Theme 1: Differential Equations

Introduction, Classification, and Mathematical Modelling

1 Introduction

Many laws in science, engineering, economics, biology, and other disciplines describe how one
quantity changes in response to another. Since derivatives measure rates of change, differential
equations arise naturally whenever we attempt to describe such phenomena mathematically.
The general mandate of this course is twofold:

1. to develop techniques for identifying and solving different types of differential equations;

2. to learn how common real-world phenomena can be translated into differential equations,
and how the resulting equations can be used to obtain useful information about the
phenomena.

Thus, throughout the course we shall repeatedly encounter three related activities:

Modelling −→ Solving −→ Interpretation.

A mathematical model is useful only if we can understand what its solutions tell us about the
original problem.

2 Differential Equations

Definition 1 (Differential equation). A differential equation is an equation involving an

unknown function together with one or more of its derivatives.

For example,
y ′ + 2y = 0,
y ′′ − 4y ′ + 3y = ex ,
and
∂u ∂2u
=
∂t ∂x2
are all differential equations.
The unknown in a differential equation is usually a function, rather than a number.
For example, in
y ′ + 2y = 0,
we seek a function y = y(x) whose derivative and function values satisfy the equation.

1
3 Classification of Differential Equations

There are many different types of differential equations. Before we develop methods for solving
them, we need a systematic way of classifying them.
In this course, we shall frequently classify differential equations according to:

1. type: ordinary or partial;

2. order;

3. linearity.

3.1 Classification by type: ODEs and PDEs

Definition 2 (Ordinary differential equation). An ordinary differential equation, abbreviated

ODE, is a differential equation involving ordinary derivatives of one or more dependent variables
with respect to a single independent variable.

Examples include
dy
+ 3y = x2 ,
dx
y ′′ + 4y = 0,
and
d3 y dy
− 2 + t y = 0.
dt3 dt
In each case, the unknown function depends on only one independent variable.

Definition 3 (Partial differential equation). A partial differential equation, abbreviated

PDE, is a differential equation involving partial derivatives of a function of two or more
independent variables.

For example, if u = u(x, t), then

∂u ∂2u
=
∂t ∂x2
is a PDE.
Other examples are
∂2u ∂2u
+ 2 =0
∂x2 ∂y
and
∂2u 2
2∂ u
= c .
∂t2 ∂x2

The distinction is determined by the number of independent variables, not simply by the
notation used for the derivative.

2
 Think about it
Classify each equation as an ODE or PDE:
dy
(a) + ty = 0;
dt
(b) ut + ux = 0;

(c) x2 y ′′ + xy ′ − y = 0;

(d) uxx + uyy = 0.

3.2 Classification by order

Definition 4 (Order). The order of a differential equation is the order of the highest derivative
appearing in the equation.

For example,
y ′ + 3y = sin x
is a first-order ODE,

y ′′ − 3y ′ + 2y = 0
is a second-order ODE, and

y ′′′ + xy ′ − y = ex
is a third-order ODE.

Remark. The order is determined by the highest derivative, not by the highest power appearing
in the equation.

For example,
(y ′ )5 + y = 0
is still a first-order differential equation.
Think about it
Determine the order of:

y (4) + y ′′ + y = 0, (y ′′ )3 + y ′ = x, uxxt + ut = 0.

3.3 Classification by linearity

Linearity is one of the most important classifications in this course.

An nth-order ODE is called linear if it can be written in the form

an (x)y (n) + an−1 (x)y (n−1) + · · · + a1 (x)y ′ + a0 (x)y = g(x),

where the coefficient functions

a0 (x), a1 (x), . . . , an (x)

3
depend only on the independent variable x, and an (x) is not identically zero.
Thus, in a linear differential equation:

ˆ y and its derivatives occur only to the first power;

ˆ products such as yy ′ or y ′ y ′′ do not occur;

√
ˆ nonlinear functions of y, such as sin y, ey , y, or y 2 , do not occur.

The coefficient functions may themselves be complicated functions of the independent variable.
For example,
ex y ′′ + sin(x)y ′ + x3 y = ln(1 + x2 )
is still a linear differential equation.

