NPTEL COURSE ON QUANTUM PHYSICS

V. Balakrishnan
Department of Physics, Indian Institute of Technology Madras
Chennai 600 036, India

Some mathematical preliminaries: Linear vector spaces

There are many textbooks and monographs that give excellent accounts of the
mathematics required for quantum mechanics, at different levels of sophistication.
Here are just a few of these references (google for their publication details !):

N. I. Akhiezer and I. M. Glazman, Theory of Linear Operators in Hilbert Space

R. Courant and D. Hilbert, Methods of Mathematical Physics, Vol. 1
P. Dennery and A. Krzywicki, Mathematics for Physicists
P. A. M. Dirac, Principles of Quantum Mechanics
B. Friedman, Principles and Techniques of Applied Mathematics
P. R. Halmos, Finite Dimensional Vector Spaces
T. F. Jordan, Linear Operators for Quantum Mechanics
D. W. Lewis, Matrix Theory

A linear vector space (or LVS) V is a set of elements |i , |i , . . . called vectors,

with an operation called addition satisfying the following properties:
(i) |i + |i = |i + |i V , |i , |i V
(ii) |i + (|i + |i) = (|i + |i) + |i
(iii) a unique null vector |i V such that |i + |i = |i , |i V

(iv) a unique vector |i V for every |i such that |i + (|i) = |i .

There is also an operation of multiplication by scalars a, b, . . . belonging to R
(the field of real numbers) or C (the field of complex numbers), such that
(v) a |i V , a R or C and |i V

(vi) a (b |i) = (ab) |i

(vii) a (|i + |i) = a |i + a |i
(viiii) (a + b) |i = a |i + b |i

(ix) 1 |i = |i
(x) 0 |i = 0 . It is because of this property that we may as well use the usual
symbol 0 for both the null vector |i and for the usual scalar zero.
If the scalars a, b, . . . are restricted to the real numbers, the LVS is a real LVS; if
the scalars are complex numbers, we have a complex LVS.1

The dual space: Every LVS has a dual that is also an LVS. The notation used
for the elements of this dual LVS is h| , h| . . . . The inner product h|i has the
general properties
1 The scalars may also be drawn from fields other than the real or complex number fields, but

we shall not consider these generalizations.

1
 (i) h|(|i + |i) = h|i + h|i
(ii) h|i = h|i
(iii) h|a i = ah|i
(iv) ha |i = a h|i
The norm of a vector is given by k k = h|i1/2 . The norm is positive definite
for every non-null vector. It vanishes if and only if |i = |i.

A basis in the LVS is a set of vectors that (i) are linearly independent and (ii) span
the space. A basis has n vectors if and only if the dimensionality of the LVS is n.
Every n-dimensional LVS is isomorphic to the Euclidean LVS Rn .

An infinite-dimensional LVS is one that has no n-dimensional basis for any pos-
itive integer n.

Orthonormal basis: Given a basis set, we can make it an orthonormal basis by

the Gram-Schmidt orthonormalization procedure. If the set of vectors {|j i}
is an orthonormal basis, we have the relations

hj |k i = jk (orthonormality)

and X
|j ihj | = I, (completeness)
j

where I is the unit operator in the LVS. It follows that any vector |i in the LVS
can be uniquely expanded in terms of the basis according to
X
|i = cj |j i, where cj = hj |i.
j

Uniquely expanded means that the coefficients cj uniquely determine |i, and vice
versa.

Change of basis: The basis set in any LVS is by no means unique, although its
dimensionality is. Given two orthonormal basis sets {|j i} and {|k i}, we have the
expansions X X
|i = cj |j i = dk |k i ,
j k

for any vector |i. The expansion coefficients cj = hj |i and dk = hk |i are

related by X X
cj = dk hj |k i, dk = cj hk |j i.
k j

1. Check whether the following sets of elements form an LVS. If they do, find the
dimensionality of the LVS.
(a) The set of all n n matrices with complex entries.
(b) The set of all polynomials (of all orders) of a complex variable z.
(c) The set of all (Cartesian) tensors of rank 2 in three dimensions.
(d) The set of all antisymmetric (Cartesian) tensors of rank 2 in three dimensions.

2
 (e) The set of all 2 2 matrices whose trace is zero.
d2 y dy
(f) The set of all solutions of the differential equation 3 + 2y = 0.
dx2 dx
(g) The set of all n n unitary matrices. (U is unitary iff U U = U U = I.)

(h) The set of all n n hermitian matrices (with multiplication by real scalars).

2. In the three-dimensional LVS with basis vectors

1 0 0
|1 i = 0 , |2 i = 1 and |3 i = 0 ,
0 0 1

find a vector |i such that hi |i = 1 for i = 1, 2 and 3.

3. In a three-dimensional LVS, consider the three vectors

1 1 1
|1 i = 1 , |2 i = 1 , |3 i = 0 .
1 0 1

(a) Show that they are linearly independent.

(b) Use the Gram-Schmidt procedure to construct an orthonormal basis {|i i}
from these three vectors.

4. The Cauchy-Schwarz inequality is of fundamental importance. It says that

|h |i| k k k k ,

the equality holding if and only if |i and |i are linearly dependent. In terms of or-
dinary vectors in Euclidean space, it amounts to saying that the cosine of the angle
between two vectors has a magnitude between 0 and 1, the limiting value of unity
occurring if and only if the vectors are collinear. Establish the Cauchy-Schwarz
inequality.

Hint: Consider the inner product h + a | + a i where |i, |i V, and a is an

arbitrary complex number. Choosing a appropriately leads to the desired inequality.

5. The triangle inequality: Use the Cauchy-Schwarz inequality to establish the

triangle inequality (or Minkowski inequality)

k + kk k + k k

for any two vectors |i and |i V.

6. Here are a couple of useful results:

(a) Let |i and |i be two linearly independent vectors in a real LVS. Find the
value of the (real) scalar that makes k k a minimum.
(b) Same problem, in a complex LVS. (Note that any complex number and its
complex conjugate are linearly independent!)

3
7. Show that the 2 2 unit matrix I and the three Pauli matrices, namely,
       
1 0 0 1 0 i 1 0
I= , 1 = , 2 = and 3 =
0 1 1 0 i 0 0 1
form a basis in the LVS of all 2 2 matrices. What is the corresponding basis if we
restrict ourselves to the set of all traceless 2 2 matrices?

8. Which of the following infinite sequences (x1 , x2 , ) listed below belong to `2 ,

the LVS of square-summable sequences?
(a) xn = (1)n (ln n)/n
(b) xn = n!/(2n)!
(c) xn = (2/n)n
(d) xn = (2n + 1)/(3n + 4)2
(e) xn = en /nn
(f) xn = 2n/2 .

9. Identify the functions that belong to L2 ( , ), the LVS of square-integrable

functions of a real variable x (, ).
(a) f (x) = (x2 + 1)1/4
(b) f (x) = ex cos x
2
(c) f (x) = e1/x
(d) f (x) = (sin x)/x
2
(e) f (x) = x3 ex
(f) f (x) = (tanh x)/x .

10. In an n-dimensional LVS, consider the vectors |k i (k = 1, 2, . . . , n) given by

 T

T 1 1
|1 i = 1 0 0 0 , |2 i = 0 0 , ,
2 2

1 T
 
1 1 1
|n i = ,
n n n n
where the superscript T denotes the transpose. Construct a vector |i such that
hk |i = 1 for every k (1 k n).

11. The set of all n n matrices (with complex entries) forms an LVS. The inner
product of two elements in this space may be defined as
def.
(A, B) = Tr (A B),
where A is the hermitian conjugate of A, and Tr denotes the trace.
(a) If A is hermitian, show that
1
(A, A) (I, A)2 .
n
(b) If A is an arbitrary n n matrix, and U is an unitary n n matrix, show that
1
(A, A) |(U , A)|2 .
n

4
Operators on a linear space
Matrices as operators in an LVS: We may view (n n) matrices as the rep-
resentations of operators in an n-dimensional LVS. Every n-dimensional LVS is
isomorphic to n-dimensional Euclidean space. The natural basis in this space is
given by the ket vectors represented by the column matrices

1 0 0
0 1 0

|1 i = 0 , |2 i = 0 , , |n i = 0 .

.. .. ..
. . .
0 0 1

This space is self-dual. The natural basis in the dual space are the bra vectors hi |
represented by the row matrices that are the hermitian conjugates of the column
matrices above. It is then easy to see that the operator |i ihj | is represented by
a matrix whose ij th element is unity, all its other elements being zero. In other
words, any matrix A with elements aij can be regarded as the representation of an
abstract operator A given by
n X
X n
A= aij |i ihj |.
i=1 j=1

Thus the n2 operators |i ihj | (where i , j = 1, 2, . . . , n) comprise the natural basis

for all operators acting on the vectors of the LVS. The orthonormality of the natural
basis immediately yields
aij = hi |A|j i.
This is why an object like hi |A|j i is called (what else!) a matrix element in
quantum mechanics, even when the LVS is infinite-dimensional, and even when the
basis set itself is a nondenumerable basis or a continuous basis. We shall exploit
this fact to use the same symbol for both the abstract operator as well as its matrix
representatione. g., we shall write A for both A and the matrix representing it
(in the natural basis, unless otherwise specified). Which of the two is meant will be
clear from the context.

Many of the definitions that follow below are obvious in the case of finite-
dimensional vector spaces. It is the case of infinite-dimensional spaces that is non-
trivial, and in which care must be exercised.

The domain of an operator: Let V denote a linear vector space. Then the
domain DA of the operator A is the subset of V such that the operator A acting
on each element |i D yields an element A|i = |i V .

The range of an operator: The set A consisting of elements |i , as |i runs

through all the elements of D , is called the range of the operator A .

The inverse of an operator: If A maps each pair of different elements of DA

into a pair of different elements of A , then A has an inverse A1 which maps
the elements of A into the elements of DA . We then have
A1 |i = |i if and only if A|i = |i.
Equality of operators: Two operators are equal if they have the same domain,
and if they have the same action on each given vector in their common domain.

5
Linear manifolds: A subset U of V is a linear manifold if the following property
is satisfied: given any pair of elements |i , |i U , any arbitrary linear combina-
tion |i + |i (where and are scalars) is also an element of U .

Cauchy sequences of operators and complete vector spaces: Recall that

a sequence of complex numbers zn is a Cauchy sequence if |zn zm | 0 when
both n and m tend to . A Cauchy sequence is guaranteed to converge to a limit,
i. e., limn zn exists, and is some complex number. A set that contains all its
limit points is called a closed set.

In the same way, we can speak of sequences of vectors in an LVS. Such a sequence
of vectors, say {|n i}, is a Cauchy sequence if limn,m kn m k = 0. Every
Cauchy sequence of vectors converges to some limiting vector, i. e., limn |n i
exists. If the limit vectors of all the Cauchy sequences of an LVS V also lie in V,
then V is said to be a complete LVS. Every finite-dimensional LVS is complete.

Subspaces: A linear manifold U in an LVS V is a subspace of the LVS if it is

complete: that is, all Cauchy sequences of vectors belonging to U also lie in U.

It turns out that it is easy to test whether a subset U of vectors in an LVS V is

a subspace or not. All that is needed for this to be so is the following: (i) U must
contain the null vector; (ii) the sum of any two vectors in U must lie in U; (iii) any
scalar multiple of any vector in U must lie in U. A subspace of an LVS is also an
LVS, with the same operations of addition and scalar multiplication as the original
LVS.

Hilbert spaces are of primary importance in quantum mechanics: the state vec-
tor of a quantum mechanical system is a vector in some Hilbert space. A Hilbert
space is a linear vector space that

(i) is complete, and

(ii) is equipped with an inner product.

A separable Hilbert space is one that has a countable or denumerable basis,

finite or infinite (as opposed to an LVS that has no countable basis, but only non-
denumerable basis sets). We shall only be concerned with separable Hilbert spaces.

Dimensionality of subspaces: If U1 and U2 are subspaces of an LVS, then so

are their sum (U1 + U2 ) and intersection (U1 U2 ). The dimensionalities of these
subspaces are related by
dim (U1 U2 ) = dim U1 + dim U2 dim (U1 + U2 ).

Linear operators: A is a linear operator acting on the vectors of an LVS if

(i) DA is a linear manifold, and

(ii) A ( |i + |i) = A|i + A|i for any scalars and .

Combinations of operators: If A and B are linear operators in V , then any lin-

ear combination C = aA + bB (where a and b are scalars) is also a linear operator
with domain DC = DA DB . The operators AB and BA are also linear operators.
If A and B are bounded linear operators in all of V , then so are AB and BA.

