An-Najah National University

Faculty of Graduate Studies

The Numerical Methods for

Solving Schrödinger Equation

By
Hadeel Muaffaq Hejja

Supervisor
Prof. Naji Qatanani

This Thesis is Submitted in Partial Fulfillment of the Requirements for

the Degree of Master of Mathematics, Faculty of Graduate Studies, An-
Najah National University, Nablus, Palestine.
2019
 III

Dedication
I dedicate this thesis to all my family members, to my

parents, my husband Moath, to my sister, and my brothers,

without their patience, understanding, support, and most
of all love, this work wouldn’t have been possible. I also

dedicate this dissertation to my homeland Palestine.

 IV

Acknowledgement
At the beginning, I am greatful to God for the completion

of this thesis, without whose mercy and clemency nothing

would have been possible.
I am thankful to my supervisor, Prof. Naji Qatanani, for his

encouragement and support to develop and understand the

topic of my research.

My thanks also to my internal examiner Dr. Adnan

Daraghmeh and my external examiner Dr. Abdelhalim
Ziqan from Arab American University for their positive
comments.

I take this opportunity to express gratitude to all my family

members for their help and support. I also thank my parents
and my husband for the unceasing encouragement, support
and attention.

Finally, I offer my thanks and blessing to all of those who

supported me in any respect during the completion of this
thesis.
 ‫‪V‬‬

‫اإلقرار‬

‫أنا الموقع أدناه مقدم الرسالة التي تحمل عنوان ‪:‬‬

‫‪The Numerical Methods for Solving Schrödinger‬‬

‫‪Equation‬‬
‫أقر بأن ما اشتملت عليه هذه الرسالة إنما هي نتاج جهدي الخاص‪ ،‬بإستثناء ما تمت اإلشارة إليه‬

‫حيثما ورد‪ ،‬وأن هذه الرسالة ككل‪ ،‬أو أي جزء منها لم يقدم من قبل لنيل أي درجة علمية أو بحث‬
‫علمي أو بحثي لدى أي مؤسسة تعليمية أو بحثية أخرى‪.‬‬

‫‪Declaration‬‬

‫‪The work provided in this thesis، unless otherwise referenced، is the‬‬

‫‪researcher’s own work، and has not been submitted elsewhere for any‬‬
‫‪other degree or qualification‬‬

‫‪Student's Name:‬‬ ‫اسم الطالب‪:‬‬

‫‪Signature:‬‬ ‫ال ّتوقيع‪:‬‬

‫‪Date:‬‬ ‫ال ّتاريخ‪:‬‬

 VI

Table of contents
Dedication ................................................................................................... III
Acknowledgement....................................................................................... IV
Declaration ................................................................................................... V
Table of contents ......................................................................................... VI
List of Figures .......................................................................................... VIII
List of Tables ................................................................................................ X
Abstract ....................................................................................................... XI
Introduction ................................................................................................... 1
Chapter One................................................................................................... 6
Introduction to Schrödinger Equation ........................................................... 7
1.1Schrödinger Equation ……………………………………………… 7
1.2Derivation of the Time-Independent Schrödinger Equation ................ 9
1.3Derivation of the Time-dependent Schrödinger Equation .................. 11
1.4Schrödinger's operators……………………………........................... 12
1.5Linear Operators…………………………………….. ....................... 14
1.6Wave function…………………………………….. ........................... 14
1.7The equation of motion of the wave function…………..................... 15
1.8Properties of the solutions of the time independent Schrödinger
equation………………………………………………………..………….16
1.9Basis of quantum mechanics...................................... ........................ 18
1.10 Stationary states………………………………………………………………………….19

Chapter Two ................................................................................................ 23

Analytical Methods ..................................................................................... 23
2.1Introduction......................................................................................... 23
2.2Infinite square well..................................................... ........................ 23
2.3Finite square well................................................................................ 27
 VII
2.3.1Bound states .................................................................................... 28
2.4The quantum harmonic oscillator…………………….. ..................... 32
Chapter Three .............................................................................................. 38
Numerical Techniques for Solving Schrödinger Equation ........................ 38
3.1 Finite Difference Method (FDM)............................... ....................... 38
3.2Taylor's Theorem........................................................... ..................... 38
3.3Strategy of Discretization............................................. ...................... 40
3.4Eigenvalue Problem...................................................... ...................... 43
3.5Finite Difference Method for infinite square well............. ................. 47
3.6Finite difference method for finite square well......... ......................... 48
3.7Pseudo-spectral method............................................. ......................... 50
3.8 Pseudo-Spectral Algorithm......................................... ....................... 53
Chapter Four................................................................................................ 55
Numerical Examples and Results ............................................................... 55
4.1 Numerical solution for infinite square well of TISE ......................... 55
4.2 Numerical solution for finite square well.................. ........................ 64
4.3 Example (4.3)………… ….………………………………….…….71

Conclusion..................................................................................... ............. 77
References ................................................................................................... 78
Appendix ..................................................................................................... 83
‫ الملخص‬........................................................................................................... ‫ب‬
 VIII

List of Figures
Fig 1.1: Eigenfunctions and the energy for a particle in a box. .................. 27
Fig 1.2: A finite square well, depth V0, width 2L. ..................................... 27
Fig 1.3: A semilog graph for the energy of finite square well. ................... 36
Fig 1.4:The first five wave functions of the quantum harmonic oscillator.36

Fig 1.5: Grid lines and mesh points of the grid using Finite difference
method ........................................................................................................41
Fig. 4.1(𝑎): A comparison between the exact and approximate solution in
example 4.1. ................................................................................................ 59
Fig 4.1(b): Absolute error between exact and numerical solution in example
4.1 ................................................................................................................ 59
Fig 4.2 (a): A comparison between The exact and approximate results of
𝐸𝑛. ............................................................................................................... 60
Fig 4.2 (b): Absolute error between The exact and approximate results of
𝐸𝑛. ............................................................................................................... 61
Fig 4.3 (a): A comparison between exact and numerical solution in example
4.1. ............................................................................................................... 63
Fig 4.3 (b): Absolute error between exact and numerical solution in example
4.1. ............................................................................................................... 64
Fig 4.4 (a): A comparison between the exact and approximate solution in
example 4.2. ................................................................................................ 68
Fig 4.4 (b): Absolute error between exact and numerical solution in example
4.2. ............................................................................................................... 69
Fig 4.5 (a):A comparison between The exact and approximate results of 𝐸𝑛.
..................................................................................................................... 70
Fig 4.5 (b): Absolute error between exact and numerical energy in example
4.2. ............................................................................................................... 70
 IX
Fig 4.6 (a): A comparison between the exact and approximate solution in
example 4.3. ................................................................................................ 74
Fig 4.6 (b): Absolute error between exact and numerical solution in example
4.3. ............................................................................................................... 75
Fig 3.7 (a):A comparison between The exact and approximate results of 𝐸𝑛.
..................................................................................................................... 76
Fig 4.7 (b): Absolute error between exact and numerical energy in example
4.3. ............................................................................................................... 76
 X

List of Tables
Table 𝟒. 𝟏: The exact and the numerical solutions using finite difference
method algorithm where 𝑵=9……………………………………….…….58
Table 𝟒. 𝟐: The exact and the numerical solutions for the energy in example
𝟒. 𝟏 where 𝑵=9. .......................................................................................... 60
Table (4.3): The exact and the numerical solutions using pseudo-spectral
method algorithm where 𝑵=9. .................................................................... 63
Table 𝟒. 𝟒: The exact and the numerical solutions using finite difference
method algorithm where 𝑵=3. ................................................................... 68
Table 𝟒. 𝟓: The exact and the numerical results for the energy in example
𝟒. 𝟐. .............................................................................................................. 69
Table 𝟒. 𝟔: The exact and the numerical solutions using finite difference
method algorithm where 𝑵=5. ................................................................... 74
Table 𝟒. 𝟕: numerical The exact and the results for the energy in example
𝟒. 𝟑. .............................................................................................................. 75
 XI
The Numerical Methods for Solving Schrödinger Equation
By
Hadeel Muaffaq Hejja
Supervisor
Prof. Dr. Naji Qatanani

Abstract

Schrödinger equation and its variants are one of the basic

equations studied in the field of partial differential equations, and have
various applications in geometry, spectral and scattering theory and

integrable systems.
In this thesis we review some basic details of quantum mechanics
such as Schrödinger equation of both types: these include time-
dependent and time-independent Schrödinger equation. Moreover, we
focus mainly on some analytical and numerical methods for solving
Schrödinger equation and its variants. For the analytical solution of the

Schrödinger equation, we use the separation of variables method and

method of characteristics. For the numerical handling of this equation,
we use the Finite Difference Method (FDM) and Pseudo-Spectral
Method (PSM), for two cases of quantum mechanics: finite and infinite
square well.
To test the efficiency of these methods, we consider some

numerical test cases. Numerical results show clearly that Pseudo-

Spectral Method is one of the most powerful numerical technique for
solving time independent Schrödinger equation in comparison with the
Finite Difference Method.
 1

Introduction

At the beginning of the twentieth century, experimental evidence

suggested that atomic particles were also wave-like in nature. For example,

electrons were found to give diffraction patterns when passed through a

double slit in a similar way to light waves. Therefore, it was reasonable to
assume that a wave equation could explain the behavior of atomic particles

[11,14].
Schrödinger equation is a type of differential equation known as a
wave-equation, which serves as a mathematical model of the moment of

waves. Solutions to Schrödinger's equation express not only molecular,

atomic and subatomic systems but also macroscopic systems, perhaps even
the whole universe. Schrödinger’s equation is central to all applications of

quantum mechanics including quantum field theory which incorporate

special relativity with quantum mechanics [14]. In mathematics, the
Schrödinger equation and its variants are one of the basic equations studied

in the field of partial differential equations, and have applications in

geometry, spectral and scattering theory, and integrable systems.
In1924, de-Broglie suggested that every moving particle has a wave

associated with it, which is also known as matter wave. Furthermore,

Erwin Schrödinger, in continuation to de-Broglie’s hypothesis,
constructed a differential wave equation of second order to rationalize the

wave nature of matter and particle that correlates to wave [11]. Thus the
equation is analogous to the equation for waves in optics, which assumes
 2

that the particle behaves as wave and yields solution in terms of a function
called the wave function. When this equation is solved, it generates two of

the following; namely the wave function 𝜑 and the energy 𝐸, of the particle
under consideration.
In 1926, Irwin Schrödinger inserted de Broglie's wave-like

representation of particles into the conservation of energy equation (total

energy = kinetic energy plus potential energy) and from this he derived an
equation to describe their behavior— which has become known as the
Schrödinger wave equation.
Once the wave function 𝜑 is known, then everything about the
particle is known or can be deduced from the wave function. Thus, the

wave function 𝜑 is the most important thing, which itself does not have
any physical significance, however the absolute square of 𝜑, i.e. |𝜑|2 gives
the probability of finding the particle in a particular region of space at a
particular instant of time [11,14].
A wave equation is an example of an 'equation of motion' which,
as the name suggests, can be used to predict the motion of an object. In

this case the object is a wave. In other words, if we know the amplitude
and velocity of the wave at a given time and place, we can project forward
(or backward) and predict the amplitude and velocity of the wave at some

other time and place. For example, if one dropps a pebble into a pond it
makes a wave of a given height (amplitude) which will decrease with time
as the wave spreads. Knowing the rate at which the wave spreads and loses
amplitude, we can predict what it will look like in ten seconds, twenty
 3

seconds and so on. Or conversely, we can look at the circular wave pattern
at a given time and run the whole thing in reverse to re-create the original

pulse created by the pebble.

Essentially, Schrödinger equation has two forms: one consisting of
time termed as time dependent equation and the other in which time factor

is eliminated and hence named as time independent equation [ 37]. The

solution of Schrödinger equation can be obtained analytically by using the
exact solvable models developed by Makowski [22]. On the other hand,
numerical methods play a very crucial rule in solving Schrödinger
equation. Sandvik [34] has obtained a numerical solution of the
Schrödinger equation by using Numerov's method. Marston [24] has

described the Fourier grid Hamiltonian method for bound state eigenvalues
and eigenfunctions. Monovasilis [28] studied the exponential-fitting
symplectic methods for the numerical integration of the Schrödinger
equation. Doescher and Rice, in their work [8] studied the infinite square
well potential with a moving wall. Jackiewicz technique [17] is used to
find the solutions by spectral collocation method and wave method form

relaxation methods. Strikwerda [36] has obtained a numerical solution of

the Schrödinger equation by using finite difference method. Aronstein and
Stroud [1] implemented the general series for finite square-well energy

levels for use in wave-packet studies. Robinson and Fairweather [33] have
obtained an orthogonal spline collocation method for Schrödinger -type
equation in one space variable. Gildener and Patrascioui [13] have shown
that the energy spectrum of a one-dimensional system by using the pseudo
 4

spectral contributions. The solution of the differential equations of

chemical physics can be obtained by using the spectral difference methods

by Mazziotti [26]. Orszag [29] made a comparison of pseudo-spectral and

spectral approximation. Also, England and Savari [9] implemented
pseudo-spectral method of solving linear ordinary differential equations.

Furthermore, Bulirsch, Miele and Stoer [4] have used the direct collocation
as numerical method to find the numerical solution of optimal control
problems.
For the numerical handling of the Schrödinger equation, we employ
the Finite Difference Method (FDM). In order to implement the FDM
method, Schrödinger equation is first transformed into a diffusion equation

by the imaginary time transformation. The resulting time-domain diffusion

equation is then solved numerically by the FDM. In this method, we
approximate derivatives using difference equation with errors of order
𝑂(ℎ𝑛 ) to solve differential equations numerically. This method was first
developed by A. Thom in the 1920s [6]. Finite difference procedures
approximate the derivative appearing in a partial differential equation by

sums and differences of function values at a set of discrete points. These

approximations are based on Talyor series expansions of a function of one
or more variables [5,6,20,36].

In addition, we will solve the Schrödinger equation by the Pseudo-

spectral method, which is a family of numerical methods for the solution
of differential equations based on the expansion of basis functions defined
on a set of grid points. A pseudo-spectral method is proposed for the
 5

numerical solution of linear Schrödinger equation. The employed method

is based on Chebyshev-Gauss-Lobbato quadrature points. Using the

pseudo-spectral differentiation matrices, the problem identified is reduced

to a system of nonlinear algebraic equations [13,26]. However, this method
has already been implemented by England and Savari, in their work [9]

studied the pseudo-spectral method of solving linear ordinary differential

equations.
This thesis is organized as follows: In chapter one, we review some
basic details of quantum mechanics such as Schrödinger equation of both
types: these include time-dependent and time-independent Schrödinger
equation. In chapter two, we use analytical methods to solve the time-

independent Schrödinger equation for three cases, namely, the finite

square well, the infinite square well and the harmonic oscillator. Two
numerical techniques for solving Schrödinger equation are presented in
chapter three, these techniques are finite difference method and pseudo-
spectral method for two cases of quantum mechanics: finite and infinite
square well. In chapter four, some numerical examples for solving

Schrödinger equation including the finite square well and infinite square
well are illustrated. Finally, Conclusion are drawn.
 6

Chapter One
Introduction to Schrödinger Equation
 7

Chapter One

Introduction to Schrödinger Equation

In this chapter, we investigate some important concepts related to
Schrödinger equation including the derivation of two types of Schrödinger
equation, namely: time-dependent and time-independent Schrödinger

equation, also we introduce some properties of the solutions of the

Schrödinger equation, probability, normalization, and expectation value.

1.1 Schrödinger Equation

For a general quantum system, the Schrödinger equation may be written

into the form:

∂ ̂ 𝜑(𝑥, 𝑡)
𝑖ħ ∂t φ(𝑥, 𝑡) = 𝐻 (1.1)

h
where ħ = is the plank constant, φ(𝑥, 𝑡) is the wave function or
2𝑚
∂
state function, 𝑖 is the imaginary unit, 𝑖ħ is the energy operator, and
∂t
ħ 2 d2
̂=−
𝐻 + V(𝑥)
2𝑚 dx2

is the Hamiltonian operator, where V(𝑥) represent the real function

that illustrates the potential energy (input) of the method.

The Schrödinger equation has two types, in the first one, the time
is obvious in the form with certain description and characterization
that will show the wave function and how it will change with time.

Therefore, the functional equation is known as time-dependent

 8

Schrödinger (TDSE) wave equation for one dimension which can be

written as:
∂ ħ2 ∂ 2
𝑖ħ φ(𝑥, 𝑡) = − φ(𝑥, 𝑡) + V(𝑥)φ(𝑥, 𝑡) (1.2)
∂t 2𝑚 ∂𝑥 2

or

∂ ħ2 ∂ 2
𝑖ħ φ(𝑥, 𝑡) = (− + V(𝑥)) φ(𝑥, 𝑡), (1.3)
∂t 2𝑚 ∂𝑥 2

where 𝑚 indicates the mass of the element.

This is typically a second-order linear differential equation, where
the visible expression on the left-hand side of the functional equation
(1.3) denotes the total energy of the visible particle. The first part on
the right-hand side sufficiently denotes the kinetic energy of the particle,
while the second part on the right-hand side denotes the potential energy

of the particle. There are three important properties of the solution for
time-dependent Schrödinger functional equation as the following:

1. The time dependent Schrödinger equation is carefully consistent with

energy reservation.