Examples

The equation
y ′′ + xy ′ − 3y = sin x
is linear.
The equation
y ′′ + y 2 = 0
is nonlinear because y occurs squared.
The equation
yy ′ + x = 0
is nonlinear because y and y ′ are multiplied together.
The equation
y ′′ + ex y ′ = x2
is linear: the factor ex depends only on the independent variable.

Equation Order Linear?

y ′′ + xy ′ − 3y = sin x 2 Yes
y ′′ 2
+y =0 2 No
yy ′ + x = 0 1 No
y ′′ + ex y ′ = x2 2 Yes
(y ′ )2 + y = 0 1 No
x2 y ′′ + xy ′ − y = 0 2 Yes

4
 Think about it
Classify each equation by type, order, and, where appropriate, linearity:

(a) y ′′ + 4y = cos x;

(b) y ′ + y 3 = 0;

(c) x2 y ′′ + ex y ′ = x;

(d) y ′′ + yy ′ = 0;

(e) ut = 3uxx .

4 Notation for Derivatives

We shall use different notations for derivatives depending on which is most convenient.
For a function y = y(x):

Leibniz notation
dy d2 y d3 y
, , , ...
dx dx2 dx3

Prime notation

y′, y ′′ , y ′′′ , ...

Thus,
dy d2 y
y′ = , y ′′ = .
dx dx2
If the independent variable is time, we may write
dy
.
dt

In applications, Leibniz notation is often especially useful because it reminds us which variable
the rate of change is being measured with respect to.

5 Methods of Solution

There is no single method that solves every differential equation. Different families of differential
equations require different techniques.
Among the methods we shall encounter in this course are:

ˆ direct integration;

ˆ separation of variables;

ˆ methods for first-order linear equations;

5
 ˆ substitution methods;

ˆ methods for higher-order linear equations;

ˆ the method of undetermined coefficients.

A major skill in differential equations is therefore not only solving an equation, but first
recognising its type and choosing an appropriate method.

6 Mathematical Modelling

We now obtain our first taste of translating scientific laws and verbal statements into differential
equations.
A mathematical model is a mathematical description of a real-world system. The purpose of the
model is not merely to produce equations, but to use those equations to understand, explain, or
predict the behaviour of the system.
A typical modelling process is:

Real-world problem
↓
Assumptions and variables
↓
Mathematical model
↓
Mathematical analysis
↓
Interpretation
↓
Comparison with reality

If the model does not adequately describe the observed behaviour, its assumptions may need to
be revised.

6.1 Translating common verbal statements

Let k > 0 be a constant.

6
 Verbal statement Mathematical statement

y is directly proportional to x y = kx
k
y is inversely proportional to x y=
x
y is directly proportional to x2 y = kx2
k
y is inversely proportional to x2 y= 2
x
√ √
y is proportional to x y=k x
dy
The rate of change of y is proportional to y = ky
dt
dy
The rate of change of y is proportional to the = k(y − A)
dt
difference y − A

Remark. The sign of the proportionality constant matters. It is often clearer to take k > 0 and
place the required sign explicitly in the model. For example,
dy
= −ky, k > 0,
dt
clearly describes decay.

Think about it
Translate each statement into an equation:

(a) P is proportional to t2 ;

(b) v is inversely proportional to r;

(c) the rate of change of Q is proportional to Q;

(d) the rate of decrease of M is proportional to M .

7 Some Common Differential-Equation Models

7.1 Newton’s Law of Heating and Cooling

Suppose that T (t) denotes the temperature of an object at time t, and A(t) denotes the
temperature of its surroundings.
Newton’s law of heating and cooling states:

The rate of change of the temperature of an object is proportional to the difference

between the object’s temperature and the temperature of its surroundings.

If k > 0, we write
dT

= −k T − A .
dt

The negative sign is important.

7
Case 1: The object is hotter than its surroundings

If
T > A,
then
T − A > 0.
Hence
dT
< 0.
dt
Thus the object’s temperature decreases, as expected.

Case 2: The object is colder than its surroundings

If
T < A,
then
T − A < 0.
Hence
dT
> 0.
dt
Thus the object’s temperature increases.

Case 3: The temperatures are equal

If
T = A,
then
dT
= 0.
dt
Thus there is no instantaneous change in temperature.
If the ambient temperature A is constant, then

T =A

is an equilibrium temperature.