6
The adjoint of an operator: Let A be an operator A acting on the vectors of a
linear space V. Let (, ) denote the inner product in V , i.e., (, ) h|i. If,
for every pair of vectors |i and |i in V, there is an operator B such that

( , A) = (B , ),

then the operator B is the adjoint of the operator A, and will be denoted by A .
Thus the adjoint of the equation

A |i = |i is h| A = h|.

It is easy to see that in the case of a finite-dimensional LVS of n dimensions, in

which operators are representable by (n n) matrices, the adjoint A is simply the
hermitian conjugate (i.e., complex conjugate transpose) of the matrix A.

Self-adjoint operators: An operator A is self-adjoint if A = A . For operators

represented by (finite-dimensional) matrices, it is trivial to verify whether A is self-
adjoint or not. In more general cases, we must not only ensure that the operator
A as identified from the relation ( , A) = (A , ) is the same as the operator
A, but also that DA = DA . Only then can we assert that A = A , i.e., that A is
self-adjoint.

In general, it turns out that DA DA , i.e., the domain of the adjoint of

an operator is larger than that of the operator itself. In that case the equality
( , A) = (A , ) for all , DA merely means that the operator A is sym-
metric.2 Then, provided certain conditions are met, it is possible to find a so-called
self-adjoint extension (or extensions) of the operator. Broadly speaking, this is
done by enlarging the domain of A and shrinking that of A till they match. The
physical importance of the property of self-adjointness arises from the fact that,
in quantum mechanics, observables are represented by self-adjoint operators. Such
operators have only real eigenvalues.

Projection operators: In general, given an orthonormal basis {|j i} in an LVS,

the set of operators {|i ihj |} forms a basis for the operators acting on the vectors
in the LVS. Of these, the diagonal members of the set, namely, Pj = |j ihj | , are
projection operators. It is easy to see that

Pj2 = Pj , Pj (I Pj ) = 0 , and Pj = Pj .

These properties serve to define projection operators, in general.

The norm of an operator: Let A be a linear operator acting on the vectors

in an LVS. The square of the norm kAk of A is defined as the least upper bound
or supremum of the ratio (A , A)/( , ) as |i runs over all the vectors in the
domain of A. That is,

(A , A ) h|A A|i
kAk2 = sup = sup .
|i DA ( , ) |i DA h|i

Hence
kA k
kAk = sup .
|i DA kk
2 In the physics literature, we often say A is hermitian. Strictly speaking, however, the term

hermitian applies to operators that are symmetric as well as bounded.

7
If a is any scalar, then the norm of the operator a A is given by
kaAk = |a| kAk.
The norm of a combination of operators satisfies the inequality
kABk kAk kBk.

Bounded and unbounded operators: An operator A is a bounded operator

if kAk < . If the norm of A is infinite, then A is an unbounded operator.

Several identities and relationships that one takes for granted based on their validity
for finite-dimensional matrices are not necessarily valid for unbounded operators.
Hence, as already mentioned, caution must be exercised when dealing with such
operators, which occur quite frequently in quantum mechanics.

1. Let A and B be linear operators in an LVS, with respective domains DA , DB

and ranges A , B . What are the domains of the operators AB and BA?

2. Let |i = (x1 , x2 , x3 , . . .) `2 . Consider operators A1 , A2 , A3 , A4 , A5 whose

actions on an element of the linear space are given, respectively, by
A1 |i = (x2 , x3 , x4 , . . . ),
A2 |i = (0 , x1 , x2 , x3 , . . . ),
A3 |i = (1! x1 , 2! x2 , 3! x3 , . . . ),
A4 |i = (x1 /1! , x2 /2! , x3 /3! , . . . ),
A5 |i = (2x1 , x2 , x3 , . . . )
Examine whether each of the operators A1 , . . . , A5 (i) has an inverse; and (ii) is
bounded; if so, find the value of the norm of the operator concerned.

3. We have seen that the momentum operator of a particle moving on the x-

axis has the representation i~ d/dx when acting on position-space wave functions
(x, t). This is a direct consequence of the fundamental canonical commutation
relation [x, p] = i~ I between the position and momentum operators, where I is the
unit operator. (We shall generally omit the superscriptthat is used sometimes to
denote operators.) More generally, for three-dimensional motion, the momentum
operator has the representation i~ when acting on position-space wave functions
(r, t). This is the underlying reason why differential operators occur so frequently
in quantum mechanics. We have also seen why the space L2 occurs naturally in
quantum mechanics.

In the vector space L2 ( , ) of square-integrable functions of a real variable

x, find the adjoints of the following operators. Here a is a real constant, and m, n
are positive integers.

(i) x d/dx (ii) x2 + d2 /dx2 (iii) exp (i a x) (iv) exp (i a d/dx) (v) xm dn /dxn .

4. Use the Cauchy-Schwarz inequality in an appropriate function space to show

that Z 2
ex
dx 2 1/2
< 3/4 .
(x + 1)

8
5. Which of the following are bounded operators in L2 ( , )?
2
(i) d/dx (ii) xn (n = 1, 2, . . . ) (iii) ex /2
(iv) (x2 + 1)1 .

6. The commutation relations

[x, I] = 0, [p, I] = 0, [x, p] = i~I

between the position operator x, momentum operator p and the unit operator I
specify a Lie algebra, called the Heisenberg algebra.3
(a) Show that all functions of x and p up to quadratic functions also form an
algebra: that is, linear combinations of the operators I, x, p, x2 , p2 , xp and px
form a closed algebra in the sense that their mutual commutators are again
linear combinations of these operators.
(b) Verify that, once polynomials of any order higher than quadratic are consid-
ered, no such closed algebra is possible with finite order polynomials: polyno-
mials of arbitrarily high order are generated by the commutators. The only
possible algebra then involves all powers of the operators x and p. (This is
called the W algebra.)

7. A linear operator a and its adjoint a in a certain LVS are found to satisfy the
commutation rule [a, a ] = I, where I is the unit operator.4

(a) Find the commutators [an , a ] and [a , (a )n ] , where n is any positive integer.
(b) Hence find a simpler expression for the operator ea a ea .

8. An extension of the uncertainty principle: Let A and B be the self-adjoint

operators corresponding to two of the observables of a system, and let

AB BA = i C and AB + BA = D.

Thus C and D are also self-adjoint operators. Consider the state given by

|i = (A + iB)|i,

where is an arbitrary complex number and |i is an arbitrary state of the system.

Using the fact that h|i 0 for all , show that

(A)2 (B)2 41 hCi2 + hDi2 .

3 Recall that the concept of a Lie algebra has been introduced in the Classical Physics course. In

that context, the Poisson bracket operation specifies the Lie algebra of functions of the phase space
variables (q1 , . . . , qn , p1 , . . . , pn ) of a classical Hamiltonian system with n degrees of freedom.
4 As you know, this commutation relation, together with the obvious ones [a, I] = 0 = [a , I],

is just the Heisenberg algebra, written in terms of suitable dimensionless linear combinations of x
and p.

9
Operator identities
Identities between mutually noncommuting operators are of great importance and
use, as they are required very frequently in quantum mechanical calculations. What
follows are some of the most common of these identities. Let A and B be linear
operators in an LVS with a common domain. In general, A and B do not commute
with each other, that is, [A, B] AB BA 6= 0.

1. Perturbation expansion for the inverse of an operator: Frequently, we

face the problem of finding the inverse of an operator of the form (A +  B), where
the inverse A1 of A is known, and  is a scalar of sufficiently small magnitude such
that k Bk  kAk. We can then write a so-called perturbation expansion for
the inverse (A +  B)1 (that is, an infinite series in powers of ), as follows:

(A +  B)1 = A1  A1 B A1 + 2 A1 B A1 B A1 + . . .

Establish this result by starting with the identity

(A +  B)1 = A1  A1 B (A +  B)1 .

2. Hadamards lemma is an extremely useful operator identity, from which a

host of other operator identities can be derived. Let denote a scalar constant.
Then
2   3 h  i
e A B e A = B + [A, B] + A, [A, B] + A, A, [A, B] + .
2! 3!
Derive this result by defining the operator-valued function

F () = e A B e A .

Now obtain a first order differential equation (in ) satisfied by F (), and solve it,
using the boundary condition F (0) = B.

Hadamards lemma may be regarded as an identity between two analytic func-

tions of the complex variable . The result is valid, by analytic continuation, for
all complex satisfying || < . Re-labeling as i, we have a useful form of the
lemma, namely,

2  3 h  i
ei A B ei A = B + i [A, B]

A, [A, B] i A, A, [A, B] + .
2! 3!
As I have already indicated, Hadamards lemma is the starting point for the deriva-
tion of a large number of operator identities and relations.

3. Another useful operator relation that is closely related to Hadamards

lemma is as follows. Let A and B be two operators that do not commute with each
other, in general. Define the operator Bt by

Bt = eiAt B eiAt ,

where t is a parameter. (A and B do not depend on t.) Verify that Bt is a solution

of the integral equation
h Z t i
Bt = B + i A , dt 0 Bt 0 .
0

10
The Zassenhaus formula: It is often required to find the exponential of a linear
combination of A and B. Since A and B do not commute with each other, this
cannot be written as the exponential of A times the exponential of B. In general,
exp (A + B) can only be expressed as an infinite product of operators of the form

eA+B = eA eB eZ1 eZ2 eZ3 . . .

where
12 [A, B]

Z1 =


1
  1 
Z2 = 6 A , [A, B] + 3 B , [A, B]



h i h i h i
1
A , A , [A, B] 81 A , B , [A, B] 18 B , B , [A, B]
  
Z3 = 24

This is the Zassenhaus formula. The subsequent terms in the product involve higher-
order multiple commutators that can be determined by a systematic but quite te-
dious iterative procedure.

The Baker-Campbell-Hausdorff (BCH) formula is the complement of the

Zassenhaus formula, in the sense that it expresses the product of exponentials,
eA eB , in terms of the exponential of an infinite sum of operators involving multiple
commutators. Setting [A, B] = C, the formula is
n o
(exp A) (exp B) = exp A+B + 12 C + 12 1 1 1
 
[A , C] 12 [B , C] 24 B , [A , C] + ,

where in the exponent on the right-hand side stands for an infinite sum of
higher-order multiple commutators. As in the case of the Zassenhaus formula, the
successive terms in the sum can be found by a recursive procedure.

4. In specific cases, if more information is available about the multiple commutators

that appear in the Zassenhaus and BCH formulas, these formulas may be simplified
somewhat. The simplest of these special cases, and the most important one, obtains
when both A and B commute with their commutator C = [A, B]. (Sometimes this
happens because C is just a constant multiple of the unit operator, but this is not
necessary for the validity of the results that follow below.) The Zassenhaus and
BCH formulas can then be derived in a fairly simple manner.
(a) Show that, if [A, C] = 0 = [B, C], then
1
eA+B = eA eB e 2 [A,B] .

You can derive this formula this by obtaining and solving a first-order differ-
ential equation for the operator F () defined as

F () = e(A+B) ,

where is a scalar (not an operator). Use the obvious boundary condition

F (0) = I. Keep track of the fact that A and B do not commute with each
other.
(b) Similarly show that, if [A, C] = 0 = [B, C], then
1
eA eB = eA+B+ 2 [A,B] .

11
 (c) Apply the result to the important case A = iax , B = ibp where a and b are
real scalars and x, p are the canonical position and momentum operators in
one dimension, to find the commutator [eiax , eibp ]. This is called the Weyl
form of the canonical commutation relation.

5. Here is a related but slightly different way of arriving at the Zassenhaus formula
when the commutator of two operators is the unit operator. Let A and B be two
operators such that [A, B] = I, the unit operator. Consider the linear combination
L = A + B, where and are arbitrary complex scalars (not operators).
Ln
(a) Show that = nALn1 21 n(n 1)Ln2 .

eL
= A 12 eL .


(b) Show that

1
(c) Integrate this last equation suitable to show that e A+B = e A e B e 2 .
Apply this to the important case B = a, A = a , so that [A, B] = [a, a ] = 1.
Setting = , we get the important relation

a
a 21 ||2
e a = e a e e .

The operator on the left-hand side is called the displacement operator. It gen-
erates coherent states, as we will see a little later.

12
Motion in one dimension
Many of the basic principles, properties and consequences of quantum mechanics
can be illustrated very effectively in the case of a particle moving in one spatial
dimension, the counterpart of a classical Hamiltonian system with one degree of
freedom. I consider below some of the prototypical problems in this regard.