2. The time dependent Schrödinger equation has linear and singular value

solution, which suggest that solutions can be typically formed by

superposition of two or more independent solutions.
3. The free-particle solution 𝑉(𝑥) = 0 is harmonious, with a single wave

of de Broglie.
 9

The second type of the Schrödinger equation is the time-independent

Schrödinger equation (TISE), which is suitable for finding energy values

for a one-dimensional system, which can be written as:

ħ2 ∂ 2
𝐸φ(𝑥) = − φ(𝑥) + V(𝑥)φ(𝑥). (1.4)
2𝑚 ∂𝑥 2

We will derive the two types of the Schrödinger equation:

1.2 Derivation of the Time-Independent Schrödinger Equation

We will start with the one-dimensional standard wave equation [10,14]

𝜕2 𝑢 1 𝜕2 𝑢
= . (1.5)
𝜕𝑥 2 c2 𝜕𝑡 2

Using separation of variables,

𝑢(𝑥, 𝑡) = 𝜓(𝑥)𝑓(𝑡), (1.6)

we have,

𝜕2𝑢 𝜕2
= 𝜑(𝑥) 𝑓(𝑡)
𝜕𝑥 2 𝜕𝑥 2
𝜕2𝑢 𝜕2
= 𝜓(𝑥) 2 𝑓(𝑡)
𝜕𝑡 2 𝜕𝑡

then we obtain,
𝜕2 1 𝜕2
𝑓(𝑡) 2
𝜓(𝑥) = 2
𝜓(𝑥) 𝑓(𝑡), (1.7)
𝜕𝑥 c 𝜕𝑡 2

when we use a standard solution of the wave equation, 𝑓(𝑡) = 𝑒 𝑖𝜔𝑡 , we

obtain
𝜕2 −𝜔2
2
𝜓(𝑥) = 𝜓(𝑥). (1.8)
𝜕𝑥 c2
 10

We want to find the standard form of the Schrödinger equation by

using the total energy which contains kinetic and potential energy[17]
𝑝2
𝐸= + 𝑉(𝑥), (1.9)
2𝑚

finally, by using 𝜔 = 2𝜋𝑣, c = 𝑣𝜆, and ℎ = 𝑝𝜆 we have

𝜔2 4𝜋2 𝑣 2 4𝜋2 2𝑚[𝐸−𝑉(𝑥)]
= = = , (1.10)
c2 c2 𝜆2 ħ2

Combining equations (1.10) and (1.8) gives

𝜕2 2𝑚[𝐸−𝑉(𝑥)]
𝜓(𝑥) = − 𝜓(𝑥)
𝜕𝑥 2 ħ2

𝜕2 2𝑚[𝐸−𝑉(𝑥)]
𝜓(𝑥) + 𝜓(𝑥) = 0 (1.11)
𝜕𝑥 2 ħ2

Rearranging equation (1.11) it becomes

𝑑2 2𝑚 2𝑚
𝜓(𝑥) + 𝜓(𝑥)𝐸 − 𝑉(𝑥)𝜓(𝑥) = 0, (1.12)
𝑑𝑥 2 ħ2 ħ2

−ħ2
Multiplying both side of equation (1.12) by gives:
2𝑚

−ħ2 𝜕2
𝜓(𝑥) − 𝐸𝜓(𝑥) + 𝑉(𝑥)𝜓(𝑥) = 0
2𝑚 𝜕𝑥 2

or
−ħ2 𝜕2
𝜓(𝑥) + 𝑉(𝑥)𝜓(𝑥) = 𝐸𝜓(𝑥). (1.13)
2𝑚 𝜕𝑥 2

This is the time-independent Schrödinger Equation, which describes the

state function of a particle with mass 𝑚 and potential energy 𝑉(𝑥).

Functional equation (1.13) can be expanded sufficiently to the

specific case of three dimensions [10]. That is,
 11
−ħ2
∇2 𝜓(𝑥) + 𝑉(𝑥)𝜓(𝑥) = 𝐸𝜓(𝑥) (1.14)
2𝑚

𝜕2 𝜕2 𝜕2
where ∇= 2
+ 2
+ .
𝜕𝑥 𝜕𝑦 𝜕𝑧 2

Equation (1.14) may take the form [10]

̂ 𝜑(𝑥) = 𝐸𝜑(𝑥).
𝐻

1.3 Derivation of the Time-dependent Schrödinger Equation

There is more than one method to derive time-dependent Schrödinger

equation. We will derive it by using the time-independent Schrödinger
equation.

The Schrödinger's time-independent equation is:

∂2 2𝑚
𝜓+ (𝐸 − 𝑉)𝜓 = 0 (1.15)
∂𝑥 2 ħ2

We assume that a wave function represented as (see [9,36]):

𝜓 = 𝐴 ∙ 𝑒 −𝑖𝜔𝑡 (1.16)

where 𝐴 is the amplitude of the wave, 𝜔 is an angular frequency which

equal 2𝜋𝑣, and 𝑡 is the time period.

Differentiating equation (1.16) with respect to 𝑡,

𝜕𝜓
= −𝑖𝜔𝐴𝑒 −𝑖𝜔𝑡 = −𝑖(2𝜋𝑣)𝐴𝑒 −𝑖𝜔𝑡 (1.17)
𝜕𝑡

since 𝐸 = ℎ𝑣
𝜕𝜓 2𝜋𝐸 2𝜋𝐸
= −𝑖 ( ) 𝐴 ∙ 𝑒 −𝑖𝜔𝑡 = −𝑖( )𝜓
𝜕𝑡 ℎ ℎ
−ℎ 𝜕𝜓 𝑖ℎ 𝜕𝜓 𝜕𝜓
𝐸𝜓 = ∙ = = 𝑖ħ (1.18)
2𝜋𝑖 𝜕𝑡 2𝜋 𝜕𝑡 𝜕𝑡
 12

Substituting equation (1.18) into equation (1.15) gives,

∂2 2𝑚 𝜕𝜓 2𝑚𝑉
𝜓 + ∙ 𝑖ħ − 2 𝜓=0
∂𝑥 2 ħ2 𝜕𝑡 ħ
2𝑚𝑖 𝜕𝜓 2𝑚𝑉
∙ = 𝜓 − ∇2 𝜓 (1.19)
ħ 𝜕𝑡 ħ2

ħ2
Multiplying equation (1.19) by , we get
2𝑚

𝜕 ħ2 ∂2
𝑖ħ 𝜓(𝑥, 𝑡) = 𝑉𝜓(𝑥, 𝑡) − 𝜓(𝑥, 𝑡) (1.20)
𝜕𝑡 2𝑚 ∂𝑥 2

this is time-dependent Schrödinger equation with the term:

ħ2 ∂ 2
− +𝑉 =𝐻
2𝑚 ∂𝑥 2

as the Hamiltonian operator.

Equation (1.20) takes the form [10, 37]

𝜕
𝑖ħ ̂ 𝜑(𝑥, 𝑡)
𝜓(𝑥, 𝑡) = 𝐻
𝜕𝑡

1.4 Schrödinger's operators

An operator is a rule for construction one function from another. In
quantum mechanics, we have just seen that the momentum becomes
associated with an operator related to the spatial derivative.
̂
𝐷
For any function: 𝑓(𝑥) → another function of 𝑥

or
̂ 𝜑 = 𝜑́ ,
𝐷
 13
̂ is an operator. Every operator in quantum mechanics can be
where 𝐷
typically structured from the fundamental factors of both position and
momentum.

𝑥̂𝜑 = 𝑥𝜑

𝑑
𝑝̂𝑥 𝜑 = −𝑖ħ 𝜑.
𝑑𝑥

̂ , if
Definition (𝟏. 𝟏)[11,14]: For an operator 𝐷
̂ 𝑓(𝑥; 𝐷) = 𝐷𝑓(𝑥; 𝐷)
𝐷
for a given 𝐷 ∈ ℂ where 𝑐 denotes any complex number, then 𝑓(𝑥)
̂ and 𝐷 is the corresponding
denotes an eigenfunction of the operator 𝐷
eigenvalue.
Operators act on the eigenfunctions in a way identical to multiplying the
eigenfunction by a constant number as seen in definition (1.1).

To every observable quantity, there is associated corresponding

operator. For instance, the momentum operator is
𝑑
𝑝̂ = −𝑖ħ ,
𝑑𝑥
the position operator is

𝑥̂ = 𝑥
the energy operator is
𝑝̂ 2 𝜕
𝐸̂ = + 𝑉(𝑥̂) = 𝑖 ħ
2𝑚 𝜕𝑥
and so on.

Note that the operator order is important. For instance,

𝑑 𝑑𝑓(𝑥)
𝑝̂ 𝑥̂𝑓(𝑥) = 𝑝̂ (𝑥𝑓(𝑥)) = −𝑖ħ
𝑑𝑥
(𝑥𝑓(𝑥)) = −𝑖ħ (𝑓 (𝑥) + 𝑥 𝑑𝑥
), (1.21)

while,
 14
𝑑𝑓(𝑥) 𝑑𝑓(𝑥)
𝑥̂𝑝̂ 𝑓(𝑥) = 𝑥̂ (−𝑖ħ ) = −𝑖ħ𝑥 . (1.22)
𝑑𝑥 𝑑𝑥

To measure the importance of order, we define the commutator of two

operators 𝐴̂ and 𝐵̂ as [11,14]

[𝐴̂, ̂𝐵] ≡ 𝐴̂𝐵̂ − 𝐵̂𝐴̂.

Subtracting equation (1.21) from equation (1.22) we get

𝑥̂𝑝̂ 𝑓 (𝑥) − 𝑝̂ 𝑥̂𝑓(𝑥) = 𝑖ħ𝑓(𝑥) (1.23)

1.5 Linear Operators

Definition (𝟏. 𝟏)[11]: An operator 𝐴̂ is said to be linear if

𝐴̂(𝑐𝑓(𝑥)) = 𝑐𝐴̂𝑓(𝑥) (1.24)

and

𝐴̂(𝑓(𝑥) + 𝑔(𝑥)) = 𝐴̂𝑓(𝑥) + 𝐴̂𝑔(𝑥) (1.25)

where 𝑓(𝑥) and 𝑔(𝑥) are any two appropriate functions and c is a
complex constant.
̂ are all linear operators.
Examples: the operators 𝑥̂, 𝑝̂ , 𝑎𝑛𝑑 𝐻

1.6 Wave function

The wave function defined as a variable quantity that describes

the wave characteristics of a particle mathematically.

Note that wave function, denoted by the Greek letter, 𝜑, may be

thought of as an expression for the amplitude of the particle wave. The
wave function 𝜑 must be single-valued, continuous, and finite.
 15

Also, the probability of finding the particle described by a specific

wave function 𝜑 at a given point and time is proportional to the value of

𝜑2.
𝑝(𝑥) = |𝜑(𝑥)|2 determines the probability that an object in the state
𝜑(𝑥) will be found at position 𝑥. The total probability is the probability of

the particle that must be unity [11,14,37],

∞
∫−∞|φ(𝑥)|2 𝑑𝑥 = 1 (1.26)

and this is the normalization requirement which can be satisfied only

if the wave function, 𝜑(𝑥), does not diverge to infinite.

1.7 Equation of motion of the wave function

Moving to another operator called the Hamiltonian operator which
plays an essential part in quantum mechanics. The Hamiltonian operator
corresponds to the total energy observable for a free particle of mass 𝑚
moving in 1-dimension in a potential 𝑉(𝑥). The Hamiltonian operator may
take the form

ħ2 d2
̂ = 𝑇̂ + 𝑉̂ = −
𝐻 + 𝑉(𝑥). (1.27)
2𝑚 d𝑥 2

In three dimensions, equation (1.27) may take the form [37]:

ħ2 𝜕2 𝜕2 𝜕2 ħ2
̂=−
𝐻 ( + 𝜕𝑦2 + 𝜕𝑧 2) + 𝑉(𝑟⃗) = − 2𝑚 ∇2 + 𝑉(𝑟⃗) (1.28)
2𝑚 𝜕𝑥 2

For many-particle systems, we need to explain

the expression for the total energy which includes the kinetic energy of
the particles and the potential energy of the system. For two particles in 3-
dimensions, functional equation (1.28) may take the form [37]:
 16
ħ2 ħ2
̂=−
𝐻 ∇1 2 − ∇2 2 + 𝑉(𝑟⃗⃗⃗⃗,
1 𝑟⃗2 ) (1.29)
2𝑚 2𝑚

where
2 𝜕2 𝜕2 𝜕2
∇1 = + +
𝜕𝑥1 2 𝜕𝑦1 2 𝜕𝑧1 2
and
2 𝜕2 𝜕2 𝜕2
∇2 = + +
𝜕𝑥2 2 𝜕𝑦2 2 𝜕𝑧2 2
The Hamiltonian operator typically plays an essential role in
the time-dependent development of the wave function. time- dependent
wave function will have its equation of motion given by the time-
dependent Schrödinger equation (TDSE):

𝜕𝜑
𝑖ħ =𝐻 ̂ 𝜑.
𝜕𝑡
This obtain a basic equation that describes the movement of a quantum
mechanical system.

1.8 Properties of the solutions of the time-independent Schrödinger

equation

Considering a free particle where V(𝑥)= 0, the wave function

solution can be carefully inserted in the form of a plane wave

𝜑(𝑥, 𝑡) = 𝐴𝑒 𝑖𝑘𝑥−𝑖𝜔𝑡 (1.30)

which as a complex function can be expanded sufficiently in the form

𝜑(𝑥, 𝑡) = 𝐴 cos(𝑘𝑥 − 𝜔𝑡) + 𝑖𝐴 sin(𝑘𝑥 − 𝜔𝑡) (1.31)

where 𝐴 is the amplitude of the wave , 𝜔 is an angular frequency and 𝑡 is

the time period. The free particle wave function is associated with a known
momentum:
 17
ℎ ℎ𝐾
𝑝= = = ħK.
𝜆 2𝜋

The time-independent Schrödinger equation is beneficial for finding

energy values for a one dimensional system

ħ2 ∂ 2
𝐸φ(𝑥 ) = − φ(𝑥, 𝑡) + V(𝑥 )φ(𝑥, 𝑡). (1.32)
2m ∂𝑥 2

From equation (1.32) we obtain the normalized eigenfunctions:

2 𝑛𝜋
𝜑𝑛 (𝑥 ) = √ sin 𝑥 𝑛 = 1,2,3, …. (1.33)
𝐿 𝐿

which we will explore it in chapter two.

The solutions of the time-independent Schrödinger equation (TISE)

have the following three properties:
1. Continuity: The possible solutions to the time-independent
Schrödinger equation 𝜑(𝑥) and its first derivative 𝜑́ (𝑥) must be
naturally extended for all values of 𝑥 (the latter holds for finite potential
𝑉(𝑥)).

2. Normalizable: The possible solutions to the time-independent

Schrödinger equation must typically square integrable, i.e. the
functional integral of the modulus squared of the wave function over
all space must be a finite constant so that the wave function can be
∞
normalized to give ∫−∞|φ(𝑥)|2 𝑑𝑥 = 1.

3. Linearity: Owing to the linearity of the time-independent Schrödinger

equation, given two independent solutions 𝜑1 (𝑥) and 𝜑2 (𝑥), we can

construct other solutions by taking an appropriate superposition of
these: 𝜑(𝑥) =𝛼1 𝜑1 (𝑥) + 𝛼2 𝜑2 (𝑥), where |𝛼1 |2 + |𝛼2 |2 = 1 to ensure
normalization.
 18

1.9 Basis of quantum mechanics

Corollary (𝟏. 𝟏) [10,36]: The function 𝜑(𝑥, 𝑡) denotes the state of a

quantum mechanical system, which depends on space and time

coordinates of the particle. This function, called the wave function or state
function.

Corollary (𝟏. 𝟐) [11,37]: To every observable, 𝐴, in classical mechanics

(e.g energy, position and momentum), there corresponds a linear
Hermitian operator, 𝐴̂ in quantum mechanics.
Corollary (𝟏. 𝟑)[𝟏𝟏, 𝟑𝟕]: In any measurement of the observable
associated with operator 𝐴̂, the only values that will ever be available are
the eigenvalues (a), which satisfy the eigenvalue equation:
𝐴̂ 𝜑𝑎 =𝑎 𝜑𝑎 , (1.34)
where 𝜑𝑎 is the eigenfunction corresponding, respectively, with the
eigenvalue 𝑎 of the operator 𝐴̂. An arbitrary state can be expanded in the
complete set of eigenvectors of 𝐴̂ (𝐴̂𝜑𝑎 = 𝑎𝜑𝑎 ) as [11,14,37]:
𝜑=∑𝑛𝑎 𝑐𝑎 𝜑𝑎 (1.35)
in this specific situation , we only know that the specific
measurement of the observable 𝐴 will be typically yield one of the values
(𝑎) with a probability |𝑐𝑎 |2 .
Corollary (𝟏. 𝟒)[𝟏𝟏, 𝟏𝟒, 𝟑𝟕]: If a system is in a state which is typically

related by a normalized wave function 𝜑, then the average value of

the observable corresponding to 𝐴̂ is given by:
∞
<𝐴̂> =∫−∞ 𝜑 ∗ (𝑥) 𝐴̂𝜑(𝑥)𝑑𝑥 (1.36)

and
∆𝐴 ≡ √〈𝐴2 〉 − 〈𝐴〉2
 19

1.10 Stationary States

By separation the space and time dependence of the Schrödinger

equation as
φ(𝑥,t)= φ(𝑥)g(t)
then the concept of a stationary state arises naturally. It then turns

out that the separation of the partial differential Schrödinger equation is

possible only if
̂ is time independent,
1- 𝐻
̂,
2- 𝜑(𝑥) is an eigenfunction of 𝐻
that is, 𝜑 must be a solution of the time-independent Schrödinger equation,
̂ 𝜑(𝑥) = 𝐸𝜑(𝑥).
𝐻

It is then easy to find 𝑔(𝑡). We consider the time-dependent

Schrödinger equation [36]:
2 2
∂ ħ ∂ ̂ φ(𝑥, 𝑡)
𝑖ħ ∂t φ(𝑥, 𝑡) = (− 2𝑚 2 + V(𝑥)) φ(𝑥, 𝑡) = 𝐻 (1.37)
∂𝑥
We assume that the potential energy in the Hamiltonian operator in
equation (1.37) to be time-independent, i.e ( 𝑉 = 𝑉(𝑥)).

