Notes on Quantum Mechanics

Module 2 of Refresher course conducted by Indian Academies of Sciences at

JSS Educational Institutions, Suttur, Mysore District, Karnataka, Dec 2-14, 2013
Govind S. Krishnaswami, Chennai Mathematical Institute
• These are very brief and incomplete notes for lectures at the above Refresher Course. Please let me know (govind@cmi.ac.in) of any
comments or corrections. 7 Apr, 2020.

Contents

1 References 2

2 Formalisms of classical and quantum mechanics 2

2.1 Setup of classical mechanics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

2.2 Framework and postulates of quantum mechanics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2.2.1 Hilbert Space of states of a quantum system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.2.2 Observables and hermitian linear operators on states . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2.3 Outer products of vectors and completeness relation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.2.4 Hermiticity of position and momentum operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2.5 Commutators of operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

2.2.6 Eigenvalue problem for hermitian operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

2.2.7 Measured value of observables in states and interpretation of expectation values . . . . . . . . . . . . . . . . . . . . . 11
2.2.8 Schrödinger time evolution of states . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.2.9 Separation of variables, Stationary states, time-independent Schrodinger equation . . . . . . . . . . . . . . . . . . . . 12
2.2.10 Conserved probability density and current . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2.11 Relation between wave function in position and momentum space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

2.3 Ehrenfest’s theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

2.4 Generators of transformations, symmetries and conservation laws . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.4.1 Generators of space translations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.4.2 Generator of time translation and time evolution operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.4.3 Angular momentum as generator of rotations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

2.4.4 Symmetries and conservation laws . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

3 Harmonic oscillator 19
3.1 Algebraic approach via creation and annihilation (ladder) operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

4 Heisenberg uncertainty principle (inequality) 23

5 Angular momentum 24
5.1 Basic properties of angular momentum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
5.2 Eigenvalues of L2 and Lz by ladder operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
5.3 Visualization of angular momentum and location in states Ylm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
5.4 Rigid Body . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

5.5 Matrix elements of L± in the |lmi basis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

5.5.1 E.g.: Matrix representation of angular momenta for l = 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

6 Spin angular momentum 29

6.1 Spin half: Pauli spin matrices and spinor wave functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

7 Addition of angular momenta 31

7.1 State space and observables for two particle systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
7.2 Addition of two spin-half angular momenta . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

7.3 Addition of two or more angular momenta . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

8 Multiparticle systems: Identical particles: bosons & fermions 35

8.1 Identical and indistinguishable particles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

1
 8.2 Application: electronic configuration of light elements and periodic table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

1 References

• Here are some books on quantum mechanics.

1. E. Wichmann, Quantum Physics, Berkeley Physics Course, Vol 4.

2. H. S. Mani and G. K. Mehta, Introduction to Modern Physics.
3. Binayak Dutta-Roy, Elements of Quantum Mechanics.
4. L I Schiff, Quantum Mechanics, 3rd Ed. McGraw-Hill (1968).
5. P J E Peebles, Quantum Mechanics, Princeton Univ Press (1992).
6. B H Bransden and C J Joachain, Quantum Mechanics, 2nd Ed., Pearson (2000).
7. D. J. Griffiths, Introduction to Quantum Mechancis.
8. R. L. Liboff, Introductory Quantum Mechancis.
9. R Shankar, Principles of Quantum Mechanics, 2nd Ed. Plenum Press (1994).
10. P M Mathews and K Venkatesan, A textbook of quantum mechanics, Tata McGraw-Hill (1977).
11. P A M Dirac, The principles of quantum mechanics, 4th Ed., Oxford (1958).
12. W. Heitler, Elementary wave mechanics with applications to Quantum Chemistry.
13. Landau and Lifshitz, Quantum mechanics: Non-relativistic theory, 3rd Edition.

