Advanced Mathematics for Engineers

Part II: Ordinary Differential Equations

Tilahun Abebaw (PhD)

Addis Ababa University

November 26, 2023

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary Differential
November Equations
26, 2023 1 / 111
1. Introduction

An equation containing one or more derivatives of the function under

consideration is known as a differential equation (DE).
To define DE, we need to identify the dependent and the independent
variable(s) involved.
If there is only one independent variable, the DE is called an Ordinary
Differential Equation (ODE), whereas, if there are more than one
independent variables, the DE is called a Partial Differential Equation
(PDE)
An nth -order ODE is linear if it can be expressed in the form of

ao (x)y (n) (x) + a1 (x)y (n−1) (x) + · · · + an (x)y (x) = f (x),

where ao (x), . . . , an (x) are functions of the independent variable x

alone.

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary Differential
November Equations
26, 2023 2 / 111
 Introduction · · · Cont’d
Classification by Order

The order of a differential equation (either ODE or PDE) is the order of

the highest derivative that appear in the equation. For example,

dy d 2y dy
4x +y =x and 2
+4 − 6y = e x
dx dx dx
are first and second-order ordinary differential equations respectively.
The general nth −order ordinary differential equation in one dependent
variable is given by the general form general form

F (x, y , y 0 , y 00 , ..., y (n) ) = 0, (1)

where F is a real-valued function of n + 2 variables x, y , y 0 , y 00 , ..., y (n) .

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary Differential
November Equations
26, 2023 3 / 111
Introduction · · · Cont’d
For both practical and theoretical reasons we shall also make the
assumption hereafter that it is possible to explicitly solve the differential
equation of the form (1) uniquely for the highest derivative y (n) in terms
of the remaining n + 1 variables x, y , y 0 , y 00 , ..., y (n−1) . Then the
differential equation (1) becomes
d ny
= f (x, y , y 0 , ..., y (n−1) ), (2)
dx n
where f is a real-valued continuous function and this is referred to as the
normal form of (1).
Example
The normal form of the first-order equation 4xy 0 + y = x is
x −y
y0 =
4x
and the normal form of the second-order equation y 00 − y + 6y = 0 is
Tilahun Abebaw (PhD) (AAU) 00
Advanced Mathematics0 for Engineers Part II: Ordinary Differential
November Equations
26, 2023 4 / 111
Introduction · · · Cont’d

The first order ordinary differential equation is generally expressed as:

F (x, y , y 0 ) = 0 or y 0 = f (x, y ).

For example, the differential equation y 0 + y = x is equivalent to

y 0 + y − x = 0. If F (x, y , y 0 ) = y 0 + y − x, then the given differential
equation becomes of the form F (x, y , y 0 ) = 0.

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary Differential
November Equations
26, 2023 5 / 111
 Introduction · · · Cont’d
Classification by Linearity

An nth -order ordinary differential equation (1) is said to be linear if F is

linear in y , y 0 , . . . , y (n) . This means that an nth -order linear ordinary
differential equation is of the form
an (x)y (n) + an−1 (x)y (n−1) + · · · + a1 (x)y 0 + a0 (x)y − b(x) = 0, (3)
where an (x) 6= 0.
If b(x) ≡ 0, the equation (3) is called a homogeneous ODE and
otherwise it is called nonhomogeneous.
Notation: We may equivalently use the notations
d ny
and y (n)
dx n
interchangeably for the nth −order derivative of y with respect to x. Using
this notation, equation (3) can be equivalently written as
d ny d n−1 y dy
an (x) n
+ an−1 (x) n−1
+ · · · + a1 (x) + a0 (x)y = b(x).
dx dx dx
Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary Differential
November Equations
26, 2023 6 / 111
Introduction · · · Cont’d

Let x := x(t) denotes a continuous function of t on some interval

[a, b]. The general nth -order ODE has the form:

F (t, x(t), x 0 (t), x 00 (t), . . . , x (n) (t)) = 0.

If F is linear in x, x 0 , . . . , x (n) , then it is called linear ODE; Otherwise

it is known as a nonlinear ODE.

Example
1. x 00 (t) + 2x(t) + sin(t) = 0 is a linear ODE,
2. x 0 + x 2 = 1 is a nonlinear ODE.

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary Differential
November Equations
26, 2023 7 / 111
 Introduction · · · Cont’d
Solution Concept

Definition
Let h(x) be a real valued function defined on an interval [a, b] and having
nth order derivative for all x ∈ (a, b). If h(x) satisfies the nth order ODE
(3) on (a, b), that is,
1. F (x, h(x), h0 (x), h00 (x), . . . , h(n) (x)) is defined for all x ∈ (a, b) and
2. F (x, h(x), h0 (x), h00 (x), . . . , h(n) (x)) = 0, for all x ∈ (a, b),
then y = h(x) is called a (an Explicit) solution of the ODE on [a, b].

Sometimes a solution of a differential equation may appear as an implicit

function, i.e. the solution can be expressed implicitly in the form:
h(x, y ) = 0, where h is some continuous function of x and y , and such
solution is called an Implicit Solution of the DE.

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary Differential
November Equations
26, 2023 8 / 111
First Order ODES
Existence and Uniqueness of Solutions

Example
The function h(x) = sin x + cos x is an explicit solution for the DE y 00 + y = 0,
where as the function h(x, y ) = x 2 + y 2 − 4 = 0 is an implicit solution of the DE
yy 0 = −x on (−2, 2), since y > 0.

Theorem
1. (Existence) Let F (t, x) be a continuous function, then the initial value
problem
dx
= F (t, x), x(to ) = xo
dt
has a solution x = f (t) defined in some neighbourhood of to .
2. (Uniqueness) If F (t, x) is a once continuously differentiable function, then
there exists a unique solution to the initial value problem,
dx
= F (t, x), x(to ) = xo
dt
Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary Differential
November Equations
26, 2023 9 / 111
First Order ODEs Separation of Variables
A First Order ODE is generally of the form

ẏ = F (x, y ), where y is a function of x.

If a given ODE can be reduced or transformed into the form

dy
g (y ) = f (x),
dx
where f (x) is a function of x only and g (y ) is a function of y only, then it
is called a separable ODE.
To solve a separable ODE, first write as

g (y )dy = f (x)dx

and integrate both sides,

Z Z
G (y ) = g (y )dy = f (x)dx = F (x) + C .

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 10 / 111
Example
1 Solve the initial value problem,
dy
= (1 − 2x)y , y (0) = 1.
dx
2 Solve the following DEs
(a) 6yy 0 + 4x = 0.
(b) y 0 = x 2 e −y

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 11 / 111
Ordinary Differential Equations Reducible to separable form
There are ODEs that are not separable, but they can be transformed to
separable form with simple change of variables (proper substitutions).
1. Linear Substitution
If a given DE can be written in the form

y 0 = g (ax + by + c).

Then use the substitution u = ax + by + c. Then the given DE is

transformed to the form
du
= bg (u) + a
dx
and this is a separable DE in u and x.
Example
Solve each of the following ODEs.
y 0 = (x + y )2

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 12 / 111
Ordinary Differential Equations Reducible to separable
form . . .
2. Quotient Substitution
If a given DE can be written in the form
 
0 y
y =g .
x

Then use the substitution u = yx . Then the given DE is transformed to the

form
dx du
=
x g (u) − u
and this is a separable DE in u and x.
Example
Solve the following DEs.

x 2 y 0 = x 2 + xy + y 2 .
Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 13 / 111
Exact Differential Equations

Definition
An ODE of the form M(x, y )dx + N(x, y )dy = 0 is said to be an exact DE
is some domain D, if there exists a function F (x, y ) such that
∂F ∂F
∂x = M(x, y ) and ∂y = N(x, y ) for all (x, y ) ∈ D.