To solve equation (1.37) we use the separation of variables method.

Let
𝜑(𝑥, t) = g(t)𝜑(𝑥)

Inserting the above equation into equation (1.37) yields:

2 2
𝜕𝑔(𝑡) ħ ∂ 𝜑(𝑥)
𝑖ħ𝜑(𝑥) 𝜕𝑡 = − 2𝑚 2 𝑔(𝑡) + 𝑉(𝑥)𝜑(𝑥)𝑔(𝑡) (1.38)
∂𝑥
 20

Dividing both sides of equation (1.38) 𝑏𝑦 𝜑(𝑥)𝑔(𝑡) gives:

2 2
1 𝜕𝑔(𝑡) ħ 1 ∂ 𝜑(𝑥)
𝑖ħ = − 2𝑚 + 𝑉(𝑥). (1.39)
𝑔(𝑡) 𝜕𝑡 𝜑(𝑥) ∂𝑥2

Both sides of equation (1.39) must be equal to a constant, since the left-
hand side is only dependent on 𝑡 and the right-hand side only on 𝑥, which
we can denote 𝐸. Thus we have two separated sides as:
1 𝜕𝑔(𝑡) 𝑑𝑔 𝑖
𝑖ħ =𝐸 ⇒ = − 𝐸𝑑𝑡
𝑔(𝑡) 𝜕𝑡 𝑔 ħ
𝑖
ln( 𝑔) = − 𝐸𝑡 + 𝑐𝑜𝑛𝑠𝑡. ⇒ 𝑔(𝑡) = 𝑐𝑜𝑛𝑠𝑡. 𝑒 −𝑖𝐸𝑡⁄ħ (1.40)
ħ

So, we have

𝜑(𝑥, 𝑡) = 𝜑(𝑥) 𝑒 −𝑖𝐸𝑡⁄ħ

This is called a stationary solution of the Schrödinger equation, because:

 The probability density is time-independent

|𝜑(𝑥, 𝑡)|2 = 𝜑 ∗ (𝑥)𝜑(𝑥)𝑒 𝑖𝐸𝑡⁄ħ 𝑒 −𝑖𝐸𝑡⁄ħ = |𝜑(𝑥)|2 (1.41)

The spatial part of the wave function satisfies the time-independent

Schrödinger equation.

 all operators which do not depend explicitly on time, like

𝑥, 𝑝𝑥 , 𝐸 𝑒𝑡𝑐, have time-independent expectation values in the
stationary state 𝜑(𝑥, 𝑡). Thus, if 𝐴 = 𝐴(𝑥, 𝑝𝑥 ), then

𝑑
〈𝐴(𝑥, 𝑝𝑥 )〉 = ∫ 𝜑 ∗ (𝑥)𝑒 𝑖𝐸𝑡⁄ħ 𝐴(𝑥, −𝑖ħ )𝜑(𝑥)𝑒 −𝑖𝐸𝑡⁄ħ 𝑑𝑥
𝑑𝑥
𝑑
= ∫ 𝜑 ∗ (𝑥) 𝐴(𝑥, −𝑖ħ )𝜑(𝑥)𝑑𝑥 = constant. (1.42)
𝑑𝑥
 21

Theorem (𝟏. 𝟏)[𝟑𝟕]: (Time-independent Schrödinger equation)

̂ φ(𝑥) = Eφ(𝑥)
H
ħ2
̂= −
where 𝐻 ∇ + 𝑉(𝑥) is the Hamiltonian.
2𝑚

Definition (𝟏. 𝟐)[ 𝟏𝟏, 𝟏𝟒, 𝟑𝟕]: A state is called stable if it is performed
by the wave function:
𝜑(𝑥, t) = 𝜑(𝑥) 𝑒 −𝑖𝐸𝑡⁄ħ
Corollary (𝟏. 𝟓)[ 𝟏𝟏, 𝟏𝟒]: As a direct result, the eigenvalues of the
Hamiltonian, which represent the potential energy levels of
the system, are clearly time-independent.
̂ (𝑥, 𝑝𝑥 ) instead of 𝐴(𝑥, 𝑝𝑥 ) in equation (1.42)
To see this, just take 𝐻

and use the time-independent Schrödinger equation (Theorem 1.1)

̂ (𝑥, 𝑝𝑥 )〉 = ∫ 𝜑 ∗ (𝑥) 𝐻 𝜑(𝑥) 𝑑𝑥 = ∫ 𝜑 ∗ (𝑥) 𝐸 𝜑(𝑥)𝑑𝑥 =
〈𝐻

𝐸(∫ 𝜑(𝑥)𝜑 ∗ (𝑥)𝑑𝑥). (1.43)

The value of this integration is finite.

 22

Chapter Two
Analytical Methods
 23

Chapter Two

Analytical Methods

2.1 Introduction

In this chapter we attempt to solve the time-independent

Schrödinger equation analytically. This include the following three cases:

the infinite potential well, the finite potential well and the quantum
harmonic oscillator.

2.2 Infinite square well

Regarding the infinite square well, the particle exists only in the
finite interval [0, L]. Such that [8],
0, 𝑥 ∈ [0, 𝐿]
𝑉(𝑥) = { (2.1)
∞, 𝑒𝑙𝑠𝑒

which sufficiently indicate that the quantum object is carefully restricted

to a specific area between 𝑥 = 0 and 𝑥 = 𝐿 it moves freely but cannot
escape. Therefore, mathematically we have.

outside region: 𝜑(𝑥) = 0 𝑓𝑜𝑟 𝑥 ∉ [0, L] (2.2)

−ħ2 𝜕2
inside region: ( 2𝑚 ) 𝜕𝑥 2 = 𝐸𝜑, this is time-independent Schrödinger

equation.
and,
𝜓(0) = 0 and 𝜓(𝐿) = 0 (2.3)
 24

this because 𝜑 must be continuous.

Taking into Consideration any sample point 𝜓(𝑥𝑘 ), 𝑥𝑘 ∉ [0, L]

, there is no 𝜓(𝑥𝑘 ) will visible in the system of linear equations, since

each 𝜓(𝑥𝑘 ) = 0.
The wave functions, 𝜓, are eigenvectors of the Hamiltonian operator,

and satisfy equation:

̂ 𝜓 = 𝐸𝜓
𝐻 (2.4)

we recall that the Hamiltonian is simply the sum of the kinetic and
potential energies, so equation (2.4) becomes

̂ 𝜓 = 𝐸𝜓
H

̂ + 𝑉̂ ]𝜓 = 𝐸𝜓
[𝐾 (2.5)

we know that the kinetic energy of the particle is

√2𝑚𝐸 2𝑚𝐸
𝐾= , 𝐾2 = (2.6)
ħ ħ2

Substituting equation (2.6) into equation (2.5), we get [24]

−ħ2 𝜕 2 𝐾 2 ħ2
( ) 𝜓(𝑥) = 𝜓(𝑥)
2𝑚 𝜕𝑥 2 2𝑚

𝜕2
2
𝜓(𝑥) = −𝐾 2 𝜓(𝑥)
𝜕𝑥
𝜕2
𝜓(𝑥 ) + 𝐾 2 𝜓(𝑥 ) = 0 (2.7)
𝜕𝑥 2

Equation (2.7) has the general solution:

𝜓(𝑥) = 𝑎 sin(𝐾𝑥) + 𝑏 cos(𝐾𝑥) (2.8)

 25

𝑎 and 𝑏 are two constants that are being specified by the boundary
conditions
𝜓(0) = 0 and 𝜓(𝐿) = 0 , (2.9)

beginning with 𝜓(0) = 0,

𝜓(0) = 𝑎 sin(0) + 𝑏 cos(0) = 0 ⇒ 𝑏=0 (2.10)

from the second boundary condition 𝜓(𝐿) = 0 then implies

𝜓(𝐿) = 𝑎 sin(𝐾𝐿) = 0 (2.11)

it is assumed that 𝑎 ≠ 0, for otherwise 𝜓(𝑥) would be zero everywhere

and the particle would disappear. The condition that sin 𝐾𝐿 = 0 implies
that
𝑛𝜋
𝐾𝐿 = 𝑛𝜋 ⇒ 𝐾= , (2.12)
𝐿

where 𝑛 = 1,2,3, … can be any natural number.

Substituting equation (2.12) into equation (2.6) to obtain

𝑛 2 𝜋 2 ħ2
𝐸𝑛 = , (2.13)
2𝑚𝐿2

from equation (2.13) we can see that:

1. the energy is finite [7],

𝜋 2 ħ2 4𝜋2 ħ2 9𝜋2 ħ2
2. 𝐸1 = , 𝐸2 = = 4𝐸1 , 𝐸3 = = 9𝐸1 , . . . ..
2𝑚𝐿2 2𝑚𝐿2 2𝑚𝐿2

3. 𝐸𝑚𝑖𝑛 = 𝐸1 ≠ 0 (𝑧𝑒𝑟𝑜 𝑝𝑜𝑖𝑛𝑡 𝑒𝑛𝑒𝑟𝑔𝑦).

4. As 𝐿 → ∞, 𝐸𝑚𝑖𝑛 = 0 .
 26

5. As 𝑚 → ∞, 𝐸𝑚𝑖𝑛 = 0 .
𝑛𝜋 2𝜋 𝑛𝜆
6. We know that 𝐾 = = , so 𝐿 = .
𝐿 𝜆 2

𝑛𝜋
The eigenfunctions in equation (2.8) with 𝑏 = 0 and 𝐾 = , is
𝐿

given by:
𝑛𝜋𝑥
𝜓𝑛 (𝑥) = 𝑎 sin , 𝑛 = 1,2,3, … (2.14)
𝐿

Finally, we get the value of the constant (𝑎) from the normalization of the
wave function.

𝐿
∫0 |𝜓𝑛 |2 𝑑𝑥 = 1 (2.15)

the integration running over the domain of the particle, 0 ≤ 𝑥 ≤ 𝐿.

Substituting equation (2.14) into equation (2.15)[9,12,30],
𝐿 𝑛𝜋 𝐿 𝑛𝜋 𝐿
|𝑎|2 ∫0 sin2 ( 𝑥) 𝑑𝑥 = |𝑎|2 ∫ sin2 𝜃 𝑑𝜃 = |𝑎|2 2 = 1. (2.16)
𝐿 𝑛𝜋 0

𝑛𝜋𝑥
We have made the substitution 𝜃 = and used the fact that the average
𝐿
1
value of sin2 𝜃 over an integral of half wavelengths equals . From
2
2
equation (2.16) we can identify the normalization constant 𝑎 = ( )1/2 for
𝐿

all values of 𝑛. Thus, we obtain the normalized eigenfunctions:

2 𝑛𝜋
𝜓𝑛 (𝑥) = √ cos( 𝑥) , 𝑛 = 1,3,5, … (2.17)
𝐿 𝐿

2 𝑛𝜋
𝜓𝑛 (𝑥) = √ sin ( 𝑥), 𝑛 = 2,4,6 … (2.18)
𝐿 𝐿

For 𝑛 = 1 we get the bound state of the energy 𝐸1 and wave

function 𝜓1 , for 𝑛 > 1 are named excited states.
 27

𝜓4 (𝑥) 𝐸4 = 16𝐸1
𝑛=4

𝜓3 (𝑥) 𝐸3 = 9𝐸1
𝑛=3

𝐸2 = 4𝐸1
𝜓2 (𝑥) n=2

𝐸1
𝜓1 (𝑥) n=1
0 𝑥 L

Figure 1.1: Eigenfunctions and the energy for a particle in a box

2.3 Finite square well

In this section we want to solve the finite square well. Consider the
potential shown in

𝑉(𝑥)

V = V0
E < V0 E > V0 E < V0

Region 1 Region 2 Region 3

V=0
−L L
0 𝑥

Figure 1.2: A finite square well, depth V0 , width 2L.

 28
Since we have a symmetric well, we must focus on the positive half
of the 𝑥 -axis since we know 𝜑(−𝑥) = 𝜑(𝑥) for the even parity states
such as (ground, second excited, and forth exited state) and 𝜑(−𝑥) =
−𝜑(𝑥) for odd parity states (such as for the first excited, third excited, and
fifth excited state) [1].

2.3.1 Bound states

Region 1:
𝑥 ≤ −𝐿, 𝑉(𝑥) = 𝑉0 ,
similarly, to region 3, the solutions are:
𝜓(𝑥) = 𝐹𝑒 𝛼𝑥 + 𝑄𝑒 −𝛼𝑥 ,
but since 𝜓 → 0 as 𝑥 → −∞, 𝑄 = 0
we get 𝜓 = 𝐹𝑒 𝛼𝑥 for region 1.

Region 2:
−𝐿 < 𝑥 < 𝐿, 𝑉(𝑥) = 0.
substituting into these equation:
−ħ2 𝜕2 𝜕2 2𝑚
( 2𝑚 ) 𝜕𝑥 2 𝜓(𝑥) = 𝐸 𝜓(𝑥)  𝜓(𝑥) = − 𝐸 𝜓(𝑥)
𝜕𝑥 2 ħ2

meaning
𝜕2 2𝑚𝐸
𝜓(𝑥) = −𝐾 2 𝜓(𝑥) with 𝐾 2 = .
𝜕𝑥 2 ħ2

The solutions to this equation is [1]

𝜓(𝑥) = 𝐴 sin(𝐾𝑥) + 𝐵 cos(𝐾𝑥)

Region 3:

𝑥 ≥ L, 𝑉(𝑥) = 𝑉0 , the Schrödinger equation becomes:

−ħ2 𝜕 2
( ) 𝜓 + 𝑉0 𝜓(𝑥) = 𝐸 𝜓(𝑥)
2𝑚 𝜕𝑥 2
 29
𝜕2 2𝑚
𝜓= (𝑉0 − 𝐸)𝜓 (2.19)
𝜕𝑥 2 ħ2

yielding
∂2 2𝑚
𝜓 = 𝛼 2 𝜓, with 𝛼 2 = (𝑉0 − 𝐸) > 0
∂x2 ħ2

So,

2𝑚
𝛼=√ (𝑉0 − 𝐸), which is real.
ħ2

The solutions to this differential equation are:

𝜓(𝑥) = 𝐶𝑒 −𝛼𝑥 + 𝐷𝑒 𝛼𝑥 ,

but since 𝜓 → 0 as 𝑥 → ∞, 𝐷 = 0

we get 𝜓 = 𝐶𝑒 −𝛼𝑥 for region 3.

We want to determine the value of the energy 𝐸 and the other
coefficients (C, D, F, Q) The symmetry of the well allow us to choose
even or odd functions, which means cosine or sine solution in the
central region depending on which n state we want.
The parity condition 𝜑(𝑥) =±𝜑(−𝑥) allows us to join the
coefficients in the area 𝑥 ≤ −𝐿 to the coefficients in the area 𝑥 ≥ 𝐿 [1].
Here is a graphic summery of which solutions apply in which regions.

𝜑(𝑥) = cos( 𝐾𝑥)

𝑜𝑟 sin( 𝐾𝑥) 𝐸

φ(𝒙) = 𝐹𝑒𝛼𝑥 φ(𝒙) = 𝐶𝑒−𝛼𝑥

+𝑄𝑒 −𝛼𝑥 V=0

+𝐷𝑒 𝛼𝑥

−𝐿 0 𝐿
 30
This leaves us with three unknowns: (𝐸, 𝐶, 𝐷)and right now we
have three boundary conditions to consider: the continuity of the
wave function and the continuity of its derivative. The last boundary
condition is that the wave function must have a sensible behavior at
infinity [1,23]. That is, 𝜑(𝑥) is finite at infinity, we require 𝜑(𝑥) to be
finite as 𝑥 → ±∞.

