1 The foundations of quantum

mechanics

Operators in quantum mechanics The whole of quantum mechanics can be expressed in terms of a small set
1.1 Linear operators of postulates. When their consequences are developed, they embrace the
1.2 Eigenfunctions and eigenvalues
behaviour of all known forms of matter, including the molecules, atoms, and
1.3 Representations
1.4 Commutation and electrons that will be at the centre of our attention in this book. This chapter
non-commutation introduces the postulates and illustrates how they are used. The remaining
1.5 The construction of operators chapters build on them, and show how to apply them to problems of chemical
1.6 Integrals over operators interest, such as atomic and molecular structure and the properties of mole-
1.7 Dirac bracket notation
cules. We assume that you have already met the concepts of hamiltonian and
1.8 Hermitian operators
The postulates of quantum wavefunction in an elementary introduction, and have seen the Schrodinger
mechanics equation written in the form
1.9 States and wavefunctions
Hc Ec
1.10 The fundamental prescription
1.11 The outcome of measurements This chapter establishes the full significance of this equation, and provides
1.12 The interpretation of the a foundation for its application in the following chapters.
wavefunction
1.13 The equation for the
wavefunction
1.14 The separation of the Schrodinger Operators in quantum mechanics
equation
The specification and evolution of
states An observable is any dynamical variable that can be measured. The principal
1.15 Simultaneous observables mathematical difference between classical mechanics and quantum mechan-
1.16 The uncertainty principle ics is that whereas in the former physical observables are represented by
1.17 Consequences of the uncertainty functions (such as position as a function of time), in quantum mechanics they
principle
are represented by mathematical operators. An operator is a symbol for an
1.18 The uncertainty in energy and
time instruction to carry out some action, an operation, on a function. In most of
1.19 Time-evolution and conservation the examples we shall meet, the action will be nothing more complicated than
laws multiplication or differentiation. Thus, one typical operation might be
Matrices in quantum mechanics multiplication by x, which is represented by the operator x  . Another
1.20 Matrix elements
operation might be differentiation with respect to x, represented by the
1.21 The diagonalization of the
hamiltonian operator d/dx. We shall represent operators by the symbol O (omega) in
The plausibility of the Schrodinger general, but use A, B, . . . when we want to refer to a series of operators.
equation We shall not in general distinguish between the observable and the operator
1.22 The propagation of light that represents that observable; so the position of a particle along the x-axis
1.23 The propagation of particles
will be denoted x and the corresponding operator will also be denoted x (with
1.24 The transition to quantum
mechanics multiplication implied). We shall always make it clear whether we are
referring to the observable or the operator.
We shall need a number of concepts related to operators and functions
on which they operate, and this first section introduces some of the more
important features.
10 j 1 THE FOUNDATIONS OF QUANTUM MECHANICS

1.1 Linear operators

The operators we shall meet in quantum mechanics are all linear. A linear
operator is one for which
Oaf bg aOf bOg 1:1
where a and b are constants and f and g are functions. Multiplication is a
linear operation; so is differentiation and integration. An example of a non-
linear operation is that of taking the logarithm of a function, because it is not
true, for example, that log 2x 2 log x for all x.

1.2 Eigenfunctions and eigenvalues

In general, when an operator operates on a function, the outcome is another
function. Differentiation of sin x, for instance, gives cos x. However, in
certain cases, the outcome of an operation is the same function multiplied by
a constant. Functions of this kind are called eigenfunctions of the operator.
More formally, a function f (which may be complex) is an eigenfunction of an
operator O if it satisfies an equation of the form
Of of 1:2
where o is a constant. Such an equation is called an eigenvalue equation. The
function eax is an eigenfunction of the operator d/dx because (d/dx)eax aeax,
2
which is a constant (a) multiplying the original function. In contrast, eax is
ax2 ax2
not an eigenfunction of d/dx, because (d/dx)e 2axe , which is a con-
2
stant (2a) times a different function of x (the function xeax ). The constant o
in an eigenvalue equation is called the eigenvalue of the operator O.