If we can find a function F (x, y ) such that

∂F ∂F
= M(x, y ) and = N(x, y ),
∂x ∂y

then the differential equation M(x, y )dx + N(x, y )dy = 0 is just

M(x, y )dx + N(x, y )dy = dF = 0.
However, for a function F of two variables, if dF = 0, then
F (x, y ) = constant. The equation F (x, y ) = c, where c is an arbitrary
constant, implicitly defines the general solution of the deferential equation
M(x, y )dx + N(x, y )dy = 0.
Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 14 / 111
Theorem (Test for Exactness)
Let M(x, y ), N(x, y ), ∂M ∂N
∂y and ∂x be all continuous functions in some
domain D in the xy -plane. Then the DE

M(x, y )dx + N(x, y )dy = 0

is exact differential in D if and only if

∂M ∂N
=
∂y ∂x
everywhere in D.

Example
Solve each of the following DEs.
1 (x 2 − 3x 2 y )dx − (x 3 +
√
y )dy = 0.
x−y +4
2 y0 = x+y .

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 15 / 111
Integrating Factors
The differential equation ydx + 2xdy = 0 is not exact. But if we multiply
this equation by y , the equation is becomes and exact equation. That is,
y 2 dx + 2xydy = 0
is exact, since
∂y 2 ∂(2xy )
= 2y = .
∂y ∂x
In this case y is called an integrating factor for the given differential
equation.
Definition
If the differential equation M(x, y )dx + N(x, y )dy = 0 is not exact but the
differential equation

µ(x, y )M(x, )dx + µ(x, y )N(x, y )dy = 0, where µ(x, y )  0

is exact, then the multiplicative function µ(x, y ) is called an integrating

factor of the DE.
Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 16 / 111
Integrating Factor...

Suppose we have a differential equation which is not exact but it can be

made exact by an integrating factor. The method of determining an
integrating µ of the given DE is described below.
µ(x, y ) is any (non-zero) solution of the equation

∂ ∂
(µN) = (µN) (by exactness of the given DE) (4)
∂y ∂x
and this is equivalent to the equation

µy M + µMy = µx N + µNx ,

which is a first-order partial differential equation in µ.

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 17 / 111
Integrating Factor...
If the integrating factor µ can be found to be a function of x alone, µ(x)
(or a function of y alone µ(y )),then the above equation is reduced to

dµ
µMy = N + µNx
dx
dµ My −Nx
which is equivalent to dx = µ( N ) and this is a separable differential
equation.
M y − Nx
This idea works correctly if the ratio is a function of x only; that
N
My −Nx
is, if p(x) = N is a function of x only. In this case
 
dµ My − Nx
= dx,
µ N

which implies R
p(x)dx
µ(x) = e .

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 18 / 111
Integrating Factor...
My − Nx
If the quotient is not a function of x alone, then the integrating
N
factor µ can not be obtained using the above procedure, but we can try to
find µ as a function of y alone, µ(y ).
Then when µ(y ) is only a function of y , equation (4) will be reduced to
dµ
M + µMy = µNx
dy
which implies

 
dµ M y − Nx
= −µ , which is a separable differential equation.
dy M
My −Nx
Hence, if instead, the fraction q(y ) = M is a function of y alone, then
R
− q(y )dy
µ(y ) = e .

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 19 / 111
Example
Consider the differential equation

dx + (3x − e −2y )dy = 0. (5)

Let M = 1 and N = 3x − e −2y . Then ∂M ∂N

∂y = 0 and ∂x = 3. This implies
∂M ∂N
∂y 6= ∂x and hence the differential equation is not exact.
Assume that the given equation has an integrating factor.

M y − Nx 0−3 −3e 2y
= =
N 3x − e −2y 3xe 2y − 1
which is not a function of x alone. Hence obtaining µ(x) is not possible.
M −N
However, yM x = 0−3 1 = −3( constant function)
can be considered
R as a function of y alone . Therefore,
µ(y ) = e − (−3)dy = e 3y .

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 20 / 111
Example ....

Example
Now to solve the problem in (5), multiplying the given equation by
µ(y ) = e 3y we have an exact DE

e 3y dx + (3x − e −2y )e 3y dy = 0.

Thus, there exists F (x, y ) such that

∂F ∂F
= e 3y and = (3x − e −2y )e 3y .
∂x ∂y

Hence F (x, y ) = xe 3y − e y + B = constant.

That means xe 3y − e y = C , where C is an arbitrary constant, defines the
implicit solution of the Differential Equation in (5).

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 21 / 111
Linear First Order Differential Equations
Consider the general first-order linear differential equation

a1 (x)y 0 + a0 (x)y = f (x), where a1 (x) 6≡ 0 (6)

By dividing both sides by a1 (x) 6= 0, we get y 0 + p(x)y = q(x), where

a0 (x) f (x)
p(x) = and q(x) = .
a1 (x) a1 (x)

Here we assume that p(x) and q(x) are continuous.

First compute R
p(x)dx
e
This is called an integrating factor for the linear equation.
Multiply both sides of the differential equation by the integrating
factor. R R R
e p(x)dx y 0 + e p(x)dx p(x)y = e p(x)dx q(x).
Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 22 / 111
Linear First Order Differential Equations...

Write the left side of the resulting equation as the derivative of the
product of y and the integrating factor. The integrating factor is
designed to make this possible. The right side is a function of just x.
That is R
de p(x)dx y R
= e p(x)dx q(x)
dx
Integrate both sides of this equation with respect to x and solve the
resulting equation for y gives the general solution.
R Z R
p(x)dx
e y = e p(x)dx q(x)dx.

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 23 / 111
Example
Solve the differential equation y 0 + 3y = 6.
The given equation is linear with p(x) = 3 and q(x) = 6.
1 We compute the integrating factor
R R
p(x)dx 3dx
e =e = e 3x

2 We multiply y 0 + 3y = 6 by e 3x to get y 0 e 3x + 3ye 3x = 6e 3x .

3 The above equation is equivalent to

d(ye 3x )
= 6e 3x .
dx
4 We integrate
d(ye 3x )
Z Z
dx = 6e 3x
dx
and get y (x)e 3x = 2e 3x + C . Then solve for y (x) to get the general
solution y (x) = Ce −3x + 2 for an arbitrary constant C .
Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 24 / 111
Reading Assignment .

Special first order Equations

Use transformation to convert the equations into separable or exact form.
For instance,
For Bernouli equation y 0 + p(x)y = q(x)y α ,
- use the transformation v = y 1−α .
For Riccati equation y 0 = p(x)y 2 + q(x)y + r (x)
- use the transformation y = S(x) + z1 .
etc.

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 25 / 111
Linear ODEs of the Highest Order

Definition
A linear ordinary differential equation of order n in the dependent variable
y and independent variable x is an equation which can be expressed in the
form:

an (x)y (n) + an−1 (x)y (n−1) + · · · + a1 (x)y 0 + a0 (x)y = f (x), (7)

where an (x) 6≡ 0 and the functions a0 , . . . , an are continuous real- valued

functions of x ∈ [a, b].
The function f (x) is called the non-homogeneous term and all the points
x ∈ [a, b] in which an (x ) = 0 are called singular points of the DE (7).
If f (x) ≡ 0, then (7) is reduced to:

an (x)y (n) + an−1 (x)y (n−1) + · · · + a1 (x)y 0 + a0 (x)y = 0 (8)

This equation is known as homogeneous Linear ODE of order n.

