An Introduction to Quantum Mechanics

. . . for those who dwell in the macroscopic world

Antonio Barletta
DIENCA
Alma Mater Studiorum – Università di Bologna
arXiv:1201.4234v2 [quant-ph] 18 Feb 2016

— Lecture Notes —
January 2012

Foreword
There is a huge number of excellent and comprehensive textbooks on quantum mechanics. They
mainly differ for the approach, more or less oriented to the formalism rather than to the phe-
nomenology, as well as for the topics covered. These lectures have been based mainly on the
classical textbook by Gasiorowicz (1974). I must confess that the main reason for my choice of
Gasiorowicz (1974) is affective, as it was the textbook where I first learned the basic principles of
quantum mechanics. Beyond my personal taste, I now recognize that Gasiorowicz (1974) is still
a very good textbook on quantum mechanics, with a rigorous theoretical approach accompanied
by a wide collection of applications. If the textbook by Gasiorowicz was my main basis, I have
taken much also from other textbooks such as Phillips (2003), as well as from the excellent classical
textbook by Dirac (1981).

Contents
1 The origin of quantum physics 3
1.1 The black–body radiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 The photoelectric effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 Wave behaviour of particles and electron diffraction . . . . . . . . . . . . . . . . . 4
1.4 The Bohr atom . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

2 Wave packets and the uncertainty principle 6

2.1 Heisenberg uncertainty principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2 The wave equation for a free particle . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.3 The statistical interpretation of ψ(x, t) . . . . . . . . . . . . . . . . . . . . . . . . . 10

3 The Schrödinger equation 13

3.1 Expectation values of physical observables . . . . . . . . . . . . . . . . . . . . . . . 13

1
 3.2 The x–space and the p–space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.3 On the general structure of quantum mechanics . . . . . . . . . . . . . . . . . . . . 16
3.4 Bases in the Hilbert space and representations . . . . . . . . . . . . . . . . . . . . 18
3.5 The statistical interpretation of quantum states . . . . . . . . . . . . . . . . . . . . 19
3.6 Operator methods and the uncertainty principle . . . . . . . . . . . . . . . . . . . 19
3.7 The Hamiltonian operator, the evolution operator, and the
stationary states . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

4 One–dimensional quantum mechanics 22

4.1 An infinite potential well . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
4.2 Quantum tunnelling through a barrier . . . . . . . . . . . . . . . . . . . . . . . . . 23
4.3 The quantum harmonic oscillator . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

Appendix A 31

Appendix B 32

2
1 The origin of quantum physics
A fundamental concept of classical physics is the particle, a point–like mass that moves along
a trajectory in the three–dimensional space. The position and the instantaneous velocity of the
particle can be determined with an arbitrarily high precision at every time, and their evolution is
perfectly predictable from the equations of motions. The position and the velocity as functions of
time can be known if one fixes the initial state of the particle, and if one knows the forces acting
on the particle.
Another fundamental concept of classical physics is the wave. The electromagnetic field is a
wave: an entity not localized in a specified position and with a velocity of propagation, also called
phase velocity.
Waves and particles are independent paradigms of classical physics: a particle is not a wave
and a wave is not a particle. As suggested by Phillips (2003), “the laws of particle motion account
for the material world around us and the laws of electromagnetic fields account for the light waves
which illuminate this world ”.
This conception of natural world started to change at the beginning of twentieth century, as new
experimental evidence was provided by the study of the black–body radiation, of the photoelectric
effect, of the Compton effect, of the diffraction and interference properties of electrons, and of the
spectra of atomic radiation.

1.1 The black–body radiation

A result well–known from the studies initiated by Gustav Kirchhoff in 1859, is that the intensity of
electromagnetic radiation emitted by a perfect absorber (black body) is a universal function of the
temperature T and of the frequency of radiation ν. In practice, the black–body radiation is the
radiation from a small hole in an isothermal cavity. The approach of classical physics provided a
theoretical formula to express the energy density (energy per unit volume) of the radiation inside
the cavity in the frequency range ν to ν + dν,

8πν 2
u(ν, T ) dν = kT dν, (1.1)
c3
where k = 1.3807 × 10−23J/K is Boltzmann constant, and c = 2.998 × 108m/s is the speed of light
in vacuum. Eq. (1.1) is now well–known as the Rayleigh–Jeans law. This law has the unphysical
feature that, when integrated over all the possible frequencies, yields a diverging result. Moreover,
it turns out to be in agreement with experimental data only for low frequencies. A completely
different approach was needed to get a satisfactory agreement with experimental data, and this
was done by Max Planck in 1900, who obtained

8πh ν3
u(ν, T ) dν = 3 hν/(kT )−1
dν. (1.2)
c e
Eq. (1.2) contains a new constant of physics, Planck’s constant,

h = 6.6261 × 10−34 Js. (1.3)

One can easily check, by a Taylor expansion with small values of hν/(kT ), that Eq. (1.2) coincides
with Eq. (1.1) at low frequencies. Eq. (1.2) displays an excellent agreement with the experimental

3
results, it can be integrated over all the frequencies with a converging result, but what is more
interesting is that Eq. (1.2) was obtained by invoking assumptions beyond the framework of classical
physics. Planck assumed that radiation could be absorbed and emitted in “quanta” of energy given
by hν. The electromagnetic radiation is treated as a gas of particles instead of a wave field. This
particles, the quanta, are the photons. The same physical entity is treated as simultaneously
having a wave and a particle character. Planck’s hypothesis of photons is what we now call the
wave–particle duality of electromagnetic radiation.

1.2 The photoelectric effect

The photoelectric effect was discovered by Heinrich Hertz in 1887. This effect consists in the
emission of electrons from a polished metal plate when irradiated by electromagnetic waves. The
emission of electrons may take place (or not) depending on the frequency of the electromagnetic
radiation. There exists a threshold frequency for the emission that depends on the metal. When
the threshold is exceeded, the emission takes place and one has an electric current that increases
with the intensity of the electromagnetic radiation.
What is incomprehensible in the framework of classical electromagnetism is that electrons may
be emitted or not depending on the frequency of the radiation. The classical electromagnetism
assigns an energy to the radiation that depends on the amplitude, but not on the frequency, of
the waves. One can understand that the energy of the radiation incident on the metal plate may
be transferred to the electrons trapped in the metal, so that the electrons may be extracted when
a sufficient energy is received. However, on the basis of classical electromagnetism, one expects a
threshold condition for the photoelectric effect to be formulated in terms of the radiation intensity
and not of the radiation frequency. Let us call W (the work function) the energy required for an
electron to escape the metallic bond, i.e. to separate the electron from the lattice of positively
charged nuclei. Then, Planck’s hypothesis of quanta allows one to express the kinetic energy of an
escaping electron as

Ekin = hν − W. (1.4)

The threshold condition is thus ν > W/h. The use of Eq. (1.4) is the basis of Einstein’s theoretical
analysis of the photoelectric effect. Robert A. Milikan, skeptical about the theory of quanta,
performed in 1915 the experiment that confirmed the validity of Eq. (1.4). He found that W was
of the order of several electron volts (1 eV = 1.602 × 10−19 J).

1.3 Wave behaviour of particles and electron diffraction

If the electromagnetic waves are also particles, the photons, is it conceivable to think of electrons as
waves? This extrapolation of the wave–particle duality was the starting point of Louis de Broglie
studies leading to the assumption that a particle having a momentum p can be thought of as a
wave having wave length
h
λ= . (1.5)
p
This idea, first proposed in his 1924 doctorate thesis, was the object of experimental verification.
In particular, in 1927, C. J. Davisson and L. H. Germer proved at Bell Telephone Laboratories

4
in New York that diffraction of electrons from a crystal lattice could take place, thus yielding the
first validation of de Broglie’s idea.

