
r
12
(5.160)
=
1
2
2
1
Z
r
1
1
2
2
2
Z
r
2
+
1
r
12
=

H
1
+

H
2
+ V
ee
, (5.161)
where
H
1
=
1
2
2
1
Z
r
1
(5.162)
and similarly for

H
2
.

H
1
can be re-written as
H
1
=
1
2
2
1

r
1
(Z )
r
1
,
=

h
(Z )
r
1
, (5.163)
where

h
is the Hamiltonian for a hydrogenic atomof nuclear charge . Thus by analogy with the hydrogen wavefunctions
100
(r
1
) =
_
_
1/2
e
r
1

h
with eigenvalue E
1
=

2
2n
2
. Hence
H
1
=
_
_
+
_
(Z )
r
1
_
,
H
1
=
2
2
+
_
(Z )
r
1
_
(5.164)
since
_
_
=
_
100
(r
1
)
100
(r
2
) [
[
100
(r
1
)
100
(r
2
)
_
=
_
100
(r
1
) [
[
100
(r
1
)
100
(r
2
)
100
(r
2
)
_
=
2
2
1.
Now
_
(Z )
r
1
_
= (Z )

3
_

0
4
1
r
1
e
2r
1
r
2
1
dr
1
= (Z )

3
_

0
4e
2r
1
r
1
dr
1
= (Z ) 4
3
1
(2)
2
= (Z ) , (5.165)
noting that
100
(r
2
)
100
(r
2
) = 1 and
_
0
re
2r
dr = 1/ (2)
2
and hence
H
1
=
2
2
(Z ) =

2
2
Z. (5.166)
81
By the interchange r
1
r
2
it follows that
H
2
=

2
2
Z
and
H
1
+
H
2
=
2
2Z. (5.167)
The electron-electron interaction expectation value is much more difcult to evaluate as it involves the relative angles
between the position vectors r
1
, r
2
as r
12
= [r
1
r
2
[ = r
2
1
+ r
2
2
2r
1
r
2
cos
12
. The result is
V
ee
=
_
_
2
_

0
dr
1
_

0
dr
2
1
r
12
e
2(r
1
+r
2
)
=
5
8
. (5.168)
Adding the three terms, we have the expectation value of the total Hamiltonian
H =
2
2Z +
5
8
. (5.169)
Since this is a quadratic form that goes to for large , it has a single minimum. To nd this, differentiating with respect
to gives
dH
d
= 2 2Z +
5
8
= 0, (5.170)
so that
= Z
5
16
=
27
16
. (5.171)
The term 5/16 reduces from the value of = Z = 2 that it would have in the absence of the electron-electron
repulsion. The value of the total energy for this value of is:
E
T
=
2
2
_
+
5
16
_
+
5
8
=
2
=
_
27
16
_
2
, (5.172)
and
E
T
= (27/16)
2
2.85 au. (5.173)
The experimental value of the total energy of the He atom is 2.905 au. The variational estimate of 2.85 au which is
slightly above the correct value (as it must be!), but the accuracy is good; the error is less than 2 %.
An example of a more elaborate trial function with two screening constants is
(r
1
, r
2
) = C
_
e
(r
1
+r
2
)/a
0
+ e
(r
2
+r
1
)/a
0
_
and gives E = 2.875 au. With improved variational wavefunctions, which include explicitly the correlation between
the electrons, essentially exact values of the ground-state energy can be obtained.
5.5 Systems of identical particles
A wide range of physical systems, e.g. electrons in atoms or molecules; protons and neutrons in nuclei; cold atoms in a
Bose-Einstein condensate (BEC) consist of many identical particles. If they interact strongly solutions of the Hamiltonian
are very difcult due to the complexity. Generally the difculty increases "exponentially" with the number of particles.
Even for "non-interacting" particles attention must be paid to the fact that the particles are indistinguishable.
In classical dynamics particles move in well dened trajectories and their positions can be tracked with arbitrary preci-
sion. Hence they can be distinguished. In quantum mechanics if their wavefunctions overlap they cannot be distinguished,
especially if the Hamiltonian does not "label" the particles in some way.
82
5.5.1 Identical particles
Identical particles are often called indistinguishable particles to emphasize that they cannot be distinguished by any
physical measurement. For example, two electrons may be at different positions, have different momenta and even have
different z-component of spin as these are dynamical parameters of the system but they have the same charge, mass and
intrinsic magnetic moment.
Any operator representing a physical measurement on the system must remain unchanged if the labels assigned to the
particles are interchanged. For example, if one writes the Hamiltonian of two identical particles as

H (1, 2) then one must
have
H (1, 2) =

H (2, 1) . (5.174)
Such a Hamiltonian is that for the helium atom
H (r
1
, r
2
) =
h
2
2m
2
1
Ze
2
4
0
r
1
h
2
2m
2
2
Ze
2
4
0
r
2
+
e
2
4
0
[r
1
r
2
[
, (5.175)
which is symmetric in the electron coordinates, i.e. invariant under the interchange of the labels 1 2. For such systems
it is meaningless to ask for the probability that helium in its ground state has electron 1 in a volume element dV
a
at
position a and electron 2 in a volume element dV
b
at position b. One can only ask for the probability that one electron is
in dV
a
and the other in dV
b
.
An immediate consequence of the symmetry in equ(5.174) is that the eigenvalues of

H are degenerate. Suppose that
H (1, 2) (1, 2) = E (1, 2) . (5.176)

Interchanging 1 2 gives
H (2, 1) (2, 1) = E (2, 1)

but

H (1, 2) =

H (2, 1) so
H (1, 2) (2, 1) = E (2, 1) . (5.177)

Thus if (1, 2) is an eigenfunction with energy eigenvalue E, then so also is (2, 1). Hence the eigenvalue is two-fold
degenerate. This is known as exchange degeneracy. Thus (1, 2), (2, 1) or any linear combination of them is also an
eigenfunction. Two particular combinations are
S
=
1
2
[ (1, 2) + (2, 1)] (5.178)
and
A
=
1
2
[ (1, 2) (2, 1)] , (5.179)
which are, respectively, symmetric and antisymmetric under the interchange 1 2.
It follows from the Schrdinger equation that the symmetry of the wavefunction is a constant of the motion. In fact
physically acceptable wavefunctions representing identical particles must be either symmetric or antisymmetric. Dene
a particle exchange operator P
12
such that P
12
acting on any function of the variables of the two particles interchanges
their labels. Then
P
12

H (1, 2) =

H (2, 1)
also
P
12
_
H (1, 2) (1, 2)
_
= H (2, 1) (2, 1) ,
=

H (1, 2) P
12
(1, 2) ,
so
_
P
12
,

H (1, 2)
_
(1, 2) = 0. (5.180)
But (1, 2) is any two-particle wavefunction so
_
P
12
,

H (1, 2)
_
= 0. (5.181)
Thus P
12
and

H (1, 2) are compatible and have a common set of eigenfunctions. If (1, 2) is such an eigenfunction,
then to be an eigenfunction of P
12
it satises
P
12
(1, 2) = p(1, 2) ,
(2, 1) = p(1, 2) .
83
Operate with P
12
again
P
12
(2, 1) = (1, 2) = pP
12
(1, 2) = p
2
(1, 2) (5.182)
so
p
2
= 1,
p = 1. (5.183)
Hence
(1, 2) = (2, 1)
and is either symmetric or antisymmetric under the interchange 1 2. Not only the Hamiltonian but any operator
representing a physical property of the system must be symmetric under 1 2 and must therefore commute with P
12
.
Thus whatever measurement is made in the system the resulting wavefunction will be an eigenfunction of P
12
. No loss of
generality occurs if one assumes that the wavefunction always has this property.
This property can be extended to a many-body wavefunction for identical particles. The wavefunction must be either
symmetric or antisymmetric with respect to interchange of any pair of particles. It follows that every particle belongs
to one of two classes depending on whether a number of them is symmetric or antisymmetric under particle exchange.
Experimental evidence is that for a given kind of particle the symmetry is always of one kind only. For example, electrons,
positrons, protons, neutrons, neutrinos are described by antisymmetric wavefunctions; photons, pions, K-mesons,
4
He
(alpha particles) have symmetric wavefunctions. Particles with antisymmetric wavefunctions are called fermions, while
those with symmetric wavefunctions are bosons. The choice of symmetry is closely related with the value of the total spin
of the particle. It an empirical fact that fermions have spin s =
1
2
,
3
2
,
5
2
, . . ., i.e. odd half-integer values, and bosons have
s = 0, 1, 2, 3, . . ., i.e. integral values. This one-to-one correspondence between spin and the interchange symmetry can
be shown to be a necessary consequence of a relativistic quantum eld theory of identical particles (and the spin-statistics
theorem).
5.5.2 Exclusion principle
Consider two identical particles which interact very weakly with one another. To a rst approximation one may neglect
their mutual interaction and hence write the Hamiltonian

H (1, 2) as the sum of the two single-particle Hamiltonians as
H (1, 2) =

H
1
(1) +

H
2
(2) . (5.184)
If

H
1
(1) has a complete set of eigenfunctions
a
(1) with eigenvalues E
a
, such that