Example 1.1 Determining if a function is an eigenfunction

Is the function cos(3x 5) an eigenfunction of the operator d2/dx2 and, if so,

what is the corresponding eigenvalue?
Method. Perform the indicated operation on the given function and see if
the function satisfies an eigenvalue equation. Use (d/dx)sin ax a cos ax and
(d/dx)cos ax a sin ax.
Answer. The operator operating on the function yields

d2 d
2
cos3x 5 3 sin3x 5 9 cos3x 5
dx dx
and we see that the original function reappears multiplied by the eigen-
value 9.
Self-test 1.1. Is the function e3x 5 an eigenfunction of the operator d2/dx2
and, if so, what is the corresponding eigenvalue?
[Yes; 9]

An important point is that a general function can be expanded in terms of

all the eigenfunctions of an operator, a so-called complete set of functions.
 1.2 EIGENFUNCTIONS AND EIGENVALUES j 11

That is, if fn is an eigenfunction of an operator O with eigenvalue on (so Ofn

on fn), then1 a general function g can be expressed as the linear combination
X
g cn fn 1:3
n

where the cn are coefficients and the sum is over a complete set of functions.
For instance, the straight line g ax can be recreated over a certain range by
superimposing an infinite number of sine functions, each of which is an
eigenfunction of the operator d2/dx2. Alternatively, the same function may be
constructed from an infinite number of exponential functions, which are
eigenfunctions of d/dx. The advantage of expressing a general function as a
linear combination of a set of eigenfunctions is that it allows us to deduce the
effect of an operator on a function that is not one of its own eigenfunctions.
Thus, the effect of O on g in eqn 1.3, using the property of linearity, is simply
X X X
Og O cn fn cn Ofn c n on f n
n n n

A special case of these linear combinations is when we have a set of

degenerate eigenfunctions, a set of functions with the same eigenvalue. Thus,
suppose that f1, f2, . . . , fk are all eigenfunctions of the operator O, and that
they all correspond to the same eigenvalue o:
Ofn ofn with n 1, 2, . . . , k 1:4
Then it is quite easy to show that any linear combination of the functions fn
is also an eigenfunction of O with the same eigenvalue o. The proof is as
follows. For an arbitrary linear combination g of the degenerate set of
functions, we can write
X
k X
k X
k X
k
Og O cn fn cn Ofn cn ofn o cn fn og
n1 n1 n1 n1

This expression has the form of an eigenvalue equation (Og og).

Example 1.2 Demonstrating that a linear combination of degenerate

eigenfunctions is also an eigenfunction

Show that any linear combination of the complex functions e2ix and e2ix is an
eigenfunction of the operator d2/dx2, where i (1)1/2.
Method. Consider an arbitrary linear combination ae2ix be2ix and see if the
function satisfies an eigenvalue equation.
Answer. First we demonstrate that e2ix and e2ix are degenerate eigenfunctions.

d2 2ix d
e 2ie2ix 4e2ix
dx2 dx

.......................................................................................................

1. See P.M. Morse and H. Feschbach, Methods of theoretical physics, McGraw-Hill, New York
(1953).
12 j 1 THE FOUNDATIONS OF QUANTUM MECHANICS

where we have used i2 1. Both functions correspond to the same eigen-
value, 4. Then we operate on a linear combination of the functions.
d2
ae2ix be2ix 4ae2ix be2ix
dx2
The linear combination satisfies the eigenvalue equation and has the same
eigenvalue (4) as do the two complex functions.
Self-test 1.2. Show that any linear combination of the functions sin(3x) and
cos(3x) is an eigenfunction of the operator d2/dx2.
[Eigenvalue is 9]

A further technical point is that from n basis functions it is possible to con-

struct n linearly independent combinations. A set of functions g1, g2, . . . , gn is
said to be linearly independent if we cannot find a set of constants c1, c2, . . . ,
cn (other than the trivial set c1 c2 0) for which
X
ci gi 0
i
A set of functions that is not linearly independent is said to be linearly
dependent. From a set of n linearly independent functions, it is possible to
construct an infinite number of sets of linearly independent combinations,
but each set can have no more than n members. For example, from three
2p-orbitals of an atom it is possible to form any number of sets of linearly
independent combinations, but each set has no more than three members.

1.3 Representations
The remaining work of this section is to put forward some explicit forms of
the operators we shall meet. Much of quantum mechanics can be developed in
terms of an abstract set of operators, as we shall see later. However, it is often
fruitful to adopt an explicit form for particular operators and to express them
in terms of the mathematical operations of multiplication, differentiation,
and so on. Different choices of the operators that correspond to a particular
observable give rise to the different representations of quantum mechanics,
because the explicit forms of the operators represent the abstract structure of
the theory in terms of actual manipulations.
One of the most common representations is the position representation,
in which the position operator is represented by multiplication by x (or
whatever coordinate is specified) and the linear momentum parallel to x is
represented by differentiation with respect to x. Explicitly:
h
 q
Position representation: x ! x  px ! 1:5
i qx
where  h h=2p. Why the linear momentum should be represented in pre-
cisely this manner will be explained in the following section. For the time
being, it may be taken to be a basic postulate of quantum mechanics.
An alternative choice of operators is the momentum representation, in
which the linear momentum parallel to x is represented by the operation of
 1.4 COMMUTATION AND NON-COMMUTATION j 13

multiplication by px and the position operator is represented by differentia-

tion with respect to px. Explicitly:
h
 q
Momentum representation: x !  px ! px  1:6
i qpx
There are other representations. We shall normally use the position repres-
entation when the adoption of a representation is appropriate, but we shall
also see that many of the calculations in quantum mechanics can be done
independently of a representation.