2 Formalisms of classical and quantum mechanics

2.1 Setup of classical mechanics

• The set of possible instantaneous locations of a classical particle is called its configuration
space. This is usually three dimensional Euclidean space R3 . The number of coordinates needed
to specify the instantaneous configuration of a system is the number of degrees of freedom. A
system consisting of a pair of particles has 6 degrees of freedom x1 , y1 , z1 , x2 , y2 , z2 , its configu-
ration space is R3 × R3 = R6 . A particle attached to a fixed support by a rod of fixed length
has two degrees of freedom, its configuration space is a sphere. The configuration space and
number of degrees of freedom are kinematical notions. They do not depend on the nature of
forces between the particles.
• If the forces acting on/between the particles are known, then we may determine the dynamical
time evolution of the system by solving Newton’s equations for the trajectories. For one particle,
mr̈ = F. Newton’s equations are second order in time, they require two sets of initial conditions,
the initial positions r(0) and initial velocities ṙ(0). In other words the initial coordinates
r(0) and initial momenta p(0) = mṙ(0) determine the future trajectory. We say that the
instantaneous state of the system is specified by giving the coordinates and momenta of all
the particles. The set of possible instantaneous states of a system is its phase space. For a
particle moving along a line, its phase space is the x − p phase plane. Newton’s equations may
be formulated as Hamilton’s equations for the time evolution of coordinates and momenta
∂H ∂H
ẋ = and ṗ = − . (1)
∂p ∂x

2
 p2
For a particle in a potential H(x, p) = 2m + V (x) and Hamilton’s equations are a pair of first
order equations
p dV
ẋ = and ṗ = − , (2)
m dx
which may be written as a single second order equation expressing Newton’s second law mẍ =
−V 0 (x). The curve in phase space (x(t), p(t)) is called the phase trajectory. Draw the phase
portrait for a free particle as well as for a simple harmonic oscillator, indicating the direction
of trajectories. A dynamical variable that is constant along trajectories is called a constant of
motion. Its value may differ from trajectory to trajectory. The hamiltonian H = T + V is a
conserved quantity for conservative systems (i.e. where the force is the negative gradient of a
scalar potential).
• Dynamical variables like angular momentum and the hamiltonian are functions of the basic
dynamical variables position and momentum. In general, any real function of position and
momentum is called an observable. Observables are simply real-valued functions on phase space.
They must be real since observables are physical quantities that may be measured.

2.2 Framework and postulates of quantum mechanics

2.2.1 Hilbert Space of states of a quantum system

• States of a quantum system are vectors in a linear space (“vector space”) called a complex
Hilbert space H . For a particle moving on a line, its configuration space is R1 , parametrized by
one coordinate x. Its quantum state space H = L2 (R) is the space of square-integrable functions
ψ(x) on the classical configuration space. ψ is called the state function or state vector or wave
function of the particle.
• By Born’s probability postulate, |ψ(x)|2 dx is interpreted as the probability of finding the
particle between x and x + dx. Since the totalR probability of the particle being somewhere
∞
should be one, we normalize the wave function 0 |ψ(x)|2 dx = 1. This is why we restrict to
square-integrable wave functions. ψ(x) itself is called a probability amplitude, its square is
a probability density.
• Unlike the classical space of states (phase space) which can be a non-linear manifold (e.g. if a
particle is constrained to move on a circle), the quantum Hilbert space is always a linear space.
The sum of two states ψ + φ is a possible state and so is a complex multiple cψ of any state.
This is the principle of linear superposition of states, used to explain the interference of matter
waves in the double slit experiment.
• A complex Hilbert space H is a vector space over the complex numbers. It is a space of
ket vectors |ψi closed under linear superposition. If |ψi and |χi are state vectors, then so is
α|ψi + β|φi, for any α, β ∈ C. A simple example is the two dimensional complex vectorspace 
ψ1
of spin states of a spin half particle which are usually denoted as column vectors |ψi = .
ψ2
Notably, the space of states of a quantum system is a complex, rather than a real vector space.
• The quantum state space is equipped with an inner or dot product. The inner product of a

3
pair of vectors ψ, χ is denoted hψ|χi. For the spin- 21 Hilbert space, the inner product is
 †    
ψ1 χ1
χ1
hψ|χi = = ψ1∗ ψ2∗ = ψ † χ = ψ1∗ χ1 + ψ2∗ χ2 . (3)
ψ2 χ2 χ2