The finite behavior at infinity is a powerful condition since it

tells us that 𝐷 = 0 and we must have φ(𝑥) = 𝐶𝑒 −𝛼𝑥 in the region 𝑥 ≥ 𝐿.
For the even parity states we select a cosine function for the well
center point. We start with the wave function continuity condition
which requires
𝜑(𝐿)𝑙𝑒𝑓𝑡 = 𝜑(𝐿)𝑟𝑖𝑔ℎ𝑡

cos(𝐾𝐿) = 𝐶𝑒 −𝛼𝐿 (2.20)

To apply the continuity of the derivative condition, we begin by taking

the derivative of the wave functions [33]:

∂ cos(K𝑥)
= −K sin(K𝑥)𝑥=L = −K sin(KL)
∂𝑥 𝑥=L

∂Ce−α𝑥
= −αCe−α𝑥 𝑥=L = −αCe−αL ,
∂𝑥 𝑥=L

∂φ(L) ∂φ(L)
hence the condition = gives :
∂𝑥 left ∂𝑥 right

−𝐾 sin(𝐾𝐿) = − 𝛼𝐶𝑒 −𝛼𝐿 (2.21)

From equations (2.20) and (2.21) we obtain:

𝐾 tan(𝐾𝐿) = 𝛼,
 31

inserting expressions for 𝐾 and 𝛼 as a function of 𝐸 we have,

2𝑚𝐸 2𝑚𝐸 2𝑚(𝑉−𝐸)
√ tan(𝐿 √ )=√ ,
ħ2 ħ2 ħ2
2𝑚
taking out the common factor of √ leaves us with the transcendental
ħ2

equation:
2𝑚𝐸
√𝐸 tan(𝐿√ ħ2
) = √𝑉 − 𝐸 (2.22)

We want to solve equation (2.22) by graphing the left and right

sides of the equation on a semilog graph. Logical solutions happened at

the intersection of the two plots. We take 𝑉= 50 𝑒𝑉.

Figure 1.3: A semilog graph for the energy of finite square well.

There are precisely two intersections, as a result, there

are two even bound states in the well n = 1 and n = 3. One of the even
bound states happens at 𝐸 = 5.72 𝑒𝑉, the other happens near the peak of
 32

the well at a power of 𝐸 = 45.4 𝑒𝑉. There is an odd bound state corres-
ponding respectively to 𝑛 = 2 at roughly 20 𝑒𝑉.

2.4 The quantum harmonic oscillator

The quantum harmonic oscillator is very important in quantum

mechanics, is a very useful solution in both approximations and in exact

solutions of various problems.

The harmonic oscillator is described by the Hamiltonian:
̂2
p 1 2 ̂2
p 1 2
̂=
𝐻 + k𝑥
̂ = + 𝑚ω2 𝑥
̂ , (2.23)
2𝑚 2 2𝑚 2
𝑘
where 𝑚 is the particle's mass, k is the force constant, ω = √ is
𝑚

the angular frequency of the oscillator, 𝑥

̂ is the position operator (given
𝜕
by 𝑥), and p̂ is the momentum operator (given by p̂ = −𝑖 ħ ). The
𝜕𝑥

initial concept in the Hamiltonian intentionally introduces the kinetic

energy of the fundamental particle [7], and the second key concept
introduces its potential energy.
First, we have to sufficiently identify what is meant by the energy
eigenstate of the possible solution of the Schrödinger equation.
ħ2
The energy quantity = ħ𝜔 = = 𝑚𝜔2 𝑎2 ,
𝑚 𝑎2

4 ħ2 ħ
thus 𝑎 = → 𝑎2 = , where 𝑎 is the length.
𝑚2 𝜔2 𝑚 𝜔

The Schrödinger equation of the harmonic oscillator is:

ħ2 𝑑 2 𝜑 1
− + 𝑚𝜔2 𝑥 2 𝜑 = 𝐸𝜑 (2.24)
2𝑚 𝑑𝑥 2 2
 33

with boundary condition:

𝜑(𝑥) → 0 as |𝑥| → ∞ . (2.25)

2
Multiplying equation (2.24) by , we get:
ħ𝜔

ħ 𝑑2 𝜑 𝑚𝜔 2𝐸
− + 𝑥2𝜑 = 𝜑, (2.26)
𝑚𝜔 𝑑𝑥 2 ħ ħ𝜔
2𝐸
let 𝜀 = be the dimensionless value of the energy, so equation (2.26)
ħ𝜔

takes the form:

ħ 𝑑2 𝜑 𝑚𝜔
− + 𝑥 2 𝜑 = εφ, (2.27)
𝑚𝜔 𝑑𝑥 2 ħ

let, 𝑥 = 𝑎𝑢 , (2.28)

where 𝑎 is the length and 𝑢 is the new variable of differential

equation.

Differentiating both sides of equation (2.28) with respect to 𝑥,

𝑑 1 𝑑 𝑑2 1 𝑑2
= → =
𝑑𝑥 𝑎 𝑑𝑢 𝑑𝑥 2 𝑎2 𝑑𝑢2

so, equation (2.27) may take the form :

𝑑2 𝜑
− + 𝑢2 𝜑 = 𝜀𝜑 (2.29)
𝑑𝑢2

when 𝜀 → ∞ , the equation becomes 𝜑́́ = 𝑢2 𝜑,

𝛼𝑢2⁄
let 𝜑 = 𝑢𝜇 𝑒 2, where 𝛼 is any number.

𝛼𝑢2⁄ 𝛼𝑢2⁄
𝜑́ = 𝛼𝑢𝑢𝜇 𝑒 2 + 𝜇𝑢𝜇−1 𝑒 2,
 34
𝛼𝑢 2
𝜑́́ = (𝛼𝑢)2 𝑢𝜇 𝑒 ⁄2
+…………..
𝛼𝑢 2
2𝜇+1 1 𝜇(𝜇−1) 1
𝜑́́ = 𝛼 2 𝑢2 𝑢𝜇 𝑒 ⁄2
[1 + + ]
𝛼 𝑢2 𝛼2 𝑢4

2𝜇+1 1 𝜇(𝜇−1) 1
= 𝛼 2 𝑢2 𝜑 [1 + + ].
𝛼 𝑢2 𝛼2 𝑢4

When 𝑢 → ∞, 𝛼 2 = 1 since 𝜑́́ = 𝑢2 𝜑.

−𝑢2⁄ 𝑢2⁄
𝜑(𝑢) = 𝐴𝑢𝜇 𝑒 2 + 𝐵𝑢𝜇 𝑒 2, (2.30)

In equation (2.30) , 𝐵 = 0 for |𝑢| → ∞.

𝑢2⁄
Let, 𝜑(𝑢) = ℎ(𝑢)𝑒 − 2 , (2.31)

where ℎ(𝑢) an arbitrary function.

Substituting equation (2.31) into equation (2.29) to obtain:

𝑑2ℎ 𝑑ℎ
− 2𝑢 + (𝜀 − 1)ℎ = 0, (2.32)
𝑑𝑢2 𝑑𝑢

solving equation (2.32) by a power series expansions, so,

ℎ(𝑢) = ∑∞ 𝑗
𝑗=0 𝑎𝑗 𝑢 , (2.33)

𝑑ℎ
= ∑∞
𝑗=0 𝑗 𝑎𝑗 𝑢
𝑗−1
, (2.34)
𝑑𝑢

𝑑2ℎ
= ∑∞
𝑗=0 𝑗(𝑗 − 1)𝑎𝑗 𝑢
𝑗−2
= ∑∞
𝑗=2 𝑗(𝑗 − 1)𝑎𝑗 𝑢
𝑗−2
, (2.35)
𝑑𝑢2

Suppose that, 𝑗 = 𝑗́ + 2, equation (2.35) become:

𝑑2ℎ
= ∑∞ 𝑗́ ∞ 𝑗
𝑗́ =0(𝑗́ + 2)(𝑗́ + 1)𝑎𝑗́ +2 𝑢 = ∑𝑗=0(𝑗 + 2)(𝑗 + 1)𝑎𝑗+2 𝑢 , (2.36)
𝑑𝑢2

Substituting equations (2.33), (2.34), (2.35) and (2.36) into equation

(2.32), we get:
 35

∑((𝑗 + 2)(𝑗 + 1) 𝑎𝑗+2 − 2𝑗𝑎𝑗 + (𝜀 − 1)𝑎𝑗 )𝑢 𝑗 = 0

𝑗=0

∑∞ 𝑗
𝑗=0((𝑗 + 2)(𝑗 + 1) 𝑎𝑗+2 − (2𝑗 + 1 − 𝜀)𝑎𝑗 )𝑢 = 0, (2.37)

from equation (2.37), we get this relation:

(2𝑗+1−𝜀)
𝑎𝑗+2 = 𝑎. (2.38)
(𝑗+2)(𝑗+1) 𝑗

Possible solutions fixed by given 𝑎0 , 𝑎1 (ℎ(0), ℎ́(0)), where from

𝑎0 we can properly fixed 𝑎2 , 𝑎4 , …. even solutions, and from 𝑎1 we can

fixed 𝑎3 , 𝑎5 , …. odd solutions.
Now, to terminate the series we can choose

2𝑗 + 1 − 𝜀 = 0, (2.39)

this will make 𝑎𝑗+2 = 0.

So the solution ℎ(𝑢) = 𝑎𝑗 𝑢 𝑗 + 𝑎𝑗−2 𝑢 𝑗−2 + ⋯

call 𝑗 = 𝑛, we get ℎ(𝑢) = 𝑎𝑛 𝑢𝑛 + 𝑎𝑛−2 𝑢𝑛−2 + ⋯

equation (2.39) become:

2𝑛 + 1 − 𝜀𝑛 = 0, (2.40)

from equation (2.40), we get

𝐸𝑛
𝜀𝑛 = 2𝑛 + 1 = ħ𝜔 ,
( )
2

where 𝐸𝑛 is the energy of the harmonic oscillator

ħ𝜔 1
𝐸𝑛 = (2𝑛 + 1) = ħ𝜔 (n + ) (2.41)
2 2
 36

𝐻𝑛 (𝑢) = 2𝑛 𝑢𝑛 + 𝑂(𝑢𝑛−2 ) (2.42)

Here, 𝐻𝑛 (𝑢) is a polynomial of degree n called a Hermite polynomial.
The first four Hermite polynomials are
𝐻0 (𝑢)=1
𝐻1 (𝑢) = 2𝑢
𝐻2 (𝑢) = 4𝑢2 − 2
𝐻3 (𝑢) = 8𝑢3 − 12𝑢
A few sample wave functions are given in Figure 1.3. As the value of
the principal number increases, the solutions alternate between even
functions and odd functions about x=0.

Figure 1.4: The first five wavefunctions of the quantum harmonic oscillator.

The solution of the quantum harmonic oscillator is:

−𝑢2
𝜑𝑛 = 𝐻𝑛 (𝑢)𝑒 ⁄2
𝑥 −𝑥 2⁄ 2
𝜑𝑛 (𝑥) = 𝐻𝑛 ( ) 𝑒 2𝑎
𝑎
 37

Chapter Three
Numerical Techniques for Solving
Schrödinger Equation
 38

Chapter Three

Numerical Techniques for Solving Schrödinger

Equation
In this chapter, we attempt to solve the Schrödinger equation and its
variants numerically. This involves using the finite difference and the

pseudo-spectral methods.

3.1 Finite Difference Method (FDM)

This method replaces the partial derivatives of the dependent variable

(unknown function) with a partial differential equation using finite
difference approximations with errors.

This procedure transforms the region (where the independent variables in

PDE are defined on) to a mesh grid of points where the dependent
variables are approximated [5,20,36]. The possible replacement of partial
derivatives with various approximation formulas depends on Taylor's
Theorem. Hence, Taylor's Theorem is presented.

3.2 Taylor's Theorem [5,36]

Let 𝜑(𝑥) has 𝑛 ∈ 𝑁 continuous derivatives over the interval (𝑎, 𝑏). Then,
for 𝑎 < 𝑥0 , 𝑥0 + ℎ < 𝑏, we can write the value of 𝜑(𝑥) and its derivatives
nearby the point 𝑥0 + ℎ as follows:
𝜑́ (𝑥0 ) 𝜑́́ (𝑥0 )
𝜑(𝑥0 + ∆𝑥) = 𝜑(𝑥0 ) + (𝑥 − 𝑥0 ) + (𝑥 − 𝑥0 )2
1! 2!
́
𝜑́ (𝑥0 )
+ (𝑥 − 𝑥0 )3 +..
3!
 39
ℎ2
𝜑(𝑥0 + ℎ) = 𝜑(𝑥0 ) + ℎ𝜑́ (𝑥0 ) + 𝜑́́ (𝑥0 )+
2!
ℎ3 ℎ𝑛−1
𝜑́́ (𝑥0 )+. . + (𝑛−1)! 𝜑 (𝑛−1) (𝑥0 ) + 𝑂(ℎ𝑛 ) (3.1)
3!

which can be written in the more compact notation as

∞
𝜑 (𝑛) (𝑥0 )
∑ (𝑥 − 𝑥0 )𝑛 + 𝑂(ℎ𝑛 )
𝑛!
𝑛=0

where

1. 𝜑́ (𝑥0 ) is the first derivative of 𝜑 with recognition to 𝑥 at the

specific point 𝑥0 .
2. 𝜑 (𝑛−1) (𝑥0 ) is the 𝑛 − 1𝑡ℎ derivative of 𝜑 with respect to 𝑥 at the

point 𝑥0 .
3. 𝑂(ℎ𝑛 ) denotes an unknown error term that satisfies the property: for

𝜑(ℎ) = 𝑂(ℎ𝑛 )
𝜑(ℎ)
lim = 𝑐, for any nonzero constant c.
ℎ→0 ℎ𝑛

When we eliminate the error term, 𝑂(ℎ𝑛 ), from the right-hand side
of functional equation (3.1), we get an approximation to 𝜑(𝑥0 + ℎ).
the forward–difference formula for approximating 𝜑́ (𝑥0 )

𝜑(𝑥0 +∆𝑥)−𝜑(𝑥0 )
𝜑́ (𝑥0 ) ≅ + 𝑂(∆𝑥),
∆𝑥

The backward−difference formula

𝜑(𝑥0 ) − 𝜑(𝑥0 − ∆𝑥)

𝜑́ (𝑥0 ) = + 𝑂(∆𝑥),
∆𝑥
 40

and the central–difference formula

𝜑(𝑥0 + ∆𝑥) − 𝜑(𝑥0 − ∆𝑥)

𝜑́ (𝑥0 ) = + 𝑂(∆𝑥)2 ,
2∆𝑥

and we have also

𝜑(𝑥0 + ∆𝑥) − 2𝜑(𝑥0 ) + 𝜑(𝑥0 − ∆𝑥)

𝜑́́ (𝑥0 ) ≅ 2
+ 𝑂(∆𝑥)2 .
∆𝑥

To find the numerical solution to partial differential equation with

finite difference method, we discretize the domain D of the given
problem [5,20,36,37].

3.3 Strategy of Discretization

We start implementing the finite difference method to Schrödinger

equation (3.2),

−ħ2 𝜕2 𝜕
𝜓(𝑥𝑖 , 𝑡𝑗 ) + 𝑉(𝑥𝑖 , 𝑡𝑗 )𝜓(𝑥𝑖 , 𝑡𝑗 ) = 𝑖ħ 𝜑(𝑥𝑖 , 𝑡𝑗 ), for(𝑥, 𝑡) ∈ 𝑅 (3.2)
2𝑚 𝜕𝑥 2 𝜕𝑡

The rectangular domain 𝑅 = {(𝑥, 𝑡)|𝑎 < 𝑥 < 𝑏, 𝑐 < 𝑡 < 𝑑} and 𝜑(𝑥, 𝑡) =
𝑔(𝑥, 𝑡) for any (𝑥, 𝑡) ∈ 𝑆, where: 𝑆 denotes the boundary of a region 𝑅,

𝑔(𝑥, 𝑡) is continuous on 𝑆.

Now, we will use the finite difference algorithm for solving the time-
independent Schrödinger equation [5,35]:
 41

The Finite Difference Algorithm

Step 1: Choose positive integers 𝑁 and 𝑀.

𝑏−𝑎 𝑑−𝑐
Step 2: Define ∆𝑥 = ℎ = and ∆𝑡 = 𝑘 = .
𝑁 𝑀

This step partitions the interval [a, b] into 𝑁 equal parts of width ℎ and
partitions the interval [c, d] into 𝑀 equal parts of width 𝑘 as step 3

illustrates.
Step 3: Define the mesh point (𝑥𝑖 , 𝑡𝑗 ) as

𝑥𝑖 = 𝑎 + 𝑖ℎ, 𝑖 = 0,1,2, … . . , 𝑁
𝑡𝑗 = 𝑗𝑘, 𝑗 = 0,1,2, … . . , 𝑀

Step 2 and step 3 are illustrated in figure 1.4.

𝑡
𝑡𝑀 = 𝑑

𝑡2
𝑡1
𝑡0 = 𝑐

𝑥
𝑥0 = 𝑎 𝑥1 𝑥2 … 𝑥𝑁 = 𝑏

Figure 1.5: Grid lines and mesh points of the grid using Finite difference method.

It is clear from figure 1.5 that we obtain horizontal and vertical lines inside
the rectangle 𝑅. These lines are called "grid lines" and
 42

their intersections are named "mesh points" of the grid. For each mesh
point inside the grid, (𝑥𝑖 , 𝑡𝑗 ), 𝑖 = 1,2, … , 𝑁 − 1 and

𝑗 = 1,2, … , 𝑀 − 1 [2,5]. We use Taylor series in the variable 𝑥 about t to

generate the central-difference formula:
𝜕2 𝜑(𝑥𝑖+1 ,𝑡𝑗 )−2𝜑(𝑥𝑖 ,𝑡𝑗 )+𝜑(𝑥𝑖−1 ,𝑡𝑗 ) ħ2 𝜕4
2
𝜑(𝑥𝑖 , 𝑡𝑗 ) = − . 𝜑(𝜉𝑖 , 𝑡𝑗 ) (3.3)
𝜕𝑥 ħ2 12 𝜕𝑥 4

𝑤ℎ𝑒𝑟𝑒 𝜉𝑖 ∈ (𝑥𝑖−1 , 𝑥𝑖+1 ).