We have obtained useful qualitative information from the differential equation without
solving it.

8
 Think about it
Suppose the surrounding temperature is 20◦ C and

T ′ = −0.2(T − 20).

Without solving the equation:

(a) Is the object heating or cooling when T = 35?

(b) Is it heating or cooling when T = 12?

(c) What happens when T = 20?

(d) Which temperature acts as an equilibrium?

7.2 Torricelli’s Law

Consider a tank containing water. Let

V (t) = volume of water in the tank at time t

and
y(t) = height of the water above the outlet.

Torricelli’s law states that the speed of water flowing through a small opening is proportional to
the square root of the height of the water above the opening.
Consequently, under suitable assumptions, the rate at which the volume of water leaves the tank
has the form
dV √
= −k y, k > 0.
dt

The negative sign indicates that the volume of water in the tank is decreasing.

A cylindrical tank

Suppose the tank is cylindrical with constant horizontal cross-sectional area A. Then

V = Ay.

Since A is constant,
dV dy
=A .
dt dt
Combining this with Torricelli’s law gives
dy √
A = −k y,
dt
and hence
dy k√
=− y.
dt A

Thus Torricelli’s law may be expressed either in terms of the rate of change of volume or, after
using the geometry of the tank, in terms of the rate of change of water height.

9
 Do not confuse
dV dy
with .
dt dt
They describe different physical quantities and generally have different units.

Think about it
Without solving the equation
dy √
= −c y, c > 0,
dt
answer the following:

(a) Is the water level increasing or decreasing?

(b) What happens to the magnitude of dy/dt as y decreases?

(c) What does the model predict when y = 0?

7.3 Population Growth and Decay

Let
P (t) = population at time t.

A simple population model assumes that the number of births per unit time is proportional to
the current population.
If β > 0 is the per-capita birth rate, then

birth rate = βP.

Similarly, if δ > 0 is the per-capita death rate, then

death rate = δP.

The net rate of change of the population is therefore

dP
= birth rate − death rate
dt
= βP − δP.

Hence
dP
= (β − δ)P.
dt

If we set
k = β − δ,
then the model becomes
dP
= kP.
dt

There are three possibilities.

10
Growth: β > δ

Then
k>0
and therefore
dP
>0
dt
whenever P > 0. The population grows.

Equilibrium: β = δ

Then
k = 0,
so
dP
= 0.
dt
The population remains constant.

Decay: β < δ

Then
k < 0,
and therefore
dP
<0
dt
whenever P > 0. The population decreases.

Remark. In more realistic models, the birth and death rates need not be constant. They may
depend on time, population size, available resources, or other factors. For example,

β = β(t), δ = δ(t)

would give
dP

= β(t) − δ(t) P.
dt

Think about it
Suppose
P ′ = 0.03P.

(a) Is the population growing or decreasing?

(b) If the population doubles, what happens to its instantaneous rate of change?

(c) What does the model predict when P = 0?

11
8 Reading Information Directly from a Differential Equation

Before trying to solve a differential equation, it is often useful to ask what the equation already
tells us.
Consider
y ′ = F (t, y).

The sign of F (t, y) immediately tells us whether the solution is locally increasing or decreasing:

F (t, y) > 0 =⇒ y ′ > 0,

so y is increasing, while
F (t, y) < 0 =⇒ y ′ < 0,
so y is decreasing.
If
F (t, y) = 0,
then the rate of change is zero at that state.
For example, if
y ′ = y(5 − y),
then:

ˆ y ′ = 0 when y = 0 or y = 5;

ˆ y ′ > 0 when 0 < y < 5;

ˆ y ′ < 0 when y > 5.

Thus, without solving the equation, we can already infer substantial information about how its
solutions behave.

9 From Models to Solutions

At this point we have learned how to:

ˆ recognise differential equations;

ˆ classify them by type, order, and linearity;

ˆ translate some common verbal laws into differential equations;

ˆ interpret qualitative information directly from a model.

The next natural question is:

What exactly does it mean for a function to solve a differential equation, and how can we check
whether a proposed function is a solution?

This leads to the study of solutions and solution verification.

12