1.4 Commutation and non-commutation

An important feature of operators is that in general the outcome of successive
operations (A followed by B, which is denoted BA, or B followed by A,
denoted AB) depends on the order in which the operations are carried out.
That is, in general BA 6 AB. We say that, in general, operators do not
commute. For example, consider the operators x and px and a specific
function x2. In the position representation, (xpx)x2 x(2 h/i)x2,
h/i)x (2
2 3 2
whereas (pxx)x pxx (3 h/i)x . The operators x and px do not commute.
The quantity AB  BA is called the commutator of A and B and is denoted
[A, B]:
A, B AB  BA 1:7
It is instructive to evaluate the commutator of the position and linear
momentum operators in the two representations shown above; the procedure
is illustrated in the following example.

Example 1.3 The evaluation of a commutator

Evaluate the commutator [x,px] in the position representation.

Method. To evaluate the commutator [A,B] we need to remember that the
operators operate on some function, which we shall write f. So, evaluate [A,B]f
for an arbitrary function f, and then cancel f at the end of the calculation.
Answer. Substitution of the explicit expressions for the operators into [x,px]
proceeds as follows:
h qf 
 h qxf
x, px f xpx  px xf x  
i qx i qx
h qf h h qf

x  f x ihf
i qx i i qx
where we have used (1/i) i. This derivation is true for any function f,
so in terms of the operators themselves,
x, px ih
The right-hand side should be interpreted as the operator multiply by the
constant ih.
Self-test 1.3. Evaluate the same commutator in the momentum representation.
[Same]
14 j 1 THE FOUNDATIONS OF QUANTUM MECHANICS

1.5 The construction of operators

Operators for other observables of interest can be constructed from the ope-
rators for position and momentum. For example, the kinetic energy operator
T can be constructed by noting that kinetic energy is related to linear
momentum by T p2/2m where m is the mass of the particle. It follows that
in one dimension and in the position representation
 
p2 1  h d 2 h d2

T x  1:8
2m 2m i dx 2m dx2

In three dimensions the operator in the position representation is

( )
h 2 q2
 q2 q2 h2 2

T 2
2 2  r 1:9
Although eqn 1.9 has explicitly
2m qx qy qz 2m
used Cartesian coordinates, the
relation between the kinetic energy The operator r2, which is read del squared and called the laplacian, is the
operator and the laplacian is true sum of the three second derivatives.
in any coordinate system; for The operator for potential energy of a particle in one dimension, V(x), is
example, spherical polar multiplication by the function V(x) in the position representation. The same is
coordinates. true of the potential energy operator in three dimensions. For example, in the
position representation the operator for the Coulomb potential energy of an
electron (charge e) in the field of a nucleus of atomic number Z is the
multiplicative operator

Ze2
V  1:10
4pe0 r

where r is the distance from the nucleus to the electron. It is usual to omit the
multiplication sign from multiplicative operators, but it should not be for-
gotten that such expressions are multiplications.
The operator for the total energy of a system is called the hamiltonian
operator and is denoted H:
H TV 1:11
The name commemorates W.R. Hamiltons contribution to the formulation
of classical mechanics in terms of what became known as a hamiltonian
function. To write the explicit form of this operator we simply substitute the
appropriate expressions for the kinetic and potential energy operators in the
chosen representation. For example, the hamiltonian for a particle of mass m
moving in one dimension is

 2 d2
h
H Vx 1:12
2m dx2
where V(x) is the operator for the potential energy. Similarly, the hamiltonian
operator for an electron of mass me in a hydrogen atom is

2 2
h e2
H r  1:13
2me 4pe0 r
 1.6 INTEGRALS OVER OPERATORS j 15

The general prescription for constructing operators in the position repres-

entation should be clear from these examples. In short:
1. Write the classical expression for the observable in terms of position
coordinates and the linear momentum.
2. Replace x by multiplication by x, and replace px by (
h/i)q/qx (and likewise
for the other coordinates).