H
1
(1)
a
(1) = E
a
a
(1) and
similarly for

H
2
(2)
b
(2) = E
b
b
(2) then
H (1, 2)
a
(1)
b
(2) =
_
H
1
(1) +

H
2
(2)
_
a
(1)
b
(2) = (E
a
+ E
b
)
a
(1)
b
(2) . (5.185)
The symmetric and antisymmetric eigenfunctions of equ(5.184) are
S
A
(1, 2) =
1
2
[
a
(1)
b
(2)
a
(2)
b
(1)] . (5.186)
For fermions, the antisymmetric wavefunction vanishes identically if states a = b, i.e. two identical fermions cannot exits
in the same single-particle state. This is known as the Pauli Exclusion Principle.
{The Exclusion Principle actually states that each single-particle eigenfunction can only be used once in constructing
products, linear combinations of which form the total wavefunction.}
This obviously generalizes to more than two fermions; such that in a quantum system at most one particle can occupy
any one single-particle state. In this form it is restricted to non-interacting particles. The requirement of antisymmetric
wavefunctions for identical particles with odd half-integer spin always holds and represents a general statement of the
Pauli Exclusion Principle.
No such restriction applies to bosons. In this case states a = b are allowed and
(1, 2) =
a
(1)
a
(2) (5.187)
and any number of bosons can occupy the same single-particle state. The helium atom is a good example to illustrate the
ideas of symmetric and antisymmetric wavefunctions.
5.5.3 N-particle states
The discussion can be extended to N-particle Hamiltonians for non-interacting particles,
H =
h
2
2m
N
i=1
2
i
+
N
i=1
V
i
(r
i
) =
N
i=1
h
i
(5.188)
84
where

h
i
=
h
2
2m
2
i
+ V
i
(r
i
) is a single particle Hamiltonian. The physics becomes more interesting - genuine many-
body - if there are particle-particle interactions
i=j
V (r
ij
) e.g. the Coulomb interaction
V (r
ij
) =
Z
2
[r
i
r
j
[
, (5.189)
or a short-range spin exchange as in a spin chain,
V
ij
= KS
i
S
j
i,j1
. (5.190)
For the i-th particle
h
i
n
i
(r
i
) =
i
n
i
(r
i
) (5.191)
with
n
(r
i
) [
m
(r
i
) =
nm
. If the total Hamiltonian is separable the solutions are the products
n
1
n
2
...n
N
(r
1
, r
2
, . . . r
N
) =
n
1
(r
1
)
n
2
(r
2
) . . .
n
N
(r
N
) , (5.192)
and
H
n
1
n
2
...n
N
(r
1
, r
2
, . . . r
N
) = E
j
n
1
n
2
...n
N
(r
1
, r
2
, . . . r
N
) (5.193)
_
N
i=1
h
i
_
n
1
(r
1
)
n
2
(r
2
) . . .
n
N
(r
N
) =
_
N
i=1
(j)
i
_
n
2
(r
2
) . . .
n
N
(r
N
) (5.194)
The j-th energy eigenvalue of the system is
E
j
=
N
i=1
(j)
i
. (5.195)
Owing to the separability of

H then if
H
n
1
n
2
...n
N
(r
1
, . . . r
k
, . . . r
, . . . r
N
) = E
j
n
1
n
2
...n
N
(r
1
, . . . r
k
, . . . r
, . . . r
N
) ; (5.196)
E
j
=
1
+ . . .
k
. . .
. . .
N
(5.197)
the interchange of the position of two particles (e.g. r
k
r
) leaves the energy unchanged,
H
n
1
n
2
...n
N
(r
1
, . . . r
, . . . r
k
, . . . r
N
) = E
j
n
1
n
2
...n
N
(r
1
, . . . r
, . . . r
k
, . . . r
N
) ; (5.198)
E
j
=
1
+ . . .
. . .
k
. . .
N
. (5.199)
This will hold for Hamiltonians which are symmetric with respect to exchange whether with or without interaction, e.g.
H (r
1
, r
2
) =

h
1
(r
1
) +

h
2
(r
2
) + V (r
1
, r
2
)
if V (r
1
, r
2
) = V (r
2
, r
1
).
Let

P
k
be the operator which interchanges particle k with particle . Then applied to equ(5.196) gives
P
k

H
n
1
n
2
...n
N
(r
1
, . . . r
k
, . . . r
, . . . r
N
) = E
j

P
k
n
1
n
2
...n
N
(r
1
, . . . r
k
, . . . r
, . . . r
N
)
P
k

H
n
1
n
2
...n
N
(r
1
, . . . r
k
, . . . r
, . . . r
N
) = E
j
n
1
n
2
...n
N
(r
1
, . . . r
, . . . r
k
, . . . r
N
) , (5.200)
but equ(5.198) can be re-written as
H

P
k
n
1
n
2
...n
N
(r
1
, . . . r
k
, . . . r
, . . . r
N
) = E
j
n
1
n
2
...n
N
(r
1
, . . . r
, . . . r
k
, . . . r
N
) . (5.201)
But the R.H.S. of equ(5.200, 5.201) are the same so from the L.H.S. one has
_
H

P
k

P
k

H
_
n
1
n
2
...n
N
(r
1
, . . . r
k
, . . . r
, . . . r
N
) = 0,
_
H

P
k

P
k

H
_
= 0,
_
H,

P
k
_
= 0, (5.202)
so

P
k
and

H commute and there are simultaneous eigenstates of

P
k
and

H for all k, combinations.
Also two operations of exchange of particles k and
P
2
k
n
1
n
2
...n
N
(r
1
, . . . r
k
, . . . r
, . . . r
N
) =
n
1
n
2
...n
N
(r
1
, . . . r
k
, . . . r
, . . . r
N
) (5.203)
85
leaves the wavefunction unchanged. Hence the eigenvalue of

P
2
k
is 1 and that of

P
k
are 1. Hence the states are either
symmetric
P
k
(S)
k
= +
(S)
k
(5.204)
or antisymmetric
P
k
(A)
k
=
(A)
k
. (5.205)
Explicitly for two non-interacting indistinguishable particles the symmetric wavefunction is
(S)
(r
1
, r
2
) =
1
2
[
1
(r
1
)
2
(r
2
) +
1
(r
2
)
2
(r
1
)] (5.206)
and the antisymmetric one is
(A)
(r
1
, r
2
) =
1
2
[
1
(r
1
)
2
(r
2
)
1
(r
2
)
2
(r
1
)] . (5.207)
With interactions the individual wavefunctions are more complex (not necessarily products) but the same symmetry
applies, e.g.
(A)
(r
1
, r
2
) =
1
2
[(r
1
, r
2
) (r
2
, r
1
)] (5.208)
and often
(r
1
, r
2
) =
m,n
c
nm
m
(r
1
)
n
(r
2
) . (5.209)
Symmetrizing for N > 2 particles
Anti-symmetrizing the wavefunction when there are more than two particles is important when doing molecular or atomic
structure calculations, e.g. "Hartree-Fock", ab-initio calculations in quantum chemistry. The generalization for N > 2 is
the Slater Determinant,
(A)
n
1
n
2
...n
N
(r
1
, r
2
, . . . r
N
) =
1
N!
n
1
(r
1
)
n
1
(r
2
)
n
1
(r
N
)
n
2
(r
1
)
n
2
(r
2
)
n2
(r
N
)
.
.
.
.
.
.
.
.
.
.
.
.
n
N
(r
1
)
n
N
(r
2
)
n
N
(r
N
)
(5.210)
which makes use of the rules for evaluating determinants. For example if N = 3 the antisymmetric state is
(A)
n
1
n
2
n
3
(r
1
, r
2
, r
3
) =
1
6
[
n
1
(r
1
)
n
2
(r
2
)
n
3
(r
3
)
n
1
(r
2
)
n
2
(r
1
)
n
3
(r
3
)
+
n
1
(r
2
)
n
2
(r
3
)
n
3
(r
1
)
n
1
(r
3
)
n
2
(r
2
)
n
3
(r
1
) (5.211)
+
n
1
(r
3
)
n
2
(r
1
)
n
3
(r
2
)
n
1
(r
1
)
n
2
(r
3
)
n
3
(r
2
)]
or equivalently
(A)
n
1
n
2
n
3
(r
1
, r
2
, r
3
) =
1
perm
(1)
P
P
n
1
(r
1
)
n
2
(r
2
)
n
3
(r
3
) (5.212)
where the summation is over all permutations of 1, 2, 3.
The symmetric wavefunction is
(S)
n
1
n
2
n
3
(r
1
, r
2
, r
3
) =
1
perm
P
n
1
(r
1
)
n
2
(r
2
)
n
3
(r
3
) (5.213)
which replaces all the signs in equ(5.212) with + signs.
Consider a two-electron state . Its wavefunction must be antisymmetric with respect to interchange of both position
and spin coordinates, i.e. exchange of the particles. The wavefunction is
n
1
n
2
(r
1
, s
1
, r
2
, s
2
) =
1
2
[
n
1
(r
1
, s
1
)
n
2
(r
2
, s
2
)
n
1
(r
2
, s
2
)
n
2
(r
1
, s
1
)] (5.214)
where s
1
, s
2
denote "spin-up" or "spin-down" states. If the space-spin states are the same, i.e. n
1
= n
2
, s
1
= s
2
,
n
1
(r, s
1
) =
n
2
(r, s
2
) then = 0 for all r. If s
1
,= s
2
there is no problem, nor if n
1
,= n
2
. The same applies to the
Slater determinant. If two rows of a determinant are identical it evaluates to zero. To include spin, replace n by n, s.
86
5.5.4 Helium atom and the exchange force
The Hamiltonian for the helium atom is
H =
h
2
2m
2
1
h
2
2m
2
2
Ze
2
4
0
r
1
Ze
2
4
0
r
2
+
e
2
4
0
[r
1
r
2
[
if small corrections due to spin-spin and spin-orbit interactions are neglected. Since