1. Wave packet dynamics of a free particle: Consider a particle moving

freely in one dimension (the x-axis). As there is no force acting upon it, in the clas-
sical case it would have a definite momentum, say p0 . The corresponding quantum
mechanical momentum eigenstate is represented, in the position basis, by a wave
function proportional to eip0 x/~ (i.e., a plane wave.) But such a wave function is
not normalizable in (, ). One way of getting around this problem is to confine
the particle to a large but finite interval on the x-axis (a box). Another way is to
give up a strict momentum eigenstate, and to use a normalizable superposition of
plane waves peaked (in wave number) about the wave number p0 /~. Lets examine
the time evolution of such a wave packet.

The most common superposition of this kind uses a Gaussian wave packet
that is centered in momentum about p0 , and in position about some point x0 , to
start with. Accordingly, we start with an initial state | (0)i of the particle at t = 0
that is represented by the position-space wave function
1 2 2
hx| (0)i (x, 0) = 2 1/4
eip0 x/~ e(xx0 ) /2 ,
( )

where is a positive constant. The positional probability density of the particle is

a Gaussian centred about the point x0 , with a width proportional to .

R
(a) Check that the wave function is normalized to unity, i.e.,
dx |(x, 0)|2 = 1.

The momentum-space wave function at any instant of time is the Fourier transform
of the position-space wave function, and is given by
Z Z
1
(p, t) hp| (t)i =
e dx hp|xihx| (t)i = dx eipx/~ (x, t).
(2~)1/2

We have used the relation hp|xi = (2~)1/2 eipx/~ . 5

(b) Show that the normalized momentum-space wave function of the particle at
t = 0 is given by
 2 1/4 2 2 2
(p,
e 0) = eix0 (pp0 )/~ e (pp0 ) /2~ .
~2

Apart from the phase factor exp [ix0 (pp0 )/~], this is again a Gaussian in p centred
about the value p0 . Note that the width of the Gaussian is now proportional to
1/ 2 , i. e., it is proportional to the reciprocal of the width of the wave packet in po-
sition space. This is in accord with the general property that a narrow distribution
in x implies a wide one in p, and vice versa.

5 This is the usual normalization convention. It follows that hx|pi = hp|xi = (2~)1/2 eipx/~ .

13
 (c) Verify that the initial expectation values of the position and momentum are
hx(0)i = x0 and hp(0)i = p0 , respectively. Evaluate the initial uncertainties
x(0) and p(0) at t = 0. Note that the uncertainty product x(0) p(0) =
1
2 ~. The initial state is therefore a minimum uncertainty state.

The time evolution of the state of the particle is given, of course, by the Schrodinger
equation, (i~) d| (t)i/dt = H| (t)i. The formal solution to this equation is
| (t)i = exp (iHt/~) | (0)i. The problem is then to evaluate the exponential
of the Hamiltonian. Since H = p2 /(2m) in the case of a free particle, and does not
involve the position operator at all, it is obvious that the problem is most easily
solved in the momentum basis. In this basis, the operator exp (iHt/~) is just
multiplication by exp [ip2 t/(2m~)]. This makes it trivial to write down the wave
function (p,
e t). We have
2
(p,
e t) = hp | (t)i = hp | eiHt/~ | (0)i = hp | eip t/(2m~)
| (0)i
2 2
= eip t/(2m~)
hp| (0)i = eip t/(2m~)
(p,
e 0).

(d) Write down (p,

e t) explicitly, and evaluate its inverse Fourier transform to
determine the position-space wave function (x, t), according to
Z
1
(x, t) = dp eipx/~ (p,
e t).
(2~)1/2
R
(e) Verify that this wave function remains normalized, i.e.,
dx |(x, t)|2 = 1.
(f) Show that the expectation values of p and x at any time t are given by
p0 t p0 t
hp(t)i = p0 = hp(0)i, while hx(t)i = x0 + = hx(0)i + .
m m
In other words, the expectation values of the position and the momentum are re-
lated to each other in exactly the way in which the position and momentum of a
classical free particle would be related. This is a special case of what is known as
Ehrenfests Theorem.

(g) Show that the uncertainty in the momentum at any time t, p(t), remains
equal to its value at t = 0. Evaluate x(t) to show that the width of the wave
packet in position space broadens as t increases.

Hence the uncertainty product x(t) p(t) increases with time, and the state of the
particle is no longer a minimum uncertainty state for any t > 0. The dispersion
of the wave packet in position space is a consequence of the fact that the relation
between (the eigenvalues of) the energy and the momentum is not a linear one for
a free non-relativistic particle: if we set E = ~ and p = ~k, the relation between
the frequency and wave number for a free particle reads (k) = ~k 2 /(2m). Hence
the wave velocity /k is not identically equal to the group velocity d/dk, and
dispersion occurs.

2. Free particle in a box: The energy levels and the corresponding normalized
position-space wave functions of a particle moving freely in a one-dimensional box
(0 < x < L) are given by
( p
n 2 2 ~2 2/L sin (nx/L), 0 < x < L
En = and n (x) =
2mL2 0, all other x

14
where n = 1, 2, . . . . Although the wave function
r 
2 einx/L einx/L 
n (x) =
L 2i
looks like a superposition of two plane waves (of wave numbers n/L and n/L,
respectively), you must not jump to the conclusion that the particle is in a su-
perposition of just two momentum eigenstates when it is in an eigenstate of the
Hamiltonian. This erroneous conclusion appears to be supported by the fact that
the particle is free inside the box, and in the classical case it merely bounces back
and forth between x = 0 and x = L, the magnitude of its momentum being con-
served in time. In fact, however, there is a continuous distribution of momentum
in an eigenstate of H, as you will see shortly.

(a) Verify that the set of wave functions {n (x)} satisfies the orthonormality
condition Z
dx n (x) l (x) = nl

and the completeness relation

X
n (x) n (x0 ) = (x x0 ).
n=1

(b) Find the corresponding set of momentum-space wave functions, given by

Z
1
en (p) = dx eipx/~ n (x).
(2~)1/2

It is important to note that en (p) is not an eigenfunction of the momentum operator.

Although the system has been termed a free particle in a box, the fact is that the
particle is in a potential that is not identically equal to zero for all x. Instead, we
have (
0 for 0 < x < L
V (x) =
for all other x.
Therefore p does not commute with the Hamiltonian of the particle, and a stationary
state of the particle is not a momentum eigenstate as well. Nor is it a superposi-
tion of just two momentum eigenstates corresponding to momentum eigenvalues
n~/L, even though the form of the position-space eigenfunction n (x) might
suggest that this is so. The explicit expression for en (p) should make it amply clear
that the momentum-space wave function is spread out over all values of p.
(c) Write down the momentum-space probability density |en (p)|2 when the par-
ticle is in the stationary state corresponding to energy En , and simplify the
expression.
(d) Sketch the probability density of the momentum in the ground state,

4L~3
 
2 2 pL
|1 (p)| = 2 2
e cos , ( < p < )
(p L 2 ~2 )2 2~

as a function of p.

15
 (e) Calculate the position and momentum uncertainties (x)n and (p)n in the
energy eigenstate n (x), and hence write down the value of the uncertainty
product (x)n (p)n . Note that the uncertainty product is larger than its
lowest allowed value 12 ~ even in the ground state of the particle.

3. Particle subjected to a constant force: A particle of mass m moves in one

dimension under a uniform, constant force F . Its Hamiltonian is thus

p2
H= F x.
2m
(a) Let E denote an eigenvalue of H (i.e., an energy eigenvalue). What are the
allowed values of E? (This should be obvious on physical grounds.)
(b) Write down the time-independent Schrodinger equation for the momentum-
space wave function eE (p) corresponding to the eigenvalue E.

(c) Normalize the solution eE (p) such that

Z
dp eE (p) eE 0 (p) = (E E 0 ).

4. Reflection and transmission in the presence of a potential barrier:

Consider a quantum mechanical particle of mass m moving on the x-axis in the
presence of a potential barrier given by

0, x<0
V (x) =
V0 , x > 0

where V0 is a positive constant.

(a) If the energy E of the particle is greater than V0 , show that the reflection and
transmission coefficients are given by
2
k k0 4kk 0

R= 0
and T =
k+k (k + k 0 )2
p
respectively, where ~k = 2mE and ~k 0 = 2m(E V0 ) .
(b) What happens to R as (i) E V0 from above, and (ii) E becomes very much
larger than V0 ?
(c) What happens in the case 0 < E < V0 ?

5. Rectangular potential barrier: Now consider the potential barrier

V0 , 0 x L

V (x) =
0 otherwise

where V0 is a positive constant.

(a) For E > V0 , show that the transmission coefficient is given by

(2kk 0 )2
T = .
(k 2 k 02 )2 sin2 (k 0 L) + (2kk 0 )2

16
(b) Calculate the reflection coefficient R and verify that R + T = 1.

(c) Pass to the limit L 0 and V0 such that lim (V0 L) = . Verify that
the transmission coefficient T is now given by

2E~2
T = .
2E~2 + m2

(d) Show that this last expression precisely the transmission coefficient in the case
of the -function potential barrier V (x) = (x).
(e) For 0 < E < V0 , show that the transmission coefficient becomes

(2k0 )2
T = ,
(k 2 + 02 )2 sinh2 (0 L) + (2k0 )2
p
where ~0 = 2m(V0 E).

17
The linear harmonic oscillator
Energy eigenvalues and eigenfunctions: Here is a quick recapitulation of the
basic properties of the quantum mechanical linear harmonic oscillator. The impor-
tance of the harmonic oscillator is not restricted to the fact that it is a completely
solvable quantum mechanical problem. It is closely related to systems with many
degrees of freedom, such as collections of bosons, quantum fields (including radia-
tion) satisfying Bose statistics, etc.

The Hamiltonian of the linear harmonic oscillator is given by

p2 1
H= + m 2 x2 ,
2m 2
where the position and momentum operators satisfy the commutation relation
[x , p] = i~I, I being the unit operator. It is very convenient to introduce the
dimensionless position and momentum according to
 m 1/2 1
=x , = p.
~ (m~)1/2
Then
1
H= ~ ( 2 + 2 ), where [ , ] = i I.
2
The operators a and a are defined as
1 1
a = ( + i), a = ( i).
2 2

The fundamental commutation relation now becomes [a , a ] = 1,6 while the Hamil-
tonian reads
H = ~ a a + 21 .

The eigenvalues of a a are nondegenerate, and are given by the set of non-
negative integers 0, 1, . . . . The combination a a is therefore called the number
operator. It follows at once that the eigenvalues of H are also nondegenerate, and
are given by
En = ~ n + 21 , where n = 0, 1, . . . .

The corresponding eigenstates {| n i}, satisfying the eigenvalue equation a a | n i =

n | n i, are called Fock states. They form a complete orthonormal set. The ground
state | 0 i is also called the vacuum state.7 The action of a and a on | n i is given
by p
a |0i = 0, a | n i = n |n 1i, a | n i = (n + 1) |n + 1i.
This explains why a and a are termed the raising and lowering operators, re-
spectively. It is most important to note that the eigenvalue spectra of a a, and
therefore H, are bounded from below, and that there exists a state | 0 i that yields
zero when acted upon by a.

The normalized position-space wave function of the oscillator in the stationary

state | n i is n (x) hx| n i. It is the regular solution of the differential equation

~2 d 2 n 1
+ m 2 x2 n = En n ,
2m dx2 2
6 In a slight abuse of notation, we often write 1 for the unit operator. No confusion should arise

as a result.
7 In the context of the radiation field, this state corresponds to the zero-photon state.

18
and is given by
1  m 1/4 mx2 /(2~) p

n (x) = e Hn x m/~ ,
2 n! ~
n

where Hn denotes the Hermite polynomial of order n. The first few of these poly-
nomials are

H0 (u) = 1, H1 (u) = 2u, H2 (u) = 4u2 2, H3 (u) = 8u3 12u, H4 (u) = 16u4 48u2 +12.

The Rodrigues formula for these polynomials is

2 dn u2
Hn (u) = (1)n eu e .
dun
The generating function of the Hermite polynomials is given by

2 X tn
et +2tu
= Hn (u) .
n=0
n!

Hn (u) is the regular solution of Hermites differential equation,

d2 Hn dHn
2
2u + 2n Hn = 0.
du du
The oscillator eigenfunctions n (x) form a complete orthonormal set of functions
in L2 ( , ). The orthonormality relation is
Z
dx n (x) l (x) = nl ,

while the completeness relation is

X
n (x) n (x0 ) = (x x0 ).
n=0

1. Consider a linear harmonic oscillator in its ground state. Calculate the total
probability that the position of the oscillator is (i) in the range (~/2m)1/2
x (~/2m)1/2 ; (ii) outside this range. Note that the regions in (i) and (ii),
respectively, are the allowed and forbidden regions for a classical simple harmonic
oscillator whose total energy is E0 .