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 26 / 111
Theorem (Basic Existence Theorem for IVP)
Consider the linear ODE

an (x)y (n) + an−1 (x)y (n−1) + · · · + a1 (x)y 0 + a0 (x)y = f (x),

where a0 , a1 , . . . , an−1 , an and f are continuous functions on the interval

[a, b] and an (x) 6= 0, ∀x ∈ [a, b]. Furthermore, let x0 be any point in [a, b]
and let c0 , c1 . . . cn−1 be arbitrary real constants. Then there exists a
unique solution function g (x) of (7) on [a, b] satisfying the initial
conditions,

g (x0 ) = c0 , g 0 (x0 ) = c1 , . . . , g n−1 (x0 ) = cn−1 .

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 27 / 111
General Solution of Homogeneous Linear ODEs

Theorem (Linear Combination of Solutions)

If y1 , y2 , . . . , yk are solutions of the homogeneous linear ODE (7) and if
c1 , c2 , . . . ck are arbitrary constants, then the linear combination
k
X
y = c1 y1 + c2 y2 + . . . + ck yk = ci yi
i=1

is also a solution of (7). That is, any linear combination of solutions of a

linear homogeneous differential equation is also a solution.

Example
y1 = cos x and y2 = sin x are two solutions of the homogenous DE
y 00 + y = 0. Then for any arbitrary constants c1 and c2 ,
y (x) = c1 cos x + c2 sin x is also a solution to the given DE.

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 28 / 111
Linearly Dependent and Linearly Independent Functions
Definition
1 Functions f1 , . . . , fn are said to be Linearly Dependent (LD) on some
interval [a, b] if there are constants c1 , . . . , cn , not all zero, such that

c1 f1 (x) + c2 f2 (x) + . . . + cn fn (x) = 0 (9)

for all x ∈ [a, b].

2 If the relation (9) is true only when c1 = . . . = cn = 0, then the
functions f1 , . . . , fn are said to be Linearly Independent (LI) on [a, b].

Example
1 The functions f1 (x) = e x and f2 (x) = 4e x are Linearly Dependent on
R since −4f1 (x) + f2 (x) = −4e x + 4e x = 0, for all x ∈ R.
2 The functions f1 (x) = x and f2 (x) = x 3 are LI on R, since for
c1 , c2 ∈ R, c1 f1 (x) + c2 f2 (x) = c1 x + c2 x 3 = 0, ∀x ∈ R\{0} implies
c1 = 0 and c2 = 0.
Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 29 / 111
Existence of Linearly Independent Solutions for a LHODE
Theorem
The Linear Homogenous Differential Equation (LHODE) (8) always has n
Linearly Independent (LI) solutions. Furthermore, if f1 (x), f2 (x), . . . , fn (x)
are n LI solutions of (8), then every solution of (8) can be expressed as a
linear combination of these solution functions. i.e. If y is a solution for
(8), then
Xn
y (x) = ci fi (x)
i=1

for some c1 , ..., cn ∈ R..

Example
Consider the second order linear homogenous DE y 00 + y = 0. Then
f1 (x) = sin x, f2 (x) = cos x are LI solutions of the given equation and the
general solution of the DE is given by y (x) = c1 sin x + c2 cos x, for
constants c1 , c2 ∈ R.
Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 30 / 111
Fundamental Set of Solutions
Definition
If f1 (x), f2 (x), . . . , fn (x) are n linearly independent solutions of (8) on
[a, b], then the set {f1 (x), f2 (x), . . . fn (x)} is called the Fundamental Set
of Solutions of (8) and the function

f (x) = c1 f1 (x) + c2 f2 (x) + · · · + cn fn (x), x ∈ [a, b],

where c1 , c2 , . . . , cn are arbitrary constants is called a General Solution of

(8) on [a, b]. and each f1 , f2 , . . . , fn are called particular solutions.

Example
Consider the third order linear homogenous DE y 000 − 2y 00 − y 0 + 2y = 0.
Then the functions e x , e −x , e 2x are LI (particular) solutions and hence the
general solution of the given equation is given by:

y (x) = c1 e x + c2 e −x + c3 e 2x , where c1 , c2 , c3 ∈ R.
Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 31 / 111
The Wronskian of Functions
Definition
Let f1 (x), f2 (x), . . . , fn (x) be n real valued functions each of which has an
(n − 1)th derivative on the interval [a, b]. The determinant:

f1 (x) f2 (x) ... fn (x)

f10 (x) f20 (x) ... fn0 (x)
W[f1 , f2 . . . , fn ] = .. .. .. = W(x)
. . .
(n−1) (n−1) (n−1)
f1 (x) f2 (x) . . . fn (x)

is called the Wronskian of these n functions.

Example
The Wronskian, W(x) of y1 (x) = e 2x and y2 (x) = xe 4x is

e 2x xe 4x
W(x) = = e 4x + 2xe 4x − 2xe 4x = e 4x
2e 2x e 4x + 4xe 4x
Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 32 / 111
Wronskian Test for Linearly Independence
Theorem
The n functions f1 , f2 , . . . , fn are Linearly Independent on an interval [a, b]
if and only if the Wronskian of f1 , f2 , . . . , fn is different from zero for
some x ∈ [a, b]. That is, f1 , f2 , . . . , fn are LI if and only if there exists
x ∈ [a, b] such that W(x) 6= 0.

Example
1 Show that x and x 2 are Linearly Independent.
Solution:
Consider the Wronskian of x and x 2 ,

x x2
W(x, x2 ) = = 2x 2 − x 2 = x 2
1 2x

This implies W(x, x 2 ) = x 2 6= 0, ∀x 6= 0 and hence x and x 2 are LI.

2 Show that e x , e −x , e 2x are Linearly Independent.
Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 33 / 111
Homogeneous LODE with Constant Coefficients

Definition
A Differential Equation

bn y (n) + bn−1 y (n−1) + · · · + b1 y 0 + b0 y = 0, (10)

where b0 , b1 , . . . , bn are all real constants, is called a Homogenous Linear

Differential Equation of constant coefficients.

Let f (x) be any solution of (10) in [a, b]. Then

bn f (n) (x) + bn−1 f (n−1) (x) + · · · + b1 f 0 (x) + b0 f (x) = 0 for all x ∈ [a, b].

Hence the derivatives of f are linearly dependent since at least one of the
coefficients b0 , b1 , . . . , bn is different from zero.
The simplest case with this property is a function f such that
f (k) (x) = cf (x), ∀x ∈ [a, b] for some constant c.

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 34 / 111
Homogeneous LODE with Constant Coefficients...
Let f (x) = e λx . Then f k (x) = λk f (x) = λk e λx which implies c = λk .
Thus we will look for the solution of (10) in the form y = e λx where the
constant λ will be chosen so that y = e λx does satisfy the equation (10).
Now inserting y = e λx into (10) gives;

(bn λn + bn−1 λn−1 + · · · + b1 λ + b0 )e λx = 0.

Hence, if e λx is a solution of the equation in (10), then λ should satisfy:

bn λn + bn−1 λn−1 + · · · + b1 λ + b0 = 0, (11)

since e λx 6= 0 for all x ∈ R.
Definition
The algebraic equation (11) is called an Auxiliary equation or
characteristic equation of the given differential equation in (10).
There are 3 different cases of the roots of (11)

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 35 / 111
Case 1. Distinct Real Roots
Suppose that (11) has n distinct roots, λ1 , λ2 , . . . λn where λi 6= λj , for
i 6= j. Then, the solutions e λ1 x , e λ2 x , . . . , e λn x are linearly independent.
If λ1 , λ2 , . . . , λn are the n distinct real roots of (11), then the general
solution of (10) is:
n
X
y (x) = c1 e λ1 x + c2 e λ2 x + · · · + cn e λn x = ci e λi x ,
i=1

where c1 , c2 , . . . , cn are arbitrary constants.