1.4 The Bohr atom

The discrete spectra of radiation emitted from atoms could not find a reasonable explanation
by means of the atomic models based on the classical electromagnetism. The discrete spectra
of radiation strongly suggest that the atomic energy levels are quantized. A transition between
different levels yields an emitted radiation with a well specified wave length leading, experimentally,
to a well specified spectral line.
Niels Bohr in 1913 conjectured that the angular momentum of the electrons orbiting around the
nucleus in an atom can be just an integer multiple of ~ = h/(2π) (the reduced Planck constant).
He further assumed that electrons do not emit radiation if not in a transition between different
orbits. In such a transition, the frequency of the emitted or absorbed radiation is given by,

E − E′
ν= . (1.6)
h
The great success of Bohr’s postulates in explaining the radiation of atoms, caused the development
of this theory up to the so–called Sommerfeld–Wilson quantization rule,
I
p dq = nh, n = 1, 2, 3, . . . , (1.7)

where the integral is over a closed path (orbit), and p is the momentum canonically conjugated with
the coordinate q of a particle (see Appendix A). Eq. (1.7) was formulated in 1915. These theoretical
studies have a role in the history of modern physics. However, they are to be considered as a mere
phenomenological theory of atomic spectra, but not yet a reformulation of the basic concepts of
mechanics.

5
2 Wave packets and the uncertainty principle
The basic ideas suggested by the failure of classical physics, when applied to microscopic objects
(electrons, atoms, nuclei, . . . ) are condensed in what we called wave–particle duality. The challenge
is to reconcile the concept of wave (a non–localized object) with that of particle (a localized object).
Configurations of waves may exist where localization emerges in the form of wave packets. The
basic mathematical theory is Fourier’s analysis.
One may imagine a function f (x) expressed as
Z ∞
f (x) = dk g(k)eikx . (2.1)
−∞

Eq. (2.1) defines f (x) as a linear superposition of infinite standing plane waves with wavelength
λ = 2π/k. Two neighbouring maxima of the real and imaginary parts of eikx are separated by a
distance 2π/k. Each wave is “weighted” by the coefficient function g(k).
One may consider a Gaussian weight function
2
g(k) = e−α(k−k0 ) , (2.2)

where α > 0. One can substitute Eq. (2.2) into Eq. (2.1) and evaluate the integral on the right
hand side of Eq. (2.1),
Z ∞ ∞ r
π −x2 /(4α)
Z
2 ′2 ′
f (x) = dk e−α(k−k0 ) eikx = eik0 x dk ′ e−αk eik x = eik0 x e . (2.3)
−∞ −∞ α

On considering the square modulus of g(k) and the square modulus of f (x),
2 π −x2 /(2α)
|g(k)|2 = e−2α(k−k0 ) , |f (x)|2 = e , (2.4)
α
one realizes that we have a Gaussian signal both in k–space and in x–space. We may easily check
√
that, when k = k0 ± ∆k/2, where ∆k = 2/ 2α, the Gaussian signal in k–space drops to 1/e times
√
its peak value. When x = ±∆x/2, where ∆x = 2 2α, the Gaussian signal in x–space drops to 1/e
times its peak value. If α becomes smaller and smaller, the signal in k–space increases its width
∆k, while the signal in x–space decreases its width ∆x. One may easily check that

∆k ∆x = 4. (2.5)

The precise numerical value of the product is not important. What is important is that the product
∆k ∆x is finite and independent of α,

∆k ∆x ∼ O(1). (2.6)

Eq. (2.6) establishes an uncertainty principle for the wave packets and its interpretation is straight-
forward: an highly localized wave packet in k–space, i.e. one with a small ∆k, means a poorly
localized wave packet in x–space, i.e. one with a large ∆x, and vice versa. It is not possible to
reduce the width of the Gaussian signal both in k–space and in x–space.
Let us now consider a wave packet made up with the superposition of travelling plane waves.
Then, Eq. (2.1) is to be replaced with
Z ∞
F (x, t) = dk g(k)ei(kx−ωt) , (2.7)
−∞

6
 Figure 1: Localization of wave packets in k–space and in x–space

where ω = 2πν is the angular frequency, while k is again related to the wavelength, k = 2π/λ.
The simplest case is that with ω = ck, where c is a constant. In this case, Eq. (2.6) describes
a superposition of plane waves having a constant phase velocity c. Then, a comparison between
Eq. (2.1) and Eq. (2.7), allows one to write

F (x, t) = f (x − ct). (2.8)

The effect of the standing waves being replaced by travelling waves is just a rigid translational
motion of the wave packet with a velocity c. No distortion of the wave packet is caused by the
time evolution.
Our objective is the use of wave packets for the modelling of particles, i.e. of localized objects.
We are therefore interested in assuming a general relationship ω = ω(k). We may assume a wave
packet strongly localized in k–space, with a marked peak at k = k0 , expressed by Eq. (2.2).
The strong localization in k–space suggests that one may express ω(k) as a Taylor expansion
around k = k0 truncated to lowest orders,

1 d2 ω 
 
dω  2
ω(k) ≈ ω(k0 ) + (k − k0 ) + (k − k0 ) . (2.9)
dk k=k0 2 dk 2 k=k0

We use the notations

1 d2 ω 
 
dω 
vg = , β= , (2.10)
dk k=k0 2 dk 2 k=k0

where vg is called the group velocity.

We now substitute Eqs. (2.2) and (2.9) in Eq. (2.7) and we obtain
Z ∞
dk e−α(k−k0 ) ei{kx−[ω(k0 )+(k−k0 )vg +(k−k0 ) β]t}
2 2
F (x, t) =
−∞
Z ∞
dk ′ e−αk ei{(k0 +k )x−[ω(k0 )+k vg +k β]t}
′2 ′ ′ ′2
=
−∞
Z ∞ (2.11)
′2 ′
i[k0 x−ω(k0 )t]
=e dk ′ e−(α+iβt)k eik (x−vg t)
−∞
Z ∞
′ ′2 ′ ′
= ei[k0 x−ω(k0 )t] dk ′ e−α k eik x ,
−∞

where α′ = α + iβt and x′ = x − vg t. We note that the integral appearing in Eq. (2.11) is just the
same as that evaluated in Eq. (2.3), with α replaced by α′ and x replaced by x′ . Thus, we may

7
write
Z ∞
′ ′2 ′ ′
i[k0 x−ω(k0 )t]
F (x, t) = e dk ′ e−α k eik x
−∞
(2.12)
π
r
2
i[k0 x−ω(k0 )t]
=e e−(x−vg t) /[4(α+iβt)] .
α + iβt

Again, we consider the square moduli of g(k) and of F (x, t) as in Eq. (2.4),
π
e−α(x−vg t) /[2(α +β t )] .
2 2 2 2 2
|g(k)|2 = e−2α(k−k0 ) , |F (x, t)|2 = p (2.13)
2 2
α +β t 2

One recognizes Gaussian signals both in k–space and in x–space. The peak of the Gaussian signal
in x is located at x = vg t, and thus it travels in the x–direction with the constant group velocity,
vg .
√
The width of the Gaussian signal in k–space is obviously ∆k = 2/ 2α, while the width of the
Gaussian signal in x–space is now a function of time,
r
√ β 2 t2
∆x = 2 2α 1 + 2 . (2.14)
α
The width of the Gaussian signal in x–space increases in time. This means that the time evolution
of the wave packet implies a spreading in x–space with a decreasing value at the peak position,
x = vg t. The latter feature can be easily inferred from Eq. (2.13).

Figure 2: Spreading of the wave packet in x–space

As a consequence, the uncertainty principle,

r
β 2 t2
∆k ∆x = 4 1 + 2 > 4, =⇒ ∆k ∆x & O(1), (2.15)
α
now implies a more and more severe indetermination of the wave packets in k–space and in x–space
as time progresses.