H is spin-independent, spin operators
S
2
= (s
1
+s
2
)
2
and S
z
commute with

H and so are constants of the motion. Operators L
2
, L
z
also commute with

H,
so the simultaneous eigenfunctions of

H , L
2
, L
z
, S
2
, S
z
can be written as products of functions. The two electrons can
be in spin state S = 0, a singlet,
[0, 0 =
1
2
(
1
2
) , (5.215)
or S = 1, a triplet,
[1, 1 =
1
2
, (5.216)
[1, 0 =
1
2
(
1
2
+
1
2
) , (5.217)
[1, 1 =
1
2
. (5.218)
The singlet is antisymmetric with respect to particle exchange, 1 2; the triplet is symmetric. Since the overall
wavefunction must be antisymmetric with respect to particle exchange, 1 2 the products are spatially symmetric wave-
functions
(S)
(r
1
, r
2
) with antisymmetric spin wavefunctions,
(A)
(s
1
, s
2
), i.e. of the form
(S)
(r
1
, r
2
)
(A)
(s
1
, s
2
)
or spatially antisymmetric wavefunctions
(A)
(r
1
, r
2
) with antisymmetric spin wavefunctions,
(S)
(s
1
, s
2
), i.e. of the
form
(A)
(r
1
, r
2
)
(S)
(s
1
, s
2
). The lowest energy single particle state would have n
1
= 1, n
2
= 1 and be the spatially
symmetric state
(S)
11
(r
1
, r
2
) =
100
(r
1
)
100
(r
2
) (5.219)
which is combined with an antisymmetric singlet spin state [0, 0 to give
(Sing)
(r
1
, r
2
) =
100
(r
1
)
100
(r
2
)
1
2
(
1
2
) (5.220)
as the ground state eigenfunction.
For the rst excited state n
1
= 1, ( = 0, m = 0) and n
2
= 2, ( = 0, 1) so there are two possibilities,
(S)
(r
1
, r
2
) =
1
2
[
100
(r
1
)
2m
(r
2
) +
2m
(r
1
)
100
(r
2
)] , (5.221)
(A)
(r
1
, r
2
) =
1
2
[
100
(r
1
)
2m
(r
2
)
2m
(r
1
)
100
(r
2
)] . (5.222)
The symmetric spatial wavefunction
(S)
must combine with the antisymmetric spin function [0, 0 to give
(Sing)
(r
1
, r
2
) =
(S)
(r
1
, r
2
) [0, 0 (5.223)
and the antisymmetric spatial function
(A)
combines with the symmetric triplet spin functions [1, 1, [1, 0, [1, 1 to
give
(Trip)
(r
1
, r
2
) =
_
_
_
(A)
(r
1
, r
2
) [1, 1
(A)
(r
1
, r
2
) [1, 0
(A)
(r
1
, r
2
) [1, 1
. (5.224)
A perturbation calculation of the rst excited state should use degenerate perturbation theory. However
(a) the perturbation
e
2
4
0
|r
1
r
2
|
does not involve spin so all perturbation matrix elements between different spin states
vanish as
_
s, m
s
[
1
r
12
[s
, m
s
_
=
ss

m
s
m
s
, i.e. S is a constant.
(b) the total angular momentum cannot change as there are no external torques, so matrix elements between different
-values are also zero.
(c) the perturbation is symmetric in exchange 1 2 so matrix elements between states of different symmetry also
vanish.
87
The expectation value of the inter-electron repulsion for the singlet and triplet states are
E
(S,T)
=
_
(Sing,Trip)
(r
1
, r
2
) [
1
r
12
[
(Sing,Trip)
(r
1
, r
2
)
_
(5.225)
=

1
2
[
100
(r
1
)
2m
(r
2
)
2m
(r
1
)
100
(r
2
)]
1
r
12
2
[
100
(r
1
)
2m
(r
2
)
2m
(r
1
)
100
(r
2
)]
, (5.226)
=
1
2
100
(r
1
)
2m
(r
2
)
1
r
12

100
(r
1
)
2m
(r
2
)
_
+
_
2m
(r
1
)
100
(r
2
)
1
r
12

2m
(r
1
)
100
(r
2
)
_
2m
(r
1
)
100
(r
2
)
1
r
12

100
(r
1
)
2m
(r
2
)
100
(r
1
)
2m
(r
2
)
1
r
12

2m
(r
1
)
100
(r
2
)
_
. (5.227)
The rst two terms are symmetric in 1 2 and so equal, similarly the last two terms are symmetric in 1 2 and so
equal so that
E
(S,T)
=
_
100
(r
1
)
2m
(r
2
)
1
r
12
100
(r
1
)
2m
(r
2
)
_
100
(r
1
)
2m
(r
2
)
1
r
12
2m
(r
1
)
100
(r
2
)
_
(5.228)
E
(S,T)
=
_
[
100
(r
1
)[
2
1
r
12
[
2m
(r
2
)[
2
dr
1
dr
2
100
(r
1
)
2m
(r
2
)
1
r
12
2m
(r
1
)
100
(r
2
) dr
1
dr
2
. (5.229)
The rst integral
J =
_
[
100
(r
1
)[
2
1
r
12
[
2m
(r
2
)[
2
dr
1
dr
2
(5.230)
is called the direct integral. It is clearly positive and represents the mutual electrostatic energy of two charge densities
[
100
(r
1
)[
2
and [
2m
(r
2
)[
2
separated by a distance r
12
, i.e. the Coulomb repulsive energy and thus has a classical
analogue.
The second term
K =
_

100
(r
1
)
2m
(r
2
)
1
r
12
2m
(r
1
)
100
(r
2
) dr
1
dr
2
(5.231)
has no classical analogue. Its origin lies in the Pauli Principle and is referred to as the exchange integral. Because of this
exchange contribution the singlet and triplet states are no longer degenerate, with
E
(S)
= J + K. (5.232)
E
(T)
= J K. (5.233)
The J term is obviously positive. For the K term, it may be evaluated for the hydrogenic wavefunctions and is also
positive. For = n 1 this is (obvious) because the wavefunctions appearing in K have no nodes. That the triplet state
has lower energy than the singlet state can be seen on qualitative grounds. For the triplet state the spatial wavefunction
is antisymmetric and so the electrons tend to stay away from each other (
(A)
0; r 0; r
12
Trip
> r
12
Sing
); the
probability of nding the two electrons at the same place is low. This makes the repulsion between the electrons in the
triplet state less than in the singlet., V
12
Trip
< V
12
Sing
. An interesting feature of the singlet-triplet energy splitting
is that although the perturbing interaction
e
2
4
0
|r
1
r
2
|
is spin-independent the symmetry of the wavefunction make the
potential behave as if it were spin-dependent.
Note that S
2
= (s
1
+s
2
) = s
2
1
+s
2
2
+ 2s
1
s
2
so that
2s
1
s
2
=
S
2
_
s
2
1
_
s
2
2
_
= S (S + 1) h
2
3
4
h
2
3
4
h
2
,
= S (S + 1) h
2
3
2
h
2
. (5.234)
Thus as S = 0 for the singlet and S = 1 for the triplet,
2
h
2
s
1
s
2
=
1
2
1

2
= S (S + 1)
3
2
=
_

3
2
sin glet
1
2
triplet
, (5.235)
88
Hence the energy shift is
E = J
1
2
(1 +
1

2
) K. (5.236)
An effective "exchange interaction" term appears having a similar magnitude to the electrostatic Coulomb repulsion.
Note that the magnetic dipole interaction which gives rise to the spin-orbit coupling is not responsible for the spin-spin
coupling - it is too weak. The exchange force arises owing to symmetry effects alone. It gives rise to ferromagnetism via
an interaction of the form

V =
N
i=1
A
i

i+1
between neighbouring atoms in a ferromagnet. (In ferromagnetism spins
tend to align because K
1

2
lowers the energy if the spins are aligned and K > 0).
For helium the direct integral depends on the average separation of the electrons. Since the probability density of the
electrons in the 100 state is spherically symmetric, the average distance between the electrons depends only on the radial
probability of the two particles, i.e. r
2
R
2
n
(r). Thus the average separation is greater for = 0 , = 0 than = 0,
= 1 (see radial density plots, e.g. Eisberg p 306), so that r
12
00
> r
12
01
, and hence V
12
00
< V
12
01
and the
p-state energy is increased more than the s-state. Singlets lie above triplets in a multiplet - an example of Hunds rule;
other things being equal, state of highest spin has lowest energy.
89
90
APPENDIX A. PROBLEM SHEETS
Appendix A
Problem Sheets
The problem sheets and model answers for the course are reproduced here
91
PHAS3226: Quantum Mechanics APPENDIX A. PROBLEM SHEETS
92
PHAS3226: Quantum Mechanics Problem Sheet 1
Quantum Mechanics: PHAS3226, Winter 2011
Problem Sheet 1
Hand in your answers by Tuesday 25 October, either at the lectures or Dr Bowlers pigeonhole in Physics. Marks per
section are shown in square brackets.
1. (a) In quantum theory, physical observables are represented by Hermitian operators. Using Dirac notation, dene
the meaning of the term Hermitian. [2]
(b) A Hermitian operator

A has eigenvalues a
n
and normalised eigenvectors [
n
. Demonstrate that if two eigen-
values a
m
and a
n
are different then the associated eigenvectors are orthogonal:
m
[
n
= 0. [2]
(c) A general normalised state vector [ is a linear superposition of the normalised eigenstates [
n
:
[ =
n
c
n
[
n
When a measurement of the observable A is made in the state [, the probability of nding the value an is
denoted by p
n
. Explain why p
n
is equal to [c
n
[
2
. (Assume that the expectation value of A in the given state is
A = [
A[.) [2]
2. The observable A is represented by operator

A. The eigenvectors of

A are [
1
and [
2
which are orthonormal.
The corresponding eigenvalues are +1 and 1 respectively. The system is prepared in a state [ given by
[ =
4
10
[
1
+ c[
2
.
(a) If the state [ is normalized what is the value of [c[? [2]
(b) Calculate the numerical value of the expectation value of

A in the state [. [4]
(c) With the system in the state [ a measurement of observable A is made. What is the probability of obtaining
the eigenvalue 1? [2]
(d) Immediately after the measurement is made, with the result 1, what is now the expectation value of

A?
Explain your answer briey. [2]
3. An electron in a hydrogen atom is described by the wavefunction:
(r) (
100
(r) + 2
210
(r) 3
322
(r) 4
411
(r))
where
nlm
l
are the orthonormal eigenfunctions of the hydrogen atom with n, l, m
l
the usual quantum numbers.
(a) Normalise this wavefunction [2]
(b) Evaluate the probability that the electron is measured to be in the ground state [2]
(c) Evaluate the expectation value of the energy (you may assume that the energy levels of the hydrgen atom are
given by E
n
= 1/2n
2
a.u.) [2]
(d) Evaluate the expectation value of L
z
[3]
(e) Evaluate the probability that the electron is found to have orbital angular momentum l = 1 [2]
(f) Evaluate the probability that, having obtained l = 1, a subsequent measurement of L
z
obtains the value
m
l
= 0 [3]
4. A bound quantum system has a complete set of orthonormal, non-degenerate energy eigenfunctions u
n
with differ-
ent energy eigenvalues E
n
. The operator