Example
For the differential equation y 000 − 4y 00 + y 0 + 6y = 0, the corresponding
characteristic equation is: λ3 − 4λ2 + λ + 6 = 0 with distinct real roots
λ1 = 2, λ2 = 3 and λ3 = −1.
Therefore, the general solution of the give equation is
y (x) = c1 e 2x + c2 e 3x + c3 e −x .

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 36 / 111
Case 2. Repeated Real Roots
Theorem
1 If the characteristic equation (11) has the real root λ occurring k
times (i.e.λ1 = λ2 = · · · = λk ) where k ≤ n, then the part of the
general solution for (10) corresponding to this k fold repeated root is

(c1 + c2 x + · · · + ck x k−1 )e λx

2 If further, the remaining roots are the distinct real roots

λk+1 , λk+2 , . . . , λn , the general solution of (10) will be:

y (x) = c1 e λx + c2 xe λx + · · · + ck x k−1 e λx + ck+1 e λk+1 x + · · · + cn e λn x .

Example
Consider the Differential Equation

y (4) − 5y 000 + 6y 00 + 4y 0 − 8y = 0.
Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 37 / 111
Solution

The corresponding characteristic equation is

λ4 − 5λ3 + 6λ2 + 4λ − 8 = 0

and the roots are

λ1 = λ2 = λ3 = 2 and λ4 = −1.

Therefore, the general solution is given by

y (x) = c1 e 2x + c2 xe 2x + c3 x 2 e 2x + c4 e −x ,

where c1 , c2 , c3 and c4 are constants.

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 38 / 111
Case 3. Conjugate Complex Roots

Suppose the equation (11) has a complex root λ = a + ib, a, b ∈ R. Then

(we know from the theory of algebraic equations that) the conjugate
λ̄ = a − ib is also a root of (11) and the corresponding part of the general
solution of (10) will be:

k1 e (a+ib)x + k2 e (a−ib)x .

But e a+ib = e a e ib = e a (cos b + i sin b), (by applying Euler’s formula) and
then

k1 e (a+ib)x + k2 e (a−ib)x = k1 e ax (cos bx + i sin bx) + k2 e ax (cos bx − i sin bx)

= e ax [(k1 + k2 ) cos bx + i(k1 − k2 ) sin bx]
= e ax (c1 cos bx + c2 sin bx),

where c1 = k1 + k2 and c2 = i(k1 − k2 ) are arbitrary constants from the

set of complex numbers C.
Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 39 / 111
Case 3. Conjugate Complex Roots...
On the other hand if a + ib = λ and a − ib = λ̄ are each k fold roots of
(11), then the part of the general solution that corresponds to this part is
    
ax k−1 k−1
e c1 + c2 x + · · · + ck x cos bx + i ck+1 + · · · + c2k x sin bx .

Example
Solve y 00 − 2y 0 + 10y = 0.
Solution
The characteristic equation of the given equation is λ2 − 2λ + 10 = 0 with
roots λ1 = 1 + 3i and λ2 = 1 − 3i. Then y1 = e x cos 3x and y2 = e x sin 3x
are two independent solutions of the given equation. Therefore,
y = c1 y1 + c2 y2 , where c1 and c2 are arbitrary constants, is a general
solution of the given equation. That means

y (x) = e x (c1 cos 3x + c2 sin 3x).

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 40 / 111
Nonhomogeneous Equations with Constant Coefficients
Recall that differential equations of the form
bn (x)y (n) + · · · + b1 (x)y 0 + b0 (x)y = f (x), where f (x) 6≡ 0 (12)
are called nonhomogeneous differential equations. In our previous
discussion we have seen how to solve homogeneous differential equations.
Now we want to see how to solve differential equations of the form

bn y (n) + bn−1 y (n−1) + · · · + b1 y 0 + b0 y = f (x), (13)

where bn , ..., b0 are constants is called a nonhomogeneous differential
equation with constant coefficients.
Theorem (Homogeneous-Nonhomogeneous Solution Relation)
Consider the nonhomogeneous differential equation

bn (x)y (n) + · · · + b1 (x)y 0 + b0 (x)y = f (x), where f (x) 6≡ 0.

If f (x) ≡ 0, then the equation becomes a homogeneous equation.

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 41 / 111
Nonhomogeneous Equations with Constant Coefficients

Theorem
theorem continued....
1. If y1 and y2 are solutions of the nonhomogeneous equation on an
interval I, then y1 − y2 is also a solution of the homogeneous equation
in the interval I.
2. If y1 is a solution of the nonhomogeneous equation and y2 is a
solution of the homogeneous equation in an interval I, then y1 + y2 is
a solution of the nonhomogeneous equation in the interval I.

From the theorem we can see that, if yh (x) denote the general solution of
the homogeneous part of (13) and yp (x) denote a particular solution of
the DE (13), then the general solution of (13) is given by

y (x) = yh (x) + yp (x).

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 42 / 111
General Solution

Theorem (Superposition Principle)

If yh (x) is a general solution of the homogeneous part of (13) on an
interval [a, b] and yp1 (x), yp2 (x), . . . , ypk (x) are particular solutions of (13)
corresponding to f1 (x), f2 (x), . . . , fk (x) respectively on the right hand side,
then the general solution of (13) where, f (x) = f1 (x) + · · · + fk (x) on
[a, b], is
y (x) = yh (x) + yp1 (x) + yp2 (x) + · · · + ypk (x).

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 43 / 111
The Method of Undetermined Coefficients

Definition
1. A function is called an undetermined coefficient function (UC
function) if it is either:
a) a function defined by (a linear combination) of the following
i) x n , n = 0, 1, 2, . . . ,
ii) e ax , where a is any non-zero constant
iii) sin(bx + c), where b, c are constants, such that b 6= 0.
iv) cos(bx + c), where b, c are constants, such that b 6= 0.
or
b) a function which is defined as a finite product of two or more functions
of the above 4 types.
2. Let f be an UC function. A set S of functions consisting of f and all
the derivatives of f which are mutually LI UC functions is said to be
the UC set of function f.

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 44 / 111
The Method of Undetermined Coefficients...

Example

1. Let f (x) = x 3 . Then f is UC function and

f 0 (x) = 3x 2 , f 00 (x) = 6x, f 000 (x) = 6.
Therefore, a UC set of f is S = {1, x, x 2 , x 3 }.
2. Let f (x) = sin(2x). Then f is an UC function and
f 0 (x) = 2 cos(2x), f 00 (x) = −4 sin(2x).
Therefore, a UC set of f is S = {sin(2x), cos(2x)}.
3. Let g (x) = 2xe −x . g is an UC function (as a product of UC function)
and g 0 (x) = 2e −x − 2xe −x and g 00 (x) = −4e −x + 2xe −x .
Therefore, a UC set of g is S = {e −x , xe −x }.
4. The function
1
f (x) =
x
is not a UC function.

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 45 / 111
The Method of Undetermined Coefficients...
We outline the solution method by using the following example.
Example
Consider the differential equation

y (4) − y 00 = 3x 2 − sin 2x (14)

First find the general solution to the homogeneous part

y (4) − y 00 = 0.

The characteristic equation of the given equation is λ4 − λ2 = 0. Then

λ2 (λ2 − 1) = 0 and hence λ1 = λ2 = 0 and λ3 = 1, λ4 = −1. Therefore,
the general solution of the homogenous part is:

yh (x) = c1 + c2 x + c3 e x + c4 e −x .

The non- homogeneous term is a combination of x 2 and sin 2x.