2.1 Heisenberg uncertainty principle

The goal of our analysis is the formulation of a wave theory of particles. Thus, we must intend
the wave packet in x–space, travelling with a velocity vg , as a particle with mass m, momentum
p, and kinetic energy E = p2 /(2m). This suggests the relationship

dω p
vg = = . (2.16)
dk m

8
Moreover, by invoking Planck’s formula E = hν = ~ω, we obtain
E p2
ω= = . (2.17)
~ 2m~
Eqs. (2.16) and (2.17) yield,
dω p 1 dω
= = , (2.18)
dp m~ ~ dk
that can be satisfied on setting
p
k= . (2.19)
~
Since λ = 2π/k, Eq. (2.19) is consistent with de Broglie’s assumption, Eq. (1.5). We now consider
the expressions of ω and k given by Eqs. (2.17) and (2.19) and the structure of the wave packet
described by Eq. (2.7) to express the wave function of a particle as
Z ∞
1
ψ(x, t) = √ dp φ(p)ei(px−Et)/~ . (2.20)
2π~ −∞
√
The constant 1/ 2π~ plays just the role of an overall normalization of the wave packet, with no
implications for the features described above. An important point is the interpretation of ψ(x, t)
as a model of a moving particle. For instance, the spreading of the wave packet associated with
the time evolution cannot be interpreted as an increase in size of the particle, say an electron or a
nucleus. On the other hand, the width of the wave packet in x–space describes the width of the
region of space where the particle is likely to be found at a given instant of time. The spreading
of the wave packet does not mean that the particle increases its size. It means that the region
of space where the particle is likely to be found becomes larger and larger in the time evolution.
Hence, its localization becomes more and more uncertain. Werner Heisenberg felt uneasy with the
new role of the physical modelling of Nature, and he wrote (see Baggott, 2011, page 94):

“When we know the present precisely, we can predict the future, is not the conclusion
but the assumption. Even in principle we cannot know the present in all detail. For
that reason everything observed is a selection from a plenitude of possibilities and a
limitation on what is possible in the future. As the statistical character of quantum
theory is so closely linked to the inexactness of all perceptions, one might be led to the
presumption that behind the perceived statistical world there still hides a real world in
which causality holds. But such speculations seem to us, to say it explicitly, fruitless
and senseless. Physics ought to describe only the correlation of observations. One can
express the true state of affairs better in this way: Because all experiments are subject
to the laws of quantum mechanics . . . , it follows that quantum mechanics established
the final failure of causality.”

We are now ready for interpreting the uncertainty relationship, Eq. (2.15), in terms of the
position x and the momentum p of the particle,

∆p ∆x & O(~), (2.21)

where Eq. (2.19) has been used. The relationship between the widths of the Gaussian signals
associated with |φ(p)|2 and |ψ(x, t)|2 means that the uncertainty in the knowledge of the position

9
of a particle is inversely proportional to the uncertainty in the knowledge of the momentum of the
particle. The constant of proportionality is O(~).
The smallness of ~ ensures that Eq. (2.21) is effective, and hence marks a breakup of classical
physics, only for microscopic systems.

2.2 The wave equation for a free particle

We start from Eq. (2.20) and evaluate the time derivative of ψ(x, t),
Z ∞
∂ψ(x, t) i
=− √ dp Eφ(p)ei(px−Et)/~
∂t ~ 2π~ −∞
Z ∞ (2.22)
i
=− √ dp p2 φ(p)ei(px−Et)/~ ,
2m~ 2π~ −∞
where the relationship E = p2 /(2m) for the kinetic energy has been used. We also evaluate the
second derivative of ψ(x, t) with respect to x,
Z ∞
∂ 2 ψ(x, t) 1
= − √ dp p2 φ(p)ei(px−Et)/~ . (2.23)
∂x2 ~2 2π~ −∞
The integrals on the right hand sides of Eqs. (2.22) and (2.23) are coincident, so that one can write
the equation
∂ψ(x, t) ~2 ∂ 2 ψ(x, t)
i~ =− . (2.24)
∂t 2m ∂x2
Eq. (2.24) is a second order partial differential equation that describes the motion of a free particle.
In fact, no force and hence no potential energy is assumed to exist, as the energy is just the kinetic
energy E = p2 /(2m). Since Eq. (2.24) was obtained from Eq. (2.20), the general solution of
Eq. (2.24) is
∞
1
Z
ψ(x, t) = √ dp φ(p)ei(px−Et)/~ . (2.25)
2π~ −∞

Eq. (2.24) is a linear differential equation, so that any linear combination of two solutions of
Eq. (2.24), ψ1 (x, t) and ψ2 (x, t),

ψ(x, t) = c1 ψ1 (x, t) + c2 ψ2 (x, t), c1 ∈ C, c2 ∈ C, (2.26)

is a solution of Eq. (2.24).

The wave function ψ(x, t) can be determined at every instant of time t > 0, by solving Eq. (2.24),
if the wave function is known at the initial time t = 0. In other words, the initial condition,

ψ(x, 0) = Ψ0 (x), (2.27)

must be prescribed. This feature is a consequence of Eq. (2.24) being first order in the time
variable.

2.3 The statistical interpretation of ψ(x, t)

We pointed out that the spreading of the wave packet defined by
Z ∞
1
dp φ(p)ei[px−(p /2m)t]/~
2
ψ(x, t) = √ (2.28)
2π~ −∞

10
does not mean an increase in size of the free particle, but just an increase in width of the region
where the particle is likely to be found. This observation implies a statistical interpretation of
the wave function ψ(x, t). The wave function is complex–valued, so that the spreading of the
wave packet was established on evaluating the width ∆x of the real distribution |ψ(x, t)|2 . Thus a
reasonable guess is that

P (x, t) dx = |ψ(x, t)|2 dx (2.29)

is the probability of finding the particle in the infinitesimal region between x and x + dx. For this
interpretation to be consistent, we must require that
Z ∞
dx |ψ(x, t)|2 = 1. (2.30)
−∞

Eq. (2.30) is a normalization condition of the wave function induced by its statistical interpretation:
the probability of finding the particle somewhere is 1. The normalization condition Eq. (2.29) makes
sense only if the integral on the left hand side of Eq. (2.30) exists,
Z ∞
dx |ψ(x, t)|2 < ∞, (2.31)
−∞

i.e. the wave function must be an element of the set L2 (R) of square integrable functions in the
real domain.
The probability density P (x, t), defined by Eq. (2.29), is conserved in the time evolution. This
statement can be proved as follows. We evaluate the complex conjugate of the wave equation
(2.24),

∂ψ(x, t) ~2 ∂ 2 ψ(x, t)
i~ = . (2.32)
∂t 2m ∂x2
We multiply Eq. (2.24) by ψ(x, t), we multiply Eq. (2.32) by ψ(x, t) and sum the two resulting
equations, so that we obtain

~2 ∂2ψ ∂2ψ
   
∂ψ ∂ψ
i~ ψ +ψ =− ψ −ψ ,
∂t ∂t 2m ∂x2 ∂x2
~2 ∂
 
∂ ∂ψ ∂ψ (2.33)
i~ |ψ|2 = − ψ −ψ ,
∂t 2m ∂x ∂x ∂x
∂P (x, t) ∂j(x, t)
+ = 0,
∂t ∂x
where the probability current density, j(x, t), is defined as
 
~ ∂ψ(x, t) ∂ψ(x, t)
j(x, t) = ψ(x, t) − ψ(x, t) . (2.34)
2im ∂x ∂x

Eq. (2.33) has the form of a conservation law for the density P (x, t) with the associated current
j(x, t). A natural consequence of the wave function ψ(x, t) being square integrable over the real
axis, is that

lim ψ(x, t) = 0, (2.35)

x→±∞

and the same obviously holds for its complex conjugate, ψ(x, t), so that it must hold for j(x, t), as
one can infer from its definition. Then, on integrating Eq. (2.33) with respect to x over the real

11
axis R, we conclude that
Z ∞
∂
dx P (x, t) = lim j(x, t) − lim j(x, t) = 0. (2.36)
∂t −∞ x→−∞ x→∞

Eq. (2.36) proves the time–independence of the integral on the left hand side of the normalization
condition Eq. (2.30), and definitely justifies the choice to set this integral to 1.
We mention that there exist quantum states such that the probability density P (x, t) = |ψ(x, t)|2
is uniform over the real axis. In these states, it is evidently not possible to normalize the wave
function as prescribed in Eq. (2.30), since the integral of |ψ(x, t)|2 over the range (−∞, ∞) is
infinite. This happens for the quantum states where the momentum of the particle is known with
certainty. In these states, as a consequence of the uncertainty principle Eq. (2.21), the position of
the particle is completely unknown, so that the probability density P (x, t) is uniformly distributed
over the real axis. The impossibility to prescribe the normalization condition Eq. (2.30) does not
affect in any sense the statistical interpretation of the probability density P (x, t) = |ψ(x, t)|2 .