2. Using operator methods, show that the expectation values of the kinetic energy
and the potential energy of the oscillator in the stationary state | n i are equal.
Hence show that their expectation values in an arbitrary normalizable state are
equal.

3. A shifted harmonic oscillator: (This is a trivial problem!) Suppose a term is

added to the Hamiltonian of thje linear harmonic oscillator, so that it is now given
by
H = ~ a a + 21 + (a + a ) ,

where is a real constant.

(a) What does this new Hamiltonian correspond to, physically?
(b) Find the eigenvalues of this Hamiltonian.

19
 (c) What value(s) should take in order that the ground state energy be exactly
zero?

4. Other perturbations of the linear harmonic oscillator: Consider the

perturbed Hamiltonian H = H0 + H 0 , where

p2 1
H0 = + m 2 x2
2m 2
is the Hamiltonian of a linear harmonic oscillator, is a positive constant of appro-
priate physical dimensions, and H 0 is given, respectively, by (i) x2 (ii) x3 (iii) x4 .
Calculate, in each case, the new energy levels correct to second order in the small
parameter , and the corresponding eigenstates correct to first order in . Compare
these with the exact expressions for these quantities in those cases, if any, in which
the problem can be solved analytically.

5. Propagator for the linear harmonic oscillator: In the foregoing, we have

focused our attention on the stationary states of the linear harmonic oscillator.
Here is an aspect of interest pertaining to the time-dependent problem. The time-
dependent Schrodinger equation for a linear harmonic oscillator in the position basis
is given by
(x, t) ~2 2 (x, t) 1
i~ = + m 2 x2 (x, t) .
t 2m x2 2
Verify that the solution to this equation is given by
Z
(x, t) = dx0 K(x, x0 ; t) (x0 , 0),

where
 m 1/2 h im  i
K(x, x0 ; t) = exp (x2 + x02 ) cos t 2xx0 .
2i~ sin t 2~ sin t
First check that (x, t) obeys the time-dependent Schrodinger equation. Next,
check to see whether the proper initial condition is satisfied at t = 0, i. e., that
(x, t) reduces to the given initial function (x, 0).

The quantity K(x, x0 ; t) is called the propagator because it takes us from the
solution at time t = 0 to the solution at any subsequent instant of time t > 0. Note
that (x, t) at any given point x depends on the initial wave function (x0 , 0) at all
points x0 .

6. A pair of coupled harmonic oscillators: Before going on to other aspects of

the linear harmonic oscillator, let us consider a pair of coupled harmonic oscillators.
The coupling is such that the total Hamiltonian can be diagonalized and the energy
levels computed exactly. This enables us to compare the results of perturbation
theory with the corresponding exact expressions, thereby gaining some insight into
the former.

The two-dimensional isotropic oscillator is given by the Hamiltonian

p21 1 p2 1
H0 = + m 2 x21 + 2 + m 2 x22 ,
2m 2 2m 2
where
[xi , xj ] = 0, [pi , pj ] = 0, [xi , pj ] = i~ ij ,

20
and the indices run over the values 1 and 2. Now consider the perturbed Hamiltonian
H = H0 + H 0 , where is a real constant and

H 0 = x1 x2 .

(a) The ground state energy of the unperturbed Hamiltonian is of course given
(0)
by E0 = ( 21 + 12 ) ~ = ~. Show that, up to the second order in the coupling
constant , the ground state energy of the perturbed oscillator is given by

2 ~
E0 = ~ + .
8m2 3

(b) The first excited state of the unperturbed Hamiltonian is doubly degener-
(0)
ate. It corresponds to the energy eigenvalue E1 = 2~. The corresponding
normalized eigenstates may be taken to be the linear combinations

| 1, 0 i + | 0, 1 i | 1, 0 i | 0, 1 i
and
2 2
where |n1 , n2 i is an eigenstate of H0 with energy ~(n1 + n2 + 1). Show
that the perturbation lifts the degeneracy of the first excited state of the
(0)
Hamiltonian, and that the energy level E1 = 2~ is split, to first order in ,
into the two energy levels
~ ~
2~ and 2~ + .
2m 2m

(c) It should be clear that the corresponding classical problem can be solved easily
by changing variables. This remains true in the quantum mechanical case as
well. Define
x1 + x2 x1 x2
X= and x = ,
2 2
and similarly
p1 + p2 p1 p2
P = and p = .
2 2
Check out the commutation relations satisfied by these new operators. Write
down the Hamiltonian H in terms of these operators, using the fact that
1 2
x1 x2 = (X x2 ).
2

(d) Hence show that the exact energy levels (eigenvalues) of H are given by

E(n1 , n2 ) = ~ n1 + 21 + ~+ n2 + 12 ,

where n1 and n2 run over the values 0, 1, . . . as usual, and

r r

= 2 , + = 2 + .
m m

Note that we must have || < m 2 in this problem. If || > m 2 , the spectrum
of the Hamiltonian is not bounded from below. (Examine what happens to the
corresponding classical potential energy.)

21
Momentum-space wave functions of the oscillator: We return to the time-
independent Schrodinger equation for the linear harmonic oscillator, and the asso-
ciated eigenstates and eigenfunctions.

We know that the Fourier transform of a Gaussian function is again a Gaussian.

This relationship is a special case of a more general fact, which is connected to yet
another interesting aspect of the quantum mechanical harmonic oscillator. From
this point onward, we set ~, m and equal to unity, for convenience and notational
simplicity.8 The Hamiltonian is then H = 21 (p2 +x2 ), with eigenvalues En = (n+ 12 ).
The position-space eigenfunctions n (x) satisfy the Schrodinger equation
 d2 
2
x + 2En n (x) = 0.
dx2
The normalized solutions are the L2 ( , ) functions
1 1 2
n (x) == ex /2 Hn (x),
1/4 n
2 n!
where Hn (x) is the Hermite polynomial of order n.

Now consider the momentum-space eigenfunctions of H. These are given by

en (p), which are also L2 ( , ) functions satisfying the Schrodinger equation
 d2 
2
p + 2E n en (p) = 0.
dp2
But this precisely the same equation as the one satisfied by the position-space
eigenfunctions n (x)! This symmetry arises because the oscillator Hamiltonian is
completely symmetrical in x2 and p2 . As a consequence, the differential operator
that acts on n (x) to yield zero is exactly the same as the one that acts on en (p)
to yield zero. The normalized momentum-space eigenfunctions must therefore be
given by
n 1 2
en (p) = 1/4 ep /2 Hn (p),
n
2 n!
where the constant n must have unit modulus, i.e., |n | = 1. In other words, the
position-space and momentum-space wave functions of the stationary states of the
harmonic oscillator are exactly the same in functional form.

Eigenfunctions of the Fourier transform operator: The fact just stated leads
to an interesting connection between the eigenfunctions {n } and the Fourier trans-
form operator.

We know that, in general, position-space and momentum-space wave functions

corresponding to the same state are just Fourier transforms of each other: that is,
Z
1
(p)
e = dx eipx (x),
2
with the inverse transformation
Z
1
(x) = dp eipx (p).
e
2
8 These factors are easily restored in any of the expressions that follow, based on dimensional

considerations.

22
(Recall that we have set ~ = 1.) What we now see is that, for the oscillator
eigenfunctions, the functional forms of n and en are exactly the same. In other
words, the functions
1 1 2
ex /2 Hn (x), n = 0, 1, 2, . . .
1/4 2n n!
are, up to a possible multiplicative constant of unit modulus, normalized eigenfunc-
tions of the Fourier transform operator in L2 ( , ).

Let us elaborate on this a little further. The Fourier transform operator, which
we may denote by F, is an integral operator that acts on functions of a real
variable to produce other functions. If we restrict our attention to functions in
L2 ( , ), then
(x) L2 ( , ) = (F)(x) L2 ( , )
as well. That is, F takes functions belonging to L2 ( , ) to other functions
in the same function space. The kernel of the integral operator F is given by
F (p, x) = (2)1/2 eipx . That is,
Z Z
1
(F)(p) (p)
e = F (p, x) (x) dx = eipx (x) dx.
2
What we have found above is that, for the oscillator eigenfunctions n , the function
and its Fourier transform Fn ( en ) are the same in functional form. This means
that these eigenfunctions are also eigenfunctions of the Fourier transform operator
F itself. n and en must therefore differ, if at all, only by an overall multiplica-
tive constant, say n . But n and en must have the same norm, by Parsevals
Theorem: Z Z
|n (x)|2 dx = |en (p)|2 dp.

Therefore, since n is normalized to unity, so is en . Hence the multiplicative con-

stant n can only be a phase factor, i.e., a complex number of unit modulus. And
this number n must be an eigenvalue of the Fourier transform operator F when it
acts on elements of the space L2 ( , ).9

7. The next task is to determine the possible values of n .

(a) Work out explicitly the Fourier transforms of the eigenfunctions n (x) for
n = 0, 1, 2, 3 and 4. You will need to use the expressions given for the first few
Hermite polynomials given in the foregoing. You will also need the Gaussian
integral
Z
2
.
eax xr dx = 21 (r + 1) 2a(r+1)/2 , (a > 0 and r > 1)
0

where is the gamma function.

(b) Verify that the eigenvalues of F in these cases are, respectively, 1, i, 1, i
and again 1. That is, check out that
e0 (p) = 0 (p), e1 (p) = i 1 (p), e2 (p) = 2 (p), e3 (p) = i 3 (p),

while e4 (p) = 4 (p) once again.

9 Another argument to support the assertion that the eigenvalues of the Fourier transform

operator must be complex numbers of unit modulus: F is a unitary operator.

23
We may therefore conjecture that the possible distinct eigenvalues of F are +1, i, 1,
and i. This is indeed the case.

(c) Check out directly that the square of the Fourier transform operator is just
the parity operator P, i.e., that

(F 2 )(x) = (P)(x) = (x).

Since P 2 = I, the fourth power of the Fourier transform operator is just the unit
operator!

(F 4 )(x) = (x), i. e., F 4 = I, the unit operator in L2 ( , ).

It is then reasonable to expect that the eigenvalues of the operator F are just the
fourth roots of unity. We conclude that:
The oscillator eigenfunction n (x) is also an eigenfunction of the Fourier trans-
form operator F in L2 ( , ), with eigenvalue eir/2 , where r = n mod 4,
so that r = 0, 1, 2, 3.

The eigenfunctions n (x) are of course eigenfunctions of P as well, since [H, P] = 0.

This property is shared by all the bound state eigenfunctions of a symmetric poten-
tial (V (x) = V (x)). What we now find is that [H, F] = 0 for the special case of
the oscillator Hamiltonian. The parity operator may be regarded as a square root
of the unit operator. The Fourier transform operator may be regarded as a square
root of the parity operator.

Representations of H, P and F: In the position basis, the oscillator Hamiltonian

is represented by the differential operator H = 21 x2 d2 /dx2 . The parity and

Fourier transform operators have the respective actions

P F
(x)
(x) (x) (x),
e (x) L2 ( , ).

F is represented by an integral operator in this basis. On the other hand, the

Hamiltonian is represented in the Fock basis (i.e., in the basis of its own eigenstates)
by the diagonal matrix H = diag (E0 , E1 , . . . ), where En = (n + 21 ). What about
the parity and Fourier transform operators P and F? As these operators commute
with H, they too are representable by diagonal matrices, with their eigenvalues on
the diagonal. In terms of projection operators, this means that we may write

X
X
X
H= En | n ih n |, P= (1)n | n ih n |, F= (i)n | n ih n |.
n=0 n=0 n=0

24
Coherent states of the harmonic oscillator
It turns out to be possible to find normalizable eigenstates of the lowering operator
a, essentially because the spectrum of the number operator a a is bounded from be-
low. (That is, there exists a state | 0 i that is annihilated by a, i.e., a state such that
a | 0 i = 0.) These state are called coherent states,10 and they have an enormous
number of very interesting and important properties. These states and their vari-
ous generalisations play a pivotal role in all of quantum optics, among other areas.11

1. Let |i denote an eigenstate of a with eigenvalue , i.e., a |i = |i. The

possible values of will be determined shortly.
P
(a) Expand |i in the Fock basis, i.e., let |i = n=0 cn | n i. Substitute this in
the eigenvalue equation for |i, and determine the coefficients cn recursively,
by equating the coefficients of each individual Fock state on either side of
the equation. Impose the normalization condition h |i = 1, to arrive at
the following expression for |i (after choosing the overall phase factor to be
unity):
n
1 2 X
|i = e 2 || | n i.
n=0
n!