ψ † is called the hermitian adjoint, it is the complex conjugate transpose, it is a row vector. So
associated with a vector space of colummn/ket vectors there is a ‘dual’ space of row/bra vectors,
the adjoints of the kets |ψi† = hψ| = ψ1∗ ψ2∗ . The inner product may also be regarded as

producing a complex number from a bra and a ket vector hψ|χi. However, the inner product of
a non-zero vector with itself is always a positive real number hψ|ψi > 0, it is called the square
of the length of the vector.
• Another example is n-dimensional complex vector space Cn with the innerRproduct hu|vi =
P ∗ 2 ∞ ∗
i ui vi . The Hilbert space of a particle moving on a line is L (R) with hf |gi = −∞ f (x)g(x) dx.
• From these examples (keep hu|vi = u∗i vi in mind) we abstract the basic properties of the inner
product (these are its defining properties in an axiomatic approach)

hαu|vi = α∗ hu|vi, hu|βvi = βhu|vi, hu + v|wi = hu|wi + hv|wi, hu|vi∗ = hv|ui. (4)

hu|vi is linear in the second vector v and anti-linear in the first vector u on account of complex
conjugation of the components of the first vector.
p
• The norm/length of a vector is ||v|| = hv|vi. The norm of a vector is unchanged upon
multiplying by a phase eiα . If hu|vi = 0 then the vectors are orthogonal.
• Two state vectors that differ by multiplication by a non-zero complex number ψ2 (x) = cψ1 (x)
represent the same physical state. We often work with unit norm states.
• A basis for the Hilbert space is a set of vectors |ei i such that any vector |vi may be expressed
as a linear combination of |ei i in a unique way. The number of basis vectors is the dimension of
the vector space. The standard basis vectors of the two dimensional spin Hilbert space C2 are
     
1 0 v1
|e1 i = , |e2 i = , so = v1 |e1 i + v2 |e2 i. (5)
0 1 v2
P
The coefficients vi in the expansion |vi = i vi |ei i are called
P the components of |vi. The
components of the adjoint are the complex conjugates: hv| = i hei |vi∗ . [We will often drop the
summation symbol and assume repeated indices are summed.] E.g. the adjoints of the basis
vectors are row bra-vectors

he1 | = e†1 = 1 0 , he2 | = e†2 = 0 1 .

(6)

Cn is an n-dimensional vector space. The state space of a particle moving on a line, L2 (R) is
infinite dimensional, it is called a function space. It is intuitively clear that this is an infinite
dimensional space since the values of the function ψ(x) at each x ∈ R can be freely specified
(subject to normalizability). x here plays the role of the index i = 1, 2 in the two dimen-
sional spin-half vector space C2 . A possible basis for a function space is the set of monomials
{1, x, x2 , x3 , x4 , · · · }. Indeed, any function ψ that has a Taylor series around x = 0 admits a
expression as a linear combination of these. The coefficients are the derivatives of ψ at x = 0:
1 1
ψ(x) = ψ(0) + ψ 0 (0)x + ψ 00 (0)x2 + ψ 000 (0)x3 + · · · (7)
2 3!

4
However this basis of monomials is a bit inconvenient. In particular, the basis vectors are not
orthogonal, in fact they are not even normalizable with respect to the above L2 inner product. A
more convenient basis for for L2 (R) consists of the energy eigenstates of the harmonic oscillator
|ni, which we will study in more detail later.
• It is often convenient to work with an orthonormal basis, i.e., a basis of vectors |ei i which are
pairwise orthogonal and each normalized to have unit norm, hei |ej i = δij . The standard basis
|ei i for C n components |ei ij = δij is orthonormal with respect to the usual inner product
P with
∗
hu|vi = i ui vj .
• A set of orthonormal vectors is said to be a complete orthonormal set if it forms a basis for
the vector space, i.e., if we may write any vector as a linear combination.