1.6 Integrals over operators

When we want to make contact between a calculation done using operators
and the actual outcome of an experiment, we need to evaluate certain
integrals. These integrals all have the form
Z
I fm Ofn dt 1:14

The complex conjugate of where fm is the complex conjugate of fm. In this integral dt is the volume
a complex number z a ib element. In one dimension, dt can be identified as dx; in three dimensions it is
is z a  ib. Complex dxdydz. The integral is taken over the entire space available to the system,
conjugation amounts to which is typically from x 1 to x 1 (and similarly for the other
everywhere replacing i by  i. coordinates). A glance at the later pages of this book will show that many
The square modulus jzj2 is given by
molecular properties are expressed as combinations of integrals of this form
zz a2 b2 since jij2 1.
(often in a notation which will be explained later). Certain special cases of this
type of integral have special names, and we shall introduce them here.
When the operator O in eqn 1.14 is simply multiplication by 1, the integral
is called an overlap integral and commonly denoted S:
Z
S fm fn dt 1:15

It is helpful to regard S as a measure of the similarity of two functions: when

S 0, the functions are classified as orthogonal, rather like two perpendicular
vectors. When S is close to 1, the two functions are almost identical. The
recognition of mutually orthogonal functions often helps to reduce the
amount of calculation considerably, and rules will emerge in later sections
and chapters.
The normalization integral is the special case of eqn 1.15 for m n.
A function fm is said to be normalized (strictly, normalized to 1) if
Z
fm fm dt 1 1:16

It is almost always easy to ensure that a function is normalized by multiplying

it by an appropriate numerical factor, which is called a normalization factor,
typically denoted N and taken to be real so that N N. The procedure is
illustrated in the following example.

Example 1.4 How to normalize a function

A certain function f is sin(px/L) between x 0 and x L and is zero elsewhere.

Find the normalized form of the function.
16 j 1 THE FOUNDATIONS OF QUANTUM MECHANICS

Method. We need to find the (real) factor N such that N sin(px/L) is norm-
alized to 1. To find N we substitute this expression into eqn 1.16, evaluate the
integral, and select N to ensure normalization. Note that all space extends
from x 0 to x L.
Answer. The necessary integration is
Z Z L
f  f dt N 2 sin2 px=Ldx 12LN 2
0
R
where we have used sin2ax dx (x/2)(sin 2ax)/4a constant. For this
integral to be equal to 1, we require N (2/L)1/2. The normalized function is
therefore
 1=2
2
f sinpx=L
L
Comment. We shall see later that this function describes the distribution of a
particle in a square well, and we shall need its normalized form there.
Self-test 1.4. Normalize the function f eif, where f ranges from 0 to 2p.
[N 1/(2p)1/2]

A set of functions fn that are (a) normalized and (b) mutually orthogonal
are said to satisfy the orthonormality condition:
Z
fm fn dt dmn 1:17

In this expression, dmn denotes the Kronecker delta, which is 1 when m n

and 0 otherwise.

1.7 Dirac bracket notation

With eqn 1.14 we are on the edge of getting lost in a complicated notation. The
appearance of many quantum mechanical expressions is greatly simplified by
adopting the Dirac bracket notation in which integrals are written as follows:
Z
hmjOjni fm Ofn dt 1:18

The symbol jni is called a ket, and denotes the state described by the function
fn. Similarly, the symbol hnj is called a bra, and denotes the complex conjugate
of the function, fn . When a bra and ket are strung together with an operator
between them, as in the bracket hmjOjni, the integral in eqn 1.18 is to be
understood. When the operator is simply multiplication by 1, the 1 is omitted
and we use the convention
Z
hmjni fm fn dt 1:19

This notation is very elegant. For example, the normalization integral

becomes hnjni 1 and the orthogonality condition becomes hmjni 0
for m 6 n. The combined orthonormality condition (eqn 1.17) is then
hmjni dmn 1:20
 1.8 HERMITIAN OPERATORS j 17

A final point is that, as can readily be deduced from the definition of a Dirac
bracket,
hmjni hnjmi

1.8 Hermitian operators

An operator is hermitian if it satisfies the following relation:
Z Z 
 
fm Ofn dt fn Ofm dt 1:21a

for any two functions fm and fn. An alternative version of this definition is
Z Z
fm Ofn dt Ofm  fn dt 1:21b

This expression is obtained by taking the complex conjugate of each term on

the right-hand side of eqn 1.21a. In terms of the Dirac notation, the definition
of hermiticity is
hmjOjni hnjOjmi 1:22

