Course Development Team

Head of Programme : Dr Lau Jing Feng

Course Developer(s) : Dr Teh Yong Liang
Technical Writer : Jolene Tan,ETP
Video Production : Danny Chin, ETP

© 2021 Singapore University of Social Sciences. All rights reserved.

No part of this material may be reproduced in any form or by any means without
permission in writing from the Educational Technology & Production, Singapore
University of Social Sciences.

ISBN 978-981-5008-16-6

Educational Technology & Production

Singapore University of Social Sciences
463 Clementi Road
Singapore 599494

How to cite this Study Guide (APA):

Teh, Y. L. (2021). MTH320 Mathematical methods II (study guide). Singapore University of

Social Sciences.

Release V1.0

Build S1.0.5, T1.5.21

 Table of Contents

Table of Contents

Course Guide
1. Welcome.................................................................................................................. CG-2

2. Course Description and Aims............................................................................ CG-3

3. Learning Outcomes.............................................................................................. CG-6

4. Learning Material................................................................................................. CG-7

5. Assessment Overview.......................................................................................... CG-8

6. Course Schedule.................................................................................................. CG-10

7. Learning Mode.................................................................................................... CG-11

Study Unit 1: Non-Homogeneous System of ODEs & Linear 2nd-

Order ODEs
Learning Outcomes................................................................................................. SU1-2

Overview................................................................................................................... SU1-3

Chapter 1: Non-Homogeneous System of Linear 1st-Order ODEs.................. SU1-4

Chapter 2: Linear Second-Order ODEs................................................................ SU1-9

Chapter 3: Mathematical Modelling of Resistor-Inductor Circuits............... SU1-17

Summary................................................................................................................. SU1-22

Formative Assessment.......................................................................................... SU1-23

i
Table of Contents

Study Unit 2: Method of Undetermined Coefficients & Laplace

Transform
Learning Outcomes................................................................................................. SU2-2

Overview................................................................................................................... SU2-3

Chapter 1: Method of Undetermined Coefficients............................................. SU2-4

Chapter 2: Laplace Transform.............................................................................. SU2-11

Chapter 3: Properties of the Laplace Transform............................................... SU2-16

Summary................................................................................................................. SU2-22

Formative Assessment.......................................................................................... SU2-23

Study Unit 3: Solving ODEs By Laplace Transform

Learning Outcomes................................................................................................. SU3-2

Overview................................................................................................................... SU3-3

Chapter 1: Inverse Laplace Transform.................................................................. SU3-4

Chapter 2: Solving ODEs by Laplace Transform................................................ SU3-8

Chapter 3: Mathematical Modelling of Vehicular Suspension........................ SU3-15

Summary................................................................................................................. SU3-20

Formative Assessment.......................................................................................... SU3-22

Study Unit 4: Fourier Series & Fourier Transform

Learning Outcomes................................................................................................. SU4-2

Overview................................................................................................................... SU4-3

Chapter 1: Introduction to Fourier Series............................................................ SU4-4

ii
 Table of Contents

Chapter 2: Periodic Extension & Half-Range Expansions............................... SU4-12

Chapter 3: The Fourier Transform...................................................................... SU4-18

Summary................................................................................................................. SU4-25

Formative Assessment.......................................................................................... SU4-26

Study Unit 5: Partial Differential Equations & Heat Equation

Learning Outcomes................................................................................................. SU5-2

Overview................................................................................................................... SU5-3

Chapter 1: Partial Differential Equations............................................................. SU5-4

Chapter 2: The Heat Equation............................................................................... SU5-8

Chapter 3: Transient Solutions of the Heat Equation....................................... SU5-14

Summary................................................................................................................. SU5-25

Formative Assessment.......................................................................................... SU5-26

Study Unit 6: Wave Equation & Laplace Equation

Learning Outcomes................................................................................................. SU6-2

Overview................................................................................................................... SU6-3

Chapter 1: The Wave Equation.............................................................................. SU6-4

Chapter 2: The Laplace Equation....................................................................... SU6-15

Summary................................................................................................................. SU6-22

Formative Assessment.......................................................................................... SU6-23

iii
Table of Contents

iv
 Course
Guide

Mathematical Methods II
MTH320 Course Guide

1. Welcome

Welcome to the course MTH320 Mathematical Methods II, a 5 credit unit (CU) course.

This Study Guide will be your personal learning resource to take you through the course
learning journey. The guide is divided into two main sections – the Course Guide and
Study Units.

The Course Guide describes the structure for the entire course and provides you with an
overview of the Study Units. It serves as a roadmap of the different learning components
within the course. This Course Guide contains important information regarding the
course learning outcomes, learning materials and resources, assessment breakdown and
additional course information.

CG-2
MTH320 Course Guide

2. Course Description and Aims

MTH320 Mathematical Methods II gives a comprehensive introduction to more advanced

topics such as solving ordinary differential equations with power series solutions, using
Laplace transforms to solve ordinary differential equations and solving partial differential
equations via Fourier series. Relevant physical models such as vibrating string, heat flow
from a body in space and membranes will be taught alongside with the mathematics.

Course Structure
This course is a 5-credit unit course presented over six weeks.

There are six Study Units in this course. The following provides an overview of each Study
Unit.

Study Unit 1 – Non-Homogeneous System of ODEs & Linear 2nd-Order

ODEs

This study unit introduces systems of linear first-order non-homogeneous ODEs which
arise naturally from mathematical modelling of real-world phenomena as well as
engineering systems. It then demonstrates the use of Duhamel's principle in the solution
process of such linear systems. The general solutions for the different types of eigenvalues
are derived for 2nd-order non-homogeneous linear ODEs. The use of the general solutions
in the analysis of a suspension system is then demonstrated. Finally, the mathematical
modelling of a resistor-inductor circuit is presented and its solutions are analysed.

Study Unit 2 – Method of Undetermined Coefficients & Laplace Transform

This study unit introduces the method of undetermined coefficients as another way to
solve non-homogeneous linear ODEs. Then, due to the difficulty in solving such ODEs
when the external excitation function is not smooth, the Laplace transform is introduced.
The Laplace transforms of elementary functions and the delta and unit-step functions

CG-3
MTH320 Course Guide

are evaluated. We also learn how piecewise functions can be rewritten using the unit-
step function. Finally, the properties of the Laplace transform and how they can be
incorporated to evaluate the Laplace transforms of some functions are shown.

Study Unit 3 – Solving ODEs by Laplace Transform

This study unit introduces the inverse Laplace transform in order to convert functions
in the s-domain back to functions in the time domain. Then, the Laplace transforms
of the derivatives of the functions of time are presented, and implemented together
with the solving of ODEs. Finally, we see how the mathematical model of a suspension
system subjected to non-smooth functions is derived and solved by the method of Laplace
transform.

Study Unit 4 – Fourier Series & Fourier Transform

This study unit introduces the Fourier series representation of a periodic function. For a
function defined in a finite interval, the use of the periodic extension and/or half-range
expansions is considered. Then, the complex Fourier series representation is derived.
Finally, for a non-periodic function, the Fourier integral representation is evaluated and
the Fourier transform is defined. The effect of the Fourier transform in transforming a
function from the time or space domain to the frequency domain is then demonstrated.

Study Unit 5 – Partial Differential Equations & Heat Equation

This study unit introduces partial differential equations (PDEs) with a focus on the heat
equation. The meaning of a solution to a PDE is defined and the evaluation of the partial
derivatives of a multivariable function is demonstrated. The heat equation is then derived
from the law of conservation of heat energy. The solution to the heat equation is shown to
be composed of a steady-state solution and a transient solution. In solving the transient
heat equation, the method of separation of variables and the Fourier series are employed.
Several analyses of the temperature solutions are presented.

CG-4
MTH320 Course Guide

Study Unit 6 – Wave Equation & Laplace Equation

This study unit introduces the wave equation which describes the motion of vibrating
medium and the Laplace equation that governs the steady-state (time-independent)
behaviour of many phenomena. The wave equation is derived from Newton's 2nd law.
Using the separation of variables and the Fourier series, the solutions to the wave and
Laplace equations are presented. We also investigate the wave propagation phenomenon
along a vibrating string and reconcile the observations with our intuition.

CG-5
MTH320 Course Guide

3. Learning Outcomes

Knowledge & Understanding (Theory Component)

By the end of this course, you will be able to:

• Determine the solutions of a given partial differential equation by Laplace

transform.
• Show the validity of given mathematical statements in differential equations.
• Explain the results/conclusions of a mathematical model.

Key Skills (Practical Component)

By the end of this course, you will be able to:

• Solve a given partial differential equation by Fourier series.

• Compute the Fourier series of a piecewise continuous function.
• Sketch solutions of a given differential equation.

CG-6
MTH320 Course Guide

4. Learning Material

To complete the course, you will need the following learning material(s):

Required Textbook(s)
Potter, M. C., Lessing, J., & Aboufadel, E. F. (2019). Advanced Engineering Mathematics

(4th ed.). Springer International Publishing.

If you are enrolled into this course, you will be able to access the eTextbooks here:

To launch eTextbook, you need a VitalSource account which can be created via
Canvas (iBookStore), using your SUSS email address. Access to adopted eTextbook is
restricted by enrolment to this course.

Website(s):
Sanderson, G. (2019, March 31). Differential Equations. https://youtube.com/playlist?

list=PLZHQObOWTQDNPOjrT6KVlfJuKtYTftqH6

CG-7
MTH320 Course Guide

5. Assessment Overview

The overall assessment weighting for this course is as follows:

Assessment Description Weight Allocation

Assignment 1 Computer Marked 10%

Assignment

Assignment 2 Tutor Marked Assignment 20%

Examination Open book exam 70%

TOTAL 100%

The following section provides important information regarding Assessments.

Continuous Assessment:

There will be continuous assessment in the form of one computer marked assignment
(CMA) and one tutor-marked assignment (TMA). In total, this continuous assessment
will constitute 30 percent of overall student assessment for this course. The two
assignments are compulsory and are non-substitutable. These assignments will test
conceptual understanding of both the fundamental and more advanced concepts and
applications that underlie mathematical methods. It is imperative that you read through
your Assignment questions and submission instructions before embarking on your
Assignment.

Examination:

The final (2-hour) written exam will constitute the other 70 percent of overall student
assessment and will test the ability to apply mathematical methods to engineering
problems. All topics covered in the course outline will be examinable. To prepare for the

CG-8
MTH320 Course Guide

exam, you are advised to review Specimen or Past Year Exam Papers available on Learning
Management System.

Passing Mark:

To successfully pass the course, you must obtain a minimum passing mark of 40 percent
for each of the two assessment components. That is, students must obtain at least a mark of
40 percent for the combined continuous assessments and also at least a mark of 40 percent
for the final exam. For detailed information on the Course grading policy, please refer to
The Student Handbook (‘Award of Grades’ section under Assessment and Examination
Regulations). The Student Handbook is available from the Student Portal.

Non-graded Learning Activities:

Activities for the purpose of self-learning are present in each study unit. These learning
activities are meant to enable you to assess your understanding and achievement of the
learning outcomes. The type of activities can be in the form of Formative Assessment,
Quiz, Review Questions, Application-Based Questions or similar. You are expected to
complete the suggested activities either independently and/or in groups.

CG-9
MTH320 Course Guide

6. Course Schedule

To pace yourself and monitor your study progress, pay special attention to your
Course Schedule. It contains study-unit-related activities including Assignments, Self-
Assessments, and Examinations. Please refer to the Course Timetable on the Student
Portal for the most current Course Schedule.

Note: Always make it a point to check the Student Portal for announcements and
updates.

CG-10
MTH320 Course Guide

7. Learning Mode

The learning approach for this course is structured along the following lines:

a. Self-study guided by the study guide units. Independent study will require at
least 3 hours per week.
b. Working on assignments, either individually or in groups.
c. Classroom Seminars (3 hours each session, 6 sessions in total).

iStudyGuide

You may be viewing the interactive StudyGuide (iStudyGuide), which is the mobile-
friendly version of the Study Guide. The iStudyGuide is developed to enhance your
learning experience with interactive learning activities and engaging multimedia. You
will be able to personalise your learning with digital bookmarking, note-taking, and
highlighting of texts if your reader supports these features.

Interaction with Instructor and Fellow Students

Flexible learning—learning at your own pace, space, and time—is a hallmark at SUSS,
and we strongly encourage you to engage your instructor and fellow students in online
discussion forums. Sharing of ideas through meaningful debates will help broaden your
perspective and crystallise your thinking.

Academic Integrity

As a student of SUSS, you are expected to adhere to the academic standards stipulated
in the Student Handbook, which contains important information regarding academic
policies, academic integrity, and course administration. It is your responsibility to read
and understand the information outlined in the Student Handbook prior to embarking on
the course.

CG-11
MTH320 Course Guide

CG-12
 1
Study
Unit

Non-Homogeneous System of
ODEs & Linear 2nd-Order ODEs
MTH320 Non-Homogeneous System of ODEs & Linear 2nd-Order ODEs

Learning Outcomes

By the end of this unit, you should be able to:

• Appreciate how Non-Homogeneous Systems of ODEs arise from Mathematical

Modelling of Real-World Phenomena & Engineering Systems
• Decompose a Higher-Order Linear ODE into a 1st-Order System of Linear ODEs
• Employ the Duhamel’s Principle in the Solution of the Non-Homogeneous System
of ODEs
• Apply the General Solutions of a 2nd-Order Homogeneous Linear ODE
• Derive and Solve a Math Model of a Simple Resistor-Inductor Circuit

SU1-2
MTH320 Non-Homogeneous System of ODEs & Linear 2nd-Order ODEs

Overview

This study unit introduces systems of linear first-order non-homogeneous ODEs which
arise naturally from mathematical modelling of real-world phenomena as well as
engineering systems. It then demonstrates the use of Duhamel's principle in the solution
process of such linear systems. The general solutions for the different types of eigenvalues
are derived for 2nd-order non-homogeneous linear ODEs. The use of the general solutions
in the analysis of a suspension system is then demonstrated. Finally, the mathematical
modelling of a resistor-inductor circuit is presented and its solutions are analysed.

SU1-3
MTH320 Non-Homogeneous System of ODEs & Linear 2nd-Order ODEs

Chapter 1: Non-Homogeneous System of Linear 1st-

Order ODEs

1.1 Recap: A System of Linear 1st-Order ODEs

From SU6 in MTH319, recall that a system of linear 1st-order ODEs has the general form

where aij(t) and fi(t) are known functions. Rewriting in matrix form, we get

or

If all fi(t) = 0 and aij(t) = aij, then we have a homogeneous system with constant coefficients

We have learnt that the solution of the above variable-separable ODE is

where the exponentiation of matrix A is achieved through diagonalisation or Jordan

decomposition.