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 46 / 111
The Method of Undetermined Coefficients...
Next find the set of UC functions corresponding to the component
functions f1 (x) = 3x 2 with S1 = {x 2 , x, 1} and f2 (x) = − sin 2x with
S2 = {sin 2x, cos 2x}.
To find a particular solution yp1 (x) corresponding to f1 (x), tentatively we
seek it to be a linear combination of the functions in S1 , i.e.
yp1 (x) = Ax 2 + Bx + C ,
where A,B and C are called the undetermined constants.
Check each term in yp1 (x) for duplication with terms in yh (x). Here
the Bx and C terms are constant multiples of c2 x and c1 respectively.
If there is any duplicate, then successively multiply each member of Si
by the lowest positive integral power of x, until (so that) the resulting
revised set contains no duplicate of the terms in the homogeneous
(and previously found particular ypi ’s) solutions.
yp1 (x) = x 2 (Ax 2 + Bx + C ) = Ax 4 + Bx 3 + Cx 2 , no more duplicate.
Substitute the final revised form into the equation and determine the
coefficients A,B and C.
Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 47 / 111
The Method of Undetermined Coefficients...

(4)
yp1 − yp001 (x) = 3x 2 which implies 24A − 12Ax 2 − 6Bx − 2C = 3x 2 .
Equating the coefficients of like terms we get:

 −8A = 3
−6B = 0
24A − 2C = 0


−3 −9
This implies, A = 8 , B = 0 and C = 2 .
Therefore,
3 −9 2
yp1 (x) = − x 4 − x .
8 2

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 48 / 111
The Method of Undetermined Coefficients...
Next, we need to find yp2 (x) which corresponds to f2 (x) = − sin 2x. We
seek yp2 (x) to be a linear combination of the elements of S2 , that is,
yp2 (x) = D sin 2x + E cos 2x.

Check for a duplicate both in yh (x) and yp1 (x). – No duplicate.

Then substitute in y (4) − y 00 = − sin 2x which implies
(4)
yp2 (x) − yp002 (x) = − sin 2x.
Hence,
(24 D sin 2x + 24 E cos 2x) − (−22 D sin 2x − 22 cos 2x) = − sin 2x.

Therefore, 20D sin 2x + 20E cos 2x = − sin 2x and then

1
20D = −1 ⇐⇒ D = − 20 and 20E = 0. Therefore,
1
yp2 (x) = − 20 sin 2x.
Hence the general solution of (14) is: y (x) = yh (x) + yp1 (x) + yp2 (x) =
C1 + C2 x + C3 e x + C4 e −x − − 38 x 4 − −9 2 1
2 x − 20 sin 2x.
Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 49 / 111
Variation of Parameters
The variation of parameters method can be used to find a particular solution of
the nonhomogeneous nth order linear differential equation
y (n) + an−1 (x)y (n−1) + · · · + a1 (x)y 0 + a0 (x)y = f (x),
provided f is continuous
Consider the following second order linear differential equation.

y 00 + b1 (x)y 0 + b2 (x)y = f (x), (15)

where b1 , b2 and f are continuous functions. Suppose that the general solution
for the homogeneous part of (15) is
yh (x) = c1 y1 (x) + c2 y2 (x).
Now we want to get a particular solution corresponding to f (x) and this can be
done by varying the constants, c1 and c2 with respect to x. If yp is a particular
solution corresponding to f (x), then
yp (x) = c1 (x)y1 (x) + c2 (x)y2 (x).
We differentiate and substitute it in (15) to get
yp00 (x) + b1 (x)yp0 + b2 (x)yp = f (x).
Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 50 / 111
Variation of Parameters ...
Since we are going to have only one equation with two variable functions c1 and
c2 , we are free to choose a condition which simplifies the equation. Therefore, we
take the condition c10 y1 + c20 y2 = 0. This will simplify the equation as
yp00 = c1 y100 + c10 y10 + c20 y20 + c2 y200
and after simplification, the equation (15) becomes
c1 (y100 + b1 y10 + b2 y1 ) + c2 (y200 + b1 y20 + b2 y2 ) + c10 y10 + c20 y20 = f .
Since y1 and y2 are linearly independent solutions for the homogeneous part of
equation (15) we have the following system of equations:
 0 0
c1 y1 + c20 y20 = f
(16)
c10 y1 + c20 y2 = 0,
which is a system of two algebraic equations in c10 and c20 . Then (17) has a unique
solution if the determinant of the coefficient matrix is non-zero, that is,

y1 (x) y2 (x)
6= 0.
y10 (x) y20 (x)
Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 51 / 111
Variation of Parameters...
However, the above determinant is the Wronskian of the functions y1 and
y2 . Since y1 and y2 are LI functions, then
W[y1 ,y2 ] (x) 6= 0.
Hence by Cramer’s rule we have:

0 y2
f y20 W1 (x)
c01 (x) = =
W (x) W (x)
and
y1 0
y0 f W2 (x)
c02 (x) = 1 =
W (x) W (x)
By integrating both sides we will get:
Z  Z 
W1 (x) W2 (x)
yp (x) = dx y1 (x) + dx y2 (x).
W(x) W(x)
Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 52 / 111
Variation of Parameters for n−th order ODE

As with second order equations, begin by assuming y1 , y2 · · · , yn are

fundamental solutions to homogeneous equation. Next, assume the
particular solution y has the form

y (x) = u1 (x)y1 + u2 (x)y2 + · · · + un (x)yn (x),

where u1 , u2 , · · · un are functions to be solved for. In order to find these n

functions, we need n equations

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 53 / 111
First let us consider the derivative of y .

y 0 = (u1 y1 + · · · + un yn0 ) + (u10 y1 + · · · + un0 yn ).

If we require
u10 y10 + · · · + un0 yn = 0,
then
y 00 = (u1 y100 + · · · + un yn00 ) + (u10 y10 + · · · + un0 yn0 ).
Next we require
u10 y10 + · · · + un0 yn0 = 0.
Continuing in this way, we require
(k−1) (k−1)
u10 y1 + · · · + un0 yn = 0, for k = 1, 2, · · · , n − 1.

and hence
(k) (k)
y (k) = u1 y1 + · · · + un yn , for k = 1, 2, · · · , n − 1.

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 54 / 111
Then
(n) (n) (n−1) (n−1)
y n = (u1 y1 + · · · + un yn ) + u10 y1 + · · · + un0 yn .

Next, substitute these derivatives into our equation

y (n) + an−1 (x)y (n−1) + · · · + a1 (x)y 0 + a0 (x)y = f (x),

Recalling that y1 , y2 , · · · , yn are solutions to homogeneous equation, and

after rearranging terms, we obtain
(n−1) (n−1)
u10 y1 + · · · + un0 yn = f (x).

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 55 / 111
The n equations needed in order to find the n functions u1 , u2 , un are
u10 y1 + · · · + un0 yn

 = 0
0 0 0 0


 u1 y1 + · · · + un yn = 0
.. .. (17)

 . .
 0 (n−1)
 (n−1)
u1 y1 + · · · + un0 yn = f (x)
Using Cramers Rule, for each k = 1, . . . , n,
f (x)Wk (x)
uk0 (x) = , where W (x) = W (y1 , y2 , · · · , yn )
W (x)
and Wk (x)is the determinant obtained by replacing k−th column of W
with (0, 0, · · · , 1). Then integrate to obtain
Z x
f (t)Wk (t)
uk (x) = dt, for k = 1, 2, ..., n.
x0 W (t)
Thus, a particular solution y is given by
n Z x 
X f (t)Wk (t)
y (x) = dt yk ,
x0 W (t)
k=1
where x0 is arbitrary.
Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 56 / 111
Example
Consider the following DE.

y 000 − y 00 − y 0 + y = e 2x .