Example 1.5 How to confirm the hermiticity of operators

Show that the position and momentum operators in the position representa-
tion are hermitian.
Method. We need to show that the operators satisfy eqn 1.21a. In some cases
(the position operator, for instance), the hermiticity is obvious as soon as the
integral is written down. When a differential operator is used, it may be
necessary to use integration by parts at some stage in the argument to transfer
the differentiation from one function to another:
Z Z
u dv uv  v du

Answer. That the position operator is hermitian is obvious from inspection:

Z Z Z 
fm xfn dt fn xfm dt fn xfm dt

We have used the facts that (f ) f and x is real. The demonstration of the
hermiticity of px, a differential operator in the position representation,
involves an integration by parts:
Z Z Z
 d
h h

fm px fn dx fm fn dx fm dfn
i dx i
 Z
x1
h  

f fn  fn dfm

i m x1
 Z 1
h  x1 d
fm fn jx1  fn fm dx
i 1 dx
18 j 1 THE FOUNDATIONS OF QUANTUM MECHANICS

The first term on the right is zero (because when jxj is infinite, a normalizable
function must be vanishingly small; see Section 1.12). Therefore,
Z Z
h
 d 
fm px fn dx  fn f dx
i dx m
Z  Z 
h d

fn fm dx fn px fm dx
i dx

Hence, the operator is hermitian.

Self-test 1.5. Show that the two operators are hermitian in the momentum
representation.

As we shall now see, the property of hermiticity has far-reaching impli-

cations. First, we shall establish the following property:

Property 1. The eigenvalues of hermitian operators are real.

Proof 1.1 The reality of eigenvalues

Consider the eigenvalue equation

Ojoi ojoi
The ket joi denotes an eigenstate of the operator O in the sense that the
corresponding function fo is an eigenfunction of the operator O and we are
labelling the eigenstates with the eigenvalue o of the operator O. It is often
convenient to use the eigenvalues as labels in this way. Multiplication from the
left by hoj results in the equation
hojOjoi ohojoi o
taking joi to be normalized. Now take the complex conjugate of both sides:
hojOjoi o
However, by hermiticity, hojOjoi hojOjoi. Therefore, it follows that
o o , which implies that the eigenvalue o is real.

The second property we shall prove is as follows:

Property 2. Eigenfunctions corresponding to different eigenvalues of an

hermitian operator are orthogonal.

That is, if we have two eigenfunctions of an hermitian operator O with

eigenvalues o and o 0 , with o 6 o 0 , then hojo 0 i 0. For example, it follows at
once that all the eigenfunctions of a harmonic oscillator (Section 2.16) are
mutually orthogonal, for as we shall see each one corresponds to a different
energy (the eigenvalue of the hamiltonian, an hermitian operator).
 1.9 STATES AND WAVEFUNCTIONS j 19

Proof 1.2 The orthogonality of eigenstates

Suppose we have two eigenstates joi and jo 0 i that satisfy the following
relations:
Ojoi ojoi and Ojo0 i o0 jo0 i
Then multiplication of the first relation by ho 0 j and the second by hoj gives
ho0 jOjoi oho0 joi and hojOjo0 i o0 hojo0 i
Now take the complex conjugate of the second relation and subtract it from
the first while using Property 1 (o 0 o 0 ):
ho0 jOjoi  hojOjo0 i oho0 joi  o0 hojo0 i
Because O is hermitian, the left-hand side of this expression is zero; so (noting
that o 0 is real and using hojo 0 i ho 0 joi as explained earlier) we arrive at
o  o0 ho0 joi 0
However, because the two eigenvalues are different, the only way of satisfying
this relation is for ho 0 joi 0, as was to be proved.

The postulates of quantum mechanics

Now we turn to an application of the preceding material, and move into the
foundations of quantum mechanics. The postulates we use as a basis for
quantum mechanics are by no means the most subtle that have been devised,
but they are strong enough for what we have to do.