SU1-4
MTH320 Non-Homogeneous System of ODEs & Linear 2nd-Order ODEs

1.2 A Non-Homogeneous System of Linear 1st-Order ODEs

Now, we shall consider situations where the vector f(t) ≠ 0, which gives a non-
homogeneous system of linear 1st-order ODEs with constant coefficients

The vector function, f(t), can arise from modelling of physical/engineering phenomena
when there exists some external excitation. For example, in the coupled heat transfer
problem presented in MTH319 SU6, if the steel plate is subjected to external heating Q(t),
applying conservation of energy and Newton’s law of cooling/heating gives

Coupled with the ODE of the air temperature given by

The system of linear ODEs for the coupled heat transfer system is

which is indeed a non-homogeneous system.

1.3 The Duhamel’s Principle

Recall from MTH319 SU5 that a linear 1st-order ODE given by

has a solution given by the method of integrating factor

SU1-5
MTH320 Non-Homogeneous System of ODEs & Linear 2nd-Order ODEs

Now, rearranging the non-homogeneous system of ODEs, we get

which is in the similar form of a linear 1st-order ODE. Therefore, the solution by applying
the method of integrating factor is

Once again, the solution depends on the exponentiation of matrix A, which can be
obtained by diagonalisation or Jordan decomposition.

Although the integral gives the general solution, it can be expressed into a more
convenient form giving the particular solution directly. Let

We then have

Applying I.C.s, we get the constant vector, C, as

So the particular solution can be worked out as

The above particular solution for a non-homogeneous 1st-order system of ODEs with
constant coefficients is known as the Duhamel's principle.

SU1-6
MTH320 Non-Homogeneous System of ODEs & Linear 2nd-Order ODEs

The Duhamel’s principle has an intuitive explanation. It says that the response x(t) of a
system is due to a combination of two effects:

1. The initial ‘disturbance’ to the system (1st RHS term = homogeneous solution).
2. The subsequent ‘disturbances’ due to the external excitation f(s) (2nd RHS term).
The integral represents the cumulative effect of f(s) from t0 to any time t.

The Duhamel’s principle is also known as the variation of parameters.

Example:

Solve the following system of ODEs:

Writing in matrix form

Firstly, we need the eigenvalues and eigenvectors. Standard operations give

Next, we compute eAt

Finally, we compute the Duhamel's principle in two parts. The 1st term representing the
homogeneous solution is

SU1-7
MTH320 Non-Homogeneous System of ODEs & Linear 2nd-Order ODEs

The 2nd term representing the effect of the external excitation is

Hence, the particular solution is

or

The student should be able to verify that the above solutions satisfy the given system of
ODEs.

Activity 1.1

Solve the following system of ODEs. What is causing the response of the system?

SU1-8
MTH320 Non-Homogeneous System of ODEs & Linear 2nd-Order ODEs

Chapter 2: Linear Second-Order ODEs

2.1 Recap: A Linear ODE

Recall from SU5 of MTH319 that a linear ODE of the nth-order has the general form

If the standalone function of the independent variable f(x) = 0, then the linear ODE is
homogeneous. Otherwise, it is non-homogeneous.

2.2 Second-Order Linear ODE with Constant Coefficients

If all ai(x) = ai and ai = 0 for i ≥ 2, then it is a 2nd-order linear ODE with constant coefficients,
i.e.

Such an ODE occurs in many physical phenomena. For example, during the flight of a
rocket, applying Newton’s 2nd Law on the free-body diagram gives

which is a non-homogeneous 2nd-order linear ODE.

Another famous example giving a 2nd-order linear ODE is the suspension problem.
Applying Newton’s 2nd Law on mass m gives

SU1-9
MTH320 Non-Homogeneous System of ODEs & Linear 2nd-Order ODEs

Source: http://www.macksprings.com/suspension-systems/

To solve such a 2nd-order linear ODE, we can employ a ‘divide and conquer’ approach
by firstly decomposing it into a system of 1st-order ODEs, and then using the solution
methods already covered. For example, letting v = y’, we have

Rearranging and writing in standard form

which is a non-homogeneous 1st-order system of linear ODEs that can be solved by

the Duhamel’s principle. If f(t) = 0, then we have a homogeneous system solved by the
separation of variables.

For higher order linear ODEs (even systems), the same ‘divide and conquer’ approach
works. For example, for a 3rd-order linear ODE, letting v = y’ and a = v’ = y’’ gives

Rearranging and writing in standard form

SU1-10
MTH320 Non-Homogeneous System of ODEs & Linear 2nd-Order ODEs

which is again a non-homogeneous 1st-order linear ODE system that can be solved by the
Duhamel’s principle. We shall only focus on solving 2nd-order ODEs in this study unit.

2.3 Solutions of a Homogeneous 2nd-Order Linear ODE

Firstly, let’s evaluate the solution for the homogeneous case

Then, eigenvalues of matrix A are solved from the characteristic polynomial

The equation on the right is also called the auxiliary equation, where its roots give the
eigenvalues of the system. Recall from MTH319 SU6 that the general solution depends on
the type of eigenvalues, exemplified as follows.

Real and distinct eigenvalues

For real and distinct eigenvalues λ1 and λ2, we get the eigenvectors as

So the general solution is

Hence, for real and distinct eigenvalues λ1 and λ2, the general solution for the
homogeneous 2nd-order ODE is

SU1-11
MTH320 Non-Homogeneous System of ODEs & Linear 2nd-Order ODEs

where the constants c1 and c2 can be found by applying initial conditions for y(0) and y’(0).

Complex eigenvalues

For complex eigenvalues, λ1,2 = ɑ ± iβ, the general solution can be further evaluated by
substituting Euler’s formula: ,

Real and repeated eigenvalues

Lastly, for real and repeated eigenvalues: λ1 = λ2 = λ, the general solution is

We shall leave the student to derive the above using the Jordan normal form.

The constants in the above general solutions are to be found from initial conditions of y(0)
and y’(0). In addition, notice that it is not necessary to evaluate eigenvectors when using
the general solutions.

Example:

For the suspension system shown, by taking reference from a static equilibrium position,
derive and solve for the displacement response for the 3 different values of damping.
Initial conditions are y(0) = 0, y’(0) = 1. Plot the various responses and analyse.

SU1-12
MTH320 Non-Homogeneous System of ODEs & Linear 2nd-Order ODEs

Firstly, taking reference from a position z = 0 where the spring is unloaded, applying
Newton’s 2nd Law gives

So z(t) gives the function of the mass displacement from an unloaded-spring position.
However, at static equilibrium where the spring is compressed by z = L, we have

Substituting into the ODE for z(t) and rearranging, we get

Notice that z - L represents the position of the mass taking reference from the static
equilibrium position. So let

SU1-13
MTH320 Non-Homogeneous System of ODEs & Linear 2nd-Order ODEs

Substitution of the above into the ODE gives

which is a homogeneous linear 2nd-order ODE.

Considering case 1, the ODE becomes

Solving for the eigenvalues from the auxiliary equation

So eigenvalues are complex. Hence, the general solution is

where the subscript for y1(t) represents case 1 here. Applying the initial conditions

Therefore, the particular solution for case 1 is

Repeating the above solution procedure for cases 2 and 3 (see Activity 1.2), we respectively
obtain

SU1-14
MTH320 Non-Homogeneous System of ODEs & Linear 2nd-Order ODEs

We plot the displacement functions for various damping constants and obtain

Desmos graph: https://www.desmos.com/calculator/kyhy8vnjze

From the graph, we observe the following:

a. All displacements begin at y = 0 with a positive velocity y’, which is in agreement

with the initial conditions.
b. For case 1, when the damping is small at c = 1 N.s/m, the mass exhibits a
vibratory behaviour. This is known as an underdamped suspension system and
happens when c2 < 4mk, resulting in complex eigenvalues and hence sinusoidal
terms in y(t). The student can verify this easily.
c. For case 3, when the damping is large at c = 10 N.s/m, there is no vibration of
the mass. After the initial motion caused by the initial velocity, the displacement
reaches a peak and recovers back to the equilibrium position relatively slowly.
This is known as an overdamped system and happens when c2 > 4mk, giving
real and distinct eigenvalues. No sinusoidal terms exist in the solution.
d. Lastly, for case 2 at c = 6 N.s/m, the motion looks similar to case 3, but the return
to equilibrium is the fastest amongst all cases. This is known as critical damping
when c2 = 4mk, giving repeated eigenvalues. In many engineering applications,
critical damping is desired as it allows the system to return to a rest state in the
shortest time, such as vehicular suspension or a door closing mechanism.

SU1-15
MTH320 Non-Homogeneous System of ODEs & Linear 2nd-Order ODEs

2.4 Solution of a Non-Homogeneous 2nd-Order Linear ODE

Let’s now consider the non-homogeneous case. Recall that after converting the 2nd-order
ODE into a system of 1st-order ODEs, we get

The solution can then be found by the Duhamel’s principle

Although this method is powerful, it can however be tedious, especially in the evaluation
of the integral term. For simple functions of f(t), there is another faster way to be presented
in SU2.

Activity 1.2

Solve the 2nd-order ODEs below subjected to initial conditions y(0) = 0, y’(0) = 1.

a. Case 2: ,
b. Case 3:

SU1-16
MTH320 Non-Homogeneous System of ODEs & Linear 2nd-Order ODEs

Chapter 3: Mathematical Modelling of Resistor-Inductor

Circuits

3.1 Governing Equation from Kirchhoff Laws

In this chapter, we shall apply mathematical modelling to a resistor-inductor (RL) circuit
as shown below, and in the process apply Duhamel's principle to solve the governing
system of ODEs.

Recall from MTH319 SU5 that in order to derive the governing equation of a phenomenon
or system, in many cases we apply laws of conservation. Here, the conservation of charge
and energy, or Kirchhoff’s voltage law, is applied. Note that the voltage change across a
resistor and an inductor are respectively given by

where R is resistance, L is inductance and I is the current passing through the electrical
component.

Applying the voltage law around loop 1: , we get

SU1-17
MTH320 Non-Homogeneous System of ODEs & Linear 2nd-Order ODEs

Repeating for loop 2: ,

By inspection, notice the rate of change in both I1 and I2 depends on both currents
themselves, hence we have a coupled system of ODEs. Writing in matrix notation, we
have

which is a non-homogeneous system of linear ODEs. Suppose the switch is closed at t = 0,

the circuit behaves like an open circuit at t = 0+ due to the current through each inductor
unable to change instantaneously (or else ). Therefore, initial conditions are

3.2 Solution of Current Response

Let’s assume the parameter values given below and solve for the current functions.

Substituting the above into the system of ODEs gives

The solution given by Duhamel’s principle is

SU1-18
MTH320 Non-Homogeneous System of ODEs & Linear 2nd-Order ODEs

which implies that the current responses are entirely caused by the voltage source since
initial conditions are zero.

Standard operations to solve for eigenvalues and eigenvectors give

Next, we compute eAt:

So the particular solution is

Plotting the current functions to analyse

Desmos graph: https://www.desmos.com/calculator/p11mj2vmzw

SU1-19
MTH320 Non-Homogeneous System of ODEs & Linear 2nd-Order ODEs

Inspecting the graph, we make the following observations:

1. Both currents increase from zero expectedly as defined by the initial conditions.
2. After about 5 seconds, I1 and I2 reach steady-state values of 1A and 0A
respectively. This is due to each inductor behaving like a short circuit (a coil of
wire) when there is no change in the current passing through it. So effectively,
the circuit at steady-state (SS) looks like:

Clearly, the SS current I2 is zero since the resistor R2 is being shorted. Then, the
SS current passing through resistor R1 is I1, which is easily calculated by

which is in agreement with our solutions.

3. Although the steady-state current for I2 is zero, there is still a ‘blip’ of current
response I2(t) upon closing of the circuit. This is because during the transient
response, there are changes in each current and therefore inductors L1 and L2
both present an opposition (reactance) to their current flow. Therefore during
this phase, resistor R2 is not shorted and some current flows through it.

SU1-20
MTH320 Non-Homogeneous System of ODEs & Linear 2nd-Order ODEs

Activity 1.3

For the RL circuit analysed above, show that by applying Kirchhoff’s voltage law
around another loop: instead of loop 2 above, one can obtain the
same system of ODEs.

SU1-21
MTH320 Non-Homogeneous System of ODEs & Linear 2nd-Order ODEs

Summary

In Study Unit 1, we covered the following:

• How Non-Homogeneous Systems of ODEs arise from Mathematical Modelling of

Real-World Phenomena & Engineering Systems
• Decomposition of a Higher-Order Linear ODE into a 1st-Order System of Linear
ODEs
• Use of Duhamel’s Principle in the Solution of the Non-Homogeneous System of
ODEs
• Application of the General Solutions of a 2nd-Order Homogeneous Linear ODE on
a Suspension System
• Math Modelling of a Resistor-Inductor Circuit

SU1-22
MTH320 Non-Homogeneous System of ODEs & Linear 2nd-Order ODEs

Formative Assessment

1. Solve the following systems of ODEs:

a.

b.

c.

2. Formulate a system of linear 1st-order ODEs from the higher order ODEs below:
a.
b.

3. Solve the following 2nd-order ODEs:

a.
b.
c.

4. Mathematical Modelling: Resistor-Inductor Circuit

SU1-23
MTH320 Non-Homogeneous System of ODEs & Linear 2nd-Order ODEs

a. Using Kirchhoff laws, derive the system of ODEs governing currents I1(t)
and I2(t) in the RL circuit above.
b. The switch is opened for a long time and closed at t = 0. Given the parameters
below, redefine the ODE system with initial conditions at t = 0.

c. Solve the ODE system and plot the current functions. Analyse.

5. (Optional) Solve the following system for x(t) and y(t). What happens to the functions
as time approaches infinity? (Hint: You will need to use Euler’s formula and
compound angle formula along the way. Warning: Insanely tedious.)

SU1-24
MTH320 Non-Homogeneous System of ODEs & Linear 2nd-Order ODEs

Solutions or Suggested Answers

Activity 1.1

Solve the following system of ODEs. What is causing the response of the system?

From the ODE system, we have

Since the initial conditions are zero, the Duhamel’s principle will be

Recognise that matrix A is already a Jordan matrix with a repeated eigenvalue of 1.

Therefore we can compute

where P in this case will be an identity matrix (to be verified by the student). Evaluating
the Duhamel’s principle

Clearly, the response of the system is entirely caused by the external excitation since there
are zero initial conditions.

SU1-25
MTH320 Non-Homogeneous System of ODEs & Linear 2nd-Order ODEs

Activity 1.2

Solve the 2nd-order ODEs below subjected to initial conditions y(0) = 0, y’(0) = 1.

a. Case 2: ,
b. Case 3:

a. For case 2, solving the eigenvalues from the auxiliary equation

So eigenvalues are real and repeated. Hence the general solution is

Applying initial conditions and solving for constants, we get

So the particular solution is

b. For case 3, solving the eigenvalues from the auxiliary equation

So eigenvalues are real and distinct. Hence the general solution is

Applying initial conditions and solving for constants, we get

SU1-26
MTH320 Non-Homogeneous System of ODEs & Linear 2nd-Order ODEs

So the particular solution is

Activity 1.3

For the RL circuit analysed above, show that by applying Kirchhoff’s voltage law around
another loop: instead of loop 2 above, one can obtain the same
system of ODEs.