Then the fundamental set of solutions of the given DE is

y1 = e x , y2 = xe x , y3 = e −x .
Thus a particular solution y is given by
3 Z x
e 2t Wk (t)
X 
y (x) = dt yk ,
x0 W (t)
k=1

where

ex xe x e −x
W (x) = e x (x + 1)e x −e −x = 4e x ,
ex (x + 2)e x e −x

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 57 / 111
Example
Example continued ...

0 xe x e −x
W1 (x) = 0 (x + 1)e x −e −x = 4e x = −2x − 1,
1 (x + 2)e x e −x

ex 0 e −x ex xe x 0
W2 (x) = e x 0 −e −x = 2 and W3 (x) = e x (x + 1)e x 0 = ex .
ex 1 e −x ex (x + 2)e x 1

3 Z x
e 2t Wk (t)
X 
Thus a particular solution is given by y (x) = dt yk
x0 W (t)
k=1

This implies
x x x
e 2t (−2t − 1) 2e 2t e 2t e t
Z Z Z
y (x) = e x
dt + xe x dt + e −x dt
x0 4e 2t x0 4e 2t x0 4e 2t
Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 58 / 111
Example
Example continued ...
This implies

e x x 2t xe x x t e −x x 4t
Z Z Z
y (x) = − e (2t + 1)dt + e dt + e dt
4 x0 2 x0 4 x0

Then choosing x0 = 0, we obtain

ex xe x e −x e 2x
y (x) = − − − + .
4 2 12 3
Therefore the general solution of the given DE is given by
1
y (x) = c1 e x + c2 xe x + c3 e −x + e 2x ,
3
where c1 , c2 , c3 are arbitrary constants.

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 59 / 111
 Solution Methods
1. The Laplace Transform Method

Definition (Laplace Transform)

The Laplace Transform of a function f (t), if it exists, denoted by L{f (t)} is
given by, Z ∞
L{f (t)} = e −st f (t)dt,
0
where s is a real number called a parameter of the transform. For short we may
write, Z ∞
F(s) to denote e −st f (t)dt.
0

Example
Find the Laplace Transform of the constant function f (t) = 1.
Z ∞
L{1} = e −st × 1dt.
0
1
Answer: L{1} = s if s > 0.
Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 60 / 111
Laplace Method · · · Cont’d.

Table of basic Laplace Transforms

Function (f (t)) Laplace Trnsf. (F (s))
1
1 s,s > 0
n n!
t ,n ∈ N s n+1
,s > 0
1
e tk s−k s > k
,
n kt n!
t e (s−k)n+1
,s > k
k
sin(kt) s 2 +k 2
,s > 0
s
cos(kt) s 2 +k 2
,s > 0
k
sinh(kt) s 2 −k 2
, s > |k|
s
cosh(kt) s 2 −k 2
, s > |k|

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 61 / 111
Laplace Method · · · Cont’d.
From the table above we have
1
L{e at } = for s > a.
s −a
1
Thus the inverse operator applied on s−a will give us back the function e at

1
i.e., L−1 { } = e at for s > a.
s −a
In general, L−1 , the inverse Laplace Operator, is given by
Z γ+i∞
−1 1
L {F (s)} = F (s)e st ds,
2πi γ−i∞

(where γ is a positive real number), which is a complex improper integral.

In solving DEs using Laplace Transform Method, we need to calculate
both the transform and the inverse operator. Therefore ne need to
understand some of their properties.
Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 62 / 111
Laplace Method · · · Cont’d.
Theorem (Linearity)
(a) If u(t), v (t) are functions and α, β are any constants, then

L{αu(t) + βv (t)} = αL{u(t)} + βL{v (t)}.

(b) For any functions U(s), V (s) and any given scalars α, β, we have

L−1 {αU(s) + βV (s)} = αL−1 {U(s)} + βL−1 {V (s)}.

Example
Evaluate the following transforms
5s 2 +3s+6
1. L{3t + 5e −2t }. Use linearity and the table. Ans. s 3 +2s 2
.
2
2. L{cos2 3t}. Use half angle formula and the table. Ans. ss3 +36s
+18
.
n 2 o
3. L−1 (s+1)
s
3 . Use partial fraction and the table. Ans.

(1 − 2t + 12 t 2 )e −t .
Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 63 / 111
Laplace Method · · · Cont’d.

Theorem (Transform of the Derivative)

Let f (t) be continuous and f 0 (t) be piecewise continuous on some interval
[0, to ] for every finite to , and let |f (t)| < Ke ct for some constants K , T ,
and c and for all t > T . Then the transform L{f 0 (t)} exists for all s > c
and
L{f 0 (t)} = sL{f (t)} − f (0).

Example (Applications of Laplace Method to solve ODE)

Use the Laplace transform Method to solve the initial-value problem.

y 0 + 2y = 0 with y (0) = 1.

Ans.: y (t) = e −2t is the solution.

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 64 / 111
Laplace Method · · · Cont’d.
We can also use the Laplace Method to solve higher order equations with
constant coefficients. For these purposes the following property of the
Transform is required.
Theorem
In general, if f (n) (t) is piecewise continuous for t ≥ 0, then we have

L{f (n) (t)} = s n L{f (t)} − s n−1 f (0) − s n−2 f 0 (0) − · · · − f (n−1) (0).

Example
Solve the following IVPs using Laplace Transform.
1 y 00 − y 0 − 6y = cos 2t, with y (0) = 2, y 0 (0) = 0. Ans.
9 −2t
y (t) = 107 3t
65 e + 20 e
43
− 260 1
cos 2t − 52 sin 2t for t > 0.
2 y 000 − 2y 00 − y 0 + 2y = te t , with y (0) = 2, y 0 (0) = 1 and y 00 (0) = 0.
1 −t
Ans.y (t) = 74 e t − 23 e t − 12 e + 13 e 2t .

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 65 / 111
Theorem (First shifting theorem)
If L{f (t)} = F(s) for Re(s) > b, then L{e at f (t)} = F(s − a) for
Re(s) > a + b.

Example
1 Find the Laplace transform for the function f (t) = e 3t cos 4t.
s
Solution: Recall that L{cos 4t} = 2 .
s + 42
Then using the first shifting theorem we get
s −3
L{e 3t cos 4t} = .
(s − 3)2 + 42
s
2 Find the inverse Laplace transform for the function F(s) = s 2 +s+1
.
Solution: First let us rewrite the function F(s) as

s s s + 21 1
2
F(s) = = 1 3
= −
2
s +s +1 (s + 2 )2 + 4 (s + 12 )2 + 3
4 (s + 1 2
2) + 3
4

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 66 / 111
Example
Example continued...
and hence,
( ) ( √ )
s + 12 3
 
−1 s −1 −1 1 2
L =L −L √ .
s2 + s + 1 (s + 12 )2 + 3
4 3 (s + 21 )2 + 3
4

Then using the first shifting theorem, we have,

  √ √
−1 s −t/2 3t 1 −t/2 3t
L =e cos −√ e sin .
s2 + s + 1 2 3 2

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 67 / 111
Consider the general Laplace transform formula
Z ∞
F(s) = e −st f (t)dt.
0
Taking the derivative with respect to s on both sides we get,
Z ∞
0
F (s) = (−t)e −st f (t)dt = L{−tf (t)}.
0
By further differentiating the above equation with respect to s, we get
F 00 (s) = L{t 2 f (t)}.
In general we have the following theorem.
Theorem (Derivative of the transform)
For a piecewise continuous function f (t) and for any positive integer n, it
holds that
L{(−1)n t n f (t)} = F (n) (s),
where, F(s) = L{f (t)}.
(This formula can be used to find transforms of functions of the form
x n f (x) when the Laplace transform of f (t) is known.)
Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 68 / 111
The Cauchy-Euler Equations
Let us consider linear differential equations with variable coefficients with
some special forms.
Definition
The linear differential equation with variable coefficient of the form:

an x n y (n) + an−1 x n−1 y (n−1) + · · · + a1 xy 0 + a0 y = F (x) (18)

where a0 , a1 , . . . , an are constants is called the Cauchy-Euler Equation.