1.9 States and wavefunctions

The first postulate concerns the information we can know about a state:
Postulate 1. The state of a system is fully described by a function C(r1,
r2, . . . , t).
In this statement, r1, r2, . . . are the spatial coordinates of particles 1, 2, . . .
that constitute the system and t is the time. The function C (uppercase psi)
plays a central role in quantum mechanics, and is called the wavefunction of
the system (more specifically, the time-dependent wavefunction). When we
are not interested in how the system changes in time we shall denote the
wavefunction by a lowercase psi as c(r1, r2, . . . ) and refer to it as the time-
independent wavefunction. The state of the system may also depend on some
internal variable of the particles (their spin states); we ignore that for now
and return to it later. By describe we mean that the wavefunction
contains information about all the properties of the system that are open to
experimental determination.
We shall see that the wavefunction of a system will be specified by a set of
labels called quantum numbers, and may then be written ca,b, . . . , where
a, b, . . . are the quantum numbers. The values of these quantum numbers
specify the wavefunction and thus allow the values of various physical
20 j 1 THE FOUNDATIONS OF QUANTUM MECHANICS

observables to be calculated. It is often convenient to refer to the state of

the system without referring to the corresponding wavefunction; the state is
specified by listing the values of the quantum numbers that define it.

1.10 The fundamental prescription

The next postulate concerns the selection of operators:
Postulate 2. Observables are represented by hermitian operators chosen to
satisfy the commutation relations
q, pq0 ihdqq0 q, q0 0 pq , pq0 0
where q and q 0 each denote one of the coordinates x, y, z and pq and pq 0 the
corresponding linear momenta.
The requirement that the operators are hermitian ensures that the observables
have real values (see below). Each commutation relation is a basic, unpro-
vable, and underivable postulate. Postulate 2 is the basis of the selection of
the form of the operators in the position and momentum representations for
all observables that depend on the position and the momentum.2 Thus, if we
define the position representation as the representation in which the position
operator is multiplication by the position coordinate, then as we saw in
Example 1.3, it follows that the momentum operator must involve differ-
entiation with respect to x, as specified earlier. Similarly, if the momentum
representation is defined as the representation in which the linear momentum
is represented by multiplication, then the form of the position operator is
fixed as a derivative with respect to the linear momentum. The coordinates
x, y, and z commute with each other as do the linear momenta px, py, and pz.

1.11 The outcome of measurements

The next postulate brings together the wavefunction and the operators and
establishes the link between formal calculations and experimental observations:
Postulate 3. When a system is described by a wavefunction c, the mean
value of the observable O in a series of measurements is equal to the expec-
tation value of the corresponding operator.
The expectation value of an operator O for an arbitrary state c is denoted hOi
and defined as
R 
c Oc dt hcjOjci
hOi R  1:23
c cdt hcjci
If the wavefunction is chosen to be normalized to 1, then the expectation
value is simply
Z
hOi c Oc dt hcjOjci 1:24

Unless we state otherwise, from now on we shall assume that the wave-
function is normalized to 1.
.......................................................................................................

2. This prescription excludes intrinsic observables, such as spin (Section 4.8).

 1.11 THE OUTCOME OF MEASUREMENTS j 21

The meaning of Postulate 3 can be unravelled as follows. First, suppose

that c is an eigenfunction of O with eigenvalue o; then
Z Z Z
hOi c Oc dt c oc dt o c c dt o 1:25

That is, a series of experiments on identical systems to determine O will give

the average value o (a real quantity, because O is hermitian). Now suppose
that although the system is in an eigenstate of the hamiltonian it is not in an
eigenstate of O. In this case the wavefunction can be expressed as a linear
combination of eigenfunctions of O:
X
c cn cn where Ocn on cn
n

In this case, the expectation value is

Z X ! ! Z
X X
hOi cm cm O cn cn dt cm cn cm Ocn dt
m n m, n
X Z
cm cn on cm cn dt
m, n

Because the eigenfunctions form an orthonormal set, the integral in the last
expression is zero if n 6 m, is 1 if n m, and the double sum reduces to a
single sum:
X Z X X 2
hOi cn cn on cn cn dt cn cn on jcn j on 1:26
n n n

That is, the expectation value is a weighted sum of the eigenvalues of O,

the contribution of a particular eigenvalue to the sum being determined by the
square modulus of the corresponding coefficient in the expansion of the
wavefunction.
We can now interpret the difference between eqns 1.25 and 1.26 in the
form of a subsidiary postulate:
Postulate 30 . When c is an eigenfunction of the operator O, the determina-
tion of the property O always yields one result, namely the corresponding
eigenvalue o. The expectation value will simply be the eigenvalue o. When c
is not an eigenfunction of O, a single measurement of the property yields
a single outcome which is one of the eigenvalues of O, and the probability that
a particular eigenvalue on is measured is equal to jcnj2, where cn is the
coefficient of the eigenfunction cn in the expansion of the wavefunction.
One measurement can give only one result: a pointer can indicate only one
value on a dial at any instant. A series of determinations can lead to a series of
results with some mean value. The subsidiary postulate asserts that a mea-
surement of the observable O always results in the pointer indicating one of
the eigenvalues of the corresponding operator. If the function that describes
the state of the system is an eigenfunction of O, then every pointer reading is
precisely o and the mean value is also o. If the system has been prepared in a
state that is not an eigenfunction of O, then different measurements give
different values, but every individual measurement is one of the eigenvalues of
22 j 1 THE FOUNDATIONS OF QUANTUM MECHANICS

O, and the probability that a particular outcome on is obtained is determined

by the value of jcnj2. In this case, the mean value of all the observations is the
weighted average of the eigenvalues. Note that in either case, the hermiticity
of the operator guarantees that the observables will be real.