B corresponds to some other observable and is such that:
Bu
1
= u
2
; Bu
2
= u
1
; Bu
n
= 0, n 3
(a) FInd the complete orthonormal set of eigenfunctions of the operator

B [Hint: expand out the eigenvectors of
B in terms of u
i
and do not neglect any solutions] [6]
(b) If B is measured and found to have the eigenvalue +1, what is the expectation value of the energy in the
resulting state ? [4]
93
Problem Sheet 1 Worked Answers
1. (a) Denition of Hermitian: an operator

A is Hermitian if, for any two arbitrary state vectors [
1
and [
2
:
1
[
A[
2
=
2
[
A[
1
(A.1)
(b) For two eigenvectors [
m
and [
n
:
m
[
A[
n
= a
n
m
[
n
(A.2)
Take complex conjugate and use the fact that

A is Hermitian:
m
[
A[
n
=
n
[
A[
m
= a
m
[
n
= a
n
n
[
m
(A.3)
since a
n
= a
n
and
m
[
n
=
n
[
m
by denition. But we also have:
n
[
A[
m
= a
m
n
[
m
It
follows that:
a
n
n
[
m
= a
m
n
[
m
(A.4)
Since a
m
,= a
n
, we must have
n
[
m
= 0.
(c) From meaning of expectation value:
A =
n
p
n
a
n
(A.5)
But the expectation value is also given by:
A = [
A[ =
m,n
c
m
c
n
m
[
A[
n
=
m,n
c
m
c
n
a
n
m
[
n
=
n
[c
n
[
2
a
n
(A.6)
Hence: for any state [, we must have p
n
= [c
n
[
2
.
2. (a) Normalization requires 1 =
16
100
+[c[
2
, so c =
84
10
0.9165.
(b) The expectation values of

A in state [ is
_
[
A[
_
=
_
4
10
1
[ + c
2
[
_

A
_
4
10
[
1
+ c[
2
_
(A.7)
Using the eigenvalue equations

A[
1
= +[
1
and

A[
2
= [
2
_
[
A[
_
=
_
4
10
1
[ + c
2
[
__
4
10
[
1
c[
2
_
(A.8)
=
16
100
[c[
2
=
16
100

84
100
=
68
100
(A.9)
Alternatively, and far quicker, can calculate the expectation value from
n
[c
n
[
2
a
n
i.e.
16
100
(1) +
84
100
(1) =
68
100
.
(c) The probability is the modulus squared of the coefcient of the state [
2
, i.e.
84
100
.
(d) Immediately after the measurement the system is in state [
2
. Hence the expectation value for

A is 1.
3. (a) The wavefunction is an expansion in terms of orthonormal eigenfunctions for the hydrogen atom of the form
n
c
n
n
. Normalization requires

n
[c
n
[
2
= 1. In this case

n
[c
n
[
2
= 30, so the normalized wavefunc-
tion is
(r) =
1
30
(
100
+ 2
210
3
322
4
411
) . (A.10)
(b) The probability that the system is measured to be in an energy state given by wavefunction
nm
is [c
n
[
2
.
Therefore, the probability that the electron is measured to be in the ground state,
100
, is
30
2
=
1
30
.
(c) The expectation value of the energy is
E =
n
[c
n
[
2
E
n
=
1
30
_
1
2
_
+
4
30
_
1
8
_
+
9
30
_
1
18
_
+
16
30
_
1
32
_
=
1
15
au = 0.066667au.
(A.11)
94
(d) The expectation value of L
z
, is
L
z
=
n
[c
n
[
2
m
n
h =
1
30
(0) +
4
30
(0) +
9
30
(2) +
16
30
(1) =
1
15
h. (A.12)
(e) Measurement of = 1 select states
210
and
411
. Thus the probability is
30
2
+
30
2
=
2
3
(f) Having obtained = 1 the resulting state
2
210
4
411
. Normalization of this state gives the wave-
function after the measurement as
=
1
20
(2
210
4
411
), so the probability that a measurement of L
z
obtains the value m
= 0 is
20
2
=
1
5
.
4. (a) The eigenvalues,
i
for B will satisfy B
i
= b
i
i
. We can expand these eigenvalues in terms of the orthonor-
mal energy eigenfunctions u
n
:
i
=
n
c
in
u
n
(A.13)
Substituting this in, and using the information given for B operating on u
n
gives:
B
n
c
in
u
n
= c
i1
u
2
+ c
i2
u
1
= b
i
n
c
in
u
n
(A.14)
Take the scalar product on the left with u
k
; I will shift to Dirac notation for simplicity:
c
i1
u
k
[u
2
+ c
i2
u
k
[u
1
= b
i
n
c
in
u
k
[u
n
(A.15)
Now, we know that u
i
[u
j
=
ij
, so
c
i2
= b
i
c
i1
, k = 1 (A.16)
c
i1
= b
i
c
i2
, k = 2 (A.17)
0 = b
i
c
ik
, k 3 (A.18)
We can now solve for b
i
and c
in
. If we add Eq. (A.16) and Eq. (A.17), we get:
c
i1
+ c
i2
= b
i
(c
i1
+ c
i2
) (A.19)
b
i
= 1, c
i1
= c
i2
, c
in
= 0n 3 (A.20)
Setting i = 1, we have b
1
= 1 and
1
= (u
1
+ u
2
)/
2 (following normalisation). If we subtract Eqs. (A.16)

and (A.17), then we nd instead:
c
i1
c
i2
= b
i
(c
i2
c
i1
) (A.21)
b
i
= 1, c
i1
= c
i2
, c
in
= 0n 3 (A.22)
Taking i = 2, we have b
1
= 1 and
1
= (u
1
u
2
)/
2 (following normalisation). Is that all ? No. We also

have a solution for b
i
= 0, which yields c
i1
= c
i2
= 0. From the conditions of the question, we must have
c
i3
= 1 for i 3, which gives
i
= u
i
, i 3.
(b) If the eigenvalue +1 has been obtained, then we know that the system has state
1
. The expectation value of
the energy is then:
E =
1
[
H[
2
(A.23)
= (u
1
+ u
2
)/
2[
H[(u
1
+ u
2
)/
2 (A.24)
=
1
2
(u
1
[
H[u
1
+u
2
[
H[u
2
=
1
2
(E
1
+ E
2
) (A.25)
as we have that u
i
[u
j
=
ij
and the u
i
are energy eigenfunctions.
95
96
Problem Sheet 2
Hand in your answers by Tuesday 8 November, either at the lectures or Dr Bowlers pigeonhole in Physics. Marks per
1. Starting from the basic commutator of position and momentum, [ x, p
x
] = ih, show that
(a)
_
x, p
2
x
= 2ih p
x
, [1]
(b)
_
x, p
3
x
= 3ih p
2
x
, [1]
(c) Prove by induction that [ x, p
n
x
] = ihn p
n1
x
where n is a positive integer. [2]
(d) By expressing the position operator x and momentum operator p within a harmonic oscillator in terms of the
raising and lowering operators a
+
and a
respectively, show that in any harmonic oscillator state [n that

x = p = 0 and

x
2
_
=
_
n +
1
2
_ _
h
m
_
. (You may assume the properties a
+
[n =
n + 1[n + 1 and [2]

[2]
a
[n =
n[n 1.)
(e) Using the Hamiltonian for the harmonic oscillator, or otherwise, show that the expectation value of the kinetic
energy is equal to the expectation value of the potential energy and half that of the total energy. [2]
2. An harmonic oscillator is in the state [ =
1
6
[0 +
2
6
[1 +
1
6
[2.
(a) Evaluate in this state [ the expectation values of x, p
x
. [4]
(b) Show, by direct substitution into the eigenvalue equation a
[ = [, that the state

[ = e
||
2
/2
n=0
n!
[n
is the eigenfunction of the lowering operator a
with eigenvalue . [4]