Example
The linear differential equation 3x 2 y 00 − 11xy 0 + 2y = sin x is a Cauchy-Euler
equation.

To solve Cauchy-Euler DEs first we reduce the given DE into a linear

differential equation with constant coefficients and solve the given
equation with the methods derived in the previous sections.
Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 69 / 111
Theorem
The transformation x = e t , t ∈ R reduces the Cauchy-Euler DE to a linear
DE with constant coefficients.
Let us consider the case when n = 2. In this case the equation is:
a2 x 2 y 00 + a1 xy 0 + a0 y = F (x) (19)
Let x = e t . Then by solving for t we get t = ln x for x > 0 (or x = −e t if
x < 0) and

dy dy dt 1 dy
= . =
dx dt dx x dt
and
d 2y
 
1 d dy dy
= + .
dx 2 x dx dt dt
1 d 2y
 
d 1
= .
dx x x dt 2
 2 
dt 1 dy 1 d y dy
− 2 = − .
dx x dt x 2 dt 2 dt
Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 70 / 111
Substituting into (19) we get:
 2 
2 1 d y dy 1 dy
a2 x 2 2
− + a1 x. + a0 y = F (e t ).
x dt dt x dt
This implies,
d 2y dy
a2 2 + (a1 − a2 ) + a0 y = F (e t ).
dt dt
Then

d 2y dy
2
A2
+ A1 + A0 y = G (t), (20)
dt dt
where A2 = a2 , A1 = a1 − a2 , A0 = a0 and F (e t ) = G (t), which is a
second order linear differential equation with constant coefficients.
Example
Solve each of the following DEs.
1. x 2 y 00 − 2xy 0 + 2y = 0.
2. 3x 2 y 00 − 11xy 0 + 2y = sin x.
3. x 2 y 00 − 2xy 0 + 2y = x 3 .
Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 71 / 111
Solution Methods. . .
2. Power Series Method

Recall that an infinite series of the form

∞
X
an (x − xo )n = a0 + a1 (x − xo ) + a2 (x − xo )2 + · · · (21)
n=0

is called a power series in (x − xo ) and it is said to converge at a point x if

the limit
X m
lim an (x − xo )n
m→∞
n=0

exists and is a real number. Clearly the power series (21) always converges
at x = xo , which is the trivial case. If (21) converges on the interval
(xo − r , xo + r ) ⊂ R for some r > 0 and diverges on |x0 − r | > 0, then we
say r is a radius of convergence of the series. If the sum of the series is ∞
except at x = xo , it is conventional to write r = 0 as its radius of
convergence, and if the series converges for all x ∈ R then we take r = ∞.
Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 72 / 111
If the power series (21) converges with radius of convergence r > 0, the
function
∞
X
f (x) = an (x − xo )n
n=0

can be integrated and differentiated pointwise and hence it a derivative of

all orders for all x such that |x − xo | < r . Then from successive
differentiation we can get that

f (n) (xo )
an = , for all n = 0, 1, 2, . . .
n!

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 73 / 111
Recall: the Taylor series of a given smooth function f about a point ao is:
∞
X f (n) (ao )
TS f |ao = (x − ao )n ;
n!
n=0

If this series converges in some interval |x − ao | < r and is equal to f , then

we call f is analytic at ao and r is the radius of convergence. If f is not
analytic at ao , we say it is singular at ao . Suppose a linear second order
differential equation

a2 (x)y 00 + a1 (x) + a0 (x)y = 0

is given. Then we can put it into the form

y 00 + p(x)y 0 + q(x)y = 0

by dividing by the leading coefficient a2 (x).

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 74 / 111
Definition
A point c0 is called an ordinary point of the differential equation

y 00 + p(x)y 0 + q(x)y = 0

if both p(x) and q(x) are both analytic at x0 . A point that is not ordinary
point is called a singular point of the differential equation.

Example
Every finite value of x is an ordinary point of

y 00 + e x y 0 + (sin x)y = 0.

In particular, we can see that x = 0 is an ordinary point since both

∞ ∞
X xn X x 2n+1
ex = and sin x = (−1)n
n! (2n + 1)!
n=0 n=0

converge for all finite values of x.

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 75 / 111
Example
The differential equation xy 00 + (sin x)y = 0 has an ordinary point at x = 0
since q(x) = sinx x has a power series
∞
X x 2n+1
q(x) = (−1)n
(2n + 1)!
n=0

that converges for all real numbers x.

Example
The differential equation y 00 + (ln x)y = 0 has a singular point at x = 0,
because q(x) = ln x has no power series in x.

In the equation a2 (x)y 00 + a1 (x)y 0 + a0 (x)y = 0, if a2 (x), a1 (x) and a0 (x)

are polynomials with no common factors, then a point x = c0 is
1 an ordinary point if a2 (c0 ) 6= 0, or
2 a singular point if a2 (c0 ) = 0.
Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 76 / 111
The Power Series Method mainly uses the following theorem

Theorem (Existence of Power Series Solution)

If p(x) and q(x) are analytic at a point co , then every solution of the DE

y 00 + p(x)y 0 + q(x)y = 0 (22)

can be found in the form ∞

X
y (x) = an (x − co )n .
n=0

Moreover, the radius of convergence of every solution is at least as large as the smaller
of the radii of convergence of TS p|co and TS q|co .

Definition
A solution of a differential equation y 00 + p(x)y 0 + q(x)y = 0 of the form given by
∞
X
y (x) = an (x − co )n .
n=0

is said to be a power series solution about the ordinary point c0 .

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 77 / 111
Power Series Solution Method · · · Cont’d

Example
Solve y 00 + y = 0.
∞ ∞
X x 2n X x 2n+1
Answer: y (x) = c0 (−1)n + c1 (−1)n
(2n)! (2n + 1)!
n=0 n=0

Example
Solve y 00 + xy = 0.
∞
!
X (−1)k
Ans.: y (x) = c0 1+ x 3k +
2.3...(3k − 1)(3k)
k=1
∞
!
X (−1)k
c1 x+ x 3k+1
3.4...(3k)(3k + 1)
k=1

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 78 / 111
Frobenius Method
Consider again a second order equation with variable coefficients,
h(x)y 00 + p(x)y 0 + q(x)y = 0 (23)
If h(x)  0 for some x we can equivalently have
p(x) 0 q(x)
y 00 + y + y = 0, for h(x) 6= 0. (24)
h(x) h(x)
If h(x) 6= 0 for all x we can simply apply the power series solution method.
But if h(x) = 0 for some x the resulting equation will be different from the
original one at those points x where h(x) = 0.
Definition
A singular point xo is said to be a regular singular point of equation (23) if
the function
p(x) q(x)
(x − xo ) and (x − xo )2
h(x) h(x)
are analytic at xo . A non regular singular point is called an irregular
singular point of equation (23).
Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 79 / 111
Frobenius Method · · · Cont’d.
Theorem (Frobenius’ Theorem)
If x = x0 is a regular singular point of the equation

a2 (x)y 00 + a1 (x) + a0 (x)y = 0,

then there exists at least one series solution of the form

∞
X ∞
X
r n
y (x) = (x − x0 ) an (x − xo ) = an (x − xo )n+r ,
n=0 n=0

where the number r is a constant that must be determined. The series will
converge at least on some interval 0 < x − x0 < R.