2.2.2 Observables and hermitian linear operators on states

An observable A in quantum mechanics (e.g. hamiltonian, position, momentum, angular mo-

mentum, spin, magnetic moment) is a hermitian (self-adjoint) linear operator on the Hilbert
space of states H . Hermiticity is the quantum analogue of classical observables being real-
valued functions. We will see that a hermitian operator has real eigenvalues, which are possible
results when A is measured. To define a hermitian operator, we first note that a linear operator
on a vector space takes vectors to vectors in a linear way: A(a|ψi + b|χi) = aA|ψi + bA|χi.
When A acts on a vector |vi it produces a new ket vector A|vi which is also denoted |Avi.
• A linear operator is an abstract concept, whose concrete realisation is a matrix. A linear
operator on C2 is simply
 a 2×2
 matrix,  once we choose
 a basis
 to represent it. For example, the
0 1 0 −i 1 0
Pauli matrices σ1 = , σ2 = , σ3 = are linear operators represented
1 0 i 0 0 −1
as matrices in the standard basis for C2 .
• If |ej i are a basis for H , then a linear operator A is determined by how it acts on the basis
vectors. Since A is linear A|ej i must be a linear combination of the basis vectors themselves
X
A|ej i = |ek iAkj . (8)
i

Akj are the components of A in this basis, they may be written as entries in a matrix, with Akj
occupying the slot in the k th row and j th column. The vector that makes up the first column
Ak1 is the ‘image’ of e1 (i.e. coefficients in the linear combination appearing in A|e1 i), the
second column Ak2 is the image of e2 and so on.
• A matrix A is hermitian if it equals its own complex conjugate transpose. The latter is called
its adjoint A† = (A∗ )t . So A is hermitian if A = A† , i.e., if it is self-adjoint. In terms of matrix
entries, A∗ij = Aji . In particular, the diagonal entries of a hermitian matrix are real, while the off
diagonal entries are complex conjugates of each other. The Pauli matrices are hermitian. Note
that the adjoint of a product is the product of adjoints in the opposite order. (AB)† = B † A†
and that (A|ψi)† = hψ|A† . We also denote A|ψi = |Aψi, so that |Aψi† = hAψ|.
• The concept of hermiticity makes sense for a linear operator, even if we have not represented
it explicitly as a matrix by choosing a basis. To explain the concept, we need the idea of matrix
elements between states. If u, v are a pair of states, then hu|A|vi is called the matrix element
of A between the states u and v . To know an operator is to know its matrix elements.

5
• The adjoint of A is the operator A† defined via its matrix elements hu|A† |vi = hAu|vi. So
if we know the matrix elements of A, then we may find the matrix elements of A† . A linear
operator is hermitian if hu|Avi = hAu|vi for all states u, v ∈ H . A hermitian operator is also
called symmetric by mathematicians since it does not matter whether A is written on the left
or on the right.
• Now let us see how this abstract definition of hermiticity reduces to the formula Aij = A∗ji
for hermitian matrices. We must equate the matrix elements of A and those of A† . Let ei be
an orthonormal basis, then the matrix element of A between the states ei and ej is just Aij ,
as is seen by taking the inner product of the above equation with ei
X X
hei |A|ej i = hei |ek iAkj = δik Akj = Aij . (9)
k k

On the other hand, what are the matrix elements of A† ? By the definition of the adjoint,

hei |A† ej i = hAei |ej i = hej |Aei i∗ = (Aji )∗ (10)

So a linear operator is self-adjoint if its matrix elements in an orthonormal basis satisfy Aij =
(Aji )∗ .

2.2.3 Outer products of vectors and completeness relation

• Outer products of vectors: Consider the vector space Cn with standard basis |ei i. Just
as we may ‘dot’ row and column n-vectors to get a scalar inner product, we may also form their
‘outer’ product (column times a row), to get an n × n matrix. For n = 2 show that
       
1 0 0 0 0 1 0 0
|e1 ihe1 | = e1 e†1 = , |e2 ihe2 | = , |e1 ihe2 | = , |e2 ihe1 | = .
0 0 0 1 0 0 1 0
(11)
More generally, check that |ei ihej | is a matrix with a 1 in the ij -entry and 0’s elsewhere. From
this we see that a matrix whose entries are Aij in the ith row and j th column, can be expressed
as X
A= Aij |ei ihej | (12)
ij

Now let us use this expression to find how a matrix acts on a vector v = vk |ek i. We get
X
Av = Aij |ei ihej |vk |ek i = Aij vk |ei ihej |ek i = Aij vk δjk |ei i = Aik vk |ei i. (13)
ij

So the ith component of Av is

P
k Aik vk .
• In particular, the identity operator I , may be expressed as
X X
I= |ei ihei | = δij |ei ihej | (14)
i ij