In addition, we use Taylor series in the variable 𝑡 about 𝑥 to generate the
forward-difference formula [2,5]:
𝜕 𝜑(𝑥𝑖 ,𝑡𝑗+1)−𝜑(𝑥𝑖 ,𝑡𝑗 ) ̃ 𝜕2
𝑘
𝜑(𝑥𝑖 , 𝑡𝑗 ) = ̃
− 𝜑(𝑥𝑖 , 𝜉𝑗 ) (3.4)
𝜕𝑡 𝑘 2 𝜕𝑡 2

where (𝑥𝑖 , 𝜉𝑗 ) ∈ (𝑡𝑗 , 𝑡𝑗+1 ).

Substituting equation (3.3) and equation (3.4) into equation (3.2), we get:
𝜑(𝑥𝑖 ,𝑡𝑗+1)−𝜑(𝑥𝑖 ,𝑡𝑗 ) 𝑘 𝜕2
𝑖ħ − ∙ 𝜑(𝑥𝑖 , 𝜉𝑗 ) =
𝑘 2 𝜕𝑡 2

−ħ2 𝜑(𝑥𝑖+1 ,𝑡𝑗 )−2𝜑(𝑥𝑖 ,𝑡𝑗 )+𝜑(𝑥𝑖−1 ,𝑡𝑗 ) ℎ2 𝜕4

− . φ(𝜉𝑗 , 𝑡𝑗 ) + 𝑉(𝑥𝑖 )𝜑(𝑥𝑖 , 𝑡𝑗 ) (3.5)
2𝑚 ℎ2 12 𝜕𝑥 4

For each 𝑖 = 1,2,3, … , 𝑛 − 1 and 𝑗 = 1,2,3, … … , 𝑚 − 1.

Rearranging equation (3.5), we get:

−𝑖ħ𝜑(𝑥𝑖 ,𝑡𝑗 ) ħ2 𝜑(𝑥𝑖 ,𝑡𝑗 ) 𝑖ħ𝜑(𝑥𝑖 ,𝑡𝑗+1 )
− ∙ − 𝑉(𝑥𝑖 )𝜑(𝑥𝑖 , 𝑡𝑗 ) + +
𝑘 𝑚 ℎ2 𝑘

ħ2 𝜑(𝑥𝑖+1 ,𝑡𝑗 )+𝜑(𝑥𝑖−1 ,𝑡𝑗 ) 𝑘 𝜕2 ℎ2 𝜕4

∙ = ∙ 𝜑(𝑥𝑖 , 𝜉𝑗 ) − ∙ 𝜑(𝜉𝑖 , 𝑡𝑗 )
2𝑚 ℎ2 2 𝜕𝑡 2 12 𝜕𝑥 4

or it can simply be written as

−𝑖ħ ħ2 𝜑(𝑥𝑖 ,𝑡𝑗+1) ħ2 𝜑(𝑥𝑖+1 ,𝑡𝑗 )+𝜑(𝑥𝑖−1 ,𝑡𝑗 ) 𝑘
[ − 2
− 𝑉(𝑥𝑖 )]𝜑(𝑥𝑖 , 𝑡𝑗 ) + 𝑖ħ + ∙ = ∙
𝑘 𝑚ℎ 𝑘 2𝑚 ℎ2 2
𝜕2 ℎ2 𝜕4
𝜑(𝑥𝑖 , 𝜉𝑗 ) − ∙ 𝜑(𝜉𝑖 , 𝑡𝑗 )
𝜕𝑡 2 12 𝜕𝑥 4
 43

Multiplying both sides by −ℎ2 , we get :

𝑖ħℎ2 ħ2 𝑖ħℎ2 ħ2
[ + + ℎ2 𝑉(𝑥𝑖 )]𝜑(𝑥𝑖 , 𝑡𝑗 ) − 𝜑(𝑥𝑖 , 𝑡𝑗+1 ) − ∙ [𝜑(𝑥𝑖+1 , 𝑡𝑗 ) +
𝑘 𝑚 𝑘 2𝑚
𝑘 𝜕2 ℎ2 𝜕4
𝜑(𝑥𝑖−1 , 𝑡𝑗 )] = −ℎ2 [ ∙ 2
𝜑(𝑥𝑖 , 𝜉𝑗 ) − ∙ 𝜑(𝜉𝑖 , 𝑡𝑗 )].
2 𝜕𝑡 12 𝜕𝑥 4

Clarifying the last equation and typically let 𝜑𝑖,𝑗 approximate 𝜑(𝑥𝑖 , 𝑡𝑗 ),
we form[2,5]:

𝑖ħℎ2 ħ2 𝑖ħℎ2 ħ2
[ + + ℎ2 𝑉(𝑥𝑖 )]𝜑𝑖,𝑗 − 𝜑𝑖,𝑗+1 − ∙ [𝜑𝑖+1,𝑗 + 𝜑𝑖−1,𝑗 ] = 0(3.6)
𝑘 𝑚 𝑘 2𝑚

for each 𝑖 = 1,2,3, … , 𝑛 − 1 and 𝑗 = 1,2,3, … … , 𝑚 − 1.

3.4 Eigenvalue Problem

The wave functions, 𝜓, are eigenvectors of the Hamiltonian

operator, and satisfy equation [16]:
̂ 𝜓 = 𝐸𝜓
H (3.7)

̂ is the Hamiltonian operator, and the eigenvalues 𝐸 represents the

where H
energies of a particle with wave function 𝜓. In the one
dimensional situation, 𝜓 is dependent only on the spatial coordinate 𝑥,
and the one dimensional Hamiltonian is performed by

2 2
̂ = (−ħ ) 𝜕 2 + 𝑉(𝑥)
𝐻
2𝑚 𝜕𝑥

where ħ is a stable constant, 𝑚 denotes the mass of the particle, and

𝑉(𝑥) determine the potential energy function of the particle. Note that
𝐻 is dependent upon via 𝑥 the 𝑉(𝑥) term. To give a dimensionless
interpretation, we establish ħ2 = 2𝑚, so that
 44
𝜕2
̂=−
𝐻 + 𝑉(𝑥) (3.8)
𝜕𝑥 2

Let the points: 𝑥𝑗 j = 0, 1, 2...n.

At each point 𝑥𝑗 , equation (3.7) holds [16], so that

̂𝑗 𝜓(𝑥𝑗 ) = 𝐸𝜓(𝑥𝑗 )
𝐻
̂𝑗 is the Hamiltonian operator evaluated at 𝑉(𝑥𝑗 ). Taking into
Where 𝐻

Consideration that each of the points 𝑥𝑗 makes the structure of

equations:

̂0 𝜓(𝑥0 ) = 𝐸𝜓(𝑥0 )
𝐻

̂𝑗 𝜓(𝑥𝑗 ) = 𝐸𝜓(𝑥𝑗 )
𝐻 (3.9)

….

̂𝑛 𝜓(𝑥𝑛 ) = 𝐸𝜓(𝑥𝑛 )
𝐻

Using equation (3.8), system (3.9) can be written as[16]:

−𝜓′′(𝑥0 ) + 𝑉(𝑥0 )𝜓(𝑥0 ) = 𝐸𝜓(𝑥0 )

−𝜓′′(𝑥1 ) + 𝑉(𝑥1 )𝜓(𝑥1 ) = 𝐸𝜓(𝑥1 ) (3.10)

….

−𝜓′′(𝑥𝑛 ) + 𝑉(𝑥𝑛 )𝜓(𝑥𝑛 ) = 𝐸𝜓(𝑥𝑛 )

We now have a system (3.10) of n equations relating the wave function of
a particle to its fundamental energy.

−𝜓′′(𝑥) + [𝑉(𝑥) − 𝐸]𝜓(𝑥) = 0 (3.11)

 45

]We will limit our analysis the evaluation of 𝜑 (𝑥) in the case of the
infinite square well, and for scattering states within a finite interval[5,16].

We now select to approximate 𝜑́́ (𝑥) in equation (3.11) using finite

difference scheme for 𝜑́́ (𝑥), such that:
1
𝜑́́ (𝑥) = ( 2 ) (𝐴𝜑(𝑥𝑗−𝑛 ) + 𝐵𝜑(𝑥𝑗−𝑛+1 ) + ⋯ + 𝐶𝜑(𝑥𝑗 ) + ⋯
ℎ
+ 𝐷𝜑(𝑥𝑗+𝑛−1 ) + 𝐺𝜑(𝑥𝑗+𝑛 ))

where A, B, C, D, and G determined constants, h denotes the step size,

and n is an integer.
In the second order, the centered finite difference approximation:
1
𝜑́́ (𝑥𝑗 ) ≈ ( 2 ) (𝜑(𝑥𝑗−1 ) − 2𝜑(𝑥𝑗 ) + 𝜑(𝑥𝑗+1 ))
ℎ
́́ (𝑥) system (3.10) becomes
Substituting the approximation for 𝜑

1 2 1
(− (ℎ2) 𝜑(𝑥−1 ) + ℎ2 𝜑(𝑥0 ) − ℎ2 𝜑(𝑥1 )) + 𝑉(𝑥0 ). 𝜑(𝑥0 ) = 𝐸𝜑(𝑥0 ),

Simplifying,
1 2 1
− ( 2 ) 𝜑(𝑥−1 ) + [𝑉 (𝑥0 ) + 2 ] 𝜑(𝑥0 ) − ( 2 ) 𝜑(𝑥1 ) = 𝐸𝜑(𝑥0 )
ℎ ℎ ℎ

1 2 1
−( ) 𝜑(𝑥0 ) + [𝑉 (𝑥1 ) + ] 𝜑(𝑥1 ) − ( ) 𝜑(𝑥2 ) = 𝐸𝜑(𝑥1 )
ℎ2 ℎ2 ℎ2
……..
1 2 1
− ( 2 ) 𝜑(𝑥𝑛−2 ) + [𝑉 (𝑥𝑛−1 ) + 2 ] 𝜑(𝑥𝑛−1 ) − ( 2 ) 𝜑(𝑥𝑛 ) = 𝐸𝜑(𝑥𝑛−1 )
ℎ ℎ ℎ

1 2 1
−( ) 𝜑(𝑥𝑛−1 ) + [𝑉 (𝑥𝑛 ) + ] 𝜑(𝑥𝑛 ) − ( ) 𝜑(𝑥𝑛+1 ) = 𝐸𝜑(𝑥𝑛 )
ℎ2 ℎ2 ℎ2
now we define the vector
 46
𝜑(𝑥0 )
𝜑(𝑥1 )
𝜑= …
𝜑(𝑥𝑛−1 )
[ 𝜑(𝑥𝑛 ) ]
Employing a suitable finite differencing scheme, as well as using a
boundary conditions, the system of linear equations may be expressed
only in terms of the sample points 𝑢(𝑥𝑗 ) such that j = 0, 1, …, n

(excluding 𝑢(𝑥−1 ), … . , 𝑢(𝑥𝑛+1 ), 𝑢(𝑥𝑛+2 ), … )[5,16].

In this case, system (3.11) can be written as

𝐻𝜑
⃗⃗ = 𝐸𝜑
⃗⃗
where 𝐻 is a matrix containing the coefficients of each 𝜑(𝑥𝑗 ) in the system
of linear equations. 𝐸 is the energy of eigenvector 𝜑
⃗⃗(𝑥𝑗 ) at each sample
points 𝑥𝑗 , and is an eigenvalue of 𝐻.
−ħ2 𝜕 2 𝜑
𝐸𝜑 = ( ) + 𝑉(𝑥)𝜑
2𝑚 𝜕𝑥 2
𝜑1 𝜑1
𝜑2 𝜑2
𝐸 . =[ 𝐻 ] .
. 𝑁×𝑁 .
[𝜑𝑛 ] [𝜑𝑛 ]
The eigenvalues of the 𝑁 by 𝑁 matrix (𝐻) can be evaluated. There will
be 𝑁 eigenvalues and 𝑁 eigenvectors. For a large value of 𝑁, Matlab can
be used to find eigenvalues and eigenvectors.

[𝑉, 𝐷] = 𝑒𝑖𝑔 (𝐻)

where 𝐷 has the eigenvalues of 𝐻 as its diagonal elements. 𝑉 has

normalized eigenvectors of 𝐻 as its columns [16,24].

3.5 Finite Difference Method for infinite square well

 47

In this part, we need to find the solution to the free time-

independent Schrödinger equation with the boundary conditions [22]
−ħ2 𝜕2
( 2𝑚 ) 𝜕𝑥 2 𝜓(𝑥) = 𝐸 𝜓(𝑥) (3.12)

Using the second order centered finite difference approximation:

1
𝜑́́ (𝑥𝑗 ) ≈ ( 2 ) (𝜑(𝑥𝑗+1 ) − 2𝜑(𝑥𝑗 ) + 𝜑(𝑥𝑗−1 ))
ℎ
each linear combination will have the form
1
− ( 2 ) (𝜑(𝑥𝑗+1 ) − 2𝜑(𝑥𝑗 ) + 𝜑(𝑥𝑗−1 )) + 𝑉(𝑥𝑗 )𝜑(𝑥𝑗 ) = 𝐸𝜑(𝑥𝑗 )
ℎ

1 2 1
−( ) 𝜑(𝑥𝑗−1 ) + [𝑉(𝑥𝑗 ) + ] 𝜑(𝑥𝑗 ) − ( ) 𝜑(𝑥𝑗+1 ) = 𝐸𝜑(𝑥𝑗 ),
ℎ2 ℎ2 ℎ2
where 𝜑(𝑥𝑘 ) = 0, ∀ 𝑥𝑘 ∉ [0, L].
Thus, the matrix 𝐻 will take a form
2 1
+ 𝑉1 −
∆𝑥 2 ∆𝑥 2
1 2 1
− 2 + 𝑉2 −
∆𝑥 ∆𝑥 2 ∆𝑥 2
⋱ ⋱ ⋱
⋱ ⋱ ⋱
1 2 1
− 2 + 𝑉𝑛−1 − 2
∆𝑥 ∆𝑥 2 ∆𝑥

1 2
− + 𝑉𝑛
∆𝑥 2 ∆𝑥 2
( )

3.6 Finite difference method for finite square well

The bound states of the finite square well which shows in figure
(1.2):
 48

Region 2:
−𝐿 < 𝑥 < 𝐿, 𝑉(𝑥) = 0.

The equation is:

𝜕2
2
𝜓(𝑥) = −𝑘 2 𝜓(𝑥)
𝜕𝑥
Use second-order centered difference formula for 𝜑́́ (𝑥𝑖 ), 𝑖 = 1,2, … ,9 and
drop the error term,

𝜑𝑖+1 − 2𝜑𝑖 + 𝜑𝑖−1

2
= −𝑘 2 𝜑𝑖
ℎ
2𝑚𝐸
where 𝑘 2 = .
ħ2

Region 1 and 3 in figure (𝟏. 𝟐):

x ≤ −L 𝑜𝑟 𝑥 ≥ 𝐿, 𝑉(𝑥) = 𝑉0.
The equation is:
𝜕2
𝜓 = 𝛼2𝜓
𝜕𝑥 2
2𝑚
where 𝛼 2 = (𝑉0 − 𝐸).
ħ2

Use second-order centered difference formula for 𝜑́́ (𝑥𝑖 ), 𝑖 = 1,2, … ,9 and
drop the error term,
𝜑𝑖+1 − 2𝜑𝑖 + 𝜑𝑖−1
2
= 𝛼 2 𝜑𝑖 .
ℎ
When we use second-order centered difference formula for 𝜑́́ (𝑥𝑖 ), and
drop the error term, we get
𝜑𝑖+1 −2𝜑𝑖 +𝜑𝑖−1
= −𝐸𝜑𝑖 .
ℎ2

Each linear combination will have a form

1
− ( 2 ) (𝜑(𝑥𝑖+1 ) − 2𝜑(𝑥𝑖 ) + 𝜑(𝑥𝑖−1 )) = 𝐸𝜑(𝑥𝑖 )
ℎ
 49
1 2 1
− ( 2) 𝜑(𝑥𝑖−1 ) + 2 𝜑(𝑥𝑖 ) − ( 2) 𝜑(𝑥𝑖+1 ) = 𝐸𝜑(𝑥𝑖 ) (3.13) for
ℎ ℎ ℎ
𝑁
𝑥𝑖 ≤ 𝑥𝑁 , 𝑖 = 1, … . , .
2 2

Thus, the matrix 𝐻 takes the form

2 1
− 0 0 0 0 0 0 0
ℎ2 ℎ2
1 2 1
− 2 − 0 0 0 0 0 0
ℎ ℎ2 ℎ2
1 2 1
0 − 2 − 0 0 0 0 0
ℎ ℎ2 ℎ2
1 2 1
0 0 − 2 − 0 0 0 0
ℎ ℎ2 ℎ2
1 2 1
0 0 0 − 2 − 0 0 0
ℎ ℎ2 ℎ2
1 2 1
0 0 0 0 − 2 − 0 0
ℎ ℎ2 ℎ2
1 2 1
0 0 0 0 0 − 2 − 0
ℎ ℎ2 ℎ2
1 2 1
0 0 0 0 0 0 − 2 −
ℎ ℎ2 ℎ2
1 2
[ 0 0 0 0 0 0 0 − 2
ℎ ℎ2 ]
Use second-order centered difference formula for 𝜑́́ (𝑥𝑖 ), and drop the error
term, we get
𝜑𝑖+1 − 2𝜑𝑖 + 𝜑𝑖−1
= 𝛼 2 𝜑𝑖 .
ℎ2
Each linear combination will have a form
1
(ℎ2) (𝜑(𝑥𝑖+1 ) − 2𝜑(𝑥𝑖 ) + 𝜑(𝑥𝑖−1 )) = 𝛼 2 𝜑(𝑥𝑖 ) (3.14)
𝑁
for ≤ 𝑖 ≤ 𝑁 + 1.
2

Thus, the matrix 𝐻 takes the form

 50
2 1
− 0 0 0 0 0 0 0
ℎ2 ℎ2
1 2 1
− 2 0 0 0 0 0 0
ℎ2 ℎ ℎ2
1 2 1
0 − 2 0 0 0 0 0
ℎ2 ℎ ℎ2
1 2 1
0 0 − 2 0 0 0 0
ℎ2 ℎ ℎ2
1 2 1
0 0 0 − 2 0 0 0
ℎ2 ℎ ℎ2
1 2 1
0 0 0 0 − 2 0 0
ℎ2 ℎ ℎ2
1 2 1
0 0 0 0 0 − 2 0
ℎ2 ℎ ℎ2
1 2 1
0 0 0 0 0 0 − 2
ℎ2 ℎ ℎ2
1 2
[ 0 0 0 0 0 0 0
ℎ2
− 2 ]
ℎ
Then we need to solve these systems.