12
3 The Schrödinger equation
The conservation of the probability density, Eq. (2.33), holds also when the particle is not free,
but subject to a potential V (x). However, the wave equation (2.24) must be generalized, so that
it is replaced by

∂ψ(x, t) ~2 ∂ 2 ψ(x, t)
i~ =− + V (x) ψ(x, t). (3.1)
∂t 2m ∂x2
Eq. (3.1) is called the one–dimensional Schrödinger equation. It can be easily extended to the
three-dimensional case,

∂ψ(r, t) ~2
i~ =− ∇2 ψ(r, t) + V (r) ψ(r, t), (3.2)
∂t 2m
where r = (x, y, z) is the position vector.
The conservation of the probability density,

P (r, t) = |ψ(r, t)|2 , (3.3)

now reads
∂P (r, t)
+ ∇ · j(r, t) = 0, (3.4)
∂t
where the probability current density is a vector field given by
~  
j(r, t) = ψ(r, t) ∇ψ(r, t) − ψ(r, t) ∇ψ(r, t) . (3.5)
2im

3.1 Expectation values of physical observables

Let us consider a physical observable expressed as a function of the position, f (x). Since P (x, t) =
|ψ(x, t)|2 = ψ(x, t) ψ(x, t) is the probability density, one can evaluate the expectation value of f (x)
as
Z ∞ Z ∞ Z ∞
2
hf (x)i = dx f (x) P (x, t) = dx f (x) |ψ(x, t)| = dx ψ(x, t) f (x) ψ(x, t), (3.6)
−∞ −∞ −∞

Let us recall the relationship between position x and momentum p in classical mechanics,
dx
p = mv = m . (3.7)
dt
A reasonable guess is that, for a quantum system, the expectation values of x and p follow the
same relationship,
dhxi
hpi = m . (3.8)
dt
Hence, on account of Eq. (3.6), we can write

dhxi d ∞
Z
hpi = m =m dx ψ x ψ
dt dt −∞
Z ∞  
∂ψ ∂ψ
=m dx x ψ+ψ x (3.9)
−∞ ∂t ∂t
Z ∞  2
∂2ψ

~ ∂ ψ
= dx x ψ−ψ x ,
2i −∞ ∂x2 ∂x2

13
where we used the Schrödinger equation (3.1). We can employ the integration by parts to rewrite
Eq. (3.9) as,
∞
~ ∞ ~ ∞

~ ∂ψ ∂ψ ∂ψ ∂ψ
Z Z
hpi = x ψ−ψ x − dx ψ+ dx ψ
2i ∂x ∂x −∞ 2i −∞ ∂x 2i −∞ ∂x
 ∞ Z ∞ Z ∞ (3.10)
~ ∂ψ ∂ψ ~ ∂ψ ~ ∂
= x ψ−ψ x −ψ ψ + dx ψ = dx ψ ψ.
2i ∂x ∂x −∞ i −∞ ∂x −∞ i ∂x

In fact, for square integrable functions, we have

 ∞
∂ψ ∂ψ
x ψ−ψ x −ψ ψ = 0. (3.11)
∂x ∂x −∞

The expectation value of p evaluated in Eq. (3.10),

Z ∞
~ ∂
hpi = dx ψ(x, t) ψ(x, t), (3.12)
−∞ i ∂x

strongly suggests that we identify the momentum with the differential operator
~ ∂
p= . (3.13)
i ∂x
Then, in general, we have
!
∞
~ ∂
Z
hf (p)i = dx ψ(x, t) f ψ(x, t). (3.14)
−∞ i ∂x

A specially interesting quantity is the Hamiltonian function of classical mechanics (see Appendix A),

p2
H(x, p) = + V (x). (3.15)
2m
On account of Eq. (3.13), this corresponds to the quantum mechanical Hamiltonian operator,

~2 ∂ 2
H=− + V (x). (3.16)
2m ∂x2
We have achieved an important point: the quantum mechanical observables are defined through
their expectation values. Thus, they are defined as operators acting on the wave function ψ(x, t).
The position operator is simple, just the multiplication of x by ψ(x, t). The momentum operator
and the Hamiltonian operator are less trivial as they involve derivatives. The classical observables
commute, since xp and px coincide, but the quantum observables in general do not. For example,
we may evaluate the commutator of the quantum operators x and p,

[x, p] = xp − px. (3.17)

We can write
~ ∂ ~ ∂ ~
[x, p] ψ(x, t) = x ψ(x, t) − [x ψ(x, t)] = − ψ(x, t) = i~ ψ(x, t), (3.18)
i ∂x i ∂x i
and this means that

[x, p] = i~. (3.19)

14
3.2 The x–space and the p–space
A representation of the wave function ψ(x, t) can be based on a Fourier transform
Z ∞
1
ψ(x, t) = √ dp φ(p, t)eipx/~ . (3.20)
2π~ −∞
The inversion theorem of Fourier transforms yields
Z ∞
1
φ(p, t) = √ dx ψ(x, t)e−ipx/~ . (3.21)
2π~ −∞
One can think of φ(p, t) as the wave function in the momentum space, while ψ(x, t) is the wave
function in the coordinate space. For the sake of brevity, the coordinate space is called x–space,
while the momentum space is called p–space. The interpretation of φ(p, t) as the wave function in
p–space is supported by the Parseval–Plancherel formula,
Z ∞ Z ∞
dp |φ(p, t)|2 = dx |ψ(x, t)|2 . (3.22)
−∞ −∞

Eq. (3.22) implies that the normalization of ψ(x, t) coincides with the normalization of φ(p, t).
Then, Eq. (2.30) yields
Z ∞
dp |φ(p, t)|2 = 1. (3.23)
−∞

From Eqs. (3.12) and (3.20), we can write the expectation value of p as
Z ∞
~ ∂
hpi = dx ψ(x, t) ψ(x, t)
−∞ i ∂x
Z ∞ Z ∞
~ ∂ 1
= dx ψ(x, t) √ dp φ(p, t)eipx/~
−∞ i ∂x 2π~ −∞
Z ∞ Z ∞
1
= dx ψ(x, t) √ dp p φ(p, t)eipx/~ (3.24)
−∞ 2π~ −∞
Z ∞ Z ∞
1
= dp p φ(p, t) √ dx ψ(x, t)eipx/~
−∞ 2π~ −∞
Z ∞ Z ∞
= dp p φ(p, t) φ(p, t) = dp p |φ(p, t)|2 .
−∞ −∞

Eqs. (3.23) and (3.24) suggest that |φ(p, t)|2 dp can be interpreted as the probability of the particle
to have a momentum between p and p + dp.
The possibility to define a wave function in the x–space, ψ(x, t), and a wave function in the
p–space, φ(p, t), leads us to the conclusion that we may have different representations of the
wave function. If we adopt the x–space representation of the wave function, then the quantum
mechanical operators for the position and for the momentum are
~ ∂
x̂ = x, p̂ = . (3.25)
i ∂x
Hereafter, the quantum mechanical operators will be denoted by the hat symbol, ˆ. This is a
common practice to distinguish the generally non–commuting quantum observables, or q–numbers,
from the commuting classical observables, or c–numbers.
If we adopt the p–space representation of the wave function, then the quantum mechanical
operators for the position and for the momentum are
∂
x̂ = i~ , p̂ = p. (3.26)
∂p

15
The form of the position operator x̂ in the p–space representation is easily proved on evaluating
the expectation value of the position x,
Z ∞
hxi = dx ψ(x, t) x ψ(x, t)
−∞
Z ∞ Z ∞
1
= dx ψ(x, t) x √ dp φ(p, t)eipx/~
−∞ 2π~ −∞
Z ∞ Z ∞
1
= dp φ(p, t) √ dx x ψ(x, t)eipx/~ (3.27)
−∞ 2π~ −∞
Z ∞ Z ∞
∂ 1
=− dp φ(p, t) i~ √ dx ψ(x, t)eipx/~
−∞ ∂p 2π~ −∞
Z ∞ Z ∞
∂ ∂
=− dp φ(p, t) i~ φ(p, t) = dp φ(p, t) i~ φ(p, t),
−∞ ∂p −∞ ∂p

where we used the obvious fact that hxi is real, so that it coincides with its complex conjugate.
In general, we have
!
∞
∂
Z
hf (x)i = dp φ(p, t) f i~ φ(p, t). (3.28)
−∞ ∂p

3.3 On the general structure of quantum mechanics

We learned from the analysis of the wave–particle duality in a quantum particle that all the
accessible information on the state of a particle at a given time can be obtained from the wave
function ψ, in terms of the expectation values of physical observables. On the other hand, the one–
dimensional state of a classical particle is determined by the pair (x, p), position and momentum,
at a given time. In classical mechanics, the space of states is the phase space, i.e. the vector space
of the pairs (x, p).
The space of the states of a quantum particle is the functional space whose elements are the
wave functions ψ. The space of states is a linear space. This feature is a consequence of the linearity
of the Schrödinger equation: a linear combination of any two solutions of the Schrödinger equation
is still a solution of the Schrödinger equation. The possibility of having different representations of
the quantum states, such as the x–space representation and the p–space representation discussed in
the preceding section, suggests that we denote the states of a quantum system in a representation–
independent form.
 