The identity operator has the components δij in any basis since it takes every vector to itself.
This ‘resolution’ of the identity operator as a sum of outer products of a set of orthonormal basis
vectors is called the completeness relation. It is quite useful in many physical problems and

6
calculations. E.g. the energy eigenstates of the Harmonic oscillator form a complete orthonormal
set and satisfy the above completeness relation. Coherent states for the harmonic oscillator also
satisfy a completeness relation even though they are not orthogonal and are in fact an over-
complete set.
• An anti-hermitian operator is one that satisfies A† = −A. A unitary operator is one whose
inverse is its adjoint, U U † = U † U = I . It is clear that the identity I is hermitian as well as
unitary. If A is anti-hermitian, then iA is hermitian since (iA)† = A† i† = −A(−i) = A.

2.2.4 Hermiticity of position and momentum operators

• Physically interesting examples of hermitian operators for a particle with one degree of freedom
moving on a line include the position operator x̂ψ(x) = xψ(x), and momentum operator p̂ψ(x) =
−i~ψ 0 (x). Check that x̂† = x̂ and dˆ = ∂x∂
is anti-hermitian. We must show hf |x̂gi = hx̂f |gi for
any two states f, g . This is seen as follows:
Z Z
hf |x̂gi = f (x)xg(x) dx = (xf (x))∗ g(x) dx = hx̂f |gi.
∗
(15)

∂
Showing hermiticity of p̂ = −i~ ∂x requires integration by parts. Let us show that dˆ = ∂x
∂
is
ˆ
anti-hermitian, from which it will follow that p̂ = −i~d is hermitian. Let us denote complex
conjugate of f by f¯ here for convenience
Z Z
ˆ = f¯(x)g 0 (x)dx = − f¯0 (x)g(x) dx + f¯g ∞ = −hdf ˆ |gi.
 
hf |dgi −∞
(16)

Here we assumed f, g vanish at ±∞, which is the case for square-integrable functions. Boundary
conditions play an important role in determining the hermiticity of momentum. If we have a
particle moving on a finite interval [a, b] (as in a square well), then
Z b Z b  b
ˆ = f¯(x)g 0 (x)dx = − ˆ |gi + f¯g b .
f¯0 (x)g(x) dx + f¯g a = −hdf
 
hf |dgi a
(17)
a a

For dˆ to be anti-hermitian, the boundary term must vanish. This happens, for instance, if the
functions vanish at the end points (f (a) = f (b) = 0, as in an infinite square well) or satisfy
‘periodic boundary conditions’ f (a) = f (b).
• Of particular importance is the concept of expectation value of an observable A in a state
ψ , which is defined as the normalzed diagonal matrix element of A in the state ψ

hψ|Aψi
hAiψ = (18)
hψ|ψi

The expectation value of a hermitian operator in any state is a real number. For, by hermiticity,
and hu|vi = hv|ui∗ , we have
hψ|Aψi = hAψ|ψi = hψ|Aψi∗ (19)
In other words, the diagonal matrix element of A is equal to its own complex conjugate. We
are familiar with this: the diagonal entries of a hermitian matrix in an orthonormal basis
hei |A|ei i = Aii are real.

7
2.2.5 Commutators of operators

• Multiplication of matrices/operators is in general not commutative AB 6= BA (in general).

The amount by which AB − BA fails to vanish is called the commutator [A, B] = AB − BA.
Any operator commutes with itself or any power of itself [A, An ] = An+1 − An+1 = 0. On the
other hand, check that xp − px = [x, p] = i~I by acting on a state ψ(x):
∂
xpψ = −i~xψ 0 (x), while pxψ = −i~ (xψ) = −i~xψ 0 (x) − i~ψ(x) ⇒ [x, p]ψ = i~ψ.
∂x
(20)
• In QM, the commutator plays the role that the Poisson bracket plays in CM. Just as the
1
Poisson bracket {f, g} of two observables is another observable, i~ [A, B] is again an observable
(i.e., hermitian) if A, B are hermitian. To show this it suffices to check that [A, B] is anti-
hermitian if A and B are hermitian.