Applying Kirchhoff’s voltage law to the new loop: , we get

Since from loop 1 we obtained

combining the last 2 equations yields

SU1-27
MTH320 Non-Homogeneous System of ODEs & Linear 2nd-Order ODEs

which is the same equation derived previously. Hence, the system of ODEs derived is the
same, regardless of which loops are analysed.

Formative Assessment
1. Solve the following systems of ODEs:
a.

b.

c.

Answer:
a.
,

b.
,

c.

2. Formulate a system of linear 1st-order ODEs from the higher order ODEs below:
a.
b.

Answer:

a.
,

b.

SU1-28
MTH320 Non-Homogeneous System of ODEs & Linear 2nd-Order ODEs

3. Solve the following 2nd-order ODEs:

a.
b.
c.

Answer:

a. ,
b. ,
c.

4. Mathematical Modelling: Resistor-Inductor Circuit

a. Using Kirchhoff laws, derive the system of ODEs governing currents I1(t)
and I2(t) in the RL circuit above.
b. The switch is opened for a long time and closed at t = 0. Given the parameters
below, redefine the ODE system with initial conditions at t = 0.

c. Solve the ODE system and plot the current functions. Analyse.

Answer:

SU1-29
MTH320 Non-Homogeneous System of ODEs & Linear 2nd-Order ODEs

a.
,

b.
,

c.

5. (Optional) Solve the following system for x(t) and y(t). What happens to the functions
as time approaches infinity? (Hint: You will need to use Euler’s formula and
compound angle formula along the way. Warning: Insanely tedious.)

Answer:

SU1-30
 2
Study
Unit

Method of Undetermined
Coefficients & Laplace Transform
MTH320 Method of Undetermined Coefficients & Laplace Transform

Learning Outcomes

By the end of this unit, you should be able to:

• Use the Method of Undetermined Coefficients in Solving Non-Homogeneous

Linear ODEs
• Appreciate that Linear ODEs with Non-Smooth Excitations require another
Solution Method
• Evaluate the Laplace Transform for Elementary Functions
• Evaluate the Laplace Transform for the Delta and Unit-step Functions
• Apply the Properties of Laplace Transform

SU2-2
MTH320 Method of Undetermined Coefficients & Laplace Transform

Overview

This study unit introduces the method of undetermined coefficients as another way to
solve non-homogeneous linear ODEs. Then, due to the difficulty in solving such ODEs
when the external excitation function is not smooth, the Laplace transform is introduced.
The Laplace transforms of elementary functions and the delta and unit-step functions
are evaluated. We also learn how piecewise functions can be rewritten using the unit-
step function. Finally, the properties of the Laplace transform and how they can be
incorporated to evaluate the Laplace transforms of some functions are shown.

SU2-3
MTH320 Method of Undetermined Coefficients & Laplace Transform

Chapter 1: Method of Undetermined Coefficients

1.1 The Complete Solution

From SU1, we understand that although the Duhamel's principle offers a solution to non-
homogeneous linear ODEs, the process can however be laborious especially if the integral
is not straightforward to evaluate. For simple excitation functions for f(t), there is a simpler
way, known as the method of undetermined coefficients.

Before diving into the faster method, we need to firstly understand the concept of a
complete solution. Notice that

The solution to the second ODE is therefore composed of 2 parts - a homogeneous solution
yh(t) that solves the RHS = 0, and a non-homogeneous solution yp(t) that solves the RHS
= f(t). We have

Therefore, adding up the two ODEs above implies that the complete solution is

Notice that the Duhamel’s principle is also comprised of the sum of Yh(t) and Yp(t)

Now, the logical question to ask is: Why do we need a complete solution? Isn’t yp(t) enough
to solve the non-homogeneous ODE?

SU2-4
MTH320 Method of Undetermined Coefficients & Laplace Transform

The reason is: Although yp(t) is indeed a solution, notice that it did not account for the
initial conditions, which will certainly affect the response of the phenomenon. Clearly,
different initial conditions will result in different solution curves.

Therefore, a complete solution to a linear ODE will be the sum of a homogeneous solution
yh(t) and a particular integral yp(t). Here, the homogeneous solution yh(t) is also known
as the complementary function.

1.2 Method of Undetermined Coefficients

The method of undetermined coefficients is a ‘trial and error’ method. For the non-
homogeneous 2nd-order linear ODE

We can choose yp(t) according to the reference to f(t) below, after which constants (P, Q,
etc.) can be found by substitution into the ODE and comparing coefficients.

Example:

Solve the ODE governing the suspension system below with initial conditions y(0) = 1,
y’(0) = 0.

SU2-5
MTH320 Method of Undetermined Coefficients & Laplace Transform

Firstly, we solve the complementary function from the homogeneous part of the ODE

Referring to Activity 1.2 Answer, we get the complementary function as

Now, for a guess of the particular integral, since f(t) = 2cos(t), we choose

Differentiating the above and substituting into the ODE

By comparing coefficients of sine and cosine terms, we get

So the particular integral is

SU2-6
MTH320 Method of Undetermined Coefficients & Laplace Transform

Hence we have the general solution

Applying initial conditions y(0) = 1, y’(0) = 0 to solve for the constants

Therefore, the particular solution is

One point to take note is that the initial conditions are applied ONLY after yp(t) has been
evaluated! For this reason, DO NOT assume yh(t) = 0 even when initial conditions are all
zero.

1.3 Linear Independence of Solutions

Consider the previous example again but with another excitation f(t) as shown below:

SU2-7
MTH320 Method of Undetermined Coefficients & Laplace Transform

Solving again using the method of undetermined coefficients, we notice something weird

So we are unable to solve for the particular integral!

The above happens because yp(t) is similar to the first term of yh(t), which solves the
homogeneous problem, but not the non-homogeneous part. This brings us to the concept
of linear independence of solutions.

The complete solution to an ODE must be comprised of linearly independent solutions,

which means

ONLY when ALL ki = 0. In the previous example, this is not the case. Notice that

when c1= -P (k1= -k3) and c2= 0 (k2= 0). Hence, our solution failed.

Fortunately, there is a simple workaround to get a complete solution that is linearly

independent.

When the trial solution yp(t) fails because it solves the homogeneous problem, simply
multiply yp(t) by t and try solving again. If it fails again, multiply by t again. Rinse and
repeat.

SU2-8
MTH320 Method of Undetermined Coefficients & Laplace Transform

For example, from the previous problem

Generally, if the external excitation f(t) consists of a similar function to any term in the
homogeneous solution, then multiply tn to that function where n is the smallest integer
that makes the complete solution linearly independent.

Example:

For the critically damped suspension system now subjected to f(t), obtain its particular
solution.

From Activity 1.2 Answer, the complementary function is

So we choose the particular integral to be

SU2-9
MTH320 Method of Undetermined Coefficients & Laplace Transform

Notice that t2 is multiplied in such that yp(t) is linearly independent from yh(t).
Differentiating the above and substituting into the ODE

Hence, the particular integral is

And the general solution is

After applying initial conditions, we obtain the particular solution as

Activity 2.1

For the rocket problem shown below, solve for the displacement function y(t) after
blast off from the ground. Let m = 10 kg, c = 10 N.s/m, g ≈ 10 m/s2, P(t) = 1000e-t N.
Plot and analyse.

SU2-10
MTH320 Method of Undetermined Coefficients & Laplace Transform

Chapter 2: Laplace Transform

2.1 Recap: Solving Non-Homogeneous Linear ODEs

So far, we have learnt two methods of solving non-homogeneous linear ODEs, such as a
2nd-order ODE given by

One method is decomposing into a 1st-order system and using the Duhamel’s principle

And another is by the method of undetermined coefficients presented in the last chapter.

While these two methods have their pros and cons, one finds difficulty in the solution
process when the external excitation, f(t), is not a smooth function, such as the unit-step
function, delta function, rectangular wave or triangular wave functions, etc., as shown
below.

The real problem is, these non-smooth excitations are not uncommon, especially in the
field of electronic circuitry. So, in this topic, we will learn a third method of solving ODEs
that can handle such functions easily, the Laplace transform.

2.2 The Laplace Transform

The Laplace transform is simply an integration (aka integral transform) defined by

SU2-11
MTH320 Method of Undetermined Coefficients & Laplace Transform

where the input is a function of a real variable (usually time), and the output is another
function of a complex variable, s = σ + iω.

For the Laplace transform of a function f(t) to exist, the improper integral must converge.
For example, the LT of f(t) = k (constant) is

The usefulness of the Laplace transform lies in its ability to transform a differential
equation into an algebraic equation (which is usually easier to solve). So, the plan is

Let’s learn the basics of the Laplace transform before applying it to solve ODEs.

2.3 Laplace Transform of Elementary Functions

The Laplace transforms of f(t) = t, t2, t3 and tn respectively are

SU2-12
MTH320 Method of Undetermined Coefficients & Laplace Transform

The Laplace transform of f(t) = sin(at) is (integration by parts twice)

Using the same technique, the Laplace transform of f(t) = cos(at) can be evaluated as

The Laplace transform of f(t) = eat is

Example:

State the Laplace transforms of the following functions:

Using the above derived Laplace transforms, we simply write

SU2-13
MTH320 Method of Undetermined Coefficients & Laplace Transform

2.4 Laplace Transform of the Delta & Unit-Step Functions

The Delta Function

As shown in the figure below, the Dirac delta function δ(t-a) is defined as one that is zero
everywhere except at t = c, where it is infinitely large. Also known as the unit-impulse,
the delta function has an area = 1. The delta function is used to model excitations such as
impact forces and voltage spikes, etc.

The Laplace transforms of f(t) = δ(t-a) and f(t) = g(t)δ(t-a) respectively are

The Unit-Step Function

The unit-step function as shown below has output values of u(t-a) = 1 for t > a and zero
otherwise. It is commonly used to model the ‘switching on/off’ of voltage or current
supply in electronic circuits. This function is also known as the Heaviside function.

SU2-14
MTH320 Method of Undetermined Coefficients & Laplace Transform

The Laplace transform of the unit-step function can be evaluated as

Activity 2.2

Sketch the graph of the function below and evaluate its Laplace transform by
integration:

SU2-15
MTH320 Method of Undetermined Coefficients & Laplace Transform

Chapter 3: Properties of the Laplace Transform

3.1 Linearity of the Laplace Transform

Since the Laplace transform is an integration, it is therefore linear, i.e.

Example:

Using the linearity property, evaluate the Laplace transform of the following function:

By inspection, the rectangular pulse can be described by the function

Applying the linearity property, the Laplace transform can be evaluated as

Revisit Activity 2.2 and reconcile with the answer above.

SU2-16
MTH320 Method of Undetermined Coefficients & Laplace Transform

3.2 Shifting Properties of the Laplace Transform

Shifting in the s-domain

When a function f(t) is multiplied by eat, its Laplace transform can be evaluated as

This property of Laplace transform is called shifting in the s-domain.

Example:

Evaluate the Laplace transform of the function below by integration and verify the above
property:

Using integration by parts, we have

Using the s-shifting property, we directly get

which agrees with the Laplace transform given by integration.

Shifting in the time (t)-domain

When a function f(t-a) is multiplied by the unit-step function u(t-a), its Laplace transform
can be evaluated as

SU2-17
MTH320 Method of Undetermined Coefficients & Laplace Transform

Let τ = t - a, so dτ = dt, the above integral becomes

This property of Laplace transform is called shifting in the time (t)-domain.

Example:

Evaluate the Laplace transform of the function below by using the time-shifting property:

In order to use the t-shifting property, the t2 term needs to be rewritten in terms of (t - 2).
Therefore we rewrite f(t) as

Notice that all terms are now of the form g(t-a)u(t-a). Hence, by linearity and the t-shifting
property,

3.3 Rewriting Piecewise Functions using u(t-a)

A piecewise function can be rewritten into a single function using the unit-step function.
Generally, we can deduce

SU2-18
MTH320 Method of Undetermined Coefficients & Laplace Transform

Example:

Rewrite f(t) using the unit-step function and evaluate its Laplace transform:

Rewriting f(t) as a single function

To obtain the Laplace transform, the t-shifting property is needed, so terms must be in the
form of g(t-a)u(t-a). Therefore,

Hence the Laplace transform of f(t) is

SU2-19
MTH320 Method of Undetermined Coefficients & Laplace Transform

3.4 Derivative of the Laplace Transform

When we differentiate a Laplace-transformed function F(s), we notice that

Therefore, when a function f(t) is multiplied by t, its Laplace transform is

Further differentiating F(s) reveals that

Example:

Evaluate the Laplace transform of the following function.

By inspection, we have f(t) = e-tsin(3t), so by the s-shifting property, we get

Then, since the power of t is 1, the Laplace transform of g(t) is

SU2-20
MTH320 Method of Undetermined Coefficients & Laplace Transform

3.5 Table of Laplace Transforms

Consolidating previous results, we create a table of Laplace transforms for easy reference:

Activity 2.3

Evaluate the Laplace transform of

SU2-21
MTH320 Method of Undetermined Coefficients & Laplace Transform

Summary

In Study Unit 2, we covered the following:

• Method of Undetermined Coefficients in Solving Non-Homogeneous Linear ODEs

• Why Linear ODEs with Non-Smooth Excitations require another Solution Method
• Definition of the Laplace Transform
• Evaluation of Laplace Transforms of Elementary Functions and the Delta and Unit-
step Functions
• Application of the Properties of Laplace Transform

SU2-22
MTH320 Method of Undetermined Coefficients & Laplace Transform

Formative Assessment

1. Solve the following 2nd-order ODEs:

a.
b.

2. Mathematical Modelling: Lift Safety Engineering

A safety engineer in Toshiba Lift Systems is designing a spring-damper shock

absorber system to cushion the impact of a falling lift cabin in case of a cable snap.

a. Assuming that the lift cabin impacts the shock absorber system at time t
= 0, by applying Newton’s 2nd Law on the sprung mass, show that the
governing equation of the mass’s motion can be expressed as

where y = spring compression relative to unloaded position (m)

k = spring constant (N/m)

c = damping constant (N.s/m)

SU2-23
MTH320 Method of Undetermined Coefficients & Laplace Transform

m = mass of lift cabin (= 1000 kg)

g = gravitational acceleration (≈ 10 m/s2)

b. If the coil springs and dampers are selected such that overall, k = 1000 N/
m, c = 2000 N.s/m, obtain the response y(t). Assume that the lift impacts the
suspension system at 25 m/s (90 km/h) at t = 0.
c. Plot the response y(t) and analyse. Do you think the spring and damper
components are selected appropriately for the lift shock absorber system?
Explain.

3. Determine the Laplace transform of the following functions:

a.
b.
c.

d.
e.
f.
g.

4. Using integration, determine the Laplace transform of f(t). Show that it is equivalent
to that obtained by shifting in s-domain as well as that using the derivative of the
Laplace transform.