3.7 Pseudo-spectral method

Another method used to solve the Schrödinger equation is the pseudo-

spectral method, which based on the expansion of basis functions
defined in a collection of grid points, a function is approximated as a
weighted sum of smooth basis functions, which are often selected to be
Legendre or Chebyshev polynomials. The pseudo-spectral method is
proposed for the numerical solution of the nonlinear Schrödinger equation
[9,26].

The fundamental concept of this method is to expand the solution

function as a finite series of smooth basis functions [8],

𝜑: [−1,1] → 𝑅
 51

𝜑𝑛 (𝑥) = ∑𝑁
𝑛=0 𝑐𝑛 𝑇𝑛 (𝑥), (3.15)

where, 𝑇𝑛 (𝑥) typically represents Chebyshev or Legendre orthogonal

polynomials defined on the interval [−1,1] [9,25,31], and 𝑐𝑛 represents

the constant coefficients vector.
𝑐𝑛 = (𝑐0 , 𝑐1 , … . . , 𝑐𝑁 )𝑇 ∈ 𝑅𝑁+1 ,

Which are computed from the formula

2 1 𝜋𝑗𝑛
𝑐𝑛 = ∑𝑁𝑗=0 𝜑(𝑥𝑗 ) cos ( ) , 𝑛 = 0,1,2, … . , 𝑁. (3.16)
𝑁𝑐̃𝑛 𝑐̃𝑗 𝑁
𝜋𝑗
Here, 𝑐̃0 = 𝑐̃𝑁 = 2, 𝑐̃𝑛 = 1, 𝑛 = 1,2, … . , 𝑁 − 1, 𝑎𝑛𝑑 𝑥𝑗 = cos ( ),
𝑁

𝑗 = 0,1, … . . , 𝑁, are the Chebyshev-Gauss-Lobatto points.

Let recall the definition of a Chebyshev polynomial:
𝑇𝑛 (𝑥) = cos(𝑛 cos −1 (𝑥)) , if |𝑥| ≤ 1 (3.17)

Let:
𝜃 = cos −1 𝑥 ⇒ 𝑥 = cos 𝜃

Then:

𝑇𝑛 (𝑥) = ∅𝑛 (𝜃) = cos(𝑛𝜃), 𝜃 ∈ [−𝜋, 𝜋] (3.18)

Using the identity defined a bove, equation (3.15) becomes:

𝜑(𝑥) = ∑𝑁
𝑛=0 𝑐𝑛 cos(𝑛𝜃) (3.19)

Deriving equation (3.19) with respect to 𝑥, we get:

𝜕𝜑(𝑥) 𝑛 sin 𝑛𝜃
= ∑𝑁
𝑛=0 𝑐𝑛 ( ) (3.20)
𝜕𝑥 sin 𝜃

𝜕2 𝜑(𝑥) 𝑛 sin 𝑛𝜃 cos 𝜃−𝑛2 cos 𝑛𝜃 sin 𝜃

= ∑𝑁
𝑛=0 𝑐𝑛 ( ) (3.21)
𝜕𝑥 2 sin3 𝜃

The basic equation is:

 52
𝜕2
𝜑(𝑥) = 𝐸𝜑(𝑥) (3.22)
𝜕𝑥 2

Substituting equation (3.21) in equation (3.22), we get:

𝑛 sin 𝑛𝜃 cos 𝜃−𝑛2 cos 𝑛𝜃 sin 𝜃
∑𝑁
𝑛=0 𝑐𝑛 ( ) = 𝐸. ∑𝑁
𝑛=0 𝑐𝑛 cos(𝑛𝜃)
sin3 𝜃

Using boundary condition to find the coefficients of the vector 𝑐𝑛 [24,36].

𝑁 𝑁 𝑁
𝜋 𝜋
𝜑(0) = ∑ 𝑐𝑛 𝑇𝑛 (0) = ∑ 𝑐𝑛 ∅𝑛 ( ) = ∑ 𝑐𝑛 (cos 𝑛 ) = 0
2 2
𝑛=0 𝑛=0 𝑛=0
𝑁

𝜑(1) = ∑ 𝑐𝑛 𝑇𝑛 (1)
𝑛=0
𝑁 𝑁 𝑁

= ∑ 𝑐𝑛 ∅𝑛 (0) = ∑ 𝑐𝑛 (cos 𝑛0) = ∑ 𝑐𝑛 (1)𝑛 = 0

𝑛=0 𝑛=0 𝑛=0

So
𝑐0
1
0 −1 0 1 0 −1 0 1 0 𝑐1
0
2
[
1 ] 𝑐 2. = [ ] (3.23)
1 1 1 1 1 1 1 1 1 ; 0
2 ;
[ 𝐶𝑁 ]

These relations form a system with two equations and 𝑁 + 1 unknowns,

to construct the remaining 𝑁 − 1 equations we collocate (3.15) at the
zeros of 𝑇𝑁−1 (𝑥), which are the interior points between 0 and 1, and are
given as [8,24,30]
(2𝑛 − 1)𝜋
𝜃𝑛 = , 𝑛 = 1, … . . , 𝑁 − 1.
𝑁−1
By using inverse Fourier transform we will find the wave function 𝜑(𝑥)
𝜑(𝑥) = ∑𝑁
𝑛=0 𝑐𝑛 cos(𝑛𝜃).
 53

3.8 Pseudo-Spectral Algorithm

Step 1: Choose positive integer N.

𝐿
Step 2: Define ∆𝑥 = .
𝑁+1

Step 3: Define the mesh point 𝑥𝑖 as 𝑥𝑖 = 𝑖ℎ 𝑖 = 0,1,2, … … , 𝑁 + 1.

Step 4: Define 𝑥 = cos 𝜃, 𝜃 = cos −1 𝑥.

Step 5: Define ∅𝑛 (𝑥) = ∅𝑛 (𝜃) = cos 𝑛𝜃.

Step 6: Define 𝜑(𝑥) = ∑𝑁
𝑛=0 𝑐𝑛 cos(𝑛𝜃).

Step7: Using boundary conditions to evaluate the coefficient of the vector

𝑐𝑛 .
From the first boundary condition :
𝜑(𝑥0 ) = ∑𝑁
𝑛=0 𝑐𝑛 ∅𝑛 (𝑥0 ),

from the second boundary condition:

φ(𝑥𝑁+1 ) = ∑𝑁
𝑛=0 𝑐𝑛 ∅𝑛 (𝑥𝑁+1 ).

Step 8: Find the matrix of the coefficient of the vector 𝑐𝑛, by using this

formula
(2𝑛 − 1)𝜋
𝜃𝑛 = , 𝑛 = 1, … … , 8
𝑁−1
Step 9: Using inverse Fourier transform to find the wave function 𝜑(𝑥).
 54

Chapter Four
Numerical Examples and Results
 55

Chapter Four

Numerical Examples and Results

In this chapter, we implement the two numerical methods, namely, the
finite difference and pseudo-spectral methods, for solving the Schrödinger
equation and its variants.

4.1 Numerical solution for infinite square well of TISE

Example (𝟒. 𝟏):

Consider that for the infinite square well, the particle whose mass 0.5 g is

only found in the infinite interval [0,1], such that

0 𝑥 ∈ [0,1]
𝑉(𝑥) = {
∞ 𝑒𝑙𝑠𝑒
consider the free time- independent Schrödinger equation
−ħ2 𝜕2
( 2𝑚 ) 𝜕𝑥 2 𝜓(𝑥) = 𝐸 𝜓(𝑥) (4.1)

with the boundary conditions

𝜓(0) = 0 and 𝜓(1) = 0 (4.2)

the exact solution of (4.1) [8] is:

𝜑(𝑥) = √2 sin(𝜋𝑥)
the following algorithm is applied to obtain the solution of the equation
(4.1) using the finite difference method.

Algorithm (𝟒. 𝟏): Finite Difference Method

1- Find the general solution for the equation
 56
𝜕2
2
𝜓(𝑥) + 𝐾 2 𝜓(𝑥) = 0
𝜕𝑥
2- Define 𝑏=0, 𝐿 = 1, ħ2 = 2𝑚.
𝑛𝜋 2
3- Find 𝑘, 𝑎, where 𝐾 = ,𝑎 = √ .
𝐿 𝐿

4- Find the energy 𝐸

𝑛 2 𝜋 2 ħ2
where 𝐸𝑛 =
2𝑚𝐿2

5- Choose positive integers 𝑁 = 9.

𝐿 1
6- Clarify ℎ = = , this act separates the interval [0,1] into 9 equal
𝑁+1 10
1
parts of width .
10

7- Define the mesh point 𝑥𝑖 as 𝑥𝑖 = 𝑖ℎ 𝑖 = 0,1,2, … … ,10.

the mesh points are
𝑥0 𝑥1 𝑥2 𝑥3 𝑥4 𝑥5 𝑥6 𝑥7 𝑥8 𝑥9 𝑥10

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

8- Use second-order centered difference formula for 𝜑́́ (𝑥𝑖 ),

𝑖 = 1,2, … ,9 and drop the error term, we get
𝜑𝑖+1 −2𝜑𝑖 +𝜑𝑖−1
= −𝐸𝜑𝑖 .
ℎ2

Each linear combination will have a form

1
− ( 2 ) (𝜑(𝑥𝑖+1 ) − 2𝜑(𝑥𝑖 ) + 𝜑(𝑥𝑖−1 )) = 𝐸𝜑(𝑥𝑖 )
ℎ
1 2 1
− ( 2) 𝜑(𝑥𝑖−1 ) + 2 𝜑(𝑥𝑖 ) − ( 2) 𝜑(𝑥𝑖+1 ) = 𝐸𝜑(𝑥𝑖 ), (4.3)
ℎ ℎ ℎ

where 𝜑(𝑥𝑖 ) = 0, ∀ 𝑥𝑖 ∉ [0, 1].

For 𝑖 = 1, equation (4.3) becomes

1 2 1
− ( 2) 𝜑(𝑥0 ) + 2 𝜑(𝑥1 ) − ( 2) 𝜑(𝑥2 ) = 𝐸𝜑(𝑥1 )
ℎ ℎ ℎ

for 𝑖 = 2,
1 2 1
− ( 2) 𝜑(𝑥1 ) + 2 𝜑(𝑥2 ) − ( 2) 𝜑(𝑥3 ) = 𝐸𝜑(𝑥2 )
ℎ ℎ ℎ
 57

for 𝑖 = 3,
1 2 1
− ( 2) 𝜑(𝑥2 ) + 2 𝜑(𝑥3 ) − ( 2) 𝜑(𝑥4 ) = 𝐸𝜑(𝑥3 )
ℎ ℎ ℎ

for 𝑖 = 4,
1 2 1
− ( 2) 𝜑(𝑥3 ) + 2 𝜑(𝑥4 ) − ( 2) 𝜑(𝑥5 ) = 𝐸𝜑(𝑥4 )
ℎ ℎ ℎ

for 𝑖 = 5,
1 2 1
− ( 2) 𝜑(𝑥4 ) + 2 𝜑(𝑥5 ) − ( 2) 𝜑(𝑥6 ) = 𝐸𝜑(𝑥5 )
ℎ ℎ ℎ

for 𝑖 = 6,
1 2 1
− ( 2) 𝜑(𝑥5 ) + 2 𝜑(𝑥6 ) − ( 2) 𝜑(𝑥7 ) = 𝐸𝜑(𝑥6 )
ℎ ℎ ℎ

for 𝑖 = 7,
1 2 1
− ( 2) 𝜑(𝑥6 ) + 2 𝜑(𝑥7 ) − ( 2) 𝜑(𝑥8 ) = 𝐸𝜑(𝑥7 )
ℎ ℎ ℎ

for 𝑖 = 8,
1 2 1
− ( 2) 𝜑(𝑥7 ) + 2 𝜑(𝑥8 ) − ( 2) 𝜑(𝑥9 ) = 𝐸𝜑(𝑥8 )
ℎ ℎ ℎ

for 𝑖 = 9,
1 2 1
− ( 2) 𝜑(𝑥8 ) + 2 𝜑(𝑥9 ) − ( 2) 𝜑(𝑥10 ) = 𝐸𝜑(𝑥9 )
ℎ ℎ ℎ

Thus, the matrix H takes the form

 58
2 1
− 0 0 0 0 0 0 0
ℎ2 ℎ2
1 2 1
− 2 − 0 0 0 0 0 0
ℎ ℎ2 ℎ2
1 2 1
0 − 2 − 0 0 0 0 0
ℎ ℎ2 ℎ2
1 2 1
0 0 − 2 − 0 0 0 0
ℎ ℎ2 ℎ2
1 2 1
0 0 0 − 2 − 0 0 0
ℎ ℎ2 ℎ2
1 2 1
0 0 0 0 − 2 − 0 0
ℎ ℎ2 ℎ2
1 2 1
0 0 0 0 0 − 2 − 0
ℎ ℎ2 ℎ2
1 2 1
0 0 0 0 0 0 − 2 −
ℎ ℎ2 ℎ2
1 2
[ 0 0 0 0 0 0 0 − 2
ℎ ℎ2 ]
Applying algorithm (4.1) for example (4.1). Table (4.1) contains both the
exact and the numerical results using the finite difference method for

example (4.1).

Table (𝟒. 𝟏): The exact and the numerical solutions using finite
difference method algorithm where 𝑵=9.
𝑥𝑖 𝜑𝑒 = 𝜑𝑒𝑥𝑎𝑐𝑡 𝜑𝑖= 𝜑𝑛𝑢𝑚𝑒𝑟𝑖𝑐𝑎𝑙 Absolute error
|𝜑𝑒− 𝜑𝑖 |
0 0 0 0
0.1 0.43701602 0.46701602 0.03000000
0.2 0.83125388 0.90015676 0.06890288
0.3 1.14412281 1.18543371 0.04131090
0.4 1.34499702 1.25307610 0.09192092
0.5 1.36500261 1.28863421 0.07636840
0.6 1.34499702 1.55307610 0.20807908
0.7 1.14412281 1.18543371 0.04131090
0.8 0.831253878 0.90015676 0.06890288
0.9 0.43701602 0.46701602 0.03000000
1.0 0 0 0
 59

It can be observed that the maximum absolute error is 0.20807908.

The exact and approximate results of 𝜑(𝑥) are shown in Fig. 4.1 (𝑎) and

the resulted error is shown in Fig. 4.1 (𝑏).

The solution of TISE by finite difference method.

1.4
Exact
1.2 Approx

0.8
𝜑(𝑥)
0.6

0.4

0.2

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
𝑥(𝑁 = 9)
Fig. 4.1 (𝒂): A comparison between the exact and approximate solution in example
4.1.

Error

𝑥(𝑁 = 9)
Fig. 4.1 (b): Absolute error between exact and numerical solution in example 4.1
 60

Table (4.2) contains both the exact and the numerical results for the
values of energy for example (4.1).

Table (𝟒. 𝟐): The exact and the numerical solutions for the energy in
example (𝟒. 𝟏) where 𝑵=9.
𝑁 𝐸𝑒𝑥𝑎𝑐𝑡 𝐸𝑛𝑢𝑚𝑒𝑟𝑖𝑐𝑎𝑙 Absolute error
|𝐸𝑒𝑥𝑎𝑐𝑡 -𝐸𝑛𝑢𝑚 |
1 9.86960440 9.27431124 0.59529316
2 39.47841760 39.05884233 0.41957527
3 88.82643961 88.11904310 0.70739651
4 157.91367042 157.8754602 0.03821022
5 246.74011003 246.64312400 0.09698603
6 355.30575844 355.07313011 0.23262833
7 483.61061565 483.89924520 0.28862955
8 631.65468167 631.00001527 0.65466640
9 799.43795649 799.08054461 0.35741188

It can be observed that the maximum absolute error is 0.70739651.

The exact and approximate results of 𝐸𝑛 are shown in Fig. 4.2 (𝑎) and the
resulted error is shown in Fig. 4.2 (𝑏).
Exact
Approx

𝐸𝑛

𝑥
Fig 4.2 (a): A comparison between The exact and approximate results of 𝐄𝐧 .
 61

Error

𝑥
Fig 4.2 (b): Absolute error between The exact and approximate results of 𝐄𝐧 .