([A, B])† = (AB − BA)† = B † A† − A† B † = BA − AB = −[A, B]. (21)

An important property of the commutator is the product or Leibnitz rule, check that

[A, BC] = [A, B]C − B[A, C]. (22)

• In three dimensions, we have three coordinate and momentum operators x, y, z and px =

∂ ∂ ∂
−i~ ∂x , py = −i~ ∂y , pz = −i~ ∂z . It is easily seen that the momenta commute with each other
and the coordinates commute among themselves, more over [x, px ] = i~ while [x, py ] = 0
etc. These so-called Heisenberg canonical commutation relations may be summarised as
[xi , pj ] = i~ δij .

2.2.6 Eigenvalue problem for hermitian operators

• The eigenvalue problem for a linear operator (hermitian or not) is the equation A|ψi =
λ|ψi. A non-zero vector |ψi 6= 0 that satisfies this equation for some complex number λ is
called an eigenvector of A with eigenvalue λ. Taking the adjoint of the eigenvalue equation we
also have
(A|ψi)† = hψ|A† = λ∗ hψ| (23)
So if |ψi is an eigen-ket of A with eigenvalue λ, then hψ| is an eigen-bra of A with eigenvalue
λ∗ . In particular, if A = A† is hermitian, then hψ|A† = hψ|A = λ∗ hψ|. In other words, if |ψi is
an eigen-ket of A, then hψ| is an eigen-bra of A with eigenvalue λ∗ . We will soon show that
λ ∈ R A is hermitian.
• The eigenstate of the position operator x̂ with eigenvalue x0 is denoted |x0 i, i.e., x̂|x0 i = x0 |x0 i.
We will see that measurement of the position of a particle that is in state |x0 i is guaranteed to
give the value x0 . The ‘position-space’ or ‘coordinate-space’ wave function of any state |ψi is
defined as the inner product hx|ψi = ψ(x). It follows that ψ ∗ (x) = hψ|xi.
• Similarly, the eigenvalue problem for momentum is p̂|ki = ~k|ki. It is conventional to write
the momentum eigenvalue in terms of wave number as ~k . We will see that |ki is a state in
which a measurement of the particle’s momentum will give ~k . The momentum space wave
function of a particle in state |ψi is defined as ψ̃(k) = hk|ψi. ψ̃ is pronounced ‘psi-tilde’.
• Here are some useful facts about hermitian matrices/operators:

8
1. The eigenvalues of a hermitian operator are real. This is because the eigenvalues of a
hermitian operator are simply the expectation values in the corresponding eigenstates
hψ|A|ψi
A|ψi = λ|ψi ⇒ hψ|A|ψi = hψ|λψi = λhψ|ψi ⇒ λ= . (24)
hψ|ψi

2. Eigenvectors |χi, |ψi corresponding to distinct eigenvalues µ 6= λ are orthogonal. To see

this, we calculate hχ|Aψi in two ways using hermiticity and reality of eigenvalues and
subtract.
hχ|Aψi = λhχ|ψi and hχ|Aψi = hAχ|ψi = µhχ|ψi (25)
Thus (λ − µ)hχ|ψi = 0. Since λ 6= µ we must have hχ|ψi = 0, i.e., eigenvectors corre-
sponding to distinct eigenvalues are orthogonal.
3. It can be shown that a hermitian operator can be diagonalised by a unitary transformation
U † HU = Λ where Λ is a diagonal matrix with eigenvalues along the diagonal. Moreover,
the eigenvectors of a hermitian operator can be chosen to form a complete orthonormal
basis for H X
A|ψi i = λi |ψi i, hψi |ψj i = δij , |ψi ihψi | = I, . (26)
i
Furthermore, two hermitian operators which commute can be simultaneously diagonalised.
In other words, there is a basis of common eigenvectors in which both are diagonal. And
if they do not commute, as in the case of [x, p] = i~I , then they cannot be simultaneously
diagonalised. Operators that commute are said to be compatible, we will see that they
can be simultaneously measured.
4. The eigenvalue problem for the momentum operator is p̂|ki = ~k|ki. The position space
eigenfunction hx|ki of the momentum operator is a plane wave. p̂ψ(x) = ~kψ(x) becomes
−i~ψ 0 = ~kψ or ψ = Aeikx . We will choose A = 1. In other words hx|ki = eikx and so
hk|xi = e−ikx .
5. The position-space or coordinate-space eigenfunctions of the position operator are delta-
functions. Let’s see why. The eigenvalue problem is
x̂ ψ(x) = xψ(x) = λψ(x) where λ is a constant. (27)
The only way this can be satisfied for all x is for ψ(x) to vanish at all x 6= λ. Now if
ψ(x) were to vanish at x = λ as well, then it would be the zero function and not qualify
as a non-trivial eigenvector. The value of ψ(x) at x = λ can be either ψ(λ) = λ or
ψ(λ) = ∞. If ψ(λ) = λ, then the state will have zero norm. So it must be infinite at
x = λ. In fact, ψ(x) is proportional to the Dirac delta function. It is normalized so that
ψ(x) = δ(x − λ). It is conventional to denote the position eigenvalue by x0 rather than
λ. So δ(x − x0 ) is an eigenfunction of the position operator with eigenvalue x0 , it is a
function of x that is zero every where except at x0 . Think of it as a limit of functions that
are sharply peaked at x = x0 . Thus the coordinate space wave function of the eigenstate
|x0 i of x̂ is hx|x0 i = δ(x − x0 ). Now if we have two position eigenstates |x0 i and |x00 i,
then their coordinate space wave functions are hx|x0 i = δ(x − x0 ) and hx|x00 i = δ(x − x00 ).
Their inner product is
Z
hx00 |x0 i = δ(x − x00 )δ(x − x0 ) dx = δ(x0 − x00 ). (28)