5. Using both integration and the t-shifting property, determine the Laplace transform
of the following function. Are they equivalent?

SU2-24
MTH320 Method of Undetermined Coefficients & Laplace Transform

6. A periodic function f(t) is defined by the following waveform. Given that g(t)
represents one cycle of f(t) in [0, 2], define g(t) using the unit-step function. Hence,
determine the Laplace transform of f(t).

SU2-25
MTH320 Method of Undetermined Coefficients & Laplace Transform

Solutions or Suggested Answers

Activity 2.1

For the rocket problem shown below, solve for the displacement function y(t) after blast
off from the ground. Let m = 10 kg, c = 10 N.s/m, g ≈ 10 m/s2, P(t) = 1000e-t N. Plot and
analyse.

Substituting in the given parameter values, the ODE becomes

Firstly, we solve for the complementary function from

Solving the eigenvalues from the auxiliary equation

So eigenvalues are real and distinct. Hence the complementary function is

Now, for the particular integral, since there are two different function types on the RHS,
we might choose

SU2-26
MTH320 Method of Undetermined Coefficients & Laplace Transform

But, realise that both terms are not linearly independent to yh(t). Therefore we multiply t
to both terms and get

Differentiating the above and substituting into the ODE

Hence, the general solution is

Let’s interpret the initial conditions. During the rocket blast off from the ground, we have
y(0) = 0 and y’(0) = 0. Solving for the constants

Finally, the particular solution is

The graph is

SU2-27
MTH320 Method of Undetermined Coefficients & Laplace Transform

From the graph, we make the following observations:

1. The rocket blasted off from the ground from rest, in agreement with the initial
conditions. Notice that the slope of the curve at t = 0 is horizontal, giving y’(0) = 0.
2. The rocket reached a max altitude of slightly above 60 m and started to drop back
to the ground. This is because the propulsion force given by P(t) = 1000e-t dies
off eventually after launch and the gravitational force takes over.
3. The downward slope during free fall approaches a straight line. The gradient
of this straight line represents the terminal velocity of the rocket. This happens
when the downward gravitational force (mg) is balanced by the upward air
resistance drag force (cv).
4. The total flight time of the rocket is around 11 seconds.

Activity 2.2

Sketch the graph of the function below and evaluate its Laplace transform by integration:

The graph is

Evaluating the Laplace transform

SU2-28
MTH320 Method of Undetermined Coefficients & Laplace Transform

Activity 2.3

Evaluate the Laplace transform of

The Laplace transform of g(t) = cos(2t) is

In f(t), since the power of t is 2, the Laplace transform of f(t) is

Formative Assessment
1. Solve the following 2nd-order ODEs:
a.
b.
Answer:
a. ,

b.

2. Mathematical Modelling: Lift Safety Engineering

A safety engineer in Toshiba Lift Systems is designing a spring-damper shock

absorber system to cushion the impact of a falling lift cabin in case of a cable snap.

SU2-29
MTH320 Method of Undetermined Coefficients & Laplace Transform

a. Assuming that the lift cabin impacts the shock absorber system at time t
= 0, by applying Newton’s 2nd Law on the sprung mass, show that the
governing equation of the mass’s motion can be expressed as

where y = spring compression relative to unloaded position (m)

k = spring constant (N/m)

c = damping constant (N.s/m)

m = mass of lift cabin (= 1000 kg)

g = gravitational acceleration (≈ 10 m/s2)

b. If the coil springs and dampers are selected such that overall, k = 1000 N/
m, c = 2000 N.s/m, obtain the response y(t). Assume that the lift impacts the
suspension system at 25 m/s (90 km/h) at t = 0.
c. Plot the response y(t) and analyse. Do you think the spring and damper
components are selected appropriately for the lift shock absorber system?
Explain.
Answer:

SU2-30
MTH320 Method of Undetermined Coefficients & Laplace Transform

b. ,
c. No. Shock absorber will bottom out.

3. Determine the Laplace transform of the following functions:

a.
b.
c.

d.
e.
f.
g.

Answer:
a. ,

b. ,

c. ,

d. ,

e. ,
f. ,

g.

4. Using integration, determine the Laplace transform of f(t). Show that it is equivalent
to that obtained by shifting in s-domain as well as that using the derivative of the
Laplace transform.

SU2-31
MTH320 Method of Undetermined Coefficients & Laplace Transform

Answer:

5. Using both integration and the t-shifting property, determine the Laplace transform
of the following function. Are they equivalent?

Answer:

6. A periodic function f(t) is defined by the following waveform. Given that g(t)
represents one cycle of f(t) in [0, 2], define g(t) using the unit-step function. Hence,
determine the Laplace transform of f(t).

Answer:

SU2-32
MTH320 Method of Undetermined Coefficients & Laplace Transform

SU2-33
MTH320 Method of Undetermined Coefficients & Laplace Transform

SU2-34
 3
Study
Unit

Solving ODEs By Laplace

Transform
MTH320 Solving ODEs By Laplace Transform

Learning Outcomes

By the end of this unit, you should be able to:

• Evaluate the Inverse Laplace Transform

• Understand the Laplace Transform of Derivatives
• Appreciate the Strategy of Using Laplace Transform in Solving ODEs
• Apply Laplace Transform to Solve ODEs with Non-Smooth Excitations
• Apply Math Modelling to a Suspension System subjected to Step & Impact
Disturbances

SU3-2
MTH320 Solving ODEs By Laplace Transform

Overview

This study unit introduces the inverse Laplace transform in order to convert functions
in the s-domain back to functions in the time domain. Then, the Laplace transforms
of the derivatives of the functions of time are presented, and implemented together
with the solving of ODEs. Finally, we see how the mathematical model of a suspension
system subjected to non-smooth functions is derived and solved by the method of Laplace
transform.

SU3-3
MTH320 Solving ODEs By Laplace Transform

Chapter 1: Inverse Laplace Transform

1.1 Recap: Table of Laplace Transforms

In the last study unit, for some functions that have their Laplace transforms defined, we
derived the following:

Similar to a table of derivatives or integrals, we can use the above table for quick reference
without performing integration when evaluating the Laplace transforms of common
functions. The same table is also used in evaluating the inverse of the Laplace transform,
described in the next section.

1.2 The Inverse Laplace Transform

The inverse Laplace transform is defined such that

Usually, the function F(s) is manipulated into one of the recognisable forms in the
table of Laplace transforms so that the corresponding f(t) can be deduced. In many

SU3-4
MTH320 Solving ODEs By Laplace Transform

cases, manipulation techniques such as factorisation, completing the square and partial
fractions, etc. will be required in order to evaluate the inverse Laplace transform.

Example:

Determine the inverse Laplace transforms of the following functions:

For F(s), notice that it resembles the form n!/sn+1 in the table. Letting n + 1 = 3 which gives
n = 2, we can manipulate

Therefore, considering linearity, the inverse Laplace transform is

The same approach applies to G(s), i.e.

Lastly, for H(s), it resembles the form n!/sn+1but with s-shifting. Letting n + 1 = 4, the
inverse Laplace transform is

Observe that in the above manipulations, factoring out appropriate constants enable the
inverse Laplace transform to be identified easily from the table. Next, we look at how
completing the square and partial fractions can play a part.

Example:

Determine the inverse Laplace transform of the following functions:

SU3-5
MTH320 Solving ODEs By Laplace Transform

From the Laplace transform table, we observe that there is no F(s) that resembles fractions
with the quadratic denominators in the given functions. A logical step would be to rewrite
the ‘big’ fraction into partial fractions. For P(s), we have

Now, considering linearity, it is clear that

The use of partial fractions works well for denominators that contain real roots. However,
notice that this is not the case for Q(s). In order to avoid dealing with the complex roots
which result in a more laborious operation, we can use completing the square on the
denominator, i.e.

Now, the function resembles the Laplace transform of a sine function with s-shifting.

Therefore

Lastly, for R(s), we can see that it is Q(s) multiplied by an exponential function. This is
shifting in the t-domain. So, the inverse Laplace transform is

SU3-6
MTH320 Solving ODEs By Laplace Transform

Activity 3.1

Evaluate the inverse Laplace transform of the following function:

SU3-7
MTH320 Solving ODEs By Laplace Transform

Chapter 2: Solving ODEs by Laplace Transform

2.1 Laplace Transform of the Derivative

Using integration by parts, we can obtain the Laplace transform of f’(t) as

Similarly, the Laplace transform of the 2nd-order derivative f’’(t) is

The above integration can be performed for any higher-order derivative f(n)(t) to obtain
its Laplace transform.

Updating our table, we have

SU3-8
MTH320 Solving ODEs By Laplace Transform

2.2 Solving ODEs by Laplace Transform

Recall from the last study unit that the motivation of the Laplace transform lies in its ability
to transform a differential equation into an algebraic equation (which is usually easier to
solve). So, the plan is

For example, if given the initial value problem

applying Laplace transform to the ODE and substituting initial conditions give

which is an algebraic equation in s. So we can isolate Y(s)

and then solve for y(t) using the inverse Laplace transform.

Example (see SU2 Section 1.3):

By Laplace transform, solve the IVP of the critically damped suspension system subjected
to f(t). (m = 1 kg, k = 9 N/m, c = 6 N.s/m.)

SU3-9
MTH320 Solving ODEs By Laplace Transform

Laplace transform of the ODE gives

Substituting initial conditions and isolating Y(s),

Applying inverse Laplace transform, we obtain the particular solution as

Reconcile the above solution with that in SU2 Section 1.3.

The method of Laplace transform can also be used to solve for general solutions of ODEs.

Example:

Obtain the general solution of the ODE below, this time by Laplace transform:

SU3-10
MTH320 Solving ODEs By Laplace Transform

Laplace transform of the ODE gives

Isolating Y(s),

By partial fractions on the first RHS term,

Since y0 is just a constant, let C = y0 - ⅖, applying inverse Laplace transform gives the
general solution as

which is the same as that solved by the method of integrating factor.

Let’s now apply Laplace transform to solve ODEs with non-smooth excitations.

Example:

The mass below is initially depressed by 2 m from its equilibrium position and pulled up
with a velocity of 2 m/s. Then, at t = 2 s, a hammer knocks it imparting an impulse of 3 Ns.
Given m = 1 kg, k = 4 N/m, c = 5 N.s/m, define the IVP. Graph the displacement response
of the mass from t = 0 and explain.

SU3-11
MTH320 Solving ODEs By Laplace Transform

The IVP is defined by

Applying Laplace transform to the ODE and standard operations gives

By partial fractions we get

Inverse Laplace transform gives the particular solution as

SU3-12
MTH320 Solving ODEs By Laplace Transform

Graphing the above solution:

Desmos Graph: https://www.desmos.com/calculator/oiynlml1jg

Analysing the above response of the mass, we observe the following:

1. The displacement at t = 0 is 2 m which is in agreement with that specified in the

initial conditions.
2. There is no vibration of the mass which indicates critical or over damping. From
c2 = 25 > 4mk = 16, the system is overdamped which agrees with the motion
observed.
3. After the initial recovery of the mass towards its equilibrium position of y = 0, the
impact excitation caused by the hammer knock displaces the mass again. Notice
the sudden increase in y(t) caused by the delta function.
4. The mass slowly recovers its motion towards its equilibrium position after the
hammer knock, approaching y = 0 in about 7 seconds.

SU3-13
MTH320 Solving ODEs By Laplace Transform

Activity 3.2

By Laplace transform, solve for the displacement response of the underdamped

suspension system subjected to the given initial conditions. (m = 1 kg, k = 5 N/m, c =
4 N.s/m.)

SU3-14
MTH320 Solving ODEs By Laplace Transform

Chapter 3: Mathematical Modelling of Vehicular

Suspension

3.1 Governing Equation of Chassis Displacement

For a (quarter) car suspension system shown below, we shall derive the governing ODE
that models the chassis response y(t) subjected to the input wheel displacement w(t).

Source: Chegg.com - A simplified model of a car suspension system (a quarter)

The free-body diagram of the mass is

where the chassis position y(t) is referenced from the static equilibrium position.

Applying Newton’s 2nd Law to the mass m, we have

SU3-15
MTH320 Solving ODEs By Laplace Transform

which yields a non-homogeneous 2nd-order linear ODE. Observe that the external
excitation to the chassis is composed of terms not only involving the wheel displacement,
but also its velocity.

We now assume that the rolling wheel encounters a step disturbance of height h at time
t = a giving

Differentiating the above to obtain the wheel velocity function

Notice that the derivative of the unit-step function is the delta function, which can be
interpreted by the fundamental theorem of calculus from

Then, assuming that the vehicle is moving steadily on a smooth road before encountering
the step, initial conditions are zero. Hence, the system is governed by the initial value
problem

Notice that the step disturbance imparts not only a unit-step function to the system, but
also an “impact shock” given by the delta function. Does it agree with your intuition?

3.2 Chassis Response to Step Disturbance

We shall now assume a critically damped suspension (usually desired in vehicles) with
parameters m = 100 kg, k = 400 N/m, c = 400 N.s/m. Substitution into the IVP, we get

Laplace transform of the ODE with standard operations yield

SU3-16
MTH320 Solving ODEs By Laplace Transform

Applying partial fractions, we have

Comparing coefficients, the constants can be solved by

So we get

Applying the inverse Laplace transform, we obtain the chassis response function as

Let’s now graph the solution and analyse

Desmos Graph: https://www.desmos.com/calculator/10buqqhhte

SU3-17
MTH320 Solving ODEs By Laplace Transform

Setting the step height at h = 10 cm and the time which the wheel encounters the step at
a = 1s, we observe the following:

1. Before the wheel encounters the step disturbance, there is no vertical motion of
the chassis as expected.
2. Once the wheel hits the step at t = 1s, the chassis experiences a sharp rise in its
displacement. Notice that although the wheel motion is an instantaneous jump as
given by the unit-step function, this sudden disturbance is not fully transmitted
to the chassis. The suspension system absorbs sudden shocks and smooths out
the chassis motion as it’s designed to be.
3. The chassis displacement overshoots the step height after the step disturbance,
partly due to the impact shock given by the delta function transmitted by the
wheel velocity, and partly due to inertia of the mass. Imagine that at t = 1s, the
wheel jumps up and the spring is instantaneously compressed, so the spring
force reacts to push up the mass, hence contributing to the overshoot.
4. After the overshoot, the chassis recovers to the higher ground at 10 cm after the
step. So the steady-state displacement of the chassis is y = 0.1 m as expected.
Since the system is critically damped, no vibration of the mass is observed.

Activity 3.3

The quarter suspension system modelled above is a simplified one where the wheel
is considered as rigid, such as when the tyre pressure is very high. The figure below
shows a more realistic model in which the tyre is deformable and hence behaves like
another spring-damper system. By applying Newton’s 2nd Law to the chassis mc and
the hub assembly mh, derive the governing system of ODEs for this model. You need
not solve.