Note that no condition whatsoever has been placed upon the eigenvalue . Since
we are able to find a normalizable eigenstate |i for any arbitrary complex value of
the eigenvalue , we must conclude that the eigenvalue spectrum of the lowering
operator a is, in fact, the whole of the finite part of the complex plane. Recall
that a is non-self-adjoint. Its eigenvalues can therefore be complex numbers, in
general. We now find that its eigenvalue spectrum is actually doubly continuous
(both 1 Re and 2 Im can take on values in (, ).)

(b) Show that a similar argument does not work for the raising operator a : that
is, there can be no normalizable eigenstate of a , in stark contrast to the case
of a. Note the reason why: there is no state | n i such that a | n i = 0, i.e.,
there is no upper bound to the eigenvalue spectrum of a a.

You can verify the foregoing statements in an alternative (but equivalent) way, by
working in the position representation. Let (x) h x |i be the position-basis
wave function corresponding to the CS |i. The eigenvalue equation then gives
h x | a |i = h x |i = (x). Put a = (x + ip)/ 2, and remember that p is
represented by id/dx in the position basis. This leads to a first-order differential
equation for (x),
d
+ (x 2 ) = 0.
dx
The solution is a shifted Gaussian in x, apart from a phase factor. It is clearly nor-
malizable. In contrast, if (x) is the wave function corresponding to an eigenvalue
of the operator a , the differential equation satisfied by it is
d
(x 2 ) = 0.
dx
The solution is obviously not normalizable in (, ).

10 I
will use the abbreviation CS for coherent state.
11 In
particular, the state of the radiation field in an ideal single mode laser is a CS. We will not
be concerned here with this aspect of coherent states.

25
 (c) Show that the normalized position-space wave function corresponding to the
CS |i is given by

1
2 1 )2
(x) = 1/4 e 2 (x ei2 (1 2 x)
.

(d) Hence show that the momentum-space wave function corresponding to the CS
|i is given by

2 2 )2 /2
e (p) = 1/4 e(p ei1 ( 2 p2 )
.

The displacement operator: Using the fact that (a )n | 0 i = n! | n i, |i can
be re-expressed as
1 2
|i = e 2 || e a | 0 i.
Now, since a | 0 i = 0, we have an | 0 i = 0 for all positive integer values of n. Hence

e a | 0 i = | 0 i. Therefore |i may be further re-written as
1 2
|i = e 2 || e a e a
| 0 i.

Using the BCH formula, we find

1 2
a def.
e 2 || e a e a
= e a = D().

The operator D() is called the displacement operator, for a reason which will
become clear shortly.12 In terms of this operator, the CS |i is simply

|i = D() | 0 i.

But the adjoint of D is D = e( a a) . Since the operator a a commutes
with itself, it follows at once that D D = D D = I, i.e., that D() is a unitary
operator. The CS |i is therefore nothing but a unitarily transformed vacuum
state.13

This helps us understand why the position-space wave function (x) and the
momentum-space wave function e (p) of a CS are also Gaussians, apart from a
phase factor in each case. The significance of the parameters 1 and 2 also be-
comes
clear: the peak of the Gaussian
is displaced from 0 in the vacuum state to
2 1 in position space, and to 2 2 in momentum space.

2. Coherent states are minimum uncertainty states: Since |i is a unitarily

transformed vacuum state, you might expect certain properties of | 0 i to be carried
over to all coherent states.
(a) In the CS |i, compute the expectation values of

a + a a a
x= , p = , x2 and p2 .
2 i 2
12 We should write D(, ) because the operator D is parametrized by as well as (or the

two independent variables 1 and 2 ). But it is customary to write just D(), in a slight abuse
of notation.
13 The vacuum state itself is also a coherent state, of course, corresponding to = 0.

26

(b) Hence show that (in the units we have chosen) x = p = 1/ 2 in this state,
independent of . Therefore the uncertainty product is (x)(p) = 12 in any
CS, i.e., all of them are minimum uncertainty states.

3. Non-orthogonality of CS: The states {|i, C} are normalized to unity,

but they are not mutually orthogonal.
(a) Show that, if and are any two distinct complex numbers, then
1 2
+||2 )+
h| i = e 2 (||

(b) Hence show that

2
|h| i|2 = e|| .
The overlap |h| i|2 therefore decreases quite rapidly as the distance ||
between the two states increases, but it is not identically equal to zero.

Over-completeness of CS: It can be shown that

Z
d2 |ih| = I,

where I is the unit operator and d2 stands for an integral over the whole of the
R

complex plane, i. e., a double integral over all values of 1 and 2 in the ranges
1 (, ) and 2 (, ). The fact that the right-hand side has a factor
(> 1) implies that the set of CS is an over-complete set of states.

It is of interest to ask whether there exist subsets of the set of CS that are
complete sets. The determination of such sets is a nontrivial task. It is known, for
instance, that the set of states for which lies on the unit circle in the complex
plane form a complete set; so does the set of states for which = m + ni, where m
and n are integers; and so on.

4. The mean and variance of a a in a CS: The expectation value of the

number operator a a in the Fock state | n i is of course h n | a a | n i = n, and
hence h 0 | a a | 0 i = 0. In the case of the CS |i, we have a |i = |i and hence
h| a = h|. It follows immediately that the expectation value of the number
operator in the CS |i is
h| a a |i = ||2 .
Show that the variance of the number operator a a in the CS |i is also equal to
||2 . Hence the standard deviation of this quantity, i.e., the uncertainty in a a, is
equal to || in the state |i.

It turns out that all the higher cumulants of the number operator are also equal to
||2 in the CS |i. In the context of radiation, this is a consequence of the fact that
the photon number in ideal, single-mode laser light has a Poisson distribution.

5. Use Hadamards lemma to show that

D() a D () = a and D() a D () = a .

Again, these relations tell us why D() is termed the displacement operator. Under
a unitary transformation by D(),
D() D()
the state | 0 i |i, while the operator a D() a D ().

27
Hence the relation a | 0 i = 0 is transformed to

D() a | 0 i = 0, or D() a D ()D() | 0 i = (a ) |i = 0.

Viewed thus, the defining eigenvalue equation for a CS is nothing but a unitarily
transformed or displaced version of a | 0 i = 0.

D() as an element of a Lie group: We have mentioned already that the op-
erators a, a and I are the generators of a Lie algebra, the Heisenberg algebra.
Exponentiating a linear combination of these generators will then yield the general
element of the corresponding Lie group. It is in this sense that D() is an element
of the Heisenberg group.14 The question of interest, then, is the group multiplica-
tion law.

6. Let and be any two complex numbers. Show that the displacement operator
satisfies the following group composition rule:

1
)
D() D() = D( + ) e 2 ( = D( + ) ei Im ( ) .

Generalizations: Various generalizations of coherent states are possible. This

topic has been studied in some detail, especially in the context of quantum op-
tics. We have the generalized coherent state | n, i obtained by applying the
displacement operator D() to the Fock state | n i (rather than the vacuum state
| 0 i) and normalizing the result; the photon-added coherent state | , n i ob-
tained by operating on |i with a n and normalizing the result; and so on. Other
transformations of the vacuum state are also of relevance. For instance,

14 It involves two real parameters and , in accord with the two nontrivial generators a and
1 2
a of the Lie algebra.

28
 QUIZ

1. Are the statements in quotation marks true or false?

(a) The position and momentum operators of a particle moving in one dimension
satisfy the commutation relation [x , p] = i~ I, where I is the unit operator.

It is possible to represent the operators x and p by hermitian (nn) matrices

where n is any positive integer.

(b) A particle moves in one dimension in the potential V (x). The energy spec-
trum of the particle is discrete. The potential is symmetric about the origin,
i. e., V (x) = V (x).

It follows that the position-space wave function corresponding to any sta-

tionary state of the particle is an even function of x.

(c) A particle moves in space under the influence of a potential V (r).

The particle can never be in a state in which its kinetic energy and potential
energy can simultaneously have precise values.

(d) Let A be the operator corresponding to a physical observable of a system with

Hamiltonian H, and let [A , H] 6= 0. Then:

No eigenstate of A can be a stationary state of the system.

(e) Let A , B and C be operators corresponding to physical observables of a sys-

tem, such that [A , B] = 0 and [A , C] = 0.

It follows that [B , C] = 0.

(f) The energy spectrum of a quantum mechanical particle moving under the
influence of an arbitrary potential must necessarily be either wholly discrete,
or wholly continuous, but not partially discrete and partially continuous.

(g) The existence of operators that commute with the Hamiltonian of a system
generally implies that the energy spectrum of the system is degenerate.

(h) The expectation value of the kinetic energy of a quantum mechanical particle
can never be negative.

(i) Let A and B be the operators corresponding to two physical observables of a

system.

If A and B do not commute with each other, then the product of uncertain-
ties in these two quantities can never be zero in any state of the system.

29
 (j) The eigenvalues of the operator corresponding to a physical observable of a
system are the only possible results of a measurement of that observable.

2. Fill in the blanks.

(a) The Hamiltonian of a particle moving in one dimension is given by H =
p2 /(2m) + V (x) where x and pdenote the position and momentum operators.
Then the double commutator x , [x , H] =

(b) A system can have two possible states, denoted by |1i and |2i, respectively.
These states are normalized, and mutually orthogonal. The Hamiltonian of
the system is given by
 
H = a |1ih1| |2ih2| + |1ih2| + |2ih1| ,

where a is a real constant. In the basis formed by the states |1i and |2i, H is
represented by the matrix

(c) A particle moving in one dimension is in a state |i. Its position-space wave
function is given by hx|i (x). Consider the state |i = ei a p |i where
a is a real constant and p is the momentum operator of the particle. The
position-space wave function in the state |i is then given by (x) =

(d) Let |i be a normalized coherent state of the linear harmonic oscillator, so

that a |i = |i. The expectation value of the number operator in this state
is h|a a|i =

(e) The energy levels of an isotropic two-dimensional harmonic oscillator are given
by E(n1 , n2 ) = ~ (n1 + n2 + 1), where n1 and n2 are non-negative integers.
The degeneracy of the energy level 6 ~ is

(f) The energy levels of an isotropic three-dimensional harmonic oscillator are

given by E(n) = ~ n + 32 , where n runs over the non-negative integers.

The degeneracy of the energy level En is

(g) The expectation value of the kinetic energy of a linear harmonic oscillator of
mass m and frequency in the energy eigenstate |ni is

(h) The dependence of the energy En on the quantum number n for a particle
moving in the potential kx6 (where k is a positive constant) is En

(i) The condition that an infinite sequence (x1 , x2 , . . .) be square-summable is

(j) If [x , p] = i~ I, then [x , f (p)] =

30
 1
Fun with spin 2
Rotations of the coordinate axes: In three-dimensional space, a general rota-
tion of the coordinate axes about the origin can be completely specified by a unit
vector n and an angle , where 0 . The rotation occurs in the plane whose
normal is n, and is the angle through which the coordinate frame is rotated. Well
denote such a rotation by R(n, ).15 The point is that R(n, ) is represented by dif-
ferent kinds of objects (e.g., matrices of different dimensions, differential operators,
etc.), depending on the sort of quantity whose behaviour under the rotation we are
interested in deducing.

Pauli matrices: As we have seen in class, an explicit representation of the angular

momentum algebra in the case j = 21 is provided by the Pauli matrices. Recall their
definition,      
0 1 0 i 1 0
1 = , 2 = , 3 = .
1 0 i 0 0 1
Setting Ji = 21 ~ i , it is easy to verify that the angular momentum algebra [Ji , Jj ] =
i~ ijk Jk is satisfied. That is, the matrices 12 ~ i (i = 1, 2, 3) provide a represen-
tation of the generators of rotations in three-dimensional space. As numerous
problems, both in quantum mechanics as well as other contexts, can be reduced to
two-level problems, the Pauli matrices are of fundamental importance, and play a
most useful role. Note the following basic properties of each of the Pauli matrices
i (i = 1, 2, 3):

(i) i is hermitian, traceless, with determinant equal to 1 .

(ii) i2 = I, where I is the 2 2 unit matrix; hence i1 = i .
1 0



(iii) i has eigenvalues 1 and 1 , with normalized eigenvectors 0
and 1
re-
spectively.
It is customary to denote the set of three matrices (1 , 2 , 3 ) by the vector
. This is not just a matter of notation. We know that the angular momentum
generators Ji themselves transform like the components of a vector under rotations
of the coordinate axes (in three-dimensional space). Under a rotation J transforms
according to
J0 = R(n, ) J R (n, ) = eiJn J eiJn .
Exactly the same transformation law applies to any other vector operator as well.

Here are some more properties of the Pauli matrices:

1. Find the matrices ei1 , ei2 and ei3 , where , and are any three arbi-
trary complex numbers.

2. The basis formed by I and : Show that an arbitrary 22 matrix M = ac db

can be written as a linear combination of I , 1 , 2 and 3 in a unique manner.