9
 So position eigenstates are orthogonal and ‘delta-normalized’. They form a complete set
in the sense that they satisfy a completeness relation
Z
dx |xihx| = I. (29)

To see this, take the matrix elements of the LHS between coordinate basis states |x0 i and
|x00 i Z Z
dx hx0 |xihx|x00 i = dx δ(x − x0 )δ(x − x00 ) = δ(x0 − x00 ). (30)
0 00
00 0 00
R elements of the identity are also the same hx |I|x i =
On the other hand, the matrix
0
hx |x i = δ(x − x ). Since dx |xihx| and I have the same matrix elements, they are
equal.
• Similarly, momentum eigenstates form a complete set
Z
dk
|kihk| = I. (31)
2π
Check this by evaluating the matrix elements between position basis states |x0 i and |x00 i.
On the rhs we get hx0 |I|x00 i = δ(x0 − x00 ). On the lhs we get the same using the Fourier
representation of the delta function
Z ∞ Z ∞
dk 0 00 dk ikx0 −ikx00
hx |kihk|x i = e e = δ(x0 − x00 ). (32)
−∞ 2π −∞ 2π

How do we get the last equality? If x0 = x00 then we are integrating the function 1, and
the answer should be infinite, and indeed δ(0) = ∞. On the other hand, when x0 6= x00 ,
then we have Z
dk
[cos(k(x0 − x00 )) + i sin(k(x0 − x00 ))] = 0 (33)
2π
Since the average value of both the sine and cosine functions is zero.
• Momentum eigenstates with distinct wave numbers are orthogonal (as we expect for the
eigenstates of a hermitian operator)
Z Z
0 00
hk |k i = dx hk |xihx|k i = dx e−ik x eik x = 2π δ(k 0 − k 00 ).
0 00 0 00
(34)

6. Among hermitian operators, the positive operators are particularly interesting physically. A her-
mitian operator is positive (or non-negative) if its expectation value in every state is non-negative
hψ|A|ψi ≥ 0 , for all ψ ∈ H . Since eigenvalues are simply the expectation values in eigenstates,
we see that positive operators have non-negative eigenvalues. If A is any linear operator, then we
check that A† A and AA† are both hermitian and positive operators.
†
E.g. (AA† )† = A† A† = AA† . (35)

To check positivity, we work out the expectation value in any state:

hψ|A† A|ψi = hAψ|Aψi = ||Aψ||2 ≥ 0 and hψ|AA† |ψi = hA† ψ|A† ψi = ||A† ψ||2 ≥ 0. (36)
1 †
An example is kinetic energy T = 2m1 2
p = 2m p p = 2m1
pp† , since p = p† is hermitian. So we may
conclude that the energy eigenvalues of a free particle must all be non-negative.

10