Paul Dirac proposed a notation based on the ket symbol: a quantum state is expressed as ψ .
The linear space of the states of a quantum system is endowed with a norm and with an inner
product,
Z ∞
k ψ k 2 = ψ ψ = dx |ψ(x, t)|2 ,

 
(3.29a)
−∞
Z ∞

 
 
 
ψ ϕ = dx ψ(x, t)ϕ(x, t), ϕψ = ψ ϕ . (3.29b)
−∞
     

The inner product between ψ and ϕ is obtained by transforming the ket ψ into a bra, ψ 

  
 
and then contracting the bra ψ  with the ket ϕ to obtain the bra–c–ket ψ ϕ . Formally the
bra is the dual vector of a corresponding ket, the relationship between the bras and the kets is
just as that between row vectors and column vectors in a Rn vector space. The main point is that
 
the vector space of the quantum states ψ is not finite–dimensional. The mathematical structure

16
of this vector space is what the mathematicians call a Hilbert space. In practice, a Hilbert space
is an infinite–dimensional vector space endowed with an inner product. We will denote as H the
Hilbert space of the quantum states. The main properties of the inner product are

 
 
ψ ϕ = ϕψ , (3.30a)

 
   
ψ ψ ≥ 0, ψ ψ = 0 =⇒ ψ = 0, (3.30b)

 
 
 
ψ c1 ϕ1 + c2 ϕ2 = c1 ψ ϕ1 + c2 ψ ϕ2 , c1 , c2 ∈ C. (3.30c)

If the quantum states are the elements of the Hilbert space H, the quantum observables are linear
operators in the vector space H. A physical observable  must be a special linear operator. In
 
fact, let a be a state such that the observable  displays the value a, namely
   
Âa = aa . (3.31)

We can collect all the possible values allowed for the observable Â, {a1 , a2 , . . . , an , . . . }, and
     
consider the quantum states, {a1 , a2 , . . . , an , . . . }, where  displays one of these allowed
values. It is straightforward, from Eq. (3.31), to identify {a1 , a2 , . . . , an , . . . } with the eigenvalues
     
of the linear operator  and {a1 , a2 , . . . , an , . . . } with the eigenvectors, or more precisely
with the eigenstates. If it must describe a physical observable, the linear operator  must have
real eigenvalues, i.e.  must be a Hermitian operator. A Hermitian operator  coincides with its
adjoint operator † , defined so that

 
†      
ψ Âϕ = Â ψ ϕ , ∀ ψ , ϕ ∈ H. (3.32)

As a consequence of Eq. (3.32), a Hermitian operator is also called self–adjoint, namely † = Â.
Very useful properties of the adjoint operators regard the sum and the product of two operators
 and B̂, as well as the product by a complex number λ,

†
†
†
 + B̂ = † + B̂ † , ÂB̂ = B̂ † † , λ = λ † . (3.33)

A quantum observable does not necessarily display a discrete sequence of eigenvalues, the generic
eigenvalue a can also vary with continuity. One can think, for instance, to the one–dimensional
position operator x̂. Evidently, the eigenvalues of x̂ are all the possible positions x in the real
axis. Depending on the spectrum of the eigenvalues being discrete or continuous, we may have
different forms of normalization for the eigenstates of an observable Â. Eigenstates corresponding
to different eigenvalues are orthogonal, i.e. their inner product is zero. For a discrete spectrum,
we have a normalization formula based on Dirac’s delta function,

 ′
aa = δ(a − a′ ), (3.34)
   
where a and a′ are any two eigenstates of Â.
For a continuous spectrum, we have a normalization formula based on Kronecker’s delta symbol,

 
an am = δnm , (3.35)
   
where an and am are any two eigenstates of Â.

17
3.4 Bases in the Hilbert space and representations
The eigenstates of any Hermitian operator  may form an orthonormal basis of the Hilbert space
H, where the term orthonormal means that the elements of the basis are mutually orthogonal,
with respect to the inner product, and that they are normalized, with respect to either Eq. (3.34)
 
or (3.35). Thus, any quantum state ψ can be expanded as a linear combination of the eigenstates
of Â,
  X  
 
ψ = a aψ . (3.36)
a

The sum over a is to be considered, entirely or partly, as an integral over a if the whole spectrum
of the eigenvalues of Â, or part of it, is continuous. From the expansion (3.36), one can define a
 
    
projector, i.e. an operator a a that yields the projection of ψ in the subspace spanned by a .
 
The completeness of the basis, i.e. the possibility to express any state ψ ∈ H by the expansion
(3.36), can be formally rephrased as,

ˆ
X  

a a = I, (3.37)
a

where Iˆ is the identity operator in H. One may easily prove that two commuting Hermitian
operators  and B̂ may have common eigenstates,
   
[Â, B̂] = 0, Âa = aa ,
   
B̂ Âa = aB̂a , (3.38)
   
ÂB̂a = aB̂a .
 
Therefore, B̂a is an eigenstate of  corresponding to the eigenvalue a. This may imply two
possible cases.
     
1. The eigenvalue a is non–degenerate, so that B̂a = ba , where b ∈ R, namely a is an
eigenstate of B̂.
 
2. The eigenvalue a is degenerate, so that B̂a can be expressed as a linear combination of differ-
 
ent eigenstates a , a′ , a′′ , . . . of the operator  corresponding to the same eigenvalue
   

a. In this case, the operator B̂, as well as any other Hermitian operator that commutes with
Â, may serve to remove the degeneracy.
 
If we consider the orthonormal basis formed with the eigenstates a of a given observable Â,
 
we obtain the a–space representation of the quantum state ψ(t) at time t,

 
ψ(a, t) = aψ(t) . (3.39)

The complex–valued function ψ(a, t) is the wave function corresponding to the quantum state
 
ψ(t) in the a–space representation. Examples are the x–space representation and the p–space
representation discussed in the preceding section, obtained when  is the position operator x̂ or
the momentum operator p̂, respectively,

 
ψ(x, t) = xψ(t) ,

  (3.40)
ψ(p, t) = pψ(t) .

18
  
At an instant t, the expectation value of an observable  in a quantum state ψ(t) is expressed,
in the Dirac notation, as

 
hÂi = ψ(t)Âψ(t) , (3.41)

so that, on account of Eq. (3.37), we can write

X
 
  X
 
  X
 
 
hÂi = ψ(t)a aÂψ(t) = ψ(t)a Âaψ(t) = a ψ(t)a aψ(t)
a a a
X X (3.42)
= a ψ(a, t) ψ(a, t) = a |ψ(a, t)|2 .
a a

Again, the sum over a can be possibly replaced by an integral when the eigenvalue a is a continuous
variable.

3.5 The statistical interpretation of quantum states

On the basis of Eq. (3.42), we can formulate the statistical interpretation of the wave function
ψ(a, t). If the spectrum of the eigenvalues {a} is continuous, then |ψ(a, t)|2 da is the probability
that a measurement of the observable  yields an outcome within a and a + da. If the spectrum
of the eigenvalues {a} is discrete, then |ψ(a, t)|2 is the probability that a measurement of the
observable  yields the outcome a. An immediate consequence of this statistical interpretation is
   
the following. If ψ(t) coincides with the eigenstate a at some instant t, then a measurement
of the observable  at that instant yields with certainty, i.e. with probability 1, the outcome a.
 
We said that two commuting observables  and B̂ may have common eigenstates. Let a be such
 
an eigenstate, displaying the eigenvalue a of  and the eigenvalue b of B̂. Then in the state a
simultaneous measurements of the observables  and B̂ yield with certainty, i.e. with probability
1, the values a and b respectively.

3.6 Operator methods and the uncertainty principle

The use of an abstract Hilbert space representation of the quantum states allows one to obtain a
rigorous proof of the uncertainty principle. Let us consider two Hermitian operators, Â and B̂, in
the Hilbert space H. We assume that  and B̂ are non–commuting. The following lemma holds.

Lemma 1. Let Ĉ be a linear operator in H, such that

[Â, B̂] = iĈ,

where  and B̂ are two Hermitian operators in H. Then, Ĉ is also Hermitian.