15 You might wonder why the range of is 0 , rather than 0 < 2. The reason is

that 0 to suffices to cover all possible rotations, because a rotation about the direction n through
an angle is exactly the same as a rotation about the opposite direction n through an angle
. (Check it out!) This seemingly simple fact has truly profound consequences. It makes the
parameter space of rotations (the space of all possible unit vectors n and all possible values of )
have a nontrivial topology. In technical terms, it makes the parameter space doubly-connected. As
a consequence, we have two distinct kinds of representations of the rotation group: the so-called
tensor representations, and the so-called spinor representations. This leads, ultimately, to the
existence of two kinds of elementary particles, namely, bosons and fermions.

31
In other words, M can be written as 0 I + 1 1 + 2 2 + 3 3 , where the el-
ements (a, b, c , d) uniquely determine the coefficients (0 , 1 , 2 , 3 ) , and vice
versa. Note, in particular, that any traceless 2 2 matrix can be expanded uniquely
as a linear combination of the three Pauli matrices.

3. Commutator and anticommutator: Show that k l = i klm m , where

the indices run over the values 1, 2 and 3. Hence verify that

k l l k [k , l ] = 2 i klm m ,

and that
k l + l k [k , l ]+ = 2 kl I .
Therefore the commutator of any two different Pauli matrices is a constant times
the third Pauli matrix, while the anticommutator of two different Pauli matrices is
identically zero.

4. The exponential of a (2 2) matrix: Use the anticommuting property of

the Pauli matrices to derive the following important relation. Let a be an ordinary
vector with Cartesian components (a1 , a2 , a3 ), and let a stand for the matrix
a1 1 + a2 2 + a3 3 . Show that
(a )
exp (ia ) = I cos a + i sin a ,
a
where a = (a21 + a22 + a23 )1/2 . This is the (2 2) matrix analog of the familiar Euler
formula ei = cos + i sin for an ordinary (real or complex) number .

The formula above is readily extended to the case of exp ( ) , where

stands for 1 1 + 2 2 + 3 3 , and (1 , 2 , 3 ) are real numbers. Simply set
i a = , i.e., a1 = i 1 , a2 = i 2 , a3 = i 3 in the formula, to show that

( )
exp ( ) = I cosh + sinh ,

where = (12 + 22 + 32 )1/2 . What has actually been done here is an analytic
continuation of the original formula from real values of the quantities (a1 , a2 , a3 )
to pure imaginary values of these variables. Finally, since any 2 2 matrix M can
be written in the form M = 0 I + , we have
n ( ) o
eM = e0 I cosh + sinh .

5. Let a and b denote any two ordinary vectors, and n any unit vector. Establish
the following identities:
(a) (a ) (b ) = (a b) I + i (a b)
(b) [(a ) , (b )] = 2i (a b)
(c) [ , (a )] = 2i (a )

(d) ( n) ( n) = i i( n) n

32
Note, incidentally, that the eigenvalues of n remain equal to 1 and 1, for all
unit vectors n.

6. Let A be a self-adjoint operator with discrete eigenvalues n , and let the corre-
sponding normalized eigenfunctions |n i form a complete set in the Hilbert space.
Then A can be written as
X
A= n Pn where Pn = |n ihn |.
n

This is called the spectral resolution of the operator A.

(a) What is the spectral resolution of the operator a , where a = (a1 , a2 , a3 )
is an ordinary vector?
The operator R(z) = (A zI)1 , where z is a complex variable, is called the
resolvent of the operator A.
(b) Find the resolvent of the operator a . Where are the singularities of R(z)
located, in the complex z-plane?

The spin of a quantum mechanical particle is its intrinsic angular momentum.

It is very important to realize that the spin of a particle is not connected to
any mechanical motion of the particle, such as rotation about an axis. The
spin of a particle is an intrinsic property of the particle, like its rest mass or
its charge, if any.
The spin is present even in the rest frame of the particle, i.e., even when its linear
momentum is zero. Like all angular momenta, the spin is also quantized. That is, if
S denotes the spin operator of a particle (this is standard notation), then the opera-
tor S2 always has the eigenvalue ~2 S(S +1), where S is the spin quantum number of
the particle. Each particle has a definite S. As in the case of all angular momenta,
the possible values of S can only be 0, 12 , 1, 32 , . . .. Particles such as electrons, pro-
tons, neutrons, etc. have S = 12 , while photons have S = 1, pions have S = 0, and
so on. Integer-spin particles are called bosons, while half-odd-integer-spin particles
are called fermions. Any particular component S n of a spin-S particle can only
have (2S +1) possible eigenvalues, given by ~S, ~(S 1), . . . , ~(S 1), ~S. This
is an equi-spaced sequence, with a spacing ~ between adjacent values. The spin-
space of a spin-S particle is thus a (2S + 1)-dimensional linear vector space. The
components of the spin operator S can thus be represented by (2S + 1) (2S + 1)
hermitian matrices, while the spin part of the state vector of the particle can be
represented by a (2S + 1) 1 column matrix in this space. Different components of
the spin operator do not commute with each other.

What is the origin of the spin of a particle? At this level, the appropriate
answer is that it is a consequence of quantum mechanics together with relativistic
(or Lorentz) invariance.
The rest mass and the spin of a quantum mechanical particle arise naturally
when we impose the requirement that its state vector (or wave function) have
a definite transformation property under inhomogeneous Lorentz trans-
formations. The latter comprise translations of the origin of the space
coordinates and of the time coordinate, rotations of the coordinate axes,
and boosts16 or velocity transformations from one inertial frame to another.
16 These (alone) are called Lorentz transformations in elementary treatments, but it is preferable

to avoid this confusing terminology.

33
As I have just said, the spin of a particle is a quantum mechanical property. (In
fact, as just stated, it requires relativistic quantum mechanics, or more accurately,
quantum field theory, to begin to understand its origin. In nonrelativistic quantum
mechanics, the spin is put in by hand.) As S is generally small for particles
such as electrons, protons, atoms, etc., ~S is infinitesimally small compared with
the magnitudes of angular momenta that occur in rotational or orbital motion of
macroscopic objects in daily life. For instance, a stone of mass m = 1 kg tied to
a string and twirled around in a circle of radius r = 1 m with a time period of
revolution equal to 1 s (or angular frequency = (2) s1 ) has an orbital angular
momentum of magnitude (2) Js. But the angular momentum must be quantized,
according to the rules of quantum mechanics. It is at once evident that the corre-
sponding quantum number ` must be enormous in this instance: it must of of the
order of 10+34 , so that the product ~` is of the order of unity. This is why we dont
see the quantization of angular momentum in daily life, when dealing with macro-
scopic objects. The quantum numbers involved are so large compared with unity
that the discreteness of the angular momentum is completely masked. The classical
limit may be obtained by letting ~ 0 and ` , such that the product ~` tends
to the classical value of the magnitude of the angular momentum. In the case of
spin, however, S is of the order of unity. Therefore there is no question of passing
to a non-zero classical limit to obtain a finite value of ~S in the limit ~ 0. It is in
this sense that one often comes across the statement, Spin has no classical analog.

Now consider the spin operator of a spin-half or S = 12 particle like the electron.
The spin space has (2 12 + 1) = 2 dimensions in this case. The spin operator is
most conveniently represented in this space in terms of the Pauli matrices, as

S = 12 ~ .

Since n (where n is a unit vector) has eigenvalues 1, this automatically ensures

that S n has eigenvalues 12 ~, as required. As is clear from the very convention we
have used for the Pauli matrices, it is customary to choose the axis of quantization
as the z-axis, so that Sz is represented by a diagonal matrix (and all other compo-
nents are not). But you must remember that this is just a matter of convention,
adopted unless explicitly stated otherwise.

We have seen that 21 ~ is indeed very small in magnitude by classical standards.

How, then, can one detect it? It turns out that intrinsic magnetic dipole mo-
ments are generally associated with particles that have nonzero spin. This also
requires that the particles be charged electrically, such as the electron, or at least
have constituents that are charged, such as the neutronwhich is electrically neu-
tral, but is made up of three charged quarks. The intrinsic magnetic dipole moment
of a particle of charge e and rest mass m is related to its spin by
ge
= S,
2m
where g is called the g-factor of the particle. For spin- 21 particles, relativistic quan-
tum mechanics leads to the result g = 2. However, quantum field theory yields
corrections to this result. The contribution from g 2 to the intrinsic magnetic
moment is called the anomalous magnetic moment of the particle concerned.
This correction can be small compared to unity (as in the case of the electron), or
indeed of the order of unity itself (as in the case of the proton and neutron). Let us
consider the electron, as this is the relevant case for most applications in condensed
matter physics. Taking g = 2 (which is adequate for most practical purposes), the

34
intrinsic magnetic moment of the electron is given by17
ge 2|e| ~ |e|~
e = S= = ,
2me 2me 2 2me
where |e| is the magnitude of the electronic charge, and me is the rest mass of
the electron. Hence the eigenvalues of any component of are |e|~/(2me ). The
quantity |e|~/(2me ) B should be familiar to you as the Bohr magneton. Thus

e = B .

The intrinsic magnetic moment of the electron gives us a way of probing its spin
states. In an applied magnetic field B, the Hamiltonian of the electron has an addi-
tional term e B. This leads to a torque on the dipole. If the field is also spatially
nonuniform, there is also a force on the dipole. This is the basis of the famous and
pivotal Stern-Gerlach experiment which originally demonstrated the existence
of spin.

Spin- 12 states and operators: In what follows, we assume that the basis for the
spin states of an electron is one in which Sz is diagonal. We shall use the convenient
notation
def. def.
| i = |S = 12 , m = 12 i , | i = |S = 21 , m = 21 i .
Orthonormalization implies that

h | i = h | i = 1, h | i = h | i = 0.

It follows that any arbitrary, normalized, time-dependent spin state of the electron
can be written in the form

|(t)i = a(t) | i + b(t) | i , where |a(t)|2 + |b(t)|2 = 1.

Note that the natural representation in the spin space is

   
1 0
| i = and | i = .
0 1

7. It is clear that the four operators | ih | , | ih | , | ih | and | ih | form

a basis for all operators acting on the spin states of an electron. Write down the
operators
Sx , Sy , Sz , Sx2 , Sy2 , Sz2 , Sx Sy , Sy Sz and Sz Sx
in this basis.

8. In the state | i, calculate the expectation values of hSx i and hSy i, as well as
hSx2 i and hSy2 i. Hence find the uncertainty product (Sx ) ( Sy ), and verify that
the uncertainty relation
 
(Sx ) ( Sy ) 12 h [Sx , Sy ] i

is satisfied. Repeat the calculation (or write down the answers!) for the state | i.

9. For a spin- 12 particle, write down the commutator [S n , S n0 ], where n and n0

are two arbitrary unit vectors. (Use the identity already established for the com-
mutator [(a ) , (b )].)

17 Note the appearance of several different factors of 2.

35
10. Precession of the spin in a magnetic field: Here is the simplest version
of the effect of a magnetic field on the spin of an electron. Suppose the electron
is prepared, with the help of a magnetic field in the z-direction, in the initial state
|(0)i = | i. The field is switched off, and in its place a constant, uniform, trans-
verse magnetic field B = B ex is applied. The (magnetic part of the) Hamiltonian
is given by
H = e B = B ( B) = B B 1 .

(a) Show that the spin state at any subsequent time t is given by

|(t)i = cos 12 c t | i i sin 12 c t | i,

where c = |e|B/me is the precession frequency.

(b) Hence sketch the probabilities P (t) and P (t) that the electron is in the spin
up state and the spin down state, respectively, at time t.

This example should help you understand why we may regard a transverse com-
ponent of the spin operator (either Sx or, equivalently, 1 ) as a spin flip operator.

11. A generalization: A straightforward generalization of the case just considered

is the following. Let the initial state of the electron be given by
 
a
|(0)i = a| i + b| i = ,
b

where a and b are given complex constants satisfying |a|2 + |b|2 = 1. It is clear that
|a|2 and |b|2 represent the respective probabilities of obtaining the results + 21 ~ and
12 ~ in a measurement of the component Sz of the spin of the electron. Note that
the expectation value of Sz in this state is
1
h(0) | Sz | (0)i = ~ (|a|2 |b|2 ).
2
In the presence of a magnetic field B = Bn where n is an arbitrary unit vector, the
Hamiltonian is H = B B ( n). Find the spin state |(t)i for all t > 0. Find also
the expectation value of Sz at time t.

Comparison with the precession of a classical dipole moment: Let us com-

pare the behavior of the quantum mechanical magnetic moment (or spin) with that
of a classical dipole moment cl in a constant, uniform magnetic field B.