(c) Show that the expectation value of the position operator x in this state is given by (you may assume that the
state [ is normalized, as it is!) [2]
x =
_
2h
m
_
1/2
Re () .
3. The ground state wave function for a particle of mass m moving with energy E in a one-dimensional harmonic
oscillator potential with classical frequency is:
u
0
(x) = N
0
e
2
x
2
/2
where N
0
is some normalisation constant and =
_
m/h.
(a) For the ground state, show explicitly that the quantum mechanical expectation values x and p are both
zero. [3]
(b) If the uncertainties x and p are given by:
(x)
2
= x
2
x
2
; (p)
2
= p
2
p
2
obtain an expression for the expectation value of the energy in terms of the uncertainties. [3]
(c) If xp = c for c a constant, deduce a value for c by minimising the ground state energy. What signicance
does the value of c have in light of the uncertainty principle ? [4]
97
1. (a) The property of commutator algebra to use is
[A, BC] = B[A, C] + [A, B] C (A.26)
Then
_
x, p
2
x
= p
x
[ x, p
x
] + [ x, p
x
] p
x
= p
x
ih + ih p
x
= 2ih p
x
. (A.27)
(b) Similarly
_
x, p
3
x
= p
x
_
x, p
2
x
+ [ x, p
x
] p
2
x
= p
x
2ih p
x
+ ih p
2
x
= 3ih p
2
x
. (A.28)
(c) If we assume that the relation is true for n, i.e.:
[ x, p
n
x
] = ihn p
n1
x
(A.29)
then we can write:
_
x, p
n+1
x
= p
x
[ x, p
n
x
] + [ x, p
x
] p
n
x
= p
x
ihn p
n1
x
+ ih p
n
x
= (n + 1)ih p
n
x
, (A.30)
so it is also true for n + 1. But we know its true for n = 1, and we have shown it for n = 2, 3 above, so by
induction it is true for all n.
(d) Raising and lowering operators are a
+
=
_
1
2 hm
_
1/2
(m x i p) and a
=
_
1
2 hm
_
1/2
(m x + i p). By
adding and then re-arranging we have
x =
_
h
2m
_
1/2
( a
+
+ a
) (A.31)
and by subtracting we have
p = i
_
mh
2
_
1/2
( a
+
a
) . (A.32)
In the harmonic oscillator state [n
x =
_
h
2m
_
1/2
n[ a
+
+ a
[n . (A.33)
But a
+
[n =
n + 1[n + 1 and a
[n =
n[n 1 so
n[ a
+
+ a
[n = n[
n + 1[n + 1 +n[
n[n 1 = 0 (A.34)
since the states [n are orthonormal, k[n =
kn
.
Similarly
p = i
_
mh
2
_
1/2
n[ a
+
a
[n = 0. (A.35)
For
x
2
_
=
_
n +
1
2
_ _
h
m
_
we use the action of x on the state [n is
x[n =
_
h
2m
_
1/2
( a
+
+ a
) [n =
_
h
2m
_
1/2
_
n + 1[n + 1 +
n[n 1
_
(A.36)
Since x is Hermitian, the Hermitian conjugate of x[n is n[ x so that
n[ x x[n =
_
h
2m
_
_
n + 1[
n + 1 +n 1[
n
__
n + 1[n + 1 +
n[n 1
_
n[ x
2
[n
_
=
_
h
2m
_
n + 1 + n =
_
n +
1
2
__
h
m
_
. (A.37)
98
(e) The Hamiltonian for the harmonic oscillator is
H =
p
2
2m
+
1
2
m
2
x
2
= T + V (A.38)
and
_
H
_
= T +V =
1
2m
p
2
_
+
1
2
m
2
x
2
_
. (A.39)
But
_
H
_
= E
n
=
_
n +
1
2
_
h and
x
2
_
=
_
n +
1
2
_ _
h
m
2
_
so
V =
1
2
m
2
_
n +
1
2
__
h
m
2
_
=
_
n +
1
2
_
1
2
h =
1
2
E
n
.
Hence
E
n
= T +V = T +
1
2
E
n
, (A.40)
T =
1
2
E
n
. (A.41)
2. (a) Since x =
_
h
2m
_
1/2
( a
+
+ a
) then its expectation value in the state [ =

1
6
[0 +
2
6
[1 +
1
6
[2 is given
by
[ x[ =
_
1
6
0[ +
2
6
1[ +
1
6
2[
__
h
2m
_
1/2
( a
+
+ a
)
_
1
6
[0 +
2
6
[1 +
1
6
[2
_
. (A.42)
Using the actions of a
+
and a
on states [n, i.e. a

+
[n =
n + 1[n + 1 and a
[n =
n[n 1 and the

orthonormality of the basis states, n[m =
nm
we get
[ x[ =
_
h
2m
_
1/2
_
1
6
0[ +
2
6
1[ +
1
6
2[
_
_
2
6
[0 +
1 +
6
[1 +
2
6
[2 +
6
[3
_
,
=
_
h
2m
_
1/2
1
6
_
20[0 + (1 +
2)0[1 + 2
20[2 +
30[3 + 41[0 + 2(1 +
2)1[1
+4
21[2 + 2
31[3 + 22[0 + (1 +
2)2[1 + 2
22[2 +
32[3
_
,
=
_
h
2m
_
1/2
1
6
_
2 + 2(1 +
2) + 2
2
_
=
2 + 2
2
3
_
h
2m
_
1/2
.
For p
x
,
[ p
x
[ =
_
1
6
0[ +
2
6
1[ +
1
6
2[
_
i
_
mh
2
_
1/2
( a
+
a
)
_
1
6
[0 +
2
6
[1 +
1
6
[2
_
(A.43)
= i
_
mh
2
_
1/2
1
6
(0[ + 21[ +2[)
_
2[0 + (1
2)[1 + 2
2[2 +
3[3
_
, (A.44)
= i
_
mh
2
_
1/2
1
6
_
2 + 2(1
2) + 2
2
_
= 0 (A.45)
(b) Since a
[n =
n[n 1 then applying a
to [ as given above leads to

a
[ = e
||
2
/2
n=0
n!
a
[n = e
||
2
/2
n=0
n!
n[n 1 (A.46)
= e
||
2
/2
n=1

n1
_
(n 1)!
[n 1 = e
||
2
/2
n=0
n!
[n = [. (A.47)
(c) Express x in terms of a
+
and a
as
x =
_
h
2m
_
1/2
( a
+
+ a
) . (A.48)
Since a
[ = [, and the state [ is normalized already, then

a
[ = [, [ a
[ = [ = .
99
Also as a
+
and a
are Hermitian conjugates of each other then taking the Hermitian conjugate
( a
[)
= ([)
, (A.49)
[ a
+
= [
, (A.50)
[ a
+
[ = [
[ =
. (A.51)
Hence
x =
_
h
2m
_
1/2
[ ( a
+
+ a
) [ =
_
h
2m
_
1/2
([ a
+
[ +[ a
[) , (A.52)
=
_
h
2m
_
1/2
(
+ ) =
_
h
2m
_
1/2
2Re () =
_
2 h
m
_
1/2
Re () . (A.53)
3. (a) Expectation values are found from:
A =
_

0
(x)Au
0
(x)dx =
_

[u
0
(x)[
2
Adx (A.54)
For x we nd x = 0 as the integrand is an odd function. For p we nd:
p =
_

0
(x)
_
ih
d
dx
_
u
0
(x)dx = [N
0
[
2
_

2
x
2
/2
_
ih
d
dx
_
e
2
x
2
/2
(A.55)
= ih[N
0
[
2
_

2
x
2
/2
(
2
x)e
2
x
2
/2
dx = 0 (A.56)
as the integrand is odd again.
(b) As x = 0, then (x)
2
= x
2
and similarly (p)
2
= p
2
. The energy E = p
2
/2m + m
2
x
2
/2 so:
E =
p
2
2m
+
1
2
m
2
x
2
=
(p)
2
2m
+
1
2
m
2
(x)
2
(A.57)
(c) We have that xp = c, so we can write p = c/x and so:
E =
c
2
2m(x)
2
+
1
2
m
2
(x)
2
(A.58)
Minimising with respect to (x)
2
gives:
E
(x)
2
=
1
2m
c
2
(x)
4
+
1
2
m
2
= 0 (A.59)
(x)
2
=
c
m
(A.60)
E
min
=
c
2
+
c
2
= c (A.61)
But we know that the ground state energy of the harmonic oscillator is E =
1
2
h, so c =
1
2
h and therefore
xp =
1
2
h, which is consistent with the Uncertainty Principle.
100
Problem Sheet 3
Hand in your answers by Tuesday 29 November, either at the lectures or Dr Bowlers pigeonhole in Physics. Marks per
1. (a) The operators

J
x
,

J
y
and

J
z
are Cartesian components of the angulare momentum operator, obeying the usual
commutation relations (
_
J
x
,

J
y
_
= ih

J
z
etc). Use these commutation relations to show that the operators
J
+
=

J
x
+ i

J
y
and

J
=

J
x
i

J
y
satisfy the following commutation relations: [3]
_
J
z
,

J
+
_
= h

J
+
and
_
J
z
,

J
_
= h

J
(b) Let [jm be a simultaneous eigenvector of

J
2
and

J
z
having eigenvalues j(j + 1) h
2
and mh, respectively.
Show that

J
[jm is an eigenvector of

J
z
, and that its eigenvalue is (m1)h. [3]
(c) With [jm correctly normalised, obtain a formula for jm[

J
+
[jm in terms of j and m. For a given value
of j, what value must m have if jm[

J
+
[jm = 0 ? What value must m have if jm[
J
+

J
[jm = 0 ? [4]
2. The raising and lowering angular momentum operators,

J
+
,

J
are dened in terms of the Cartesian components
J
x
,

J
y
,

J
z
of angular momentum

J by

J
+
=

J
x
+ i

J
y
and

J
=

J
x
i

J
y
.
(a) Obtain the matrix representation of

J
y
for the state with j = 1 in terms of the set of eigenstates of

J
z
. [6]
(b) Solve the eigenvalue equation, i.e. J
y
c = c, using the matrix representation of

J
y
and the basis states
[j, m [1, 1 =
_
_
1
0
0
_
_
; [1, 0 =
_
_
0
1
0
_
_
; [1, 1 =
_
_
0
0
1
_
_
; (A.62)
(i) to nd ALL the eigenvalues , [2]
(ii) and ANY ONE of the eigenvectors c . (Make the rst element of the eigenvector real and positive.) [2]
3. (a) Calculate explicitly the commutator [
z
,
x
] and show how it is related to
y
[2]
(b) The eigenvectors of

S
z
are denoted by [
1
2
and [
1
2
. A system is in a spin state represented by the state
vector:
[ =
1
2
_
[
1
2
+[
1
2
_
(A.63)
(i) If a measurement is made of S
z
, what are the possible outcomes of the measurement, and what are the
probabilities of these outcomes ? [2]
(ii) If a measurement is made on the same state of S
x
, what are the possible outcomes of the measurement
and the probabilities of these outcomes ? [2]
(c) The basis states [ and [ are the eigenstates of

S
z
with eigenvalues h/2 and h/2 respectively. Show that
the state [
+
n
= cos (/2) [ + sin (/2) [ is an eigenstate of the spin operator

S
n
=

S n =

S
x
sin +
S
z
cos , the component of spin S of a spin-
1
2
particle in the direction of the unit vector n = (sin , 0, cos )
lying in the x-z plane, with eigenvalue +h/2. [2]
(d) If a beam of electrons polarised with spin +h/2 along the z-axis moves through a Stern-Gerlach magnet
oriented along a direction n = (sin , 0, cos ), expand the initial state in terms of the state [
+
and the
equivalent state [
. How many beams emerge, and what are their relative intensities ? [2]
101
1. (a) We will need to expand out explicitly. First