Example 1.6 How to use Postulate 3 0 .

An operator A has eigenfunctions f1, f2, . . . , fn with corresponding eigenvalues

a1, a2, . . . , an. The state of a system is described by a normalized wavefunction
c given by
1=2 1=2
c 12 f1  38 f2 38 i f3
What will be the outcome of measuring the observable A?
Method. First, we need to determine if c is an eigenfunction of the operator A.
If it is, then we shall obtain the same eigenvalue of A in every measurement.
If it is not, we shall obtain different values in a series of different measure-
ments. In the latter case, if we have an expression for c in terms of the
eigenfunctions of A, then we can determine what different values are possible,
the probabilities of obtaining them, and the average value from a large series
of measurements.
Answer. To test whether c is an eigenfunction of the operator A we proceed as
follows:
h 1=2 1=2 i
Ac A 12 f1  38 f2 38 i f3
1=2 1=2
12a1 f1  38 a2 f2 38 i a3 f3 6 constant  c
Therefore, c is not an eigenfunction of A. However, because c is a linear
combination of f1, f2, and f3 we will obtain, in different measurements,
the values a1, a2, and a3 (the eigenvalues of the eigenfunctions of A that
contribute to c). The probabilities of obtaining a1, a2, and a3 are, respectively,
1 3 3
4, 8, and 8. The average value, given by eqn 1.26, is

hAi 14 a1 38 a2 38 a3
Comment. The normalization of c is reflected in the fact that the probabilities
sum to 1. Because the eigenfunctions f4, f5, . . . do not contribute here to c,
there is zero probability of finding a4, a5, . . . .
1=2
Self-test 1.6. Repeat the problem using c 13 f2 79 f4  13 if5 :
[hAi 19 a2 79 a4 19 a5

1.12 The interpretation of the wavefunction

The next postulate concerns the interpretation of the wavefunction itself, and
is commonly called the Born interpretation:

Postulate 4. The probability that a particle will be found in the volume

element dt at the point r is proportional to jc(r)j2dt.
 1.14 THE SEPARATION OF THE SCHRODINGER EQUATION j 23

As we have already remarked, in one dimension the volume element is dx.

In three dimensions the volume element is dxdydz. It follows from this
interpretation that jc(r)j2 is a probability density, in the sense that it
yields a probability when multiplied by the volume dt of an infinitesimal
region. The wavefunction itself is a probability amplitude, and has no direct
physical meaning. Note that whereas the probability density is real and non-
negative, the wavefunction may be complex and negative. It is usually con-
venient to use a normalized wavefunction; then the Born interpretation
becomes an equality rather than a proportionality. The implication of the
Born interpretation is that the wavefunction should be square-integrable;
that is
Z
jcj2 dt < 1

because there must be a finite probability of finding the particle somewhere in

the whole of space (and that probability is 1 for a normalized wavefunction).
This postulate in turn implies that c ! 0 as x ! 1, for otherwise the inte-
gral of jcj2 would be infinite. We shall make frequent use of this implication
throughout the text.

1.13 The equation for the wavefunction

The final postulate concerns the dynamical evolution of the wavefunction
with time:
Postulate 5. The wavefunction C(r1, r2, . . . , t) evolves in time according
to the equation
qC
ih HC 1:27
qt
This partial differential equation is the celebrated Schrodinger equation
introduced by Erwin Schrodinger in 1926. At this stage, we are treating the
equation as an unmotivated postulate. However, in Section 1.24 we shall
advance arguments in support of its plausibility. The operator H in the
Schrodinger equation is the hamiltonian operator for the system, the operator
corresponding to the total energy. For example, by using the expression
in eqn 1.12, we obtain the time-dependent Schrodinger equation in one
dimension (x) with a time-independent potential energy for a single particle:

qC h2 q2 C

ih  VxC 1:28
qt 2m qx2
We shall have a great deal to say about the Schrodinger equation and its
solutions in the rest of the text.