SU3-18
MTH320 Solving ODEs By Laplace Transform

Source: Chegg.com - A more realistic model of a car suspension system (a

quarter)

SU3-19
MTH320 Solving ODEs By Laplace Transform

Summary

In Study Unit 3, we covered the following:

• Evaluation of the Inverse Laplace Transform

• The Laplace Transform of Derivatives
• The Strategy of Using Laplace Transform in Solving ODEs
• Application of the Laplace Transform to Solve ODEs with Non-Smooth Excitations
• Math Modelling of a Suspension System subjected to Step & Impact Disturbances

SU3-20
MTH320 Solving ODEs By Laplace Transform

SU3-21
MTH320 Solving ODEs By Laplace Transform

Formative Assessment

1. Solve the following IVPs by Laplace transform:

a.
b.
c.
d.
e.
f.

2. A damped mass-spring system has the following equation of motion. The initial
displacement and velocity are 2 m and 4 m/s respectively. Solve for the displacement
response, graph it and explain. What kind of damping does the system exhibit?

3. Mathematical Modelling: Resistor-Inductor Circuit

For the RL circuit shown as follows, derive the IVP governing the current flow I(t),
given that the voltage source E(t) follows the rectangular pulse shown. If E = 12 V, R
= 2 Ω , L = 1 H and t0 = 5 s, solve for I(t) and explain its graph.

4. Mathematical Modelling: Ride Comfort Engineering - Chassis Response to Pothole

SU3-22
MTH320 Solving ODEs By Laplace Transform

The ODE governing the simplified quarter-car suspension has been derived in
Chapter 3 to be

a. The wheel encounters a pothole of depth d cm at t = 1s and exits it at t = 2s.

Assuming a critically damped system with m = 100 kg, k = 900 N/m, c = 600
N.s/m, show that the IVP of this problem is

b. Solve the IVP to obtain the chassis response y(t) to the pothole.
c. If the pothole has a depth of 10 cm, plot the chassis response and explain.
d. Generally, what adjustments can you make to the spring and damper to
‘soften’ the road disturbances to the passengers?

SU3-23
MTH320 Solving ODEs By Laplace Transform

Solutions or Suggested Answers

Activity 3.1

Evaluate the inverse Laplace transform of the following function:

Multiplying out F(s) gives

where the second RHS term can be put into partial fractions as

So F(s) can be rewritten as

The inverse Laplace transform is therefore

Activity 3.2

By Laplace transform, solve for the displacement response of the underdamped

suspension system subjected to the given initial conditions. (m = 1 kg, k = 5 N/m, c = 4
N.s/m.)

SU3-24
MTH320 Solving ODEs By Laplace Transform

Laplace transform of the ODE followed by standard operations give

Since the denominator has complex roots, we complete the square and rewrite

Now, we split the fractions into two parts

which makes it evident that the inverse Laplace transform gives the solution as

Activity 3.3

The quarter suspension system modelled above is a simplified one where the wheel is
considered as rigid, such as when the tyre pressure is very high. The figure below shows
a more realistic model in which the tyre is deformable and hence behaves like another
spring-damper system. By applying Newton’s 2nd Law to the chassis mc and the hub
assembly mh, derive the governing system of ODEs for this model. You need not solve.

SU3-25
MTH320 Solving ODEs By Laplace Transform

Source: Chegg.com - A more realistic model of a car suspension system (a quarter)

Considering Newton’s 3rd Law, the free-body diagram of the two masses is

Applying Newton’s 2nd Law to the individual masses

SU3-26
MTH320 Solving ODEs By Laplace Transform

Notice that the two resultant ODEs are coupled and in 2nd-order. So we decompose them
by letting

and rewrite the ODEs as

So combining the above four 1st-order ODEs into matrix form, we get

which is a nonhomogeneous system of linear ODEs that governs the suspension system.
It can be solved by Duhamel’s principle or Laplace transform.

Formative Assessment
1. Solve the following IVPs by Laplace transform:
a.
b.
c.
d.
e.

SU3-27
MTH320 Solving ODEs By Laplace Transform

f.
Answer:
a. ,
b. ,
c. ,
d. ,
e.
,

f.

2. A damped mass-spring system has the following equation of motion. The initial
displacement and velocity are 2 m and 4 m/s respectively. Solve for the displacement
response, graph it and explain. What kind of damping does the system exhibit?

Answer: , System is overdamped.

3. Mathematical Modelling: Resistor-Inductor Circuit

For the RL circuit shown as follows, derive the IVP governing the current flow I(t),
given that the voltage source E(t) follows the rectangular pulse shown. If E = 12 V, R
= 2 Ω , L = 1 H and t0 = 5 s, solve for I(t) and explain its graph.

Answer:

SU3-28
MTH320 Solving ODEs By Laplace Transform

4. Mathematical Modelling: Ride Comfort Engineering - Chassis Response to Pothole

The ODE governing the simplified quarter-car suspension has been derived in
Chapter 3 to be

a. The wheel encounters a pothole of depth d cm at t = 1s and exits it at t = 2s.

Assuming a critically damped system with m = 100 kg, k = 900 N/m, c = 600
N.s/m, show that the IVP of this problem is

b. Solve the IVP to obtain the chassis response y(t) to the pothole.
c. If the pothole has a depth of 10 cm, plot the chassis response and explain.
d. Generally, what adjustments can you make to the spring and damper to
‘soften’ the road disturbances to the passengers?

Answer:

b.
c. Plot y(t) in (b) and explain the changes in displacement.
d. Reduce the spring stiffness and/or reduce the damping coefficient.

SU3-29
MTH320 Solving ODEs By Laplace Transform

SU3-30
 4
Study
Unit

Fourier Series & Fourier Transform

MTH320 Fourier Series & Fourier Transform

Learning Outcomes

By the end of this unit, you should be able to:

• Evaluate the Fourier Series Representation of a Periodic Function

• Determine the Fourier Periodic Extension & Half-Range Expansion of a Function
• Evaluate the Complex Fourier Series of a Periodic Function
• Evaluate the Complex Fourier Integral of a Non-Periodic Function
• Determine the Fourier Transform of a Temporal or Spatial Signal

SU4-2
MTH320 Fourier Series & Fourier Transform

Overview

This study unit introduces the Fourier series representation of a periodic function. For a
function defined in a finite interval, the use of the periodic extension and/or half-range
expansions is considered. Then, the complex Fourier series representation is derived.
Finally, for a non-periodic function, the Fourier integral representation is evaluated and
the Fourier transform is defined. The effect of the Fourier transform in transforming a
function from the time or space domain to the frequency domain is then demonstrated.

SU4-3
MTH320 Fourier Series & Fourier Transform

Chapter 1: Introduction to Fourier Series

1.1 Recap: Non-smooth Functions

In SU2 and SU3, we introduced the Laplace transform as a way to handle non-smooth
functions in the process of solving differential equations. Examples of such non-smooth
functions are

In many cases, the difficulty in handling non-smooth functions arises because they
are not differentiable at discontinuities and cusps. In this study unit, we shall learn
another method of representing such functions, known as the Fourier series, such that
the differentiability problem is somewhat mitigated. More importantly, the Fourier series
forms a pivotal step in the solution of partial differential equations to be presented in SU5
and SU6.

1.2 The Fourier Series

The Fourier series is a representation of a periodic function (of period 2L) using an infinite
series of sine and cosine functions, each weighted by a coefficient, giving the form

where a0, an and bn are known as the Fourier coefficients. Although it has a sum to
infinity, in practice it is summed to a finite number of terms such that the Fourier series
representation is ‘close enough’ to the original function. For example, the Fourier series
representations (red curves) of a square wave with up to four terms look like

SU4-4
MTH320 Fourier Series & Fourier Transform

Source: https://commons.wikimedia.org/wiki/File:Fourier_Series.svg#/media/
File:Fourier_Series.svg

In order to obtain the Fourier series of a function, one only needs to evaluate the Fourier
coefficients a0, an and bn. The derivation is described as follows:

If we multiply the Fourier series by sin(mπx/L) and integrate over one period, we get

SU4-5
MTH320 Fourier Series & Fourier Transform

Since the integration of an odd function over any symmetric interval is zero, we have the
first two RHS terms as

Now, for the third RHS term, we can show that (in Activity 4.1)

Therefore, when m = n, we obtain

which gives a formula to obtain coefficients bn.

Repeating the above process but this time multiplying the Fourier series by cos(mπx/L)
instead and integrating yields

This time, the third RHS term is clearly zero because the integrand is odd. The first RHS
term can be easily evaluated to give

SU4-6
MTH320 Fourier Series & Fourier Transform

As for the second RHS term, we can show that (similar to Activity 4.1)

Combining the previous three relations, when m = 0, we get

which gives a formula for coefficient a0. When m = n ≠ 0,

which gives the formula for coefficient an. Notice that in the process of obtaining the
coefficients, we made use of several integration results over a symmetric interval. These
are known as the orthogonality relations.

While the above derivation seems tedious, we can summarise to give the complete
definition of the Fourier series as

where

In a nutshell, obtaining a Fourier series representation of a function f(x) mainly consists

of evaluating the Fourier coefficients.

SU4-7
MTH320 Fourier Series & Fourier Transform

Example:

Determine the Fourier series representation of the square wave defined over one period
below and analyse the plots for the first few terms.

Firstly, note that the period of the square wave is 2, so the half-period L = 1. Evaluating
the Fourier coefficients

Now, notice that a0 and an are zero, which is expected because the square wave f(x) is
an odd function, and f(x)cos(nπx) is also odd. So integration over any symmetric interval
gives zero. And

Substituting bn into the Fourier series gives

To plot the Fourier series, we simply expand the summation to the first few terms

SU4-8
MTH320 Fourier Series & Fourier Transform

Plotting in Desmos:

Plots of Fourier series up to 4 sinusoids (upper), 100 sinusoids (lower)

Desmos graph: https://www.desmos.com/calculator/rvr2lljfog

Analysis of the graphs reveals the following observations:

1. As the number of terms in the Fourier series increases, it seems to approximate

the square wave more accurately.
2. In representing an odd function by a Fourier series, there are no cosine or non-
zero constant terms since they are even functions. Their graphs cannot possibly
approximate the odd f(x). This agrees with a0 and an being zero.

SU4-9
MTH320 Fourier Series & Fourier Transform

3. By the same logic, for an even function, there will not be any sine terms giving
all bn = 0. Hence, only functions that are neither even nor odd might have both
sine and cosine terms in their Fourier series.
4. Near the jump discontinuities of the square wave, there seems to be large
oscillations of the Fourier series resulting in poor approximation to f(x). This
observation is known as the Gibbs phenomenon.

1.3 Fourier Sine & Cosine Series

As we observed in the previous example, we can simplify the Fourier series for even and
odd functions into one that contains only either sine or cosine functions. We have the
following:

Fourier Sine Series

For an odd function f(x), we can define a Fourier sine series given by

where

Notice that the original integral for bn is now simplified to integrating over half of the
period. This is because f(x)sin(nπx) is even, so the original area under the curve can be
found by multiplying the area over a symmetric half by two.

Fourier Cosine Series

For an even function f(x), we can define a Fourier cosine series given by

where

SU4-10
MTH320 Fourier Series & Fourier Transform

Again, since f(x) and f(x)cos(nπx) are both even, we only need to integrate over half of the
period to obtain the coefficients.

Activity 4.1

Verify the orthogonality relation

Activity 4.2

Determine the Fourier series representation of the rectangular wave defined over one
period below. Plot to verify the agreement.

Activity 4.3

Using trigonometric relations, determine the Fourier series of .

SU4-11
MTH320 Fourier Series & Fourier Transform

Chapter 2: Periodic Extension & Half-Range Expansions

2.1 Periodic Extension of a Function

For a function that is defined over a symmetric finite interval, we can extend the function
into a periodic one by ‘copying and pasting’ the function to other intervals so that its
Fourier series can be defined. For example, the periodic extension of g(x) defined in (-L,
L) is simply

Graphically, this means

So the Fourier series representation of g(x) in the specified interval is simply the same as
that of the periodic f(x), just with the finite interval defined

where

SU4-12
MTH320 Fourier Series & Fourier Transform

Example:

Determine the Fourier series representation of the function below and plot it:

The periodic extension of g(x) has a period of 2π. Since it is an even function, we can define
a Fourier cosine series for it. Evaluating the coefficients

To evaluate the above integral, we first consider the case if n = 1. We have

For n ≥ 2, we make use of the compound angle formula

such that

or

SU4-13
MTH320 Fourier Series & Fourier Transform

So the Fourier series representation of g(x) is

The plot below shows good agreement between g(x) and its Fourier cosine series (periodic)
for just a small number of terms.

Desmos graph: https://www.desmos.com/calculator/x89xarg323

2.2 Half-Range Expansions of a Function

The half-range expansion is analogous to the periodic extension of a function. For a
function g(x) that is defined in a finite interval (0, L), we can either choose to extend it to
an odd function

SU4-14
MTH320 Fourier Series & Fourier Transform

or to an even function

For example, the odd and even extensions of the function g(x) are illustrated as follows:

After the original function g(x) has been expanded into either an odd or even function,
we can apply the periodic extension (with period 2L) which then gives the Fourier series
representations as:

SU4-15
MTH320 Fourier Series & Fourier Transform

For the odd extension

where

For the even extension

where

Note that although both the odd and even half-range expansions can be used to
approximate the original function, the even expansion is preferred because the periodic
function in the end will be fully continuous. For the odd expansion, jump discontinuities
might occur at the endpoints of the given interval which result in the Gibbs phenomenon,
as encountered in the example in Chapter 1.2.

Example:

Sketch the odd and even extensions of g(x). Evaluate its Fourier half-range expansions

The odd and even extensions of g(x) are plotted as follows:

SU4-16
MTH320 Fourier Series & Fourier Transform

For the odd extension, it is clear that

which is equal to its Fourier sine series. So the Fourier sine series of g(x) is itself.

For the even extension, we have

which is similar to the previous example. Replacing x by 2x, we get Fourier cosine series

Activity 4.4

For the odd extension of g(x) given below, evaluate its half-range Fourier series:

SU4-17
MTH320 Fourier Series & Fourier Transform

Chapter 3: The Fourier Transform

3.1 Complex Fourier Series

The Fourier series developed in Chapter 1 can be represented in a more concise form by
using complex numbers. Recall that the Fourier series representation of a function f(x) is

If we let the angular frequencies ωn = nπ/L, we have

Using Euler’s formula

we can write

noting that i-1 = -i.

Substituting the above into the Fourier series gives

We can define the complex coefficients as

SU4-18
MTH320 Fourier Series & Fourier Transform

thus giving

Letting ω0 = 0, the above becomes

Notice that the three RHS terms can be combined in one summation giving the complex
Fourier series representation

The complex coefficients given by cn can be derived to be

Example:

Determine the complex Fourier series of the square wave defined over one period

Noting that f(x) is an odd function and L = 1, the complex Fourier coefficients are

Since ωn = nπ/L = nπ,

SU4-19
MTH320 Fourier Series & Fourier Transform

Letting n = 2k - 1, the complex Fourier series is

If we cast the complex Fourier series back to the real form, we get

Notice that the final result is the same as that given in Chapter 1.2, which must be the case
since the function f(x) is the same.