J
+
:
_
J
z
,

J
+
_
= (

J
z
_
J
x
+ i

J
y
_
J
x
+ i

J
y
_

J
z
(A.64)
=

J
z

J
x

J
x

J
z
+ i

J
z

J
y
i

J
y

J
z
=
_
J
z
,

J
x
_
+ i
_
J
z
,

J
y
_
(A.65)
= ih

J
y
+ i
_
ih

J
x
_
(A.66)
= h(

J
x
+ i

J
y
) = h

J
+
(A.67)
as required. Now

J
:
_
J
z
,

J
_
= (

J
z
_
J
x
i

J
y
_
J
x
i

J
y
_

J
z
(A.68)
=

J
z

J
x

J
x

J
z
i

J
z

J
y
+ i

J
y

J
z
=
_
J
z
,

J
x
_
+ i
_
J
y
,

J
z
_
(A.69)
= ih

J
y
+ i
_
ih

J
x
_
(A.70)
= h(

J
x
i

J
y
) = h

J
(A.71)
again, as required.
(b) We will need to act on

J
[jm with

J
z
; we will use the commutator relations just derived as well:
J
z

J
[jm =
_
h

J
+

J

J
z
_
[jm (A.72)
= h

J
[jm +

J
mh[jm (A.73)
= (m1)h

J
[jm (A.74)
which shows that

J
[jm is an eigenvector of

J
z
with eigenalue (m1) h, as required.
(c) We will expand out

J
+
and

J
:
jm[

J
+
[jm = jm[
_
J
x
i

J
y
__
J
x
+ i

J
y
_
[jm (A.75)
= jm[
J
2
x
+

J
2
y
+ i

J
x

J
y
i

J
y

J
x
[jm (A.76)
= jm[
J
2

J
2
z
+ i
_
J
x
,

J
y
_
[jm (A.77)
= jm[
J
2

J
2
z
+ i(ih

J
z
)[jm (A.78)
= jm[
J
2

J
2
z
h

J
z
[jm (A.79)
= j(j + 1)h
2
m
2
h
2
mh
2
= (j(j + 1) m(m + 1)) h
2
(A.80)
where we have used:

J
2
[jm = j(j + 1)h
2
[jm and

J
z
[jm = mh[jm. Then if m = j we will have
jm[

J
+
[jm = 0. The only change if we look at jm[
J
+

J
[jm will be the sign of the commutator after

expansion, so we will have jm[
J
+

J
[jm = (j(j + 1) m(m1)) h

2
so if m = j then we will nd
jm[
J
+

J
[jm = 0.
2. By adding, we have
J
y
=
1
2i
_
J
+
_
. (A.81)
The action of

J
+
,

J
on [1, m are
J
+
[1, m =
_
2 m(m + 1)h[1, m + 1 (A.82)
[1, m =
_
2 m(m1)h[1, m1 (A.83)
and so
_
1, m
J
+
[1, m
_
=
_
2 m(m + 1)h1, m
[1, m + 1 =
_
2 m(m + 1)h
m
m+1
, (A.84)
_
1, m
[1, m
_
=
_
2 m(m1)h1, m
[1, m1 =
_
2 m(m1) h
m
m1
, (A.85)
_
1, m
J
y
[1, m
_
=
1
2i
_
_
2 m(m + 1)h
m
m+1
_
2 m(m1) h
m
m1
_
. (A.86)
102
(a) The matrix can be constructed by choosing m
= 1, 0, 1 and m = 1, 0, 1 in turn in eq(A.86) and noting

that the Kronecker delta
kn
= 1 if k = n and
kn
= 0 if k ,= n. Clearly the diagonal elements are zero and
the table below can be built up,
m = 1 m = 0 m = 1
m
= 1 0
2
2i
h 0
m
= 0
2
2i
h 0
2
2i
h
m
= 1 0
2
2i
h 0
and so the matrix representation of

J
y
is

J
y
=
h
2
_
_
0 i 0
i 0 i
0 i 0
_
_
.
(b) We have to solve the matrix equation

J
y
c = c for the eigenvectors c. This equation has a non-trivial solution
for c if
det
J
y
I
= 0. (A.87)
Writing =
h
2
we have to solve
h
2
_
_
0 i 0
i 0 i
0 i 0
_
_
_
_
c
1
c
2
c
3
_
_
=
_
_
c
1
c
2
c
3
_
_
=
h
_
_
c
1
c
2
c
3
_
_
, (A.88)
h
2
_
_
i 0
i i
0 i
_
_
_
_
c
1
c
2
c
3
_
_
= 0 (A.89)
This equation has a non-trivial solution for c if
det
i 0
i i
0 i
= 0. (A.90)
i
i
+ i
i i
0
=
_
2
1
_
+ i(i) =
3
+ 2 =
_
2
2
_
= 0. (A.91)
Thus =
2, 0 and the eigenvalues are = + h, 0 h, h (As is to be expected as there is not difference

between the y-axis and the z-axis for a free particle.)
To nd the eigenvectors we have to nd the components c
1
, c
2
, c
3
.
For = h, ( = +
2)
h
2
_
_
0 i 0
i 0 i
0 i 0
_
_
_
_
c
1
c
2
c
3
_
_
= h
_
_
c
1
c
2
c
3
_
_
, (A.92)
_
_
0 i 0
i 0 i
0 i 0
_
_
_
_
c
1
c
2
c
3
_
_
=
2
_
_
c
1
c
2
c
3
_
_
. (A.93)
Multiplying out gives
_
_
ic
2
ic
1
ic
3
ic
2
_
_
=
2
_
_
c
1
c
2
c
3
_
_
so
c
2
= i
2c
1
; c
1
c
3
= i
2c
2
; c
2
= i
2c
3
, (A.94)
so c
1
= c
3
= c
2
/(i
2). Since the eigenvector must be normalized, i.e. [c

1
[
2
+ [c
2
[
2
+ [c
3
[
2
= 1 then
using the above [c
2
[
2
/2 +[c
2
[
2
+[c
2
[
2
/2 = 1, and c
2
= 1/
2. Taking c
2
= 1/
2 then the eigenvector for

eigenvalue +h is
c
+1
=
_
_
i/2
1/
2
i/2
_
_
=
1
2
_
_
i
2
i
_
_
. (A.95)
103
For eigenvalue = h then we have
h
2
_
_
0 i 0
i 0 i
0 i 0
_
_
_
_
c
1
c
2
c
3
_
_
= h
_
_
c
1
c
2
c
3
_
_
, (A.96)
_
_
0 i 0
i 0 i
0 i 0
_
_
_
_
c
1
c
2
c
3
_
_
=
2
_
_
c
1
c
2
c
3
_
_
. (A.97)
and
ic
2
=
2c
1
; i(c
1
c
3
) =
2c
2
; ic
2
=
2c
3
, (A.98)
so as before c
1
= c
3
= ic
2
/
2. Normalizing as before, c
2
= 1/
2. Taking c
2
= 1/
2 then the
eigenvector for eigenvalue h is
c
1
=
_
_
i/2
1/
2
i/2
_
_
=
1
2
_
_
i
2
i
_
_
. (A.99)
Finally for = 0 we have
h
2
_
_
0 i 0
i 0 i
0 i 0
_
_
_
_
c
1
c
2
c
3
_
_
= 0. (A.100)
Then
ic
2
= 0; i(c
1
c
3
) = 0; ic
2
= 0,
so c
1
= c
3
and normalization gives c
1
=
1
2
. Hence the eigenvector for eigenvalue 0h is
c
0
=
_
_
1
2
0
1
2
_
_
. (A.101)
3. (a) We have:
[
z
,
x
] =
_
1 0
0 1
__
0 1
1 0
_
_
0 1
1 0
__
1 0
0 1
_
(A.102)
=
_
0 1
1 0
_
_
0 1
1 0
_
=
_
0 2
2 0
_
(A.103)
This last matrix is, of course, 2i
y
so we can write:
[
z
,
x
] = 2i
y
(A.104)
(b) (i) The eigenvalues of

S
z
are mh, so the possible outcomes of a measurement are
1
2
h. The probabilities
are 50% for each (as the components of the spin state are equal (N.B. a probability of
1
2
is also a good
way to say this)).
(ii) Using matrix notation, we can write:
[ =
1
2
_
1
0
_
+
1
2
_
0
1
_
=
1
2
_
1
1
_
(A.105)
But this is an eigenvector of
x
with eigenvalue 1. We have that

S
x
=
1
2
h
x
so the outcome of measuring
S
x
in this state is certainly
1
2
h
(c) The easiest way to show that

S
n
[
+
n
=
h
2
[
+
n
with

S
n
=

S
x
sin +

S
z
cos and [
+
n
= cos (/2) [ +
sin (/2) [ is to substitute. Noting that

S
x
=
1
2
_
S
+
_
then
_
1
2
_
S
+
_
sin +

S
z
cos
_
(cos (/2) [ + sin (/2) [)
=
h
2
sin cos (/2) [ +
h
2
sin sin (/2) [ +
h
2
cos cos (/2) [
h
2
cos sin (/2) [,
=
h
2
(cos cos (/2) + sin sin (/2)) [ +
h
2
(sin cos (/2) cos sin (/2)) [,
=
h
2
cos (/2) [ +
h
2
sin (/2) [ =
h
2
(cos (/2) [ + sin (/2) [) =
h
2
[
+
n
. (A.106)
104
(d) Initially the beam is in the state [. The Stern-Gerlach measures the component of spin along the direction n
in the x-z plane, i.e. it measures