†
Proof. On account of Eq. (3.33), we may write iĈ = −iĈ † . Moreover,

†
†
[Â, B̂]† = ÂB̂ − B̂  = B̂ † † − † B̂ † = B̂  − ÂB̂ = [B̂, Â] = −iĈ.

This proves that Ĉ = Ĉ † .

We define the uncertainties associated with the observables  and B̂ in terms of the square
deviations from their average values,

2 

2 
∆A2 = Â − hÂi

= ψ  Â − hÂi ψ , (3.43a)

2 

2 
∆B 2 = B̂ − hB̂i

= ψ  B̂ − hB̂i ψ . (3.43b)

19
We can define the operators

V̂ = Â − hÂi, Ŵ = B̂ − hB̂i, (3.44)

so that

∆A2 = ψ V̂ 2 ψ , ∆B 2 = ψ Ŵ 2 ψ .

 
 
(3.45)

We note that





[V̂, Ŵ ] =  − hÂi B̂ − hB̂i − B̂ − hB̂i  − hÂi = [Â, B̂]. (3.46)

We now evaluate the scalar product of the state

   
V̂ ψ + iλŴ ψ , λ ∈ R,

with itself, namely

V̂ ψ V̂ ψ + λ2 Ŵ ψ Ŵ ψ + iλ V̂ ψ Ŵ ψ − iλ Ŵ ψ V̂ ψ ≥ 0,

 
 
 
 

 2 
ψ V̂ ψ + λ2 ψ Ŵ 2 ψ + iλ ψ [V̂, Ŵ ]ψ ≥ 0,

 
 
(3.47)
∆A2 + λ2 ∆B 2 + iλ ψ [Â, B̂]ψ ≥ 0.

 

The inequality (3.47) must hold for every λ ∈ R. The choice of λ that ensures the most restrictive
form of the inequality (3.47) is obtained by seeking the minimum of the left hand side of Eq. (3.47)
with respect to λ. This implies that we force the derivative of

∆A2 + λ2 ∆B 2 + iλ ψ [Â, B̂]ψ

 

with respect to λ to be zero. Thus, we obtain

 
i ψ [Â, B̂]ψ
λ=− , (3.48)
2 ∆B 2
so that, Eq. (3.47) yields

 2
 2
2 ψ [Â, B̂]ψ ψ [Â, B̂]ψ
∆A − + ≥ 0,
4 ∆B 2 2 ∆B 2 (3.49)
2 2 1
 2
∆A ∆B ≥ − ψ [Â, B̂]ψ .
4
As a consequence of Lemma 1, the right hand side of Eq. (3.49) is a non–negative real number.
Eq. (3.49) is a generalized form of Heisenberg’s uncertainty principle. An important application
is for the pair of observables x̂ and p̂. We know in fact from Eq. (3.19) that

[x̂, p̂] = i~, (3.50)

so that the inequality (3.49) yields

1 2
∆x2 ∆p2 ≥ ~ , (3.51)
4
or equivalently
1
∆x ∆p ≥ ~, (3.52)
2
Eq. (3.49) reveals an important fact about the uncertainty principle: it is not possible to achieve
simultaneously and with certainty the values of two non–commuting observables. On the other
hand, one can achieve simultaneously and with certainty the values of two commuting observables.
The second fact is a consequence of the possibility to determine a basis of common eigenstates of
two commuting observables for the Hilbert space H.

20
3.7 The Hamiltonian operator, the evolution operator, and the
stationary states
The expression of the Hermitian operator that describes a quantum observable is obtained from the
corresponding expression of classical mechanics. For instance, as already pointed out in Eqs. (3.15)
and (3.16), the Hamiltonian is given by
p̂2
Ĥ = + V (x̂). (3.53)
2m
And the Schrödinger equation is expressed as
∂    
i~ ψ = Ĥψ . (3.54)
∂t
On integrating Eq. (3.54), one obtains

ψ(t) = e−i(t−t0 )Ĥ/~ ψ(t0 ) ,

   
(3.55)

where the exponential of an operator is defined through its series expansion

∞
X Ân
e = . (3.56)
n=0
n!

Eq. (3.55) implies that one can define an evolution operator,

Û (t, t0 ) = e−i(t−t0 )Ĥ/~ . (3.57)

The operator Û (t, t0 ) is not a Hermitian operator, so that it cannot be associated to an observable
quantity. On the other hand, Û (t, t0 ) is an unitary operator, i.e. it satisfies the mathematical
identity

ˆ
Û (t, t0 ) Û (t, t0 )† = Û (t, t0 )† Û (t, t0 ) = I. (3.58)

The eigenstates of the evolution operator Û (t, t0 ) coincide with the eigenstates of Ĥ, as it can be
 
inferred from its definition. The eigenstates of Ĥ, denoted as {ε }, are also called the energy
eigenstates,
   
Ĥε = E ε , (3.59)
 
where E is the eigenvalue of Ĥ in the eigenstate ε . Eq. (3.59) plays a special role in the study
of quantum systems and it is well–known as the time–independent Schrödinger equation.
The time evolution of an energy eigenstate is obtained from Eq. (3.55)

ε(t) = e−i(t−t0 )Ĥ/~ ε(t0 ) = e−i(t−t0 )E/~ ε(t0 ) ,

     
(3.60)

The energy eigenstates are quite peculiar as the expectation value of an arbitrary observable Â
does not depend on time, as it can be deduced from Eq. (3.60),

 
 
ε(t)Âε(t) = ε(t0 )Âε(t0 ) , ∀t, t0 ∈ R. (3.61)

In this sense, the energy eigenstates are termed stationary states; they are time–independent except
for a phase factor,

e−i(t−t0 )E/~ ,

ineffective with respect to the expectation values of any observable.

21
4 One–dimensional quantum mechanics
Simple examples of the quantum mechanical behaviour can be given by considering a particle in a
one–dimensional potential field expressed through a given function V (x). The x–space represen-
tation is adopted, so that the quantum state is represented by means of the wave function ψ(x, t).
The time–independent Schrödinger equation for the stationary states

ψ(x, t) = e−itE/~ Ψ(x), (4.1)

can be written as
~2 d2 Ψ(x)
− + V (x)Ψ(x) = EΨ(x). (4.2)
2m dx2

4.1 An infinite potential well

Let us consider a potential given by,



 ∞, x < 0,

V (x) = 0, 0 < x < a, (4.3)



∞, x > a.

The infinite value of V (x) outside the interval 0 < x < a means that the space outside this
interval is not accessible to the particle. One can think of a particle in a box having width a.
Since the particle is not allowed to escape the region 0 < x < a, the continuity of the probability

Figure 3: The infinite potential well

distribution at x = 0 and x = a imposes the wave function to vanish at these positions. Thus the
time–independent Schrödinger equation, Eq. (4.2), is subject to the boundary conditions,

Ψ(0) = 0 = Ψ(a). (4.4)

For 0 < x < a, Eq. (4.2) can be written as

d2 Ψ(x) 2mE
= − 2 Ψ(x). (4.5)
dx2 ~

22
Eq. (4.5), with the boundary conditions Eq. (4.4), admits the non–trivial solutions, i.e. the eigen-
functions,

Ψn (x) = Cn sin(nπx/a), n = 1, 2, 3, . . . , (4.6)

provided that the energy E assumes the eigenvalues,

n2 π 2 ~2
En = , n = 1, 2, 3, . . . . (4.7)
2ma2
Eq. (4.7) reveals an important fact about the quantum particle: its energy cannot assume arbitrary
values, but just those defined by the discrete spectrum of the eigenvalues En .
As it usually happens in the solution of an eigenvalue problem, the normalization constants
appearing in Eq. (4.6) can be chosen arbitrarily. However, we should not forget the physical
meaning of the wave function ψ(x, t). In fact, the probability distribution |ψ(x, t)|2 must be such
that its integral over the interval 0 < x < a is equal to 1,
Z a Z a
dx |ψ(x, t)|2 = 1, =⇒ dx |Ψn (x)|2 = 1, (4.8)
0 0
p
and this yields Cn = 2/a, so that
r
2
Ψn (x) = sin(nπx/a), n = 1, 2, 3, . . . , (4.9)
a
A notable feature of the energy spectrum, Eq. (4.7), for the particle in the infinite potential well is
that the relative increment between two neighbouring energy levels is a decreasing function of n,
∆En En+1 − En 2 1 ∆En
= = + 2, lim = 0. (4.10)
En En n n n→∞ En
This result can be rephrased as follows: when the energy of the particle is such that

π 2 ~2
E≫ ,
2ma2
the energy spectrum becomes continuous. In other words, when the energy assumes sufficiently
high values, the quantum behaviour tends to mimic the classical behaviour of a particle in a box.
Another notable feature of the energy spectrum, Eq. (4.7), is the existence of a non–vanishing
minimum value for the energy of the particle. It is the value of the energy for the level n = 1. This
value,

π 2 ~2
E1 = , (4.11)
2ma2
is usually called the zero–point energy. The quantum particle cannot display an energy lower than
the zero–point energy.