1.14 The separation of the Schrodinger equation

The Schrodinger equation can often be separated into equations for the time
and space variation of the wavefunction. The separation is possible when the
potential energy is independent of time.
24 j 1 THE FOUNDATIONS OF QUANTUM MECHANICS

In one dimension the equation has the form

 2 q2 C
h qC
HC  VxC ih
2m qx2 qt
Equations of this form can be solved by the technique of separation of
variables, in which a trial solution takes the form
Cx, t cxyt
When this substitution is made, we obtain

 2 d2 c
h dy
 y Vxcy i 
hc
2m dx2 dt
Division of both sides of this equation by cy gives

 2 1 d2 c
h 1 dy
 Vx i 
h
2m c dx2 y dt
Only the left-hand side of this equation is a function of x, so when x changes,
only the left-hand side can change. But as the left-hand side is equal to the
right-hand side, and the latter does not change, the left-hand side must be
equal to a constant. Because the dimensions of the constant are those of an
energy (the same as those of V), we shall write it E. It follows that the time-
dependent equation separates into the following two differential equations:

 2 d2 c
h
 Vxc Ec 1:29a
2m dx2

dy
ih Ey 1:29b
dt
The second of these equations has the solution

y / eiEt=h 1:30
Therefore, the complete wavefunction (C cy) has the form
Cx, t cxeiEt=h 1:31
The constant of proportionality in eqn 1.30 has been absorbed into the
normalization constant for c. The time-independent wavefunction satisfies
eqn 1.29a, which may be written in the form
Hc Ec
This expression is the time-independent Schrodinger equation, on which
much of the following development will be based.
This analysis stimulates several remarks. First, eqn 1.29a has the form of a
standing-wave equation. Therefore, so long as we are interested only in the
spatial dependence of the wavefunction, it is legitimate to regard the time-
independent Schrodinger equation as a wave equation. Second, when the
potential energy of the system does not depend on the time, and the system
is in a state of energy E, it is a very simple matter to construct the time-
dependent wavefunction from the time-independent wavefunction simply by
 1.15 SIMULTANEOUS OBSERVABLES j 25

multiplying the latter by eiEt/h. The time dependence of such a wavefunction

is simply a modulation of its phase, because we can write
We have used Eulers relation, eiEt=h cosEt=
h  i sinEt=
h
ix
e cos x i sin x It follows that the time-dependent factor oscillates periodically from 1 to i
as well as sin(x)  sin(x) and to 1 to i and back to 1 with a frequency E/h and period h/E. This behaviour
cos(x) cos(x). is depicted in Fig. 1.1. Therefore, to imagine the time-variation of a wave-
function of a definite energy, think of it as flickering from positive through
imaginary to negative amplitudes with a frequency proportional to the energy.
Although the phase of a wavefunction C with definite energy E oscillates in
Re

time, the product C C (or jCj2) remains constant:
C C c eiEt=h ceiEt=h c c

x States of this kind are called stationary states. From what we have seen so far,
it follows that systems with a specific, precise energy and in which the
potential energy does not vary with time are in stationary states. Although
their wavefunctions flicker from one phase to another in repetitive manner,
the value of C C remains constant in time.
Im

Fig. 1.1 A wavefunction

corresponding to an energy E rotates The specification and evolution of states
in the complex plane from real to
imaginary and back to real at a
Let us suppose for the moment that the state of a system can be specified as
circular frequency E/
h.
ja,b, . . . i, where each of the eigenvalues a, b, . . . corresponds to the operators
representing different observables A, B, . . . of the system. If the system is in
the state ja,b, . . . i, then when we measure the property A we shall get exactly
a as an outcome, and likewise for the other properties. But can a state be
specified arbitrarily fully? That is, can it be simultaneously an eigenstate of all
possible observables A, B, . . . without restriction? With this question we are
moving into the domain of the uncertainty principle.

1.15 Simultaneous observables

As a first step, we establish the conditions under which two observables may
be specified simultaneously with arbitrary precision. That is, we establish the
conditions for a state jci corresponding to the wavefunction c to be simul-
taneously an eigenstate of two operators A and B. In fact, we shall prove the
following:
Property 3. If two observables are to have simultaneously precisely defined
values, then their corresponding operators must commute.
That is, AB must equal BA, or equivalently, [A,B] 0.

Proof 1.3 Simultaneous eigenstates

Assume that jci is an eigenstate of both operators: Ajci ajci and

Bjci bjci. That being so, we can write the following chain of relations:
ABjci Abjci bAjci bajci abjci aBjci Bajci BAjci