S
n
. Thus expand [ in terms of the eigenstates [
+
n
, and [
n
of

S
n
with
eigenvalues +h/2 and h/2 respectively, as
[ = a[
+
n
+ b[
n
. (A.107)
with [a[
2
+[b[
2
= 1 for normalization. The magnet splits the beam into 2 components.
The relative intensities are [a[
2
and [b[
2
. But a =
+
n
[ = ([ cos (/2) +[ sin (/2)) [ = cos (/2).
Hence the relative probabilities are [a[
2
= cos
2
(/2) and [b[
2
= 1 [a[
2
= sin
2
(/2).
105
106
Problem Sheet 4
Hand in your answers by Tuesday 13 December, either at the lectures or Dr Bowlers pigeonhole in Physics. Marks per
1. The vector operator

S representing the total spin angular momentum of two spin-1/2 particles is:

S =

S
1
+

S
2
,
where

S
1
and

S
2
are spin angular momenta of particle 1 and 2. The x, y and zcomponents of

S
1
and

S
2
are
denoted by

S
1x
,

S
1y
,

S
1z
and

S
2x
,

S
2y
,

S
2z
, respectively, and the components of

S are

S
x
=

S
1x
+

S
2x
, etc...
The state vectors
1
2
,
1
2
,
1
2
and
1
2
represent the states in which both particles are spin-up, particle 1 is
spin-up and particle 2 is spin-down, etc...
(a) With

S
1+
=

S
1x
+ i
S
1y
and

S
1
=

S
1x
i
S
1y
being the raising and lowering operators for particle 1 and a
similar notation being used for particle 2, show that: [2]
S
1+

S
2
+

S
1

S
2+
= 2
_
S
1x

S
2x
+

S
1y

S
2y
_
(A.108)
(b) Hence, show that the scalar product of

S
1
and

S
2
is: [2]
S
1
S
2
=
1
2
_
S
1+

S
2
+

S
1

S
2+
_
+

S
1z

S
2z
(A.109)
(c) Use the fact that

S
2
=

S
2
1
+

S
2
2
+ 2
S
1

S
2
to examine the effect of

S
2
acting on the state vector
1
2
, and
show that
1
2
is an eigenvector of

S
2
. What is the eigenvalue? [3]
(d) Using the same method as in part (c), show that the state vector
1
2
(
1
2
+
1
2
) is an eigenvector of

S
2
,
and nd the eigenvalue. If a measurement of

S
1
S
2
is made, what will be the outcome ? [3]
2. A particle of mass m is conned to move in a one-dimensional "innite" potential well dened by V (x) = 0, [x[
a, V (x) = otherwise. The energy eigenvalues are E
n
=
n
2
2
h
2
8ma
2
, with n = 1, 2, 3, . . . and the orthonormal
eigenfunctions are the even and odd functions
n
(x) =
_
1
a
cos
_
nx
2a
_
for n = 1, 3, 5 . . .
1
a
sin
_
nx
2a
_
for n = 2, 4, 6 . . .
.
The potential is modied between a < x < a to V (x) =

2
h
2
8ma
2
sin
_
3x
2a
_
with < 1.
(a) Determine the ground state energy to rst-order in . [4]
(b) Determine the ground state energy to second-order in . [6]
[HINT: (a) Note 2 sin Acos B = sin(A + B) + sin (AB), (b) the eigenfunctions
n
(x) are orthonormal. ]
3. Solve the above problem with =
1
2
, using the variational method. With b a variational parameter, use a trial
function: [10]
T
(x) =
1
a
cos
_
x
2a
_
+ b
1
a
sin
_
x
a
_

1
(x) + b
2
(x)
[HINT: (a) When evaluating
H with the trial function, make use of the fact that

1
(x) and
2
(x) are eigenfunc-
tions of the one-dimensional innite potential well, (b) the eigenfunctions
n
(x) are orthonormal.]
4. A system consists of two indistinguishable spin-1 fermions, both conned inside the same box of length L centred
on the origin. The particles do not interact with each other.
(a) What is the total energy of the ground state of this system ? (Use the standard formulas for the energy
eigenvalues of a single particle in a box.) [1]
(b) What is the spin angular momentum of the system in the ground state? [2]
(c) How many distinct and mutually orthogonal space-spin energy eigenstates of the whole system are there in
which there is one particle in the single-particle ground state and one particle in the single-particle rst excited
state? Choosing these eigenstates so that they are also eigenvectors of the square and the z-component of total
spin angular momentum, write explicit formulas for all these space-spin eigenstates. (Use , notation for
spin-up and spin-down states.) [4]
(d) With no interaction between the particles, the energy eigenstates described in part (c) all have exactly the same
energy (they are degenerate). We now introduce a repulsive interaction between the two particles. Describe
the qualitative effect that this interaction has on the relative energies of the singlet and triplet eigenstates. [3]
107
1. (a) We have:
S
1+

S
2
=
_
S
1x
+ i
S
1y
__
S
2x
i
S
2y
_
=

S
1x

S
2x
+

S
1y

S
2y
+ i
_
S
1y

S
2x

S
1x

S
2y
_
S
1

S
2+
=
_
S
1x
i
S
1y
__
S
2x
+ i
S
2y
_
=

S
1x

S
2x
+

S
1y

S
2y
+ i
_
S
1x

S
2y

S
1y

S
2x
_
Hence:
S
1+

S
2
+

S
1

S
2+
= 2
_
S
1x

S
2x
+

S
1y

S
2y
_
(b) Writing out, we nd:
S
1
S
2
=

S
1x

S
2x
+

S
1y

S
2y
+

S
1z

S
2z
=
1
2
_
S
1+

S
2
+

S
1

S
2+
_
+

S
1z

S
2z
(c) Expanding out, we have:
S
2
2
=

S
2
1
2
+

S
2
2
2
+ 2
S
1
S
2
2
We know that:
S
2
1
1
= s(s + 1)h
2
1
=
3
4
h
2
S
2
2
2
= s(s + 1)h
2
2
=
3
4
h
2
2
since s = 1/2 in both cases. Also:
2
S
1
S
2
2
= 2
S
1z

S
2z
2
+
_
S
1+

S
2
+

S
1

S
2+
_
2
= 2(
1
2
h)
2
2
+ 0
since anything with

S
1+
or

S
2+
acting on
1
2
is zero. Hence:
S
2
2
=
_
2
3
4
+ 2
1
4
_
h
2
2
= 2 h
2
2
So
1
2
is an eigenvector of

S
2
with eigenvalue 2 h
2
.
(d) In this case:
S
2
1
2
(
1
2
+
1
2
) =
_
S
2
1
+

S
2
2
+ 2
S
1
S
2
_
1
2
(
1
2
+
1
2
)
Now
1
2
and
1
2
are eigenvectors of

S
2
1
and

S
2
2
with the eigenvalues in both cases being
3
4
h
2
. Turning to
S
1
S
2
, we see that:
2
S
1
S
2
2
=
_
2
S
1z

S
2z
+ (
S
1+

S
2
+

S
1

S
2+
)
_
2
=
1
2
h
2
2
+ h
2
2
Similarly:
2
S
1
S
2
2
=
_
2
S
1z

S
2z
+ (
S
1+

S
2
+

S
1

S
2+
)
_
2
=
1
2
h
2
2
+ h
2
2
Hence:
S
2
1
2
(
1
2
+
1
2
) =
_
2
3
4
h
2
1
2
h
2
+ h
2
_
1
2
(
1
2
+
1
2
) = 2
1
2
(
1
2
+
1
2
)
so the given vector is an eigenvector of

S
2
with eigenvector 2 h
2
. If a measurement of

S
1
S
2
is made then we
will nd:
S
1
S
2
1
2
(
1
2
+
1
2
) =
_
1
4
h
2
+
1
2
_
1
2
(
1
2
+
1
2
)
so the result will be denite, with value
1
4
h
2
.
108
2. (a) The ground state is non-degenerate and has n = 1, E
1
=

2
h
2
8ma
2
and the unperturbed eigenfunction is [
(0)
1
=
1
a
cos
_
x
2a
_
for [x[ < a, and [
(0)
1
= 0, for [x[ > a. The rst-order correction to the ground state energy,
E
(1)
1
is obtained from
E
(1)
1
=
(0)
1
[

2
h
2
8ma
2
sin
_
3x
2a
_
[
(0)
1
=

2
h
2
8ma
2
(0)
1
[ sin
_
3x
2a
_
[
(0)
1
, (A.110)
=

2
h
2
8ma
2
_
a
a
(0)
1
(x) sin
_
3x
2a
_
1
a
cos
_
x
2a
_
dx,
=

2
h
2
8ma
2
_
a
a
(0)
1
(x)
1
2
_
1
a
sin
_
4x
2a
_
+
1
a
sin
_
2x
2a
__
dx,
=

2
h
2
8ma
2
_
a
a
(0)
1
1
2
_
(0)
4
+
(0)
2
_
dx =

2
h
2
16ma
2
_
(0)
1
[
(0)
4
+
(0)
1
[
(0)
2

_
= 0, (A.111)
since the states [
(0)
1
, [
(0)
2
, [
(0)
4
are mutually orthogonal. Thus there is no change to rst-order in .
This could also be seen immediately from the form of
E
(1)
1
=
(0)
1
[

2
h
2
8ma
2
sin
_
3x
2a
_
[
(0)
1
(A.112)
since
(0)
1
(x) is an even function of x and the perturbation term sin
_
3x
2a
_
is an odd function of x.
(b) The second-order correction E
(2)
1
is obtained from the expression
E
(2)
1
=
n=2
(0)
n
[H
[
(0)
1