4.2 Quantum tunnelling through a barrier

I do not know anybody in the macroscopic world that is able to jump over a fence without a kinetic
energy sufficient to overcome the gravitational potential energy associated with the height of the
fence. A quantum particle can do it: this effect is called quantum tunnelling.

23
To illustrate the quantum tunnelling, let us consider the potential barrier defined by the function



 0, x < 0,

V (x) = V0 , 0 < x < a, (4.12)



0, x > a,

where V0 > 0 is a given constant. Unlike the preceding example, the region of space allowed to

Figure 4: The potential barrier

the quantum particle is in this case the whole real axis. For a classical particle, it is impossible
to penetrate the potential barrier if its energy is lower than the barrier height V0 . For a quantum
particle, things are different as we will see.
On the left of the barrier, x < 0, the time–independent Schrödinger equation yields
r
d2 Ψ(x) 2 2mE
2
= −k Ψ(x), k= . (4.13)
dx ~2
The general solution is the superposition of incident and reflected travelling waves

Ψ(x) = eikx + R e−ikx . (4.14)

In the barrier, 0 < x < a, the time–independent Schrödinger equation is written as

r
d2 Ψ(x) 2 2m(E − V0 )
= −kB Ψ(x), kB = . (4.15)
dx2 ~2
If E > V0 , the general solution is

Ψ(x) = C eikB x + C ′ e−ikB x . (4.16)

If E < V0 , the general solution is

r
βx ′ −βx 2m(V0 − E)
Ψ(x) = C e +C e , β= . (4.17)
~2
On the right of the barrier, x > a, the time–independent Schrödinger equation yields
r
d2 Ψ(x) 2 2mE
2
= −k Ψ(x), k= . (4.18)
dx ~2

24
We may have, in this region, a transmitted wave given by

Ψ(x) = T eikx . (4.19)

We can obtain the reflection coefficient R, the transmission coefficient T , and the integration
constants C and C ′ , by imposing the smooth matching of the wave function at the positions x = 0
and x = a, namely
 
− + dΨ  dΨ 
Ψ(0 ) = Ψ(0 ), = ,
dx x=0− dx x=0+
  (4.20)
− + dΨ  dΨ 
Ψ(a ) = Ψ(a ), = .
dx  −
x=a dx  + x=a

The continuity of Ψ(x) is an obvious consequence of the continuity of the probability density
|Ψ(x)|2 . On the other hand, the continuity of the first derivative dΨ(x)/dx is a more subtle
matter. If one integrates Eq. (4.2) across a point of discontinuity of the potential V (x), say x = 0,
one obtains
δ Z δ
~2 dΨ(x)

− = dx [E − V (x)] Ψ(x), δ > 0. (4.21)
2m dx −δ −δ

If we let δ → 0, then the right hand side of Eq. (4.21) must vanish unless we have a Dirac’s delta
behaviour of V (x) at x = 0, which is not the present case. Thus, we immediately obtain the
continuity of dΨ(x)/dx across the point x = 0. Obviously, the same holds at x = a.
We now focus our attention on the case E < V0 , i.e. the case such that the penetration of the
potential barrier would be impossible for a classical particle. We can determine the constants R,
T , C and C ′ by employing Eqs. (4.14), (4.17), (4.19), and (4.20),

k 2 + β 2 sinh(aβ)


R= ,
2ikβ cosh(aβ) + (k 2 − β 2 ) sinh(aβ)
2k(k − iβ)
C= ,
(k − iβ)2 − e2aβ (k + iβ)2
(4.22)
2e2aβ k(k + iβ)
C′ = ,
e (k + iβ)2 − (k − iβ)2
2aβ

4iea(−ik+β) kβ
T = .
e2aβ (k + iβ)2 − (k − iβ)2

Physically meaningful quantities are |R|2 and |T |2 , as they express the probability that the particle
be reflected or transmitted from the potential barrier, respectively. We obtain

(k 2 + β 2 )2 sinh2 (aβ) (k 2 + β 2 )2 sinh2 (aβ)

|R|2 = 2 = 2 ,
4k 2 β 2 cosh2 (aβ) + (k 2 − β 2 ) sinh2 (aβ) (k 2 + β 2 ) sinh2 (aβ) + 4k 2 β 2
(4.23)
8k 2 β 2 4k 2 β 2
|T |2 = 2 = 2 .
(k 2 + β 2 ) cosh(2aβ) − k 4 + 6k 2 β 2 − β 4 (k 2 + β 2 ) sinh2 (aβ) + 4k 2 β 2

From Eq. (4.23), one can easily see that

|R|2 + |T |2 = 1, (4.24)

with the obvious meaning that the particle can be either reflected or transmitted from the barrier.
The important point is that |T |2 > 0. This means that, even if E < V0 , the particle has a non–zero

25
probability of penetrating the potential barrier. This phenomenon of quantum tunnelling through
a potential barrier is quite important. For instance, the thermonuclear fusion of protons in the
stars is possible in most cases only as a consequence of the tunnel effect of protons through the
repulsive Coulomb potential barrier. We note that |T |2 decreases with a. This is an expected fact,
as the larger is the potential barrier the lower is the probability of quantum tunnelling through it.

4.3 The quantum harmonic oscillator

As is well known from classical mechanics the elastic potential of a simple harmonic oscillator is
expressed as
1
V (x) = mω 2 x2 , (4.25)
2
where ω is the angular frequency of the oscillator. Thus, the time–independent Schrödinger equa-
tion can be written as
~2 d2 Ψ(x) 1
− + mω 2 x2 Ψ(x) = EΨ(x). (4.26)
2m dx2 2
Eq. (4.26) is defined on the whole real axis, and can be rewritten in a more compact form as

d2 Ψ
r
2 mω 2E
− 2 + q Ψ = εΨ, with q = x , ε= . (4.27)
dq ~ ~ω
The conditions at infinity assigned to the solutions of Eq. (4.27) are to make Ψ compatible with
the normalization condition Eq. (2.30),

lim Ψ(x) = 0. (4.28)

x→±∞

We express Ψ(x) as
2
/2
Ψ = N e−q ϕ(q), (4.29)

where N is a normalization constant, so that Eq. (4.27) can be rewritten as

d2 ϕ(q) dϕ(q)
2
− 2q + (ε − 1)ϕ(q) = 0. (4.30)
dq dq
Eq. (4.30) is well known to mathematicians as the Hermite equation. It is a second order differential
equation with variable coefficients. Its general solution1 can be expressed as a linear combination
of the confluent hypergeometric function of the first kind,
2
1F1 (−γ/4; 1/2; q ), with γ = ε − 1, (4.31)

and of the Hermite function,

Hγ/2 (q). (4.32)

However, we note that the confluent hypergeometric function of the first kind can be approximated
for q 2 ≫ 1 as
√
2 π 2
1F1 (−γ/4; 1/2; q ) ∼ q −γ/2−1 eq , (4.33)
Γ(−γ/4)
1 A good description of this point can be found at the web page:

http://mathworld.wolfram.com/HermiteDifferentialEquation.html

26
where Γ is Euler’s gamma function. Thus, if ϕ(q) would involve a term proportional to the confluent
hypergeometric function of the first kind, on account of Eqs. (4.29) and (4.33), the conditions at
infinity (4.28) cannot be satisfied. As for the Hermite function, its asymptotic expression when
q → −∞ is expressed as

2 Γ(1 + γ/2) sin(πγ/2)

Hγ/2 (q) ∼ 2γ/2 q γ/2 − eq q −1−γ/2 √ . (4.34)
π
2
Here, the dangerous term is that proportional to eq . This term disappears, thus ensuring the
validity of the conditions at infinity (4.28), if and only if

sin(πγ/2) = 0, =⇒ γ = 2n, n = 0, 1, 2, 3, . . . . (4.35)