Pseudo-Spectral Method

Solving example (𝟒. 𝟏) by the pseudo-spectral method.

Algorithm (𝟒. 𝟐)[ 𝟒, 𝟏𝟑]: The Pseudo- Spectral Method

1. Define 𝐿=1, 𝑚=0.5, 𝑛=1, 𝜃 ∈ [−𝜋, 𝜋].
2. Choose positive integer 𝑁=9.
𝐿 1
3. Define ∆𝑥 = = .
𝑁+1 10

4. Define the mesh point 𝑥𝑖 as 𝑥𝑖 = 𝑖ℎ 𝑖 = 0,1,2, … … ,10.

the mesh points are

𝑥0 𝑥1 𝑥2 𝑥3 𝑥4 𝑥5 𝑥7 𝑥10
𝑥6 𝑥8 𝑥9

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

5. Define 𝑥 = cos 𝜃, 𝜃 = cos −1 𝑥.

 62

6. Define ∅𝑛 (𝑥) = ∅𝑛 (𝜃) = cos 𝑛𝜃.

7. Define 𝜑(𝑥) = ∑𝑁
𝑛=0 𝑐𝑛 cos(𝑛𝜃).

8. Using boundary conditions to evaluate the coefficient of the vector

𝑐𝑛 .
From the first boundary condition:
𝑁 𝑁

𝜑(0) = ∑ 𝑐𝑛 ∅𝑛 (0) = ∑ 𝑐𝑛 0𝑛 = 0
𝑛=0 𝑛=0

From the second boundary condition:

𝑁 𝑁

φ(1) = ∑ 𝑐𝑛 ∅𝑛 (1) = ∑ 𝑐𝑛 1𝑛 = 0
𝑛=0 𝑛=0

9. Find the matrix of the coefficient of the vector 𝑐𝑛, by using this

formula
(2𝑛 − 1)𝜋
𝜃𝑛 = , 𝑛 = 1, … … , 8
𝑁−1
10. Using inverse Fourier transform to find the wave function 𝜑(𝑥).

Applying algorithm (4.2) for example (4.1). Table (4.3) contains both the
exact and the numerical results using the pseudo-spectral method for
example (4.1).
 63

Table (4.3): The exact and the numerical solutions using pseudo-
spectral method algorithm where 𝑵=9.

𝑥𝑖 𝜑𝑒 = 𝜑𝑒𝑥𝑎𝑐𝑡 𝜑𝑖 = 𝜑𝑛𝑢𝑚𝑒𝑟𝑖𝑐𝑎𝑙 Absolute error|𝜑𝑒 -𝜑𝑖 |

0 0 0 0
0.1 0.31880021 0.40000852 0.08120831
0.2 0.60775210 0.72194022 0.11418812
0.3 0.83703748 0.98861174 0.15157426
0.4 0.98442201 1.02520004 0.10026000
0.5 1.03530001 1.08468201 0.04938200
0.6 0.98483305 1.01227759 0.02744454
0.7 0.83780103 0.90463332 0.06683229
0.8 0.60870000 0.65430703 0.04560703
0.9 0.32000000 0.37000032 0.05000032
1 0 0 0

It can be observed that the maximum absolute error is 0.15157426.

The exact and approximate results of 𝜑(𝑥) are shown in Fig. 4.3
(𝑎) and the resulted error is shown in Fig. 4.3 (𝑏).
Exact
Approx

𝜑(𝑥)

𝑥
Fig 4.3 (a): A comparison between exact and numerical solution in example 4.1.
 64

Error

Fig 4.3 (b): Absolute error between exact and numerical solution in example 4.1.

Second case of quantum mechanics: the finite square well.

4.2 Numerical solution for finite square well

Region 2 in figure (𝟏. 𝟐):

−𝐿 < 𝑥 < 𝐿, 𝑉(𝑥) = 0.

The equation is:

𝜕2
2
𝜓(𝑥) = −𝑘 2 𝜓(𝑥)
𝜕𝑥

Use second-order centered difference formula for 𝜑́́ (𝑥𝑖 ), 𝑖 = 1,2, … ,9 and

drop the error term,

𝜑𝑖+1 − 2𝜑𝑖 + 𝜑𝑖−1
2
= −𝑘 2 𝜑𝑖
ℎ
2𝑚𝐸
where 𝑘 2 = .
ħ2
 65

Region 1 and 3 in figure (𝟏. 𝟐):

x ≤ −L 𝑜𝑟 𝑥 ≥ 𝐿, 𝑉(𝑥) = 𝑉0.
The equation is :

𝜕2
𝜓 = 𝛼2𝜓
𝜕𝑥 2
2𝑚
where 𝛼 2 = (𝑉0 − 𝐸),
ħ2

Use second-order centered difference formula for 𝜑́́ (𝑥𝑖 ), 𝑖 = 1,2, … ,9 and
drop the error term,
𝜑𝑖+1 − 2𝜑𝑖 + 𝜑𝑖−1
= 𝛼 2 𝜑𝑖 .
ℎ2
Example (𝟒. 𝟐):
Consider that for the finite square well 1 nm, the particle with mass
=0.5 g is found in the interval [0,1], such that 𝑉(𝑥) = 𝑉0 = 50 𝑒𝑉,
consider the free time independent Schrödinger equation:
−ħ2 𝜕2
( 2𝑚 ) 𝜕𝑥 2 𝜓(𝑥) + 𝑉(𝑥)𝜓(𝑥) = 𝐸 𝜓(𝑥) (4.4)

with the boundary conditions

𝜓(𝑥) → 0 𝑎𝑠 |𝑥| → ∞ (4.5)

the exact solution of (4.4) [1,23] is:

𝜑(𝑥) = 𝑐 𝑒 −𝛼𝑥
the following algorithm is applied to obtain the solution of the

equation (4.4) using the finite difference method.

Algorithm (𝟒. 𝟑): Finite Difference Method

1- Find the general solution for the equation
 66
𝜕2
𝜓(𝑥) − 𝛼 2 𝜓(𝑥) = 0
𝜕𝑥 2

2- Define 𝐿 = 1 𝑛𝑚, ħ2 = 2𝑚, 𝑉0 = 50.

2𝑚
3- Define 𝛼 = √ (𝑉0 − 𝐸)
ħ2
𝑛 2 𝜋 2 ħ2
4- Find the energy 𝐸, where 𝐸𝑛 =
2𝑚𝐿2

5- Choose positive integers 𝑁 = 3.

𝐿2 −𝐿1 1
6- Clarify ℎ = = = 0.25 , this act separates the interval [0,1]
𝑁+1 4

into 4 equal parts of width 0.25.

7- Define the mesh point 𝑥𝑖 as 𝑥𝑖 = 𝑖ℎ, 𝑖 = 0,1,2,3.

The mesh points are
𝑥0 𝑥1 𝑥2 𝑥3 𝑥4

0 0.25 0.5 0.75 1

8- Use second-order centered difference formula for 𝜑́́ (𝑥𝑖 ),

𝑖 = 1,2 and drop the error term, we get
𝜑𝑖+1 −2𝜑𝑖 +𝜑𝑖−1
= −𝐸𝜑𝑖 .
ℎ2

Each linear combination will have a form

1
−( ) (𝜑(𝑥𝑖+1 ) − 2𝜑(𝑥𝑖 ) + 𝜑(𝑥𝑖−1 )) = 𝐸𝜑(𝑥𝑖 )
ℎ2
1 2 1
− ( 2) 𝜑(𝑥𝑖−1 ) + 2 𝜑(𝑥𝑖 ) − ( 2) 𝜑(𝑥𝑖+1 ) = 𝐸𝜑(𝑥𝑖 ) (4.6)
ℎ ℎ ℎ

where 𝜑(𝑥𝑖 ) = 0, ∀ 𝑥𝑖 ∉ [0, 1].

For 𝑖 = 1, equation (4.6) becomes

1 2 1
−( ) 𝜑(𝑥0 ) + 𝜑(𝑥1 ) − ( ) 𝜑(𝑥2 ) = 𝐸𝜑(𝑥1 )
ℎ2 ℎ2 ℎ2
 67

for 𝑖 = 2,
1 2 1
−( ) 𝜑(𝑥1 ) + 𝜑(𝑥2 ) − ( ) 𝜑(𝑥3 ) = 𝐸𝜑(𝑥2 )
ℎ2 ℎ2 ℎ2
9- Use second-order centered difference formula for 𝜑́́ (𝑥𝑖 ), 𝑖 =
3,4 and drop the error term, we get
𝜑𝑖+1 − 2𝜑𝑖 + 𝜑𝑖−1
= 𝛼 2 𝜑𝑖 .
ℎ2
Each linear combination will have a form
1
( 2 ) (𝜑(𝑥𝑖+1 ) − 2𝜑(𝑥𝑖 ) + 𝜑(𝑥𝑖−1 )) = 𝛼 2 𝜑(𝑥𝑖 )
ℎ
for 𝑖 = 3,
1 2 1
( ) 𝜑(𝑥2 ) − 𝜑(𝑥3 ) + ( ) 𝜑(𝑥4 ) = 𝛼 2 𝜑(𝑥3 )
ℎ2 ℎ2 ℎ2
for 𝑖 = 4,
1 2 1
( ) 𝜑(𝑥3 ) − 𝜑(𝑥4 ) + ( ) 𝜑(𝑥5 ) = 𝛼 2 𝜑(𝑥4 )
ℎ2 ℎ2 ℎ2
Thus, for 𝑖 = 1,2 the matrix H will take form
2 1
2
−
( ℎ ℎ2 )
1 2
− 2
ℎ ℎ2
and for 𝑖 = 3,4 the matrix H will take form
2 1
− 2
( ℎ ℎ2 )
1 2
−
ℎ2 ℎ2
Applying algorithm (4.3) for example (4.2). Table (4.4) contains
both the exact and the numerical results using the finite difference method
for example (4.2).
 68

Table (𝟒. 𝟒): The exact and the numerical solutions using finite
difference method algorithm where 𝑵=3.

N 𝜑𝑒 = 𝜑𝑒𝑥𝑎𝑐𝑡 𝜑𝑖 = 𝜑𝑛𝑢𝑚𝑒𝑟𝑖𝑐𝑎𝑙 Absolute error

|𝜑𝑒 − 𝜑𝑖 |
1 0.18945941 0.18945027 0.00000914
2 0.25428327 0.24638761 0.00789566
3 0.58497260 0.58556111 0.00058851
4 0.90483742 0.90097764 0.00385978

It can be observed that the maximum absolute error is 0.00789566.

The exact and approximate results of 𝜑(𝑥) are shown in Fig.

4.4 (𝑎) and the resulted error is shown in Fig. 4.4 (𝑏).

Exact
Approx

𝜑(𝑥)

𝑥 (N=3)
Fig 4.4 (a): A comparison between the exact and approximate solution in example
4.2.
 69

Error

Fig. 4.4 (b): Absolute error between exact and numerical solution in example 4.2.

Table (4.5) contains the exact and the numerical results for
the energy in example (4.2).

Table (𝟒. 𝟓): The exact and the numerical results for the energy in
example (𝟒. 𝟐).

N 𝐸𝑒𝑥𝑎𝑐𝑡 𝐸𝑛𝑢𝑚𝑒𝑟𝑖𝑐𝑎𝑙 Absolute error

| 𝐸𝑒𝑥𝑎𝑐𝑡 -𝐸𝑛𝑢𝑚 |
1 5.72000000 3.77780000 1.94220000
2 20.00000000 10.65000000 9.35000000
3 45.40000000 39.45000000 5.95000000
4 49.84000000 47.76900000 2.07100000

It can be observed that the maximum absolute error is 9.35000000.

The exact and approximate results of 𝐸𝑛 are shown in Fig. 4.5 (𝑎)
and the resulted error is shown in Fig. 4.5 (𝑏).
 70

Approx
Exact

𝐸𝑛

𝑥 (N=3)

Fig 4.5 (a):A comparison between The exact and approximate results of 𝑬𝒏 .

Error

Fig 4.5 (b): Absolute error between exact and numerical energy in example 4.2.
 71

Example (𝟒. 𝟑):

An electron in a finite square well 3 nm and 25 eV deep, consider

the free time independent Schrödinger equation:

−ħ2 𝜕2
( 2𝑚 ) 𝜕𝑥 2 𝜓(𝑥) + 𝑉(𝑥)𝜓(𝑥) = 𝐸 𝜓(𝑥) (4.7)

with the boundary conditions

𝜓(𝑥) → 0 𝑎𝑠 |𝑥| → ∞ (4.8)

the exact solution of (4.7) [1,23] is:

𝜑(𝑥) = 𝑐 𝑒 −𝛼𝑥
the following algorithm is applied to obtain the solution of the
equation (4.7) using the finite difference method.

Algorithm (𝟒. 𝟑): Finite Difference Method

1- Find the general solution for the equation

𝜕2
𝜓(𝑥) − 𝛼 2 𝜓(𝑥) = 0
𝜕𝑥 2

2- Define 𝐿 = 3 𝑛𝑚, ħ2 = 2𝑚, 𝑉0 = 25, interval=[0,3].

2𝑚
3- Define 𝛼 = √ (𝑉0 − 𝐸)
ħ2

4- Find the energy 𝐸

𝑛 2 𝜋 2 ħ2
where 𝐸𝑛 =
2𝑚𝐿2

5- Choose positive integers 𝑁 = 5.

𝐿2 −𝐿1 3
6- Clarify ℎ = = = 0.5 , this act separates the interval [0,3]
𝑁+1 6

into 6 equal parts of width 0.5.

7- Define the mesh point 𝑥𝑖 as 𝑥𝑖 = 𝑖ℎ, 𝑖 = 0,1,2,3.
 72

The mesh points are

𝑥0 𝑥1 𝑥2 𝑥3 𝑥4 𝑥5 𝑥6

0 0.5 1 1.5 2 2.5 3

8- Use second-order centered difference formula for 𝜑́́ (𝑥𝑖 ),

𝑖 = 1,2,3 and drop the error term, we get

𝜑𝑖+1 −2𝜑𝑖 +𝜑𝑖−1
= −𝐸𝜑𝑖 .
ℎ2

Each linear combination will have a form

1
− ( 2 ) (𝜑(𝑥𝑖+1 ) − 2𝜑(𝑥𝑖 ) + 𝜑(𝑥𝑖−1 )) = 𝐸𝜑(𝑥𝑖 )
ℎ
1 2 1
− ( 2) 𝜑(𝑥𝑖−1 ) + 2 𝜑(𝑥𝑖 ) − ( 2) 𝜑(𝑥𝑖+1 ) = 𝐸𝜑(𝑥𝑖 ) (4.8)
ℎ ℎ ℎ

where 𝜑(𝑥𝑖 ) = 0, ∀ 𝑥𝑖 ∉ [0, 3].

For 𝑖 = 1, equation (4.8) becomes
1 2 1
− ( 2 ) 𝜑(𝑥0 ) + 2 𝜑(𝑥1 ) − ( 2 ) 𝜑(𝑥2 ) = 𝐸𝜑(𝑥1 )
ℎ ℎ ℎ
for 𝑖 = 2,
1 2 1
−( 2
) 𝜑(𝑥1 ) + 2 𝜑(𝑥2 ) − ( 2 ) 𝜑(𝑥3 ) = 𝐸𝜑(𝑥2 )
ℎ ℎ ℎ
for 𝑖 = 3,
1 2 1
−( 2
) 𝜑(𝑥2 ) + 2 𝜑(𝑥3 ) − ( 2 ) 𝜑(𝑥4 ) = 𝐸𝜑(𝑥3 )
ℎ ℎ ℎ
9- Use second-order centered difference formula for 𝜑́́ (𝑥𝑖 ),
for 𝑖 = 4,5,6 and drop the error term, we get
𝜑𝑖+1 − 2𝜑𝑖 + 𝜑𝑖−1
2
= 𝛼 2 𝜑𝑖 .
ℎ
Each linear combination will have a form
1
( 2 ) (𝜑(𝑥𝑖+1 ) − 2𝜑(𝑥𝑖 ) + 𝜑(𝑥𝑖−1 )) = 𝛼 2 𝜑(𝑥𝑖 )
ℎ
for 𝑖 = 4,
 73
1 2 1
( ) 𝜑(𝑥3 ) − 𝜑(𝑥4 ) + ( ) 𝜑(𝑥5 ) = 𝛼 2 𝜑(𝑥4 )
ℎ2 ℎ2 ℎ2

for 𝑖 = 5,
1 2 1
( ) 𝜑(𝑥4 ) − 𝜑(𝑥5 ) + ( ) 𝜑(𝑥6 ) = 𝛼 2 𝜑(𝑥5 )
ℎ2 ℎ2 ℎ2

for 𝑖 = 6,
1 2 1
( ) 𝜑(𝑥5 ) − 𝜑(𝑥6 ) + ( ) 𝜑(𝑥7 ) = 𝛼 2 𝜑(𝑥6 )
ℎ2 ℎ2 ℎ2

Thus, for 𝑖 = 1,2,3 the matrix H will take form

2 1
− 0
ℎ2 ℎ2
1 2 1
− 2 −
ℎ ℎ2 ℎ2
1 2
( 0 − 2
ℎ ℎ2 )

and for 𝑖 = 4,5,6 the matrix H will take form

2 1
− 0
ℎ2 ℎ2
1 2 1
− 2
ℎ2 ℎ ℎ2
1 2
( 0 ℎ2
− 2)
ℎ

Applying algorithm (4.3) for example (4.3). Table (4.6) contains

both the exact and the numerical results using the finite difference method
for example (4.3)
 74

Table (𝟒. 𝟔): The exact and the numerical solutions using finite
difference method algorithm where 𝑵=5.