2
E
(0)
1
E
(0)
n
=
n=2
(0)
n

2
h
2
8ma
2
sin
_
3x
2a
_

(0)
1

2
E
(0)
1
E
(0)
n
. (A.113)
The matrix elements are
H
n1
=
(0)
n

2
h
2
8ma
2
sin
_
3x
2a
_
(0)
1
=

2
h
2
8ma
2
(0)
n
sin
_
3x
2a
_
(0)
1
. (A.114)
Since the ground state has even parity, the perturbation has odd parity, the only non-zero elements are for the
odd parity states, [
(0)
n
i.e. for n = 2, 4, 6 . . .. Consider
(0)
n
sin
_
3x
2a
_
(0)
1
=
_
a
a
(0)
n
(x) sin
_
3x
2a
_
1
a
cos
_
x
2a
_
dx, (A.115)
=
_
a
a
(0)
n
(x)
1
2
_
1
a
sin
_
4x
2a
_
+
1
a
sin
_
2x
2a
__
dx,
=
_
a
a
(0)
n
1
2
_
(0)
4
+
(0)
2
_
dx =
1
2
_
(0)
n
[
(0)
4
+
(0)
n
[
(0)
2

_
,
=
1
2
n4
+
n2
. (A.116)
Hence the only terms that contribute to the summation in the expression for E
(2)
1
are for n = 2 and n = 4.
Thus
E
(2)
1
=
(0)
2
[H
[
(0)
1

2
E
(0)
1
E
(0)
2
+
(0)
4
[H
[
(0)
1

2
E
(0)
1
E
(0)
4
(A.117)
=
_

2
h
2
8ma
2
_2
_
_
1
2
_
2
2
h
2
8ma
2
(1 2
2
)
+
_
1
2
_
2
2
h
2
8ma
2
(1 4
2
)
_
=

2
4
2
h
2
8ma
2
_
1
3
+
1
15
_
=

2
4
2
h
2
8ma
2
_
2
5
_
=
2
10
_
2
h
2
8ma
2
_
. (A.118)
and the energy to second-order in is
E
1
=
_
2
h
2
8ma
2
__
1

2
10
_
. (A.119)
109
3. The variational method says that for any trial function
T
satisfying the boundary conditions
_
H
_
=

T
[
H[
T
T
[
T
E
0
(A.120)
where E
0
is the energy of the ground state.
The trial function is (x, b) =
1
(x) + b
2
(x) where
1
(x) and
2
(x) are the eigenfunctions for a particle in
the innite potential well dened above. Normalisation gives, since [
1
and [
2
are orthonormal.
T
[
T
= (
1
[ + b
2
[) ([
1
+ b[
2
) =
1
[
1
+ b
2
2
[
2
+ b
1
[
2
+ b
2
[
1
,
= 1 + b
2
. (A.121)
The Hamiltonian for the system

H can be written as

H =

H
0
+

H
where

H
0
is a particle in the innite potential
well dened above, and

H
= V (x) =

2
h
2
8ma
2
sin
_
3x
2a
_
. Then
T
[
H[
T
=
T
[
H
0
[
T
+
T
[V (x) [
T
, (A.122)
=
1
+ b
2
[
H
0
[
1
+ b
2
+
T
[V (x) [
T
. (A.123)
But

H
0
[
1
= E
(0)
1
[
1
and

H
0
[
2
= E
(0)
2
[
2
so
T
[
H[
T
= E
1
+ b
2
E
2
+
T
[V (x) [
T
. (A.124)
Expand
T
[V (x) [
T
as
T
[V (x) [
T
=
1
+ b
2
[V (x) [
1
+ b
2
,
=
1
[V (x) [
1
+ b
2
2
[V (x) [
2
+ b
1
[V (x) [
2
+
2
[V (x) [
1
.(A.125)
But V (x) and
2
(x) are odd functions of x, whereas
1
(x) is even. Thus the terms
1
[V (x) [
1
and
2
[V (x) [
2
are zero and

T
[V (x) [
T
reduces to
T
[V (x) [
T
= b
1
[V (x) [
2
+
2
[V (x) [
1
= 2b
2
[V (x) [
1
(A.126)
= 2b
2
[

2
h
2
8ma
2
sin
_
3x
2a
_
[
1
= 2b

2
h
2
8ma
2
2
[ sin
_
3x
2a
_
[
1
. (A.127)
and
(0)
2
sin
_
3x
2a
_
(0)
1
=
_
a
a
(0)
2
(x) sin
_
3x
2a
_
1
a
cos
_
x
2a
_
dx, (A.128)
=
_
a
a
(0)
2
(x)
1
2
_
1
a
sin
_
4x
2a
_
+
1
a
sin
_
2x
2a
__
dx,
=
_
a
a
(0)
2
1
2
_
(0)
4
+
(0)
2
_
dx =
1
2
_
(0)
2
[
(0)
4
+
(0)
2
[
(0)
2

_
(A.129)
=
1
2
(A.130)
giving
T
[V (x) [
T
= b

2
h
2
8ma
2
= bE
1
. (A.131)
Hence
_
H
_
=
E
1
+ b
2
E
2
+ bE
1
1 + b
2
= E
1
_
1 + 4b
2
+ b
1 + b
2
_
= E (b) . (A.132)
To nd the minimum value, differentiate E (b) with respect to the variational parameter, b, and set the result to zero
E (b)
b
= E
1
_
6b b
2
+
(1 + b
2
)
2
_
= 0, (A.133)
giving b =
6
36+4
2
2
. Taking =
1
2
, then b = 6 6
_
_
1 +
1
36
_
So b
1
= 0.08 276 or b
2
= 12. 08276. Either
by differentiating again or noting that + 6b b
2
= (b b
1
) (b b
2
) shows that b
1
= 0.08 276 corresponds
to a minimum. With this value of b the best upper bound on the ground state energy with this trial function is
E 0.9793E
1
. (A.134)
{Note the correct eigenstate [ for the ground state can be expanded in terms of the complete set of eigenstates
of the innite potential well as [ =
n=1
c
n
[
(0)
n
. This calculation has taken only the rst two terms. The
estimate could be improved by taking more terms.}
110
4. (a) The ground-state energy of a single particle of mass m in a box of length L is h
2
/8mL
2
. The ground state for
the 2-fermion system has both particles in the single-particle spatial ground state, one with spin-up and one
with spin-down. Hence, total energy is h
2
/4mL
2
.
(b) Since the two particles are in the same spatial state, the space part of the two-particle wave-function is sym-
metric. Therefore the spin part is antisymmetric. But an antisymmetric spin state for two spin-
1
2
particles has
total spin = 0.
(c) Call single-particle spatial ground-state
1
(x), spatial rst-excited state
2
(x). There are two kinds of 2-
particle states:
Spatial state exchange symmetric, spin state exchange anti-symmetric
Spatial state exchange anti-symmetric, spin state exchange symmetric
For the rst kind, there is only one anti-symmetric spin state:
1
2
(
1
2
)
So the state of the rst kind is:
1
2
(
1
(x
1
)
2
(x
2
) +
2
(x
1
)
1
(x
2
))
1
2
(
1
2
)
For the second kind, there are three symmetric spin states:
2
,
1
2
(
1
2
+
1
2
) ,
1
2
So there are three states of the second kind:
1
2
(
1
(x
1
)
2
(x
2
)
2
(x
1
)
1
(x
2
))
1
2
1
2
(
1
(x
1
)
2
(x
2
)
2
(x
1
)
1
(x
2
))
1
2
(
1
2
+
1
2
)
1
2
(
1
(x
1
)
2
(x
2
)
2
(x
1
)
1
(x
2
))
1
2
The total number of distinct, orthogonal space-spin energy eigenstates is four.
(d) A repulsive interaction between the particles raises the energy of the system. But it raises it by different
amounts for singlet and triplet states. In the triplet spin states, the spatial state is antisymmetric, so that the
probability of nding the particles at the same position is zero. So the spatial antisymmetry keeps the particles
away from each other. But in the singlet spin states, the spatial state is symmetric, and the particles are not
kept away from each other. Hence, the repulsive interaction raises the energy of the triplet state less than that
of the singlet state. The repulsive interaction breaks the degeneracy, so that the triplet energy is below the
singlet energy.
111

FullNotes PHAS3226

Uploaded by

Copyright:

Available Formats

FullNotes PHAS3226

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

FullNotes PHAS3226

Uploaded by

Copyright:

Available Formats

PHAS3226: Quantum Mechanics

(r) is the complex conjugate of (r).)

can be found as (r, t) evolves according to the time-dependent Schrdinger equation

[. The bra corresponding to c

A[. The Hermitian conjugate

. This means that the quantity [

A with ket [. This equivalence is an elegant feature of the Dirac notation.

A[ is called the matrix element of the operator

[ constitute a complex-valued function of the real variable x

. This function is identied with the

) is merely just one of many

and the momentum wavefunction (p

[ would be obtained in the momentum representation.

so the integral version of the closure relation is

(x) = (x) then

and so its expectation value is

and so S is unitary, i.e.

have the same spectrum To nd the eigenvalues of H

does the reverse. Note that

rather than solving second-order partial differential equations.

has the correct energy spectrum, thus

I) = 0 to get the eigenvalues E

are Hermitian conjugates of each

commutes with the Hamiltonian

. Consider the state [

[, then from above

is called a lowering operator.

N[n = n[n (2.30)

Assume that the eigenstates [n of

, the momentum operator p is

Suppose that [ is an eigenstate of a

with eigenvalue , then

possess a non-zero (non-trivial) solution if

where is a real parameter. Clearly

(x). The functions () are normalized so that

and are given in terms of the Legendre polynomials P

raise or lower respectively the eigenvalues of

[ are also eigenfunctions of

produces a sequence of eigenfunctions of

raise and lower the eigenvalues in steps of h, then

The state [j, m is an eigenstate of

(j, m) [j, m1. (3.69)

to equ(3.68) and taking the scalar product with state [j, m,

are Hermitian adjoint operators such that

are Hermitian adjoints of each other then as

k and the outgoing wavefunction is

is symmetric in the label 1, 2 of the two electrons. Hence operating on

applied to it must give zero, i.e.

. One way to construct all the eigenstates of

and act repeatedly with the total lowering operator

to it generates the state of j = +

2h[, 1[ +[, h[, (4.99)

is in some sense a "small" change to

is small enough that E

for k ,= n. In general the diagonal

. That is the rst-order shift E

. It will be assumed that this has already been

possesses only axial