3.2 Complex Fourier Integral

So far, we looked at the Fourier series representation for periodic functions with a finite
period of 2L. However, many functions are not periodic, or, of an infinite period. In this
case, instead of a Fourier series, we can derive a Fourier integral representation of the
function as follows.

Notice that the difference between two successive angular frequencies is

From the complex Fourier series, we can write

where

SU4-20
MTH320 Fourier Series & Fourier Transform

If we let

we can write

The summation term on the RHS can be recognised as a Riemann sum with respect to
ωn. Now, if the half-period L approaches infinity, the Riemann sum becomes a Riemann
integral, giving

The above relation is known as the complex Fourier integral for a non-periodic function
f(x), with the complex Fourier coefficient given by

which is also known as the Fourier transform (to be elaborated later).

We can interpret that for a non-periodic function, the Fourier representation is one in
which the sinusoid frequencies (ω) can be any value, instead of only discrete values (ωn=
nπx/L) in the Fourier series for a periodic function.

Example:

Given that the complex Fourier coefficient of a certain function, f(x), is

where δ(ω-a) is the Dirac delta function, determine the function f(x).

SU4-21
MTH320 Fourier Series & Fourier Transform

Since the Fourier coefficient is known, the function f(x) can be evaluated from the Fourier
integral

3.3 The Fourier Transform

The Fourier transform of a function is traditionally denoted by and has been derived in
the preceding section to be

The Fourier transform has the effect of decomposing a function of space or time into its
constituent frequencies with their corresponding amplitudes, as exemplified later. The

inverse Fourier transform of is the complex Fourier integral derived in the preceding
section

Together, the last two transforms enable transformations between functions in the space
(or time) domain and those in the frequency domain. Schematically, we have

In the last example, we observe the Fourier transform pair

SU4-22
MTH320 Fourier Series & Fourier Transform

Now, suppose we have a pure tone with an angular frequency of 300 rad/s (approx. 48
Hz). In the time domain, the sound signal can be described by the function of time

The Fourier transform of this pure tone is

Graphing both functions

Notice that the Fourier transform of the pure tone reveals its angular frequency of 300
rad/s. For a real-valued signal, the negative frequency can be ignored.

Extending the above demonstration, if a signal is comprised of a sum of pure tones, say

then, its Fourier transform is

If we let ω1 = 300 rad/s and ω2 = 700 rad/s, then the plots are

SU4-23
MTH320 Fourier Series & Fourier Transform

From the above, we observe that the Fourier transform has the ability to decompose a
temporal (or spatial) signal into its constituent frequencies. Because of this, the Fourier
transform is used widely in the area of signal processing. We shall leave the student to
explore more about the Fourier transform as a self-learning opportunity.

Activity 4.5

Determine the Fourier transform of the rectangular pulse shown below. Then, write
down its complex Fourier integral representation.

SU4-24
MTH320 Fourier Series & Fourier Transform

Summary

In Study Unit 4, we covered the following:

• How to Evaluate the Fourier Series Representations of a Periodic Function

• Periodic Extensions & Half-Range Expansions of a Function
• How to Evaluate the Complex Fourier Series of a Periodic Function
• The Complex Fourier Integral of a Non-Periodic Function
• The Fourier Transform of a Temporal or Spatial Signal and its Effects

Watch the following videos for more appreciation of the Fourier series and the Fourier
transform:

• https://youtu.be/r6sGWTCMz2k
• https://youtu.be/spUNpyF58BY

SU4-25
MTH320 Fourier Series & Fourier Transform

Formative Assessment

1. Determine the Fourier series representation of the following functions. Plot to verify
your answers.

a.
b.
c.
d.

2. Without evaluating the Fourier coefficients, determine the Fourier series

representation of the following functions. (Hint: Use trigonometric identities.)
a.
b.
c.

d.

3. Prove the orthogonality relation

4. Determine the half-range Fourier expansions of the sawtooth pulse given below. Plot
to verify your answers.

5. Evaluate the complex Fourier series representation of

6. For the piecewise-defined function

SU4-26
MTH320 Fourier Series & Fourier Transform

a. Evaluate its Fourier transform.

b. Hence, state its complex Fourier integral representation.

SU4-27
MTH320 Fourier Series & Fourier Transform

Solutions or Suggested Answers

Activity 4.1

Verify the orthogonality relation

For m ≠ n, we make use of the compound angle formula

which allows us to evaluate the integral

And for m = n, we use the double-angle formula and the integral is

Combining the above results, we have

Activity 4.2

Determine the Fourier series representation of the rectangular wave defined over one
period below. Plot to verify the agreement.

SU4-28
MTH320 Fourier Series & Fourier Transform

Notice that the rectangular wave is neither an even nor odd function, so we should
evaluate all coefficients. Noting that the half-period L = 2, we have

Substituting the above into the Fourier series representation of f(t), we obtain

The plot below shows the agreement between the rectangular wave and its Fourier series
for 50 sinusoids.

Desmos graph: https://www.desmos.com/calculator/ulsoagugfc

SU4-29
MTH320 Fourier Series & Fourier Transform

Activity 4.3

Using trigonometric relations, determine the Fourier series of

Using the half-angle formula, the Fourier series of g(x) is

which is a cosine series since g(x) is an even function (see Chapter 2.1).

Activity 4.4

For the odd extension of g(x) given below, evaluate its half-range Fourier series:

For an odd extension, we have a half-range sine series. The coefficient is

So the Fourier sine series of g(x) is

Activity 4.5

Determine the Fourier transform of the rectangular pulse shown below. Then, write down
its complex Fourier integral representation.

SU4-30
MTH320 Fourier Series & Fourier Transform

The rectangular pulse can be defined as the piecewise function

The Fourier transform is

The complex Fourier integral of f(t), or the inverse Fourier transform of is

Formative Assessment
1. Determine the Fourier series representation of the following functions. Plot to verify
your answers.

a.
b.
c.
d.

Answer:

a.
,

SU4-31
MTH320 Fourier Series & Fourier Transform

b.
,

c.
,

d.

2. Without evaluating the Fourier coefficients, determine the Fourier series

representation of the following functions. (Hint: Use trigonometric identities.)
a.
b.
c.

d.

Answer:
a. ,
b. ,

c. ,

d.

3. Prove the orthogonality relation

Student is to derive the equation/s stated in this question.

4. Determine the half-range Fourier expansions of the sawtooth pulse given below. Plot
to verify your answers.

SU4-32
MTH320 Fourier Series & Fourier Transform

Answer:

5. Evaluate the complex Fourier series representation of

Answer:

6. For the piecewise-defined function

a. Evaluate its Fourier transform.

b. Hence, state its complex Fourier integral representation.
Answer:
a. ,

b.

SU4-33
MTH320 Fourier Series & Fourier Transform

SU4-34
 5
Study
Unit

Partial Differential Equations &

Heat Equation
MTH320 Partial Differential Equations & Heat Equation

Learning Outcomes

By the end of this unit, you should be able to:

• Evaluate Partial Derivatives of Multivariable Functions

• Determine if a Function is a Solution to a Partial Differential Equation
• Derive the Heat Equation
• Solve the Heat Equation for its Steady-State Solution
• Solve the Heat Equation for its Transient Solution
• Determine the Complete Solution to the Heat Equation

SU5-2
MTH320 Partial Differential Equations & Heat Equation

Overview

This study unit introduces partial differential equations (PDEs) with a focus on the heat
equation. The meaning of a solution to a PDE is defined and the evaluation of the partial
derivatives of a multivariable function is demonstrated. The heat equation is then derived
from the law of conservation of heat energy. The solution to the heat equation is shown to
be composed of a steady-state solution and a transient solution. In solving the transient
heat equation, the methods of separation of variables and the Fourier series are employed.
Several analyses of the temperature solutions are presented.

SU5-3
MTH320 Partial Differential Equations & Heat Equation

Chapter 1: Partial Differential Equations

1.1 Recap: Partial Derivatives

For a function of more than one independent variable, also known as a multivariable
function, one can define the derivative of the function with respect to any one independent
variable at a time. For example, for a function f(x,y), we have

which represents the partial derivative of f(x,y) with respect to x. Notice that the
independent variable, y, is being held constant during this partial differentiation.

Similarly, the partial derivative of f(x,y) with respect to y is

in which the independent variable, x, is being held constant this time.

Therefore, a partial derivative describes the rate of change of a multivariable function

with respect to one of its independent variables. A popular shorthand notation of writing
partial derivatives is

In the process of performing a partial differentiation, the usual rules such as sum/
difference, product, quotient and chain rules, etc. still apply.

Example:

Evaluate the partial derivatives of the function

Partial differentiation with respect to x gives

SU5-4
MTH320 Partial Differential Equations & Heat Equation

Notice that since y is treated as a constant, a function of only y is also a constant which
gives a derivative of zero.

Partial differentiation with respect to y gives

in which x is treated as a constant and the product rule has been used when differentiating
the second RHS term with respect to y.

1.2 Recap: Partial Differential Equations (PDEs)

In SU5 of MTH319, we first introduced the partial differential equation (PDE) in which
the unknown function is dependent on more than one independent variable. For example,
the heat conduction model along the length of a rod has the PDE

The solution to the PDE above is the temperature function, T(x,t), where x & t are two
independent variables. This means temperature changes along the length of the rod
and also varies with time. The LHS represents the partial derivative of the temperature
function with respect to time and the 2nd-order derivative on the RHS represents the
partial derivative of the temperature function with respect to position. Using a shorthand
notation, we can also write

SU5-5
MTH320 Partial Differential Equations & Heat Equation

Similar to ordinary differential equations, PDEs arise naturally from the mathematical
modelling of physical phenomena as well as engineering systems. Some of the famous
PDEs and the phenomena they govern are given below:

Heat Equation in 3 dimensions - Describes heat flow and particle diffusion in matter.

Wave Equation - Describes wave motion such as in water, vibrating strings, sound and
light, etc.

Laplace Equation - Describes steady-state equilibrium situations such as in the heat and
wave equations.

Navier-Stokes Equation - Describes motion of fluid particles in viscous flow.

1.3 Solution to a PDE

The solution to a PDE is the unknown multivariable function that satisfies the equality
between both sides of the PDE. For example,

SU5-6
MTH320 Partial Differential Equations & Heat Equation

To verify the solution

substitute the partial derivatives into the PDE

The solution methods of some PDEs will be the focus of the final two study units in
MTH320.

Activity 5.1

Evaluate the first and second order partial derivatives of

Activity 5.2

Show that the function

is a solution to the one-dimensional wave equation

SU5-7
MTH320 Partial Differential Equations & Heat Equation

Chapter 2: The Heat Equation

2.1 Derivation of the Heat Equation

The heat equation can be derived from the conservation of heat energy. For ease of
illustration, we shall consider heat conduction in one dimension only, but the same
principle can be applied to two or three dimensions. The mathematical modelling process
is as follows.

Suppose we have a rod shown below:

By applying the conservation of heat energy to the rod segment of mass, m, specific heat
capacity, c, and length, Δx, we have

in which q represents the rate of heat flow in/out of the rod segment. Fourier’s law of heat
conduction states

which means the rate of heat energy flow is proportional to the temperature gradient and
always in the direction of decreasing temperature, in agreement with our experience and/
or intuition (heat flows from hot to cold). Assuming a uniform thermal conductivity, k,
and a constant cross-sectional area, A, along the rod, with homogeneous density, ρ, the
conservation of energy can be rewritten as

SU5-8
MTH320 Partial Differential Equations & Heat Equation

in which α = k/(ρc) is known as the thermal diffusivity. If we let Δx approach zero, taking
limits on both sides of the equation yields

which is the one-dimensional heat equation introduced at the beginning.

In a similar approach, when the conservation of heat energy is applied to three spatial
dimensions, one obtains

or using the Laplacian, ∇2 = ∇.∇,

2.2 Steady-State Solution

Before the solution method of the heat equation is presented, it will be useful to solve the
steady-state heat equation first. The term ‘steady-state’ means a state of a system in which
its variables are no longer changing with time. For example, the steady-state position of a
swinging pendulum is vertically below the pivot point, after oscillations have diminished
due to friction and air-resistance.

For our heat conduction problem, the steady-state condition means that the temperature
is no longer changing with time, giving

SU5-9
MTH320 Partial Differential Equations & Heat Equation

Hence, for the one-dimensional heat equation derived in the preceding section, the steady-
state solution is a temperature function of only position, Ts(x), which must satisfy

in which the subscript, s, denotes the steady-state condition.

One can also logically deduce that after a long enough time, if the rod is held under
the same external conditions, the temperature function will approach the steady-state
temperature function, meaning

Solving the steady-state ODE is straight forward. By direct integration twice, we obtain
the general solution

For a rod of length L, boundary conditions can be given at the ends of the rod

which means each rod end is being held at a constant temperature. Solving for constants
C and D

Therefore, the steady-state temperature function for a rod with fixed end temperatures is

Assuming the rod is 2 m long and the end temperatures are

plotting the steady-state temperature function in Desmos shows:

SU5-10
MTH320 Partial Differential Equations & Heat Equation

Desmos graph: https://www.desmos.com/calculator/uqqdeogzgc

Inspection of the graph gives the following observations:

a. Temperatures at the rod ends agree with the given boundary conditions as
logically expected.
b. The temperature drops from the higher rod-end temperature at x = 0 to the
lower rod-end temperature at x = L, so the heat transfer is from left to right.
The temperature follows a linear profile where the temperature gradient is the
same everywhere along the rod. This is logical since the heat transfer across every
section of the rod must be the same in order to satisfy the conservation of energy.
Since this model does not consider any heat loss or gain along the rod, we can
calculate the rate of heat conduction along the rod from

2.3 Transient Heat Equation

A complete definition of the one-dimensional heat equation involves the partial
differential equation with boundary conditions and an initial condition. For example, a
heat conduction problem can be defined by

SU5-11
MTH320 Partial Differential Equations & Heat Equation

The boundary conditions in the second line are representing fixed end temperatures at all
time t > 0. The initial condition in the third line represents some temperature function f(x)
at t = 0.

A method of solving the above heat equation invokes a ‘divide and conquer’ approach.
The temperature function can be written as

where one can see there are two parts of the solution. The first RHS term represents
a transient part of the solution given by Q(x,t) and the second RHS term represents
the steady-state solution described in the preceding section. The term ‘transient’ is
appropriate here because Q(x,t) eventually diminishes to zero as the temperature function
T(x,t) evolves to reach a steady-state. This means

Since we have already learnt how to solve of the steady-state problem, our focus now will
be solving the transient problem, which is defined by

Differentiating the above gives

noting that Ts’’ = 0 as evaluated in the steady-state problem. Substituting the above into
the heat equation, we get the transient heat equation

And, the boundary conditions for Q(x, t) can also be evaluated from

which are both homogeneous boundary conditions. Similarly, the initial condition is
worked out as

SU5-12
MTH320 Partial Differential Equations & Heat Equation

Consolidating the above, we can state the transient heat conduction problem

Solving the above will yield Q(x,t), the transient temperature function. Combining with
the steady-state solution Ts(x), we have the complete solution for the one-dimensional
heat conduction problem.