𝑁 𝜑𝑒 = 𝜑𝑒𝑥𝑎𝑐𝑡 𝜑𝑖 = 𝜑𝑛𝑢𝑚𝑒𝑟𝑖𝑐𝑎𝑙 Absolute error

|𝜑𝑒 − 𝜑𝑖 |
1 0.08688196 0.084587199 0.00229476
2 0.10372639 0.10373944 0.00001305
3 0.14332877 0.14432878 0.00100001
4 0.24664101 0.24742102 0.00078001
5 0.79179532 0.74039755 0.05139777

It can be observed that the maximum absolute error is 0.05139777.

The exact and approximate results of 𝜑(𝑥) are shown in Fig. 4.6
(𝑎) and the resulted error is shown in Fig. 4.6 (𝑏).

Exact
Approx

𝜑(𝑥)

𝑥(𝑁 = 5)
Fig 4.6 (a): A comparison between the exact and approximate solution in example 4.3.
 75
Error

Fig. 4.6 (b): Absolute error between exact and numerical solution in example 4.3.

Table (4.7) contains the exact and the numerical results for the energy in
example (4.3).

Table (𝟒. 𝟕): numerical The exact and the results for the energy in
example (𝟒. 𝟑).

𝑁 𝐸𝑒𝑥𝑎𝑐𝑡 𝐸𝑛𝑢𝑚𝑒𝑟𝑖𝑐𝑎𝑙 Absolute error

| 𝐸𝑒𝑥𝑎𝑐𝑡 -𝐸𝑛𝑢𝑚 |
1 1.12300000 1.09551111 0.02748889
2 4.46100000 4.38649084 0.07450916
3 9.90500000 9.86960440 0.03539560
4 17.16200000 17.545963379 0.383963379
5 24.78200000 27.4155677808 2.63356778

It can be observed that the maximum error is 2.63356778.

The exact and approximate results of 𝐸𝑛 are shown in Fig. 4.7 (𝑎)
and the resulted error is shown in Fig. 4.7 (𝑏).
 76

𝐸𝑛

Fig 4.7 (a):A comparison between The exact and approximate results of 𝑬𝒏 .

Fig 4.7 (b): Absolute error between exact and numerical energy in example 4.3.
 77

Conclusion

In this work we have used some analytical methods, namely, the

separation of variables method and the method of characteristics to solve

the Schrödinger equation. However, our main focus was on implementing
two numerical methods, these are: the finite difference and pseudo-spectral

methods, for solving the Schrödinger equation.

Numerical results show clearly that the pseudo-spectral method
(collocation) gives more efficient results than the finite difference method.
One major advantage of the pseudo-spectral method is that it does not
require a tedious steps of evaluating the unknown coefficients of the
approximating function. It is also seen to be suitable for any class of linear

differential equations with or without analytical solutions. Another

advantage of pseudo-spectral method is that the time independent
Schrödinger equation reduces to a system of algebraic equations which can
be solved by many iterative methods. This provides an accurate
approximation for a smooth solution with relatively few degree of
freedom. Finally, less grid points are needed with the pseudo-spectral

method than with finite difference methods to achieve the same accuracy.
 78

References

[1] D. Aronstein and C. Stroud, General series solution for finite

square-well energy levels for use in wave-packet studies,

American Journal of Physics, Vol. 68, No. 10, pp. 943-949. (2000).
[2] K. Atkinson, An Introduction to Numerical Analysis, John Wiley

Sons, (1978).
[3] V. Baskakov and A. Popov, Implementation of Transparent

Boundaries for the Numerical Solution of the Schrödinger

Equation, Wave Motion, Vol. 14, No. 2, pp. 123-128, (1991).
[4] R. Bulirsch, A. Miele and J. Stoer, Numerical Solution of Optimal
Control Problems by Direct Collocation, Optimal Control,

Switzerland, Vol. 111, No. 12, pp. 129-143, (1993).

[5] R. Burden and J. Faries, Numerical Analysis, PWS Publishing
Company, Boston, (1985).
[6] M. Davis, Finite Difference Methods, MSc Course in Mathematics
and Finance, Imperial College London, (2010).
[7] H. Dekker, Classical and quantum mechanics of the damped

harmonic Oscillator, Physics Reports, Vol. 80, No. 1, pp. 1-110,

(1981).
[8] SW. Doescher, MH. Rice , Infinite square well potential with a

moving Wall, American Journal of Physics, 37(12), 1246-1249,

(1969).
 79

[9] R. England and M. Savari, On the Pseudo-Spectral Method of

Solving Linear Ordinary Differential Equations, Journal of

Mathematics and Statistics, Vol. 5, No. 2, pp. 136-140, (2009).

[10] H. Erbil, Nonlinear Schrödinger equations and their solutions,
Journal of Theoretical and Applied Mechanics, Vol. 25, No. 1, pp.

16-43, (𝟐𝟎𝟎𝟑).
[11] V. Fock, Fundamentals of Quantum Mechanics, Mir Publishers,
Moscow, (1978).
[12] C. Gear, Numerical Initial Problems in Ordinary Differential
Equations, Prentice Hall, (1971).
[13] E. Gildener, A. Patrascioui, Pseudo Spectral Contributions to the

Energy Spectrum of a One-Dimensional System, Physical Review

D, 16(2), pp. 16-423, (1977).
[14] W. Greiner, Relativistic quantum mechanics, Berlin: Springer,
(1990).

[15] M. Holmes, Introduction to Numerical Methods in Differential

Equations, Springer-Verlag New York, (2007).

[16] H. Ishikawa, Numerical Methods for the Eigenvalue

Determination of Second-Order Ordinary Differential

Equations, Journal of Computational and Applied Mathmatics,

208(2), 404-424, (2007).

[17] Z. Jackiewicz, and B. Zubik-Kowal, Spectral collocation and
waveform relaxation methods for nonlinear delay partial
differential equation, Applied Numerical Mathematics, 56(3-4),
433-443, (2006).
 80

[18] H. Langtangen, Computational Partial Differential Equations,

Springer Verlag New York, (2001).

[19] S. Larsson and V. Thom´ee, Partial Differential Equations with

Numerical Method, Springer, 1st edition, (2008).
[20] R. LeVeque, Finite Difference Methods for Ordinary and Partial

Differential Equations: Steady State and Time Dependent

Problems, Society for Industrial and Applied Mathematics; 1 st
edition, (2007).
[21] R. Mahheij and R. Russell, Numerical Solution of Boundary Value
Problem for Ordinary Differential Equations, Society for
Industrial and Applied Mathematics; 1st edition, (1988).

[22] AJ. Makowski, ST. Dembinski, Exactly solvable models with time
dependent boundary conditions, Physics Letters A, 154(5-6), pp.
217-220, (1991).
[23] J. Mallow, “Simple graphical solution for the finite square well
with no change of variables,” Am. J. Phys., Vol. 64, No. 9, pp. 1072-
1073, (1996).

[24] C. Marston and G. Balint-Kurti, The Fourier grid Hamiltonian

method for bound state eigenvalues and eigenfunctions, The
Journal of Chemical Physics, 91(6), 3571-3576, (1989).

[25] J. Mason, Chebyshev polynomial approximations for the L-

membrane eigenvalue problem, SIAM, J. Applied Mathmatics,
Vol. 15, No. 1, pp. 172-186, (1967).
 81

[26] D. Mazziotti, Spectral difference methods for solving the

Differential equations of chemical physics, J. Chem. Phys., 117(6),

2455-2468, (2002).
[27] R. Mcowen, Partial Differential Equations, Methods and
Applications, Prentice-Hall, (2003).

[28] Th. Monovasilis and Z. Kalogiratou, Exponential-fitting symplectic

methods for the numerical integration of the Schrödinger

equation. Computational Methods In Sciences And Engineering,
Vol. 37, No. 3, pp. 446-450, (2003).
[29] S. Orszag, Comparison of pseudospectral and spectral
approximation, Studies in Applied Mathematics, 51(3), 253-259,
(1972).

[30] I. Parker, and L. Fox, Chebyshev Polynomials in Numerical

Analysis, 2nd Edn., Oxford University Press, Vol. 31, No. 519, pp.
881-891, (1968).
[31] A. Quarteroni and F. Saleri, Scientific Computing with Matlab,
Springer-Verlag Berlin Heidelberg, (2003).

[32] L. Ram-Mohan, Finite Element and Boundary Elemen

Applications in Quantum Mechanics, Oxford, (2002).
[33] M. Robinson and G. Fairweather, Orthogonal Spline Collocation

Methods for Schrodinger Type Equation in One Space Variable,

Numerisch Mathematik, 68(3), 355-376, (1994).
[34] A. Sandvik, Numerical solutions of the Schrödinger equation,
Department of Physics, Boston University (2013).
 82

[35] A. Stahel, Numerical Methods, Bern University of applied science,

(2012).

[36] J. Strikwerda, Finite Difference Schemes and Partial Differential

Equations, Second Edition, Philadelphia (Pa.) : SIAM, (2004).
[37] W. Thirring and P. Urban, The Schrödinger Equation: Proceedings

of the International Symposium “50 Years Schrödinger Equation” in

Vienna. Springer Science & Business Media, (2012).
[38] J. Thomas, Numerical Partial Differential Equations, Finite
Difference Methods, Applied Mathematics, Springer-Verlag New
York, (1995).
[39] A. Tveito and R. Winther, Introduction to Partial Differential

Equations a Computational Approach, Springer-Verlag Berlin

Heidelberg, (2005).
 83

Appendix
 84

Appendix (A)
Matlab code for finite difference method for time-independent
Schrödinger equation (infinite square well).

Clear all
L= input('\n Please enter the value of L');

n= input('\n Please enter the value of n');

m= input('\n Please enter the value of m');
N= input('\n Please enter the value of N');
(% since h=2m=1 )
h= (2*m)^(1/2) ;
(% find the values of the energy )

En=zeros(1,N);
for i=1:N+1
En(i)=(i*i*(22/7)*(22/7)*2*m)/(2*m*L*L)
end
En=[En1 En2 En3 En4 En5 En6 En7 En8 En9 En10 ];
k=((2*m*En)/(h*h))^(1/2)
̂)
(%enter the values of the matrix Hamiltonian 𝐻
h1=(2/h*h)
h2=-1/(h*h)

H=[h1 h2 0 0 0 0 0 0 0; h2 h1 h2 0 0 0 0 0 0;0 h2 h1 h2 0 0 0 0 0;0 0 h2

h1 h2 0 0 0 0;0 0 0 h2 h1 h2 0 0 0;0 0 0 0 h2 h1 h2 0 0;0 0 0 0 0 h2 h1 h2
0;0 0 0 0 0 0 h2 h1 h2;0 0 0 0 0 0 0 h1 h2]
̂)
(% find the eigenvector and eigenvalue of 𝐻
 85

[V,D] = eig(H)
(% the values of the diagonal of the eigenvalue)

app=[0 D(1,1) D(2,2) D(3,3) D(4,4) D(5,5) D(6,6) D(7,7) D(8,8) D(9, 9)
0];
(% the exact value of the wavefunction)

ress=zeros(1,N);
for j=1:N
ress(j)=(2)^(.05) *(sin((22/7)*j/10))
end
real= [0 ress 0]
(% enter the values of 𝑥)

x=[0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1]
(%plot the exact solution and approximation of wavefunction)
plot(x,real)
hold on
plot(x,app)
(%the error between exact and approximation solutions)

diff=real-d1;
plot(x,diff);
(%the absolute error)

AbDiff= abs(diff);
plot(x,AbDiff);
 86

Appendix (B)
Matlab code for finite difference method for time-independent
Schrödinger equation (finite square well).

L= input('\n Please enter the value of L');

v= input('\n Please enter the value of v');

m= input('\n Please enter the value of m');

N= input('\n Please enter the value of N');

if mod(N,2)==1
else N= input('\n Please renter the value of N, you entered even number')
end

f1= input('\n Please enter the value of start of interval');

f2= input('\n Please enter the value of end of interval');

h=(f2-f1)/(N+1);

; h1=(2/h*h)
;h2=-1/(h*h)
 87

number_of_X= N+1;

H=zeros(number_of_X/2);
for(i=1:number_of_X/2)
for(j=1:number_of_X/2)

if (i==j)
H(i,j)=h2;
else if (abs(i-j)==1)
H(i,j)=h1;
else
H(i,j)=0;

end
end
end
end

H1=-H;

[V,D] = eig(H)

[V1,D1] = eig(H1)

En=zeros(1,N);
for i=1:N
 88

En(i)=(-1* alpha(1,i) *alpha(1,i)*(h)*(h)*2*m)+v

end

for(i=1:number_of_X/2)
for(j=1:number_of_X/2)
c(i,j)=D(i,i)

end

for(i=1:number_of_X/2)
for(j=1:number_of_X/2)
alpha1(i,j)=D1(0,0);
end

alpha=[c, alpha1];

plot(c,V)

plot(alpha,En);

a1=alpha(1,5:8)

plot(a1,V1)

En1=En(1:4)
En2=En(5:8)
 89

Appendix (c)
Matlab code for Pseudo-spectral method for time-independent
Schrödinger equation.

n= input('\n Please enter the max value of n');

N= input('\n Please enter the value of N');

theta= zeros(1,n);
for ii=1:n
theta(ii)=((2*n-1)*(22/7))/(N-1);
tt=[0.5 0 0 0 0 0 0 0 0; 0.5 theta(1) theta(1)^2 theta(1)^3 theta(1)^4
theta(1)^5 theta(1)^6 theta(1)^7 theta(1)^8;0.5 theta(2) theta(2)^2

theta(2)^3 theta(2)^4 theta(2)^5 theta(2)^6 theta(2)^7 theta(2)^8;0.5

theta(3) theta(3)^2 theta(3)^3 theta(3)^4 theta(3)^5 theta(3)^6 theta(3)^7
theta(3)^8;0.5 theta(4) theta(4)^2 theta(4)^3 theta(4)^4 theta(4)^5
theta(4)^6 theta(4)^7 theta(4)^8;0.5 theta(5) theta(5)^2 theta(5)^3
theta(5)^4 theta(5)^5 theta(5)^6 theta(5)^7 theta(5)^8;0.5 theta(6)
theta(6)^2 theta(6)^3 theta(6)^4 theta(6)^5 theta(6)^6 theta(6)^7

theta(6)^8;0.5 theta(7) theta(7)^2 theta(7)^3 theta(7)^4 theta(7)^5

theta(7)^6 theta(7)^7 theta(7)^8;0.5 theta(8) theta(8)^2 theta(8)^3
theta(8)^4 theta(8)^5 theta(8)^6 theta(8)^7 theta(8)^8].
 ‫جامعة النجاح الوطنية‬
‫كـلـيـة الـدراسـات العلـيـا‬

‫الحلول العددية لمعادلة شرودينجر‬

‫اعداد‬
‫هديل موفق حجة‬

‫اشراف‬
‫أ‪ .‬د‪ .‬ناجي قطناني‬

‫قدمت هذه األطروحة استكماال لمتطلبات الحصول على درجة الماجستير في الرياضيات المحوسبة‬
‫في كلية الدراسات العليا‪ ،‬نابلس‪ ،‬فلسطين‪.‬‬
‫‪2019‬‬
 ‫ب‬
‫الحلول العددية لمعادلة شرودينجر‬
‫اعداد‬
‫هديل موفق حجة‬
‫اشراف‬
‫أ‪ .‬د‪ .‬ناجي قطناني‬

‫الملخص‬

‫تعتبر معادلة شرودنغر ومتغيراتها من المعادالت المحورية مجال البحث في مجال المعادالت‬
‫التفاضلية الجزئية ولها تطبيقاتها المختلفة في هندسة الرياضيات ونظرية الضوء والطيف واالنظمة‬
‫المتكاملة‪ .‬تهدف هذه الدراسة الى عرض المفاهيم االساسية لميكانيكا الكم ومنها معادلة شرودنغر‬

‫بشكليها‪ :‬معادلة شرودنغر المرتبطة بالزمن وتلك غير المرتبطة بالزمن‪ .‬اضافة الى ذلك‪ ,‬ستركز‬
‫الباحثة وبشكل رئيسي على بعض الطرق التحليلية والعددية لحل معادلة شرودنغر ومتغيراتها‪ .‬فبالنسبة‬
‫الى الحل التحليلي لمعادلة شرودنغر‪ ،‬فان الباحثة ستعمل على مبدا فصل المتغيرات ومبدا‬

‫الخصائص‪ .‬اما بالنسبة للمعالجة العددية للمعادلة‪،‬فان الباحثة ستستخدم مبدا الفروق المنتهية ومبدا‬
‫الطيف الزائف لحالتين من ميكانيكا الكم‪ :‬الحالة المحدودة وغير المحدودة‪ .‬ولكي يتم اختبار فاعلية‬
‫هذه الطرق‪ ،‬تم االخذ بعين االعتبار بعض حاالت االختبار العددية‪ .‬وقد اظهرت النتائج العددية‬
‫بوضوح ان مبدا الطيف الزائف هو الطريقة االقوى لحل معادلة شرودنغر غير المرتبطة بالزمن‬
‫بالمقارنة مع التفاضل المحدود‪.‬‬