A classical magnetic dipole moment is generated by a current loop. Consider

a classical particle of charge e and mass m moving in a circular orbit of radius r
with a time period T . The orbital angular momentum is given by L = mr2 n,
where = 2/T is the angular speed of the particle and n is the unit normal to
the plane of the orbit. The magnetic moment associated with the current loop has
a magnitude equal to the area of the loop multiplied by the current, by Amperes
rule. Hence cl = r2 (e/T )n = 12 er2 n. The dipole experiences a torque cl B
in the presence of the magnetic field. Equating this torque to the rate of change of
angular momentum, we get the equation of motion
dL e
= cl B = (L B).
dt 2m
The factor e/(2m) is the gyromagnetic ratio in this case. It follows at once that
the dipole moment executes precessional motion around the direction of the mag-
netic field, with an angular frequency (called the Larmor frequency) equal to

36
|e|B/(2m).

12. Now consider the quantum mechanical counterpart of this problem. To be

specific, consider the spin of an electron when it is placed in a magnetic field. As we
have already found, the intrinsic magnetic moment operator of the electron may be
written in the form e = B , while the Hamiltonian is given by H = e B =
B ( B). It is most convenient to work in the Heisenberg picture.

(a) Write down the Heisenberg equation of motion for the spin operator S = 21 ~.
(b) Use one of the identities established in the foregoing in connection with Pauli
matrices, to show that the equaiton of motion simplifies to
dS e
= (S B) = e B.
dt me

Note that this equation of motion is precisely of the form that is obtained in the
classical case. Taking expectation values (in any arbitrary state of the electron) on
both sides,
dhSi
= he i B.
dt
Thus the expectation values of the angular momentum and the magnetic moment
are related precisely as they are in the classical case. This is yet another instance
of Ehrenfests Theorem.

But there is a difference of a factor of 2 between the angular frequency of pre-

cession of the classical dipole cl and the quantum mechanical expectation value
he i(!) The classical frequency of precession is |e|B/(2m), while in the quantum
mechanical case it is |e|B/me . Inspection shows that this difference arises as a
result of the fact that the g-factor of the electron is equal to 2, so that the gyromag-
netic ratio is now ge/(2me ) = e/me rather than e/(2m). I have already mentioned
that the result g = 2 follows from the relativistic quantum mechanics of an electron.

13. Consider the spin state

 
a
|i = a| i + b| i = ,
b

where a and b are given constants (complex numbers) satisfying |a|2 + |b|2 = 1. We
know that |a|2 and |b|2 are the probabilities that a measurement of Sz will yield the
result + 12 ~ and 12 ~ respectively. That is, they are the probabilities that the spin
is up or down in the state |i.
(a) Show that the expectation values of the components Sx and Sy in the state
|i are given by

hSx i = h | Sx | i = ~ Re (a b) and hSy i = h | Sy | i = ~ Im (a b),

respectively. Note that these expectation values vanish if either a or b is zero,

as one might expect intuitively (the spin is then oriented along the z-axis!)

There is another way to regard the state |i. Suppose we measure the component
(S n) of the spin of an electron along an arbitrary direction in space, given by the
unit vector n with respect to our fixed coordinate system (in which Sz is represented
by a diagonal matrix, etc.). Let the direction % of the vector n be specified by the

37
spherical polar coordinates ( , ) in the fixed coordinate system. The result of the
measurement will be either 21 ~ or 12 ~. We could then say that, immediately after
the measurement, the electron spin is either along n or along n, respectively.
Let us denote the corresponding states by | %i and | .i, respectively.18
(b) Using the fact that the Cartesian components of n are given by
n = (sin cos , sin sin , cos ),
show that | %i can be represented as
!
cos 1 ei/2 cos 21
 
| %i = i 2 1 or ,
e sin 2 ei/2 sin 21

apart from an arbitrary overall multiplicative phase factor ei .

a

The general state b with |a|2 + |b|2 = 1 can thus be regarded as the up state
in terms of an axis of quantization n that is different from the usual z-axis. The
modulus of a can be regarded as the cosine of half the polar angle of n, while that
of b is the sine of this angle. The phase difference between the complex numbers a
and b is just the azimuthal angle of the direction n.

Since


n = sin ( ) cos ( + ) , sin ( ) sin ( + ) , cos ( )
= ( sin cos , sin sin , cos ),
we find !
sin 21 ei/2 sin 12
 
| .i = i or .
e cos 21 ei/2 cos 21

(c) Use the result just found to show that the normalized up and down states,
if we consider the x-axis to be the axis of quantization, are given by
   
1 1 1 1 1 1
|Sx = 2 ~i = and |Sx = 2 ~i = .
2 1 2 1
(You have only to note that the positive x-direction corresponds to =
1 1
2 , = 0, while the negative x-direction corresponds to = 2 , = .)

(d) Similarly, if the y-axis is taken to be the axis of quantization, show that the
up and down states are given by
   
1 1 1 1 1 1
|Sy = 2 ~i = and |Sy = 2 ~i = .
2 i 2 i
a

(e) Lets return to the general state |i = b , with |a|2 + |b|2 = 1 as usual.
Suppose we measure the component Sx when the electron is in the state |i.
Show that the probabilities of obtaining the values + 12 ~ and 12 ~, respectively,
are given by
1
2 |a + b|
2
and 12 |a b|2 .
Hint: Expand |i in a linear combination of the basis states |Sx = 12 ~i and
|Sx = 12 ~i. The square of the modulus of each coefficient gives the proba-
bility required.

18 Three cheers for the power of LaTex!

38
Note that if
cos 1
   
a
= i 2 1 ,
b e sin 2
then these probabilities become
1 1
2 (1 + sin cos ) and 2 (1 sin cos ),

respectively. It is easily checked that these values reduce correctly to 1 and 0,

respectively, when = 12 , = 0 (that is, when the state |i is just the state
|Sx = 12 ~i itself). Similarly, they reduce correctly to 0 and 1, respectively, when
= 12 , = (that is, when the state |i is just the state |Sx = 12 ~i itself).

(f) Suppose we measure the component Sy when the electron is in the state |i
defined in the preceding part. Show that the probabilities of obtaining the
values + 12 ~ and 12 ~, respectively, are given by
1 1 1 1
2 |a + ib|2 = 2 (1 + sin sin ) and 2 |a ib|2 = 2 (1 sin sin ).

Once again, these expressions reduce correctly to the values 1 and 0 for = 12 , =
1 1
2 (that is, when the state |i = |Sy = 2 ~i); and to the values 0 and 1, for
= 2 , = 2 (that is, when the state |i = |Sy = 12 ~i).
1 3

14. Let S be the spin quantum number of a certain particle. Consider a system of
two such particles, and let S1 and S2 be their spin operators. All the components
of S1 commute with those of S2 .

Show that the eigenvalues of the operator (S1 S2 ) are given by 21 ~2 [j(j + 1)
2S(S + 1)], where j runs over the values 0, 1, 2, . . . , 2S in steps of unity. Hence
(S1 S2 ) has (2S + 1) possible eigenvalues. The smallest eigenvalue is ~2 S(S + 1),
and the largest is ~2 S 2 .

39
Orbital angular momentum
1. The operator corresponding to the orbital angular momentum of a particle about
the origin of coordinates is defined as L = r p, where r and p are its position
and linear momentum operators, respectively. The Cartesian components of L are
Li = ijk xj pk , where the indices run over the values 1, 2, 3. Note that we dont
need to worry about the noncommutativity of xi and pi in defining the operator L,
since the cross product r p never involves a product of the same component of the
position and the momentum. The components of L satisfy the angular momentum
algebra [Li , Lj ] = i~ ijk Lk .19

(a) Find the commutators [Li , xj ] and [Li , pj ].

(b) It is clear that, classically, all the quantities r L, L r, p L and L p vanish
identically. This follows from the vector identity a (a b) = 0 and the fact
that a b = b a. But these identities are not necessarily true in the case of
non-commuting operators that happen to be vectors as well. Show, however,
that the corresponding quantum mechanical operators r L, L r, p L and
L p also vanish identically.
(c) Show that
(L r) + (r L) = 2i~ r.

(d) Suppose an operator A commutes with both Lx and Ly . Find the commuta-
tor [A , Lz ]. (Hint: Use the Jacobi identity.)

As you know, the eigenvalues of L2 are given by ~2 l(l + 1), where the allowed values
of the orbital angular momentum quantum number are l = 0, 1, 2, . . . . For a given
value of l, the eigenvalues of any one component of L (along any arbitrary direction,
conventionally taken to be the z-direction) are given by ~ m, where the magnetic
quantum number m can take on the (2l + 1) possible values from l to l in steps
of unity. The corresponding set of common eigenstates of L2 and Lz , denoted by
| l, mi, form a basis set of angular momentum states. We have

L2 | l, mi = ~2 l(l + 1) | l, mi, Lz | l, mi = ~m | l, mi.

For every given l,

h l, m | l, m0 i = mm0 .
The raising and lowering operators L are defined as L+ = Lx + iLy and L =
Lx iLy , respectively. In terms of these operators, the angular momentum algebra
of commutators reads

[L+ , L ] = 2~ Lz , [Lz , L+ ] = ~ L+ , [Lz , L ] = ~ L .

It is clear that L+ |l, li = 0 and L |l, li = 0.

(e) Show that p

L | l, mi = ~ l(l + 1) m(m 1) |l, m 1i.
19 This set of relations is sometimes written in the form L L = i~ L, where we must remember

that L stands for a set of three noncommuting operators, rather than an ordinary vector. Nothing
is gained by this formula, as the proper version [Li , Lj ] = i~ ijk Lk is just as compact.

40
It is also useful to note that

L+ L = L2x + L2y + ~ Lz = L2 L2z + ~ Lz .

Similarly,
L L+ = L2x + L2y ~ Lz = L2 L2z ~ Lz .
Hence the (basis) set of eigenstates of L2 and Lz is also the set of eigenstates of
L+ L and L L+ . (That is, the operators L2 , Lz , L+ L and L L+ can be di-
agonalized simultaneously.)

2. Representation in position space: In position space, the Cartesian compo-

nents of L are represented, of course, by the differential operators

Lx = y z , Ly = z x , Lz = x y .
z y x z y x
In spherical polar coordinates, these components are represented by the differential
operators


Lx = i~ sin + cot cos ,





 
Ly = i~ cos + cot sin ,




Lz = i~

.


Hence the raising and lowering operators are represented in spherical polar coordi-
nates by


L+ = Lx + i Ly = ~ ei + i cot ,



 
L = Lx i Ly = ~ ei + i cot .

Verify that these operators, when acting on any (differentiable) function f (, ) of

the angular coordinates, satisfy the commutation relations given earlier.

The square of the total angular momentum is of course represented by a multiple

of the angular part of the Laplacian,
h 1   1 2 i
L2 = ~2 sin + .
sin sin2 2

Spherical harmonics: The position-space representatives (or wave functions)

corresponding to the angular momentum eigenstates | l, mi are called spherical
harmonics, and are denoted by Ylm (, ). Thus
def.
Ylm (, ) = hn | l, mi h, | l, mi,

where the direction in space represented by the unit vector n is specified by the
polar angle and azimuthal angle . The spherical harmonics are related to the
Legendre polynomials as follows. Recall that the Legendre polynomial Pl (cos ) of
order l is given by the Rodrigues formula (denoting cos by , say)

1 dl 2
Pl () = ( 1)l .
2l l! d l

41
The associated Legendre polynomial Plm (cos ) is defined (for m 0) as

dm 1 dl+m
Plm () = (1 2 )m/2 m
Pl () = l (1 2 )m/2 l+m ( 2 1)l .
d 2 l! d

Note that Pl0 (cos ) Pl (cos ). The spherical harmonic Ylm (, ) is then defined
as s
(2l + 1) (l m)! m
Ylm (, ) = P (cos ) eim .
4 (l + m)! l
For m < 0, the spherical harmonics are given by

Yl,m (, ) = (1)m Ylm (, ),

where (as usual) the asterisk denotes complex conjugation. That is, for m 0,
s
(2l + 1) (l m)! m
Yl,m (, ) = (1)m P (cos ) eim .
4 (l + m)! l

We have

L2 Ylm (, ) = ~2 l(l + 1) Ylm (, ) and Lz Ylm (, ) = ~ m Ylm (, ),

where L2 and Lz are the respective differential operators given above. The or-
thonormality relation for the spherical harmonics is
Z

d Ylm (, ) Yl0 m0 (, ) = ll0 mm0

where d = sin d d is the solid angle element, and the integration is over all
solid angles. The completeness relation is
X
l
X
1
Ylm (, ) Ylm (0 , 0 ) = ( 0 ) = ( 0 ) ( 0 ).
sin
l=0 m=l

3. Show explicitly (from the corresponding differential operator representations)

that the action of the raising and lowering operators is given by
p
L Ylm (, ) = ~ l(l + 1) m(m 1) Yl,m1 (, ).