The motivation of getting the transient problem is to make use of its homogeneous
boundary conditions, to be exemplified in the following chapter.

Activity 5.3

Determine the steady-state solution of the heat equation below subjected to given
boundary conditions. What do the boundary conditions physically mean? Explain the
solution logically.

SU5-13
MTH320 Partial Differential Equations & Heat Equation

Chapter 3: Transient Solutions of the Heat Equation

3.1 Fixed End Temperatures

We now present the method of solving the transient heat conduction problem with
homogeneous boundary conditions

Firstly, suppose that the solution Q(x,t) can be written in the variable-separable form

If this assumption is not true, then we will fail at obtaining the solution down the road.
So let's attempt.

Differentiating the above gives

Substituting into the PDE and rearranging, we obtain

Now, notice that the LHS and RHS of the last equation depend only on x and t respectively.
Hence, the only way that they can be equal is that both sides must be a constant. So we
write

which yields two ODEs

SU5-14
MTH320 Partial Differential Equations & Heat Equation

where the independent variables have been omitted for simplicity. So the original problem
of solving a PDE is now converted to solving ODEs.

Applying the variable-separable solution to the boundary conditions gives

By inspection, it cannot be Y(t) = 0 since that would mean Q(x,t) = 0. So we should have
X(0) = 0 and X(L) = 0. Hence, the ODE in the spatial variable x is defined by the boundary
value problem (BVP)

We now consider various cases of the constant p in this BVP.

Case 1: p > 0

When p is positive, we can let p = ω2 and write

The auxiliary equation is

which gives the eigenvalues

Recalling Chapter 2.3 of SU1, for real and distinct eigenvalues, the general solution is

Applying the boundary conditions yields

Since L ≠ 0, (eωL - e-ωL) = 0 when ω = 0, which is invalid because p = ω2 > 0. That means

SU5-15
MTH320 Partial Differential Equations & Heat Equation

which is unacceptable because we would have Q(x,t) = 0. So p cannot be positive.

Case 2: p = 0

When p = 0, the ODE simplifies to

which has the general solution

But, applying the zero boundary conditions yields

which is again unacceptable.

Case 3: p < 0

Lastly, when p is negative, we can let p = -ω2 and write

Solving the auxiliary equation, the eigenvalues are ±iω. For the imaginary eigenvalues,
the general solution is

Applying boundary conditions

Since the constant d2 cannot be zero or else we will have X(x) = 0, we get the condition

For each integer value of n, we will have a solution

SU5-16
MTH320 Partial Differential Equations & Heat Equation

in which the original constant d2 has been replaced by cn for each integer value of n.

Now, recall from SU2 Chapter 1.3 that the complete solution to an ODE must comprise all
linearly independent solutions. This means the spatial function is

However, since the sine function is an odd function, we get

which implies the negative integers of n do not add any new functions to those already
given by positive integers of n. And noting that when n = 0, we have X0(x) = 0, we can
discard the sinusoids for n ≤ 0 and write

Finally, now we get to the temporal ODE

which is easily solved by the separation of variables

Since when solving for the spatial function we found that p = -ωn2, the temporal function
for each integer value of n is

Therefore, the transient solution of the heat conduction problem is

SU5-17
MTH320 Partial Differential Equations & Heat Equation

in which constants bn = cnAn. To evaluate bn, we apply the initial condition

Notice that the above relation is a Fourier sine series representation for g(x). Recalling
Chapter 1.3 of SU4, bnare the Fourier coefficients given by

This completes the method of solution for the transient temperature function, Q(x,t).
Combined with the steady-state solution, Ts(x), we obtain the solution to the heat
conduction problem with fixed end temperatures as

Example:

Solve the following heat conduction problem for a 2-meter steel rod. If the thermal
diffusivity of steel is α = 1.2 X 10-5 m2/s, plot the temperature function at various times to
demonstrate its time progression. Explain your observations.

Firstly, we can write down the steady-state solution of the above problem as

Therefore, the corresponding transient heat conduction problem is

SU5-18
MTH320 Partial Differential Equations & Heat Equation

The solution is

Applying the initial condition gives

The Fourier coefficients bn can be evaluated by integration, but by inspection we can

deduce

such that

Therefore, the complete temperature solution to the heat conduction problem is

Plotting in Desmos:

Desmos graph of T(°C) vs x(m): https://www.desmos.com/calculator/o75xiayzos

(Play the button for t to see the time progression of the temperature distribution.)

SU5-19
MTH320 Partial Differential Equations & Heat Equation

Temperature distribution T(°C) vs x(m) along the rod at various time t (seconds)

SU5-20
MTH320 Partial Differential Equations & Heat Equation

Inspection of the graphs reveals the following observations:

a. At t = 0, the initial temperature distribution agrees with T(x,0) given.

b. As time progresses, the temperature distribution along the rod smooths out to
the steady-state temperature Ts(x).
c. The boundary conditions of fixed end temperatures are enforced for t > 0 as
expected.
d. The temperature variation evens out at a faster rate at earlier times due to larger
curvatures of the temperature distribution. As time progresses, the rate of change
of temperature diminishes due to smaller temperature curvatures.

3.2 Insulated Ends

We shall now look at another scenario of the heat conduction problem, this time in a rod
with both ends being insulated. The heat equation is then defined by

In this problem, observe that the boundary conditions are already homogeneous, so there
is no need to split the problem into steady-state and transient parts. We can directly solve
for the temperature function T(x,t).

Once again, we assume that the solution can be written in the separable form

Differentiating the above followed by standard operations gives the ODEs:

Again, considering the 3 cases of the constant p, we will get unacceptable solutions for p
≥ 0. Therefore we let p = -ω2 < 0 and write

SU5-21
MTH320 Partial Differential Equations & Heat Equation

The general solution is therefore

Differentiating the above, we get

Applying the insulated boundary conditions

Since ω ≠ 0 and constant d1 cannot be zero or else we will have X(x) = 0, we must have

For each integer value of n, we will have a solution

Again, the complete solution to an ODE must be composed of all linearly independent
solutions. This means the spatial function is

However, since the cosine function is an even function, we get

which implies the negative integers of n do not add any new functions to those already
given by positive integers of n. And noting that for n = 0, we have ω0 = 0 and X0(x) = c0,
we can discard the sinusoids for negative integers of n and write

SU5-22
MTH320 Partial Differential Equations & Heat Equation

Now, for the temporal ODE

The general solution has been derived previously to be

Therefore, the solution of the heat conduction problem with insulated ends is

in which constants a0/2 = c0A0 and an = cnAn. To evaluate a0 and an, we apply the initial
condition

which results in a Fourier cosine series representation for f(x). From Chapter 1.3 of SU4,
the Fourier coefficients are given by

This completes the temperature solution for one-dimensional heat conduction in a rod
with insulated ends.

Notice that the steady-state temperature can be evaluated as

SU5-23
MTH320 Partial Differential Equations & Heat Equation

This indicates that the temperature everywhere along the rod approaches a uniform
constant temperature, which is expected because the rod is insulated, so heat conduction
must take place until the temperature everywhere along the rod is the same.

To this end, besides the conditions we have considered for the one-dimensional heat
conduction problem, there are other types of conditions such as a semi-infinite rod, infinite
rod as well as heat conduction with heat loss/gain. We shall leave the student to explore
further as a self-learning opportunity.

Activity 5.4

A meter-long copper rod has the initial temperature distribution given by

If the rod is insulated, define and solve the one-dimensional heat conduction problem
along the rod’s length. If copper has a thermal diffusivity of α = 0.0001 m2/s, plot the
temperature function at various times to demonstrate its time progression. Explain
your observations.

SU5-24
MTH320 Partial Differential Equations & Heat Equation

Summary

In Study Unit 5, we covered the following:

• Partial Derivatives of Multivariable Functions

• Meaning of Solution to a Partial Differential Equation
• Derivation of the Heat Equation
• Steady-State Solution of the Heat Equation
• Transient Solution of the Heat Equation
• Complete Solution of the Heat Equation

Watch the following series for more appreciation of PDEs and the heat equation:

• https://youtube.com/playlist?
list=PLZHQObOWTQDNPOjrT6KVlfJuKtYTftqH6

SU5-25
MTH320 Partial Differential Equations & Heat Equation

Formative Assessment

1. Determine all first and second partial derivatives of the following functions:

a.

b.

2. Show that the function

is a solution to the PDE

3. For the heat conduction problem defined by

show that the steady-state solution is

and the corresponding transient problem is

4. Solve the heat conduction problem

where f(x) has the function

SU5-26
MTH320 Partial Differential Equations & Heat Equation

a.
b.

5. Solve the following heat conduction problem for a 10-meter long rod. Plot to analyse.

6. Solve the heat conduction problem

where β is a constant. What is the steady-state solution?

7. The heat conduction problem with heat loss/gain can be derived as

where β is a constant and T∞ represents a constant temperature of the external

cooling/heating medium. Given that the steady-state solution is Ts(x), show that the
corresponding transient problem is

SU5-27
MTH320 Partial Differential Equations & Heat Equation

Solutions or Suggested Answers

Activity 5.1

Evaluate the first and second order partial derivatives of

The first order partial derivatives are

The second order partial derivatives are

Activity 5.2

Show that the function

is a solution to the one-dimensional wave equation: .

The partial derivatives are

Substitution of the above 2nd-order derivatives into the PDE completes the proof.

SU5-28
MTH320 Partial Differential Equations & Heat Equation

Activity 5.3

Determine the steady-state solution of the heat equation below subjected to given
boundary conditions. What do the boundary conditions physically mean? Explain the
solution logically.

The steady-state solution must satisfy the ODE

At the left end of the rod at x = 0, the boundary condition is a fixed temperature. At the
right end at x = L, the boundary condition represents an insulated end, since the heat
transfer

Solving the ODE gives the general solution

Applying boundary conditions

So the steady-state solution is

The constant steady-state temperature at T0 is expected because of the insulated end of the
rod at x = L. Since no heat can be transferred out at this end, the entire rod will eventually
reach thermal equilibrium with the fixed temperature at x = 0.

SU5-29
MTH320 Partial Differential Equations & Heat Equation

Activity 5.4

A meter-long copper rod has the initial temperature distribution given by

If the rod is insulated, define and solve the one-dimensional heat conduction problem
along the rod’s length. If copper has a thermal diffusivity of α = 0.0001 m2/s, plot the
temperature function at various times to demonstrate its time progression. Explain your
observations.

The heat conduction problem is

For insulated ends, the solution for the one-meter long rod is

in which the Fourier coefficients can be evaluated to be

Therefore, the temperature solution is

SU5-30
MTH320 Partial Differential Equations & Heat Equation

For a thermal diffusivity of α = 0.0001 m2/s, the plots are

Desmos graph of T(°C) vs x(m): https://www.desmos.com/calculator/a0bkeuwtgf

(Play the button for t to see the time progression of the temperature distribution. 50
sinusoids applied.)

SU5-31
MTH320 Partial Differential Equations & Heat Equation

Temperature distribution T(°C) vs x(m) along the rod at various time t (seconds)

Inspection of the graphs reveals the following observations:

a. At t = 0, the initial temperature distribution agrees with f(x) given.

b. As time progresses, the temperature variation along the rod evens out to a
uniform temperature due to the heat conduction taking place.

SU5-32
MTH320 Partial Differential Equations & Heat Equation

c. The boundary conditions of insulated ends (Tx = 0) are enforced for t > 0 as
expected.
d. The steady-state temperature is uniform at Ts(x) = 100/3 °C. This value would
obey the conservation of heat energy since no heat is gained or lost due to the
insulated condition enforced. The heat energy of the rod initially and at steady-
state can be calculated by

Clearly, heat energy is conserved within the rod.

e. The temperature variation evens out at a faster rate at earlier times due to larger
curvatures of the temperature distribution. As time progresses, the rate of change
of temperature diminishes due to smaller temperature curvatures.

Formative Assessment
1. Determine all first and second partial derivatives of the following functions:

a.

b.

Answer:

a. ,

b.

2. Show that the function

is a solution to the PDE

SU5-33
MTH320 Partial Differential Equations & Heat Equation

Student is to derive the equation/s stated in this question.

3. For the heat conduction problem defined by

show that the steady-state solution is

and the corresponding transient problem is

Student is to derive the equation/s stated in this question.

4. Solve the heat conduction problem

where f(x) has the function

a.
b.

Answer:

a.
,

b.

5. Solve the following heat conduction problem for a 10-meter long rod. Plot to analyse.

SU5-34
MTH320 Partial Differential Equations & Heat Equation

Answer:

6. Solve the heat conduction problem

where β is a constant. What is the steady-state solution?

Answer:

7. The heat conduction problem with heat loss/gain can be derived as

where β is a constant and T∞ represents a constant temperature of the external

cooling/heating medium. Given that the steady-state solution is Ts(x), show that the
corresponding transient problem is

SU5-35
MTH320 Partial Differential Equations & Heat Equation

Student is to derive the equation/s stated in this question.

SU5-36
 6
Study
Unit

Wave Equation & Laplace Equation

MTH320 Wave Equation & Laplace Equation

Learning Outcomes

By the end of this unit, you should be able to:

• Derive the Wave Equation

• Evaluate Solutions of the Wave Equation subjected to various Initial Conditions
• Understand the Laplace Equation as the Steady-State (Time-Independent)
Representation of the Heat Equation and the Wave Equation
• Evaluate Solutions of the 2D Laplace Equation with Partially Homogeneous
Boundary Conditions
• Evaluate Solutions of the 2D Laplace Equation with Nonhomogeneous Boundary
Conditions

SU6-2
MTH320 Wave Equation & Laplace Equation

Overview

This study unit introduces the wave equation which describes the motion of vibrating
medium and the Laplace equation that governs the steady-state (time-independent)
behaviour of many phenomena. The wave equation is derived from Newton's 2nd law.
Using the separation of variables and the Fourier series, the solutions to the wave and
Laplace equations are presented. We also investigate the wave propagation phenomenon
along a vibrating string and reconcile the observations with our intuition.

SU6-3
MTH320 Wave Equation & Laplace Equation

Chapter 1: The Wave Equation

1.1 The Big Picture

In the last study unit, we derived the heat equation from the law of conservation of heat
energy and presented the solution method by the separation of variables and the Fourier
series. Through the solutions of the heat equation applied to various boundary and initial
conditions, we analysed how the temperature distribution along a rod changes with time,
thereby empowering us with an understanding of how heat conduction occurs in nature.
This understanding is not only qualitative, but also quantitative because the solutions
enable us as scientists and engineers to do calculations, predictions and design.