Therefore, the conditions at infinity, Eq. (4.28), are satisfied by considering just the Hermite
function with γ an even natural number. In this case, the Hermite function is a Hermite polynomial
with degree equal to γ/2. Therefore, Eq. (4.29) now reads
2
/2
Ψ = Ψn = Nn e−q Hn (q), with ε = 2n + 1, n = 0, 1, 2, 3, . . . . (4.36)

This means that the allowed energy levels for the harmonic oscillator are given by
 
1
En = n + ~ω, n = 0, 1, 2, 3, . . . . (4.37)
2

As in the case of the infinite potential well, we have a zero–point energy of the harmonic oscillator
given by the lowest energy level,
1
E0 = ~ω. (4.38)
2
The first 10 Hermite polynomials are

H0 (q) = 1,
H1 (q) = 2q,
H2 (q) = −2 + 4q 2 ,
H3 (q) = −12q + 8q 3 ,
H4 (q) = 12 − 48q 2 + 16q 4 ,
(4.39)
H5 (q) = 120q − 160q 3 + 32q 5 ,
H6 (q) = −120 + 720q 2 − 480q 4 + 64q 6 ,
H7 (q) = −1680q + 3360q 3 − 1344q 5 + 128q 7,
H8 (q) = 1680 − 13440q 2 + 13440q 4 − 3584q 6 + 256q 8 ,
H9 (q) = 30240q − 80640q 3 + 48384q 5 − 9216q 7 + 512q 9 .

Eq. (4.39) shows that the Hermite polynomials are even functions of q if n is even, while they are
odd functions of q if n is odd. The normalization constants Nn are found so that the normalization
condition, Eq. (2.30), is satisfied
r
2 1 mω
Nn = n . (4.40)
2 n! ~π

27
On account of Eqs. (4.36), (4.39), and (4.40), the ground state corresponding to the zero–point
energy, Eq. (4.38), is described by a Gaussian wave function,
r
mω −q2 /2
Ψ0 = e . (4.41)
~π

The lowest three stationary states are represented in Fig. 5. An important fact about the quantum

Figure 5: Stationary states of the harmonic oscillator with n = 0, 1, 2

harmonic oscillator is that the particle can be everywhere on the real axis, with decreasing proba-
bility for large values of |x|. On the other hand, the classical harmonic oscillator is confined in a
region −A ≤ x ≤ A, where A is the amplitude of the oscillation. The amplitude A is classically
related to the energy E of the oscillator,
1
E= mω 2 A2 ,
2

28
so that the classical particle with energy E is confined within the region
r r
2E 2E
− ≤x≤ .
mω 2 mω 2
Thus, the quantum particle is allowed to escape the region where it is classically confined, for a
given value of the energy. This phenomenon can be traced back to the quantum tunnel effect,
discussed in the preceding section.

29
Suggestions for further reading
Among the several treatises on quantum mechanics, the following are strongly suggested for a
deeper insight into the behaviour of quantum systems. All these textbooks are at an adequate
level for a reader with a basic mathematical education on calculus, Fourier series and integral
transforms.

Dirac, The Principles of Quantum Mechanics, 1981

It is a classical textbooks whose first edition dates back to 1930. In this book the formal structure of
quantum mechanics based on the properties of the vectors (bras and kets) and the linear operators
in a Hilbert space is clearly described.

Messiah, Quantum Mechanics, 1999

Published in 1958, this book was originally in two volumes. It covers several topics from the basic
description of quantum systems to its extension in the domain of relativistic quantum mechanics.

Gasiorowicz, Quantum Physics, 1974

The edition of this classical textbook that is cited in these notes is of 1974, but there exists the
third edition published in 2003 with substantial changes.

Peres, Quantum Theory: Concepts and Methods, 1995

In this textbook, the focus is on the conceptual foundations of quantum mechanics and of its
statistical interpretation. Topics such as the Einstein–Podolsky–Rosen (EPR) paradox and the
fundamental questions about the measurement process on a quantum system are extensively dis-
cussed.

Sakurai, Modern Quantum Mechanics, 1994

This book is a modern classic. Dirac’s approach to the description of quantum mechanics is adopted
from scratch. It is a textbook at an advanced level with applications of the group theory to the
study of angular momentum, and of the symmetries in quantum systems.

Phillips, Introduction to Quantum Mechanics, 2003

A recent textbook with several qualities. It is very clear, simple and rigorous, definitely suggested
to people looking for a good introductory treatise on quantum physics.

30
Appendix A
A quick glance into analytical mechanics
Let us recall the basic facts about the analytical mechanics of a classical system.
Every mechanical system with N degrees of freedom is described by N generalized coordinates
q = (q1 , q2 , . . . , qN ), with associated generalized velocities q̇ = (q̇1 , q̇2 , . . . , q̇N ). Here, the dot
˜ ˜
stands for time derivative. The equations of motion for the mechanical system are formulated in
terms of the Lagrangian function,

L(q, q̇) = T − V, (A1)

˜ ˜
where T denotes the kinetic energy, and V the potential energy. In Eq. (A1), the Lagrangian
L is intended as a function of the generalized coordinates qi and of the generalized velocities q̇i ,
while it is considered as independent of time. The time independence of L is appropriate for the
conservative mechanical systems, i.e. systems that are either isolated or subject to an external
stationary force field.
The motion of the mechanical system is described by the Euler–Lagrange equations,
 
d ∂L ∂L
− = 0, i = 1, 2, . . . , N. (A2)
dt ∂ q̇i ∂qi

The generalized momentum, pi , associated with the generalized coordinate qi is defined as

∂L
pi = . (A3)
∂ q̇i

A widely used (alternative) terminology for pi is momentum canonically conjugated to qi . Obvi-

ously, one can define N generalized momenta, p = (p1 , p2 , . . . , pN ).
˜
The Hamiltonian function is defined through the Legendre transformation of the Lagrangian
function, namely
N
X
H= pi q̇i − L. (A4)
i=1

If L depends only on q and q̇, one may presume that H is a function of q, q̇ and p. However, on
˜ ˜ ˜ ˜ ˜
account of Eq. (A3), one may easily prove that H is independent of q̇. Then, if L = L(q, q̇), one
˜ ˜ ˜
may write

H = H(q, p). (A5)

˜ ˜
For a conservative mechanical system, the Hamiltonian function coincides with the energy E of
the system, namely

H(q, p) = T + V = E. (A6)
˜ ˜
The motion of the mechanical system is described by the Hamilton equations,
∂H ∂H
q̇i = , ṗi = − . (A7)
∂pi ∂qi

31
Appendix B
Suggested homework

EXERCISE 1
Consider a quantum particle in the following one–dimensional potential,

∞, x < 0,
V (x) =
λx, x > 0,

where λ is a positive constant.

(a) Determine the energy spectrum. Is it continuous or discrete?
(b) Plot the wave functions and the probability densities of the first 3 stationary states.
(c) Could V (x) be a good representation of some physically interesting system?

EXERCISE 2
Consider a quantum particle in the following one–dimensional potential,

∞, x < 0,




0,

0 < x < a,
V (x) =



 V0 , a < x < b,


0, x > b,

where V0 is a positive constant, and b − a > 0 is the thickness of the potential barrier.
(a) Determine the energy spectrum. Is it continuous or discrete?
(b) Determine the transmission coefficient for E < V0 .

EXERCISE 3
Consider a free quantum particle, i.e. a particle with Hamiltonian,

p̂2
Ĥ = .
2m
Evaluate the commutator

[x̂, Ĥ]

and determine the lower bound on the product of the two uncertainties ∆x and ∆E, where ∆x is
the uncertainty associated with x̂ and ∆E is the uncertainty associated with Ĥ.
Comment on the result.

32
References
S. Gasiorowicz. Quantum Physics. Wiley, 1974.

A. C. Phillips. Introduction to Quantum Mechanics. Wiley, 2003.

P. A. M. Dirac. The Principles of Quantum Mechanics. Clarendon Press, 1981.

J. Baggott. The Quantum Story: A History in 40 Moments. Oxford University Press, 2011.

A. Messiah. Quantum Mechanics. Dover, 1999.

A. Peres. Quantum Theory: Concepts and Methods. Kluwer Academic Publishers, 1995.

J. J. Sakurai. Modern Quantum Mechanics. Addison-Wesley, 1994.

33