4. The Hamiltonian of a particle of mass m moving in a central potential V (r) is

p2
H= + V (r).
2m
(a) Show that the commutators [Li , p2 ] = 0 and [Li , r2 ] = 0, where p2 = p p =
pi pi and r2 = r r = xi xi .
(b) Hence show that [Li , H] = 0.

(c) Use the position-space representation pi = i~ /xi of the linear momentum

operator to show that
[pi , r1 ] = i~ r3 xi .

42
 (d) Hence write down the commutator [pi , H] in the case of the Coulomb poten-
tial V (r) = K/r, where K is a constant.

5. The radial momentum operator: Classically, the radial momentum of a

particle is simply the component of its linear momentum in the direction of its
position vector, i.e., pr = er p = p er , where er = r/r. Quantum mechanically,
we must take into account the fact that er and p do not commute with each other.
The radial momentum operator must be defined as

def. 1 r r 
pr = p + p ,
2 r r
which ensures that pr is a hermitian operator.

(a) Using the fact that p is represented by the differential operator i~ in the
position representation, show that pr is represented by the differential operator
 
1
pr = i~ + .
r r

(b) Hence find the differential operator that represents p2r .

43
 TEST
1. Are the statements in quotation marks true or false?
(a) The state vector | (t)i of a quantum mechanical system with a hermitian
Hamiltonian is obtained by a unitary transformation of its initial state vector
| (0)i.

(b) No eigenstate of the position operator x of a quantum mechanical particle

can be an eigenstate of the linear momentum px of the particle.

(c) If r and p denote the position and linear momentum operators of a particle,
then the operator (r p) commutes with the orbital angular momentum oper-
ator L = r p.

(d) A particle of mass m moves in the one-dimensional potential well

V0 for 0 < x < L

V (x) =
0 for all other x.

There exists at least one bound state of the particle for all positive values of
V0 and L.

(e) A particle moves in three dimensions in a general attractive central potential

V (r).

The energy eigenvalue of any bound state depends only on the principal quan-
tum number n, and not on the orbital angular momentum quantum number
l and the magnetic quantum number m.

(f) The Hamiltonian of a particle of mass m moving in three dimensions in an

attractive Coulomb potential is given by
p2 K
H= , (K > 0).
2m r

Then, if A = (p L) mKr/r , we have [A , H] = 0.

(g) The operators a, its adjoint a , and the unit operator I satisfy the algebra
[a, I] = [a , I] = 0, [a, a ] = I.

There are no normalizable eigenstates of the operator a .

(h) Consider the bound states of a particle moving in space under the influence
of a central potential V (r).

In general, for a given value of the principal quantum number n, the energy
eigenvalue of a bound state increases as the orbital angular momentum quan-
tum number l increases.

44
 (i) Let | i and | i denote the normalized eigenstates of Sz of a spin-half particle,
corresponding to the eigenvalues 12 ~ and 21 ~, respectively.

The expectation value of Sx is zero in every state that is a linear combination

of | i and | i.

(j) The normalized state | i + | i / 2 corresponds to the singlet spin state
of a two-electron system.

(k) A particle of mass m moving from left to right on the x-axis is incident upon
a potential barrier given by

V0 for 0 < x < L
V (x) =
0 for all other x,
where V0 is a positive constant.

If the energy E of the incident particle is greater than V0 , the reflection

and transmission coefficients of the particle are given by R = 0 and T = 1,
respectively.

(0) (0)
(l) A Hamiltonian H0 has discrete, non-degenerate eigenvalues E0 < E1 <
(0)
E2 < . . . . The Hamiltonian is perturbed, so that the new Hamiltonian is
H = H0 + H 0 , where is a real constant. The new energy levels are given
(0) (1) (2)
by En = En + En + 2 En + terms of higher order in .

(2)
The coefficient E0 can never be positive.

(m) In the linear vector space `2 of square-summable sequences, consider the op-
erator A whose action on any vector is given by

A (x1 , x2 , x3 , . . .) = (0 , x1 , x2 , . . .).

A is a linear operator with norm kAk = 1.

(n) RIf the position-space wave function of a particle is normalized such that
|(r, t)|2 d3 r = 1 for all t, then its momentum-space wave function is also
e t)|2 d3 p = 1 for all t.
R
automatically normalized such that |(p,

(o) Consider a perturbed Hamiltonian H = H0 + H 0 (t).

In the interaction picture, the time evolution of the state vector of the system
is governed by the unperturbed Hamiltonian H0 , while that of observables is
governed by the perturbation H 0 (t) .

(p) In the limit of large quantum number n ( 1), the energy levels of a particle
moving in the potential V (x) = K |x| (where K is a positive constant) become
approximately equally-spaced.

45
(q) Let A be a linear self-adjoint operator acting on the state vectors of a Hilbert
space H, and let the domain and range of A be H itself.

If all the eigenvalues of A are finite in magnitude, we may conclude that A

is a bounded operator, i. e., that its norm is finite.

(r) We know that the set of functions

1 1 2
n (x) = ex /2 Hn (x),
1/4 n
2 n!
where Hn (x) is the Hermite polynomial of order n, forms an orthonormal ba-
sis in the space L2 ( , ).

The set of functions n (x) with n = 1, 3, 5, . . . forms an orthogonal basis in

the space L2 (0 , ).

(s) The Hamiltonian of a particle of mass m and charge e placed in a uniform,

constant applied magnetic field is

(p eA)2
H= ,
2m
where A is the vector potential.

Since the vector potential is dependent on the gauge we choose, the energy
levels of the particle are also gauge-dependent.

(t) Continuation of the preceding question:

In the presence of a magnetic field, the different Cartesian components of

the momentum p of the particle commute with each other, but the different
Cartesian components of its velocity v = dr/dt do not commute with each
other.

(u) Consider a system with a hermitian but time-dependent Hamiltonian H(t).

Even though the Hamiltonian is time-dependent, the state vector | (t)i at

any time t > 0 is given by a unitary transformation applied to the initial state
vector | (0)i.

(v) Consider a particle of mass m in a one-dimensional box: V (x) = 0 for

0 < x < L, and infinite for all other x. Let E1 denote the ground state
energy of the particle. A potential V (x) = (x 21 L) (where is a positive
constant) is now switched on inside the box.

The ground state energy of the particle will now be greater than E1 .

46
 (w) Proper rotations of the coordinate axes in three-dimensional space can be
represented by (2 2) matrices that are unitary, and have determinant equal
to unity.

(x) The ground state of an electron in a hydrogen atom (the state with quantum
numbers n = 1, l = 0, m = 0) is a minimum uncertainty state.

2. Fill in the blanks in the following.

(a) In the space L2 ( , ) of square-integrable functions of a real variable x,
the adjoints of the operators xn and dn /dxn (where n is a positive integer)
are and

(b) Let pr denote the radial momentum operator of a particle moving in three
2

the operator pr is represented by the
dimensions. Then, in the position basis,
2 2 2
differential operator ~ /r + A , where A =

(c) Let S1 and S2 be the spin operators of two independent spin- 21 particles. The
two possible eigenvalues of the operator (S1 S2 )/~2 are then and

(d) Let and be two normalized coherent states of the linear harmonic oscilla-
2
tor, where and are arbitrary complex numbers. Then |h | i| =

(e) If |i is a normalized state vector of a system, the possible eigenvalues of the

projection operator |ih| are and

(f) The normalized probability that there are n photons of a given frequency
and given polarization in blackbody radiation contained in a cavity at absolute
temperature T is given by
1  n n
P (n) = , where n = (eh/kB T 1)1 .
n+1 n+1
In terms of n, the variance of n is equal to

(g) Let | j1 , m1 i and | j2 , m2 i be the respective normalized angular momentum

eigenstates of two particles with given angular momentum quantum numbers
j1 and j2 . We write the angular momentum eigenstates of the total system
as
| j1 , m1 i | j2 , m2 i | j1 , j2 ; m1 , m2 i.
These are common eigenstates of the mutually commuting set of operators
J12 , J1z , J22 and J2z . Call this set of states basis I.
Adding the angular momenta, let J = J1 + J2 , and let m denote the quantum
number corresponding to Jz . We may then work in terms of the alternative
set of states | j1 , j2 ; j , m i. Call this basis II. These states are common eigen-
states of the mutually commuting set of operators

47
(h) Continuing the preceding question, the basis-I states | j1 , j2 ; j1 , j2 i and
| j1 , j2 ; j1 , j2 i must be the same as the basis-II states and re-
spectively. (You must specify the values of j and m in writing down the
corresponding | j1 , j2 ; j , m i states.)

(i) Continuing further, the orthonormality relation for basis-II states is

(j) Finally, suppose j1 = j2 = 21 . Then the basis-II state | 12 , 1

2 ; 0 , 0 i, when
written as a superposition of basis-I states, is given by

(k) Consider the oscillator raising and lowering operators satisfying the commu-
tation relation [a , a ] = I. In the basis formed by the eigenstates |ni of a a,
operators are represented by infinite-dimensional matrices. The general ma-
trix element of the operator a is given by aij = where i and j run over
the values 0, 1, 2, . . . ad inf. (You must use appropriate Kronecker deltas to
write down an expression for aij .)

(l) A spin-half particle is in the state | i = 10 that is an eigenstate of Sz with

eigenvalue 21 ~. Let | i and | i denote eigenstates of Sx corresponding to

the eigenvalues 21 ~ and 12 ~, respectively. The expansion of | i in terms of
these two eigenstates is | i =

(m) The Hamiltonian of a perturbed linear harmonic oscillator is given by

p2 1 1
H = H0 + H 0 = + m 2 x2 + x4 ,
2m 2 4
where is a small positive constant. The eigenvalue En of H is, correct to
first order in , En =

(n) The Schrodinger equation for the position-space wave function (r, t) of a
particle of mass m moving in a potential V (r) is

~2 2
i~ = + V (r).
t 2m
The wave function satisfies the continuity equation /t + j = 0, where
(r, t) = |(r, t)|2 , and the probability current density is given by j(r, t) =

(o) Continuing the preceding question, if the wave function has the asymptotic
form (for very large values of r)
A ikr iEt/~
(r, t) = e e ,
r
where A is a normalization constant, and k and E are positive constants, then
j(r, t) =

(p) The Hamiltonian of a particle moving in one dimension is

p2
H= + V (x).
2m

48
  
Then x , [x , H] =

(q) Continuing with the preceding question: assume that the spectrum {En } of H
is discrete and non-degenerate, with a complete set of orthonormal eigenfunc-
tions {|n i}. Take the matrix element of both sides of the equation
P derived
above between hn | and |n i. Insert a complete set of states k |k ihk | = I
appropriately, to get the identity
X
(Ek En ) |hk | x |n i|2 =
k

Note that the right-hand side is independent of the potential!

Solutions

1. True (T) or false (F):

(a) T
(b) T
(c) F
(d) T
(e) F
(f) T
(g) T
(h) T
(i) F
(j) F
(k) F
(l) T
(m) T
(n) T
(o) F
(p) F
(q) T
(r) F
(s) F
(t) T
(u) T
(v) T
(w) T
(x) F

49
2. Fill in the blanks:
dn
(a) xn and (1)n .
dxn

2
(b) A = .
r r

(c) 34 and 1
4 .

2 2
(d) |h | i| = e|| .

(e) 0 and 1.

(f) n2 + n.

(g) J12 , J22 , J 2 , Jz .

(h) | j1 , j2 ; j1 + j2 , j1 + j2 i and | j1 , j2 ; j1 + j2 , j1 j2 i respectively.

(i) h j1 , j2 ; j , m | j1 , j2 ; j 0 , m0 i = jj 0 mm0 .

| 21 , 1
2 ; 1
2 , 12 i | 12 , 1
2 ; 12 , 1
2 i (| i | i
(j) , which is precisely .
2 2

(k) aij = j i+1,j , where i and j run over the values 0, 1, 2, . . . ad inf.

| i + | i
(l) | i = .
2

1

3~2
(m) En = ~ n + 2 + (2n2 + 2n + 1).
16m2 2

~
(n) j(r, t) = [ ].
2mi

~k |A|2
(o) j(r, t) = er .
m r2

  ~2
(p) x , [x , H] = .
m

X ~2
(q) (Ek En ) |hk | x |n i|2 = (the Thomas-Reiche-Kuhn sum rule).
2m
k

50