It is easy to indulge or lose focus in the sea of equations in math modules, but in this
last study unit, let us remind ourselves of the big picture of learning the mathematical
methods covered in MTH319 and MTH320. That is, to apply mathematical modelling to
understand and solve real world problems. Recalling SU5 of MTH319, the general flow of
mathematical modelling is shown as follows:

General Flow of Math Modelling

In this study unit, we are going to apply mathematical modelling to the propagation of
waves and understand the motion of a string subjected to various initial conditions. Then

SU6-4
MTH320 Wave Equation & Laplace Equation

we will look at the Laplace equation that describes the steady-state solutions of many
PDEs.

1.2 Derivation of the Wave Equation

The wave equation can be derived from Newton’s 2nd law. For ease of illustration, we
shall consider the wave equation in one dimension only, but the same principle can be
applied to two or three dimensions.

Suppose we have a string of length L fixed at its end points as shown below:

In the free-body diagram of a small string segment with length Δx, there are tension forces
acting in the tangential direction along the deflected string. We shall also consider the
existence of an external vertical force, F(x,t), acting on the string segment. Since the string
segment is only moving in the vertical direction, Newton’s 2nd law gives

SU6-5
MTH320 Wave Equation & Laplace Equation

The slope of the tangent at any position along the string satisfies

Substitution of the above into the vertical force equation yields

Now, if we only consider small deflections of the string, the angle θ will also be small,
which gives cos(θ) ≈ 1. Then, from the horizontal force equilibrium we have

which implies that the tension force T is constant everywhere along the string for small
deflections. Then, the mass of the small string segment can be approximated by

Combining the above, the vertical force equation can be simplified to be

We denote

where f(x,t) represents the external force per unit length acting on the string and a positive
constant c2 which represents the tension force per unit linear density (ρA) of the string.
So, we get

If we let Δx approach zero, taking limits on both sides of the equation yields the wave
equation with external excitation

SU6-6
MTH320 Wave Equation & Laplace Equation

In the absence of the external force, we arrive at the classical one-dimensional wave
equation

The wave equation has an intuitive explanation. Consider a small string segment in a
string with small deflections where the tension is constant as shown below:

Notice that when the concavity, uxx, is negative, the net tension force is downward.
Since the net force is proportional to acceleration, utt, this means the string segment is
accelerating downward. Similarly, when the concavity is positive, the string segment
accelerates upward due to net upward tensile forces.

So we observe that the concavity, uxx, is related to the acceleration, utt, where both are of
the same polarity. This relation is given by the wave equation.

SU6-7
MTH320 Wave Equation & Laplace Equation

1.3 Solution to the Wave Equation

The wave equation is fully defined with boundary conditions and initial conditions. For
example, in a string fixed at its end points, we have

Notice that there are two initial conditions now, one for initial displacement and another
for initial velocity. This must be so in order to satisfy the 2nd-order partial derivative in
time in order to have a particular solution.

We shall now proceed to derive the solution, using the same approach shown in solving
the heat equation. Suppose that the solution can be written in variable-separable form

Differentiating the above gives

Substituting into the wave equation and rearranging, we obtain

Since the LHS and RHS of the last equation depend only on x and t respectively, the only
way that they can be equal is that both sides must be a constant. So we write

Then, applying the variable-separable solution to the boundary conditions gives

SU6-8
MTH320 Wave Equation & Laplace Equation

By inspection, it cannot be Y(t) = 0 since that would mean u(x,t) = 0. So we should have
X(0) = 0 and X(L) = 0. Hence, the ODE in the spatial variable x is defined by the boundary
value problem (BVP)

By now, it should be clear that this BVP for the spatial function, X(x), is addressed first
because of its homogeneous boundary conditions.

Recalling SU5 Chapter 3.1, standard operations in the consideration of the values of p will
result in unacceptable solutions for p ≥ 0. And for p < 0, we let

and obtain the spatial solution (similar to the transient heat equation)

Now, for the temporal ODE we have

Recalling Chapter 2.3 of SU1, solving the auxiliary equation gives the eigenvalues

and the general solution for each integer n is

Therefore, the solution of the wave equation takes the form

SU6-9
MTH320 Wave Equation & Laplace Equation

in which we denoted constants an = cnAn and bn = cnBn. To evaluate the constants, we

apply the initial conditions

and obtain

The above can be recognised as the Fourier sine series for f(x) and g(x) respectively. Hence,
the Fourier coefficients are

This completes the solution for the wave equation with fixed ends.

We observe from the solution

that the ‘angular frequencies’ in the spatial and time domains are ωn and ωnc respectively.
Therefore, the wavelength and period can be evaluated from

So the transverse velocity of the wave is given by

SU6-10
MTH320 Wave Equation & Laplace Equation

which has been defined earlier to be

A dimensional analysis reveals that

Indeed, c represents the transverse velocity of the wave. From the relationship above,
notice that a string with a larger tension has a faster transverse velocity, which agrees with
intuition since a tighter string vibrates faster.

Example:

Consider a string tautly fixed at its ends. Then, the string is lifted in the centre by height
h and let go. Determine the solution of the string deflection, u(x,t). Given that L = 1m, h =
0.1m and c = 1 m/s, plot the time-progression of the deflection function.

This problem satisfies the wave equation defined by

where the initial displacement has the piecewise function

Since the initial velocity is zero, we will have all bn = 0. Then, the solution is

SU6-11
MTH320 Wave Equation & Laplace Equation

where the Fourier coefficients an can be evaluated from

Using integration by parts, we can get

Therefore, the complete solution is

Substitution of L = 1m, h = 0.1m and c = 1 m/s yields

Plotting in Desmos:

SU6-12
MTH320 Wave Equation & Laplace Equation

Desmos graph of u(m) vs x(m): https://www.desmos.com/calculator/aaa5wglxju

(Play the button for t to see the time progression of the string deflection.)

Deflection u(m) vs x(m) along the string at various time t (seconds)

SU6-13
MTH320 Wave Equation & Laplace Equation

An inspection of the plots reveals the following observations:

a. At t = 0, the deflection function, u(x), agrees with the given initial condition, f(x),
as logically expected.
b. Throughout the time-progression, the boundary conditions of fixed ends of the
string are being enforced.
c. The string vibrates about the mean deflection of u(x) = 0 continuously, since
there is no dissipation of energy because friction and/or air resistance are not
considered in the classical wave equation.

The solution above has been validated through a high-speed camera footage at https://
youtu.be/Qr_rxqwc1jE.

Activity 6.1

Solve the wave equation problem given below. Plot the solution and observe the time-
progression of the wave. What happens when the tension of the string is increased?

SU6-14
MTH320 Wave Equation & Laplace Equation

Chapter 2: The Laplace Equation

2.1 Introduction to the Laplace Equation

The Laplace equation (aka the potential equation) is a 2nd-order PDE that governs the
steady-state equilibrium of many physical processes, such as in heat conduction and
vibration of a medium. For example, in SU5, recall that the steady-state time-independent
temperature function, Ts(x), is solved from

which is the Laplace equation in one dimension. Similarly, for the wave equation in the
last chapter, if the deflection is not changing with time, we will get

which is the 1D Laplace equation again. For heat conduction in a plate and vibration of
a membrane, the steady-state equilibrium is governed by the Laplace equation in two-
dimensions:

Generally, using the divergence (∇.) and gradient operator (∇), the Laplace equation is
commonly written as

in which ∇2is known as the Laplace operator or Laplacian.

2.2 Solution for Partially Homogeneous Boundary Conditions

The Laplace equation is fully defined with boundary conditions. Suppose we have a
rectangular boundary of side lengths L1 and L2, such that

SU6-15
MTH320 Wave Equation & Laplace Equation

Notice that the boundary conditions in x are homogeneous. Since the separation of
variables is successful in establishing the solution for both the heat and wave equations,
let’s try it again. So we assume that the solution has the form

Differentiating the above gives

Substituting into the Laplace equation and rearranging, we obtain

Since the LHS and RHS of the last equation depend only on x and y respectively, the only
way that they can be equal is that both sides must be a constant. So we write

Then, applying the variable-separable solution to the boundary conditions in the x

variable gives

By inspection, it cannot be Y(y) = 0 since that would mean u(x,y) = 0. So we should have
X(0) = 0 and X(L1) = 0. Hence, the ODE in the spatial variable x is defined by the boundary
value problem (BVP)

SU6-16
MTH320 Wave Equation & Laplace Equation

Again, we address this BVP for the spatial function, X(x), first because of its homogeneous
boundary conditions. Standard operations in the consideration of the values of p will
result in unacceptable solutions for p ≥ 0. And for p < 0, we denote

and obtain the solution in x as

Now, along the y direction, we have

The eigenvalues can be worked out from auxiliary equation

Recalling Chapter 2.3 of SU1, for real and distinct eigenvalues, the general solution for
each integer value of n is

Here, we make use of the hyperbolic functions

and rewrite

We shall see that this form of Y(y) allows us to apply boundary conditions more easily
later.

SU6-17
MTH320 Wave Equation & Laplace Equation

To this end, the complete solution is

in which constants an = cnAn and bn = cnBn, to be found by applying the non-homogeneous

boundary conditions in the y spatial variable

Noting that cosh(0) = 1 and sinh(0) = 0, the above relations yield

Again, we have the Fourier sine series for f(x) and g(x) respectively, where the Fourier
coefficients can be recognised to be

The last equation above can be rearranged to obtain bn as

This completes the solution for the Laplace equation in a rectangle with homogeneous
boundary conditions in x.

Analogously, if the homogeneous boundary conditions are in y instead, such that we have

SU6-18
MTH320 Wave Equation & Laplace Equation

then we can expect the complete solution would be

where

Example:

Solve the Laplace equation problem in a 1m by 1m square region

The Fourier coefficients are

So bn has the expression

SU6-19
MTH320 Wave Equation & Laplace Equation

The solution is

with an and bn defined above.

2.3 Solution for Non-Homogeneous Boundary Conditions

In cases when all boundary conditions are not homogeneous such that the Laplace
equation is defined by

then we can apply the ‘divide and conquer’ approach similar to that in the heat equation,
where the complete solution is written as a sum of the transient and steady-state solutions.
Let

such that we get two Laplace equation problems, but each with partially homogeneous
boundary conditions

The partial solutions, u1(x,t) and u2(x,t), have been evaluated in the preceding section.

SU6-20
MTH320 Wave Equation & Laplace Equation

Activity 6.2

Show that the sum of partial solutions, u1(x,y) and u2(x,y), defined above solves the
original problem with all non-homogeneous boundary conditions

Activity 6.3

Obtain the steady-state solution of the heat conduction problem in a rectangular 1m

by 2m plate described below:

SU6-21
MTH320 Wave Equation & Laplace Equation

Summary

In Study Unit 6, we covered the following:

• Derivation of the Wave Equation

• Solutions of the Wave Equation subjected to various Initial Conditions
• How the Laplace Equation represents the Steady-State (Time-Independent)
Problem of the Heat Equation and the Wave Equation
• Solutions of the 2D Laplace Equation with Partially Homogeneous Boundary
Conditions
• Solutions of the 2D Laplace Equation with Nonhomogeneous Boundary Conditions

SU6-22
MTH320 Wave Equation & Laplace Equation

Formative Assessment

1. Solve the wave equation problem

2. From Newton’s 2nd law, derive the wave equation for a vibrating string in a viscous
medium subjected to a damping force per unit length, but, where b is the damping
coefficient.

3. For a vibrating string in viscous medium, if the damping coefficient, b, is small, apply
the separation of variables and show that the general solution of the string deflection
is

where

4. Show that each function given below is a solution of the Laplace equation. What are
the boundary conditions satisfied by each solution?

a.
b.
c.

5. Given that the solution to the Laplace equation in a unit square is

SU6-23
MTH320 Wave Equation & Laplace Equation

Determine the solution for Y(y) such that it satisfies the boundary conditions

6. Solve the potential equation problem

7. Determine the steady-state solution of a vibrating membrane in a rectangle governed

by

where g(x) has the function

SU6-24
MTH320 Wave Equation & Laplace Equation

Solutions or Suggested Answers

Activity 6.1

Solve the wave equation problem

The solution is

But since there is no initial displacement, we will get all an = 0 and bn can be evaluated as

Hence, the solution is

Observe the wave propagation in Desmos:

https://www.desmos.com/calculator/xigz2wvd3y

The initial velocity causes the string to deflect upwards at t = 0, with higher velocities
towards the right of the string. Notice that when the tension is increased, the wave
velocity c is increased which results in a faster propagation of the wave, but with smaller
amplitudes due to a larger ‘stiffness’ of the string. This agrees with our intuition.

SU6-25
MTH320 Wave Equation & Laplace Equation

Activity 6.2

Show that the sum of partial solutions, u1(x,y) and u2(x,y), defined above solves the
original problem with all non-homogeneous boundary conditions

From

Partial differentiation gives

Substitution into the Laplace equation gives

The boundary conditions for u(x,y) become

Rearranging the above to isolate u2(x,y), we obtain

If we define the Laplace equation problem for u1(x,y) as

SU6-26
MTH320 Wave Equation & Laplace Equation

then combining the above gives

which is the Laplace equation problem for u2(x,y). Hence, the sum of partial solutions,
u1(x,y) and u2(x,y), solves the original problem with all non-homogeneous boundary
conditions.

Activity 6.3

Obtain the steady-state solution of the heat conduction problem in a rectangular 1m by

2m plate described below:

Since Tt = 0 at steady-state, the heat conduction problem is governed by the Laplace

equation below, where u(x,y) represents the steady-state temperature distribution over
the plate:

Since the boundary conditions are not fully homogeneous in each of the spatial
coordinates, the solution can be written as

where we solve

SU6-27
MTH320 Wave Equation & Laplace Equation

for u1(x,y) as

Noting that all bn = 0 for zero initial velocity, we have

and, for u2(x,y), we solve

The solution is, noting all bn = 0 for zero initial velocity again

Therefore, the complete steady-state temperature function is

Formative Assessment
1. Solve the wave equation problem

SU6-28
MTH320 Wave Equation & Laplace Equation

Answer:

2. From Newton’s 2nd law, derive the wave equation for a vibrating string in a viscous
medium subjected to a damping force per unit length, but, where b is the damping
coefficient.
Answer:

3. For a vibrating string in viscous medium, if the damping coefficient, b, is small, apply
the separation of variables and show that the general solution of the string deflection
is

where

Student is to derive the equation/s stated in this question.

4. Show that each function given below is a solution of the Laplace equation. What are
the boundary conditions satisfied by each solution?

a.

SU6-29
MTH320 Wave Equation & Laplace Equation

b.
c.

Answer:

a.
,

b.
,

c.

5. Given that the solution to the Laplace equation in a unit square is

Determine the solution for Y(y) such that it satisfies the boundary conditions

Answer:

6. Solve the potential equation problem

Answer:

SU6-30
MTH320 Wave Equation & Laplace Equation

7. Determine the steady-state solution of a vibrating membrane in a rectangle governed

by

where g(x) has the function

Answer:

SU6-31
MTH320 Wave Equation & Laplace Equation

SU6-32