If equation (23) has a regular singular point at xo , then use the power
series:
∞
X
y (x) = an (x − xo )n+r
n=0
Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 80 / 111
Indicial Equations
Then differentiating and substituting this series for y (x) in to the modified
equation (24), we get,

∞
X
am (m + r )(m + r − 1)(x − xo )m+r −2
m=0
∞ ∞
" #
X X
+ pm (x − xo )m am (m + r )(x − xo )m+r −1
m=0 m=0
∞ ∞
" #
X X
m
+ qm (x − xo ) am (x − xo )m+r = 0,
m=0 m=0

where
∞ ∞
p(x) X q(x) X
= pm (x − xo )m and = qm (x − xo )m
h(x) h(x)
m=0 m=0

are the power series expansions, as they are analytic at xo .

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 81 / 111
Indicial Equations . . .

Then combining the corresponding powers of (x − xo ) and equating the

coefficients of each powers of (x − xo ) to zero we will arrive at a system of
equations in the variables am and r . Taking only the coefficient for the
constant part we have,

a0 [r (r − 1) + p0 r + q0 ] = 0 (25)

But since a0 6= 0 was assumed we have

r (r − 1) + p0 r + q0 = 0, (26)

which is known as the indicial equation of the differential equation (23).

Its roots, say r1 and r2 , define the two linearly independent solutions y1 (x)
and y2 (x) of the differential equation. Using these values of the roots for
(26), we can determine the remaining coefficients am , m = 0, 1, 2, 3, . . ..
This procedure is called the Frobenius Method.

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 82 / 111
Frobenius Method · · · Cont’d.

Example
Use Frobenius method to solve

x 2 y 00 + 5xy 0 + (x + 4)y = 0

∞
X 1 n−2
Ans.: y (x) = ao (−1)n x
(n!)2
n=0

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 83 / 111
Systems of ODE of the First Order

A system of n linear first-order equations in the n unknowns

x1 (t), x2 (t), . . . , xn (t) is a system that can be written in the form:

x10 = a11 (t)x1 + a12 (t)x2 + · · · + a1n (t)xn + f1 (t)

x20 = a21 (t)x1 + a22 (t)x2 + · · · + a2n (t)xn + f2 (t)
.. (27)
.
xn0 = an1 (t)x1 + an2 (t)x2 + · · · + ann (t)xn + fn (t),

which is called the normal form. In vector form this system becomes:

X0 = AX + F(t),

where

X = (x1 , x2 , . . . , xn )T , A = (aij )n×n , and F(t) = (f1 (t), f2 (t), . . . , fn (t))T

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 84 / 111
Systems of ODE of the First Order · · · continued

The system (27) is called homogeneous if F(t) ≡ 0, so that the

system becomes X0 = AX.
If F(t) 6≡ 0 for some t, the system is nonhomogeneous.

Definition
A solution vector of the system of differential equation in (27) over some
interval I is a a vector (x1 (t), x2 (t), . . . , xn (t))T whose entries are
differentiable functions that satisfies the system in (27) on the interval I .

We see how to solve such systems of equations.

The Eigenvalue Method is considered for this discussion.

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 85 / 111
Eigenvalue Method for Homogeneous Systems with
Constant Coefficients
Consider the system
y0 = Ay, (28)
where A = (aij )n×n is a constant matrix. That is, all the entries of A are
constants.
Given a first order ODE y 0 = ky , we have seen that its solution is given by
y = ce kt , where c is a constant.
Let y = xe λt , where x = (x1 , x2 , . . . , xn )T be a solution of the system in
(28).
Substituting this into (28) we have:

λxe λt = y0 = Ay = Axe λt and e λt 6= 0.

This implies,
Ax = λx,
which is an eigenvalue problem.
Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 86 / 111
Eigenvalue Method for Homogeneous Systems with
Constant Coefficients
Once we find the eigenvalues λi and a corresponding eigenvector xi , the
general solution will be
y(t) = c1 x1 e λ1 t + · · · + cn xn e λn t ,
where c1 , ..., cn are constants.
Example
Solve the following systems of linear differential equations.
 0
y1 = −3y1 + y2
y20 = y1 − 3y2

The system can be written as:

 0    
y1 −3 1 y1
=
y20 1 −3 y2
Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 87 / 111
Example · · · continued
Example
Let y(t) = xe λt . Then the corresponding eigenvalue problem will be:
    
−3 1 x1 x
Ax = λx ⇐⇒ =λ 1
1 −3 x2 x2

The eigenvalues of the coefficient matrix are λ1 = −2 and λ2 = −4

a) An eigenvector corresponding to λ1 = −2. is (1, 1)t .

b) An eigenvector corresponding to λ2 = −4 is (1, −1)t

Therefore, the general solution of the given system is

   
1 −2t 1
y(t) = c1 e + c2 e −4t
1 −1
which is equivalent to
y1 (t) = c1 e −2t + c2 e −4t
y2 (t) = c1 e −2t − c2 e −4t
Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 88 / 111
Nonlinear Ordinary Differential Equations

Many Nonlinear equations cannot be solved in closed form.

Hence we need to develop qualitative methods to determine
properties of solutions without having them explicitly in hand.
The properties tell us the way how all the trajectories (solution
curves) behave near to some points.

Read on Phase portrait and Phase plane Analysis for

ODEs.

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 89 / 111
Nonlinear Systems

Consider the following second order equation which models the motion of
a particle of unit mass moving on the x-axis with a force f (x(t), x 0 (t))
acting on it:
d 2x
 
dx
= f x, (29)
dt 2 dt
In this equation the values of x and x 0 (t) (the position and velocity of the
particle) characterize the state of the system at each instant of time,
hence are called the phases of the system.
The plane of the variables x and x 0 (t) is known as the phase-plane, and
the set of solutions of the equation shown in this plane (for different initial
conditions) is called the phase plane diagram.

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 90 / 111
Nonlinear Systems · · · continued

dx
If we introduce a variable y = , then equation (29) can be equivalently
dt
written as the system
dx


 =y
dt (30)
 dy = f (x, y ).

dt
By regarding t as a parameter, a solution of this system is a pair of
functions (x(t), y (t)) defining a curve in the xy -plane (which is the phase
plane).
Thus we will be interested in the overall picture formed by these curves in
the phase plane.

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 91 / 111
Nonlinear Systems · · · continued

Consider the authonomous Nonlinear system in two variables

dx
= P(x, y )
dt
dy
= Q(x, y ) (31)
dt
If we assume that y is dependent on x, we can equivalently get

dy Q(x, y )
=
dx P(x, y )

Any point (xo , yo ) such that both P and Q vanishes is called a critical (or
singular or equilibrium) point of the system (31).
At a singular point (xo , yo ), since ẋ = 0 and ẏ = 0, a particular solution of
equation (31) is simply the constant values x(t) = xo , y (t) = yo .

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 92 / 111
Stability of Equilibrium points
An equilibrium point Xo = (xo , yo ) of system (31) is said to be stable if
motions (or trajectories) that start sufficiently close to Xo remain close to
Xo .
Mathematically we have the following definition of stability an n
equilibrium point Xo = (xo , yo ) of system (31).

Definition
Let d(P1 , P2 ) denote the distance between any two points P1 = (x1 , y1 )
and P2 = (x2 , y2 ) and let P(t) = (x(t), y (t)) denote the representative
point in the phase plane corresponding to system (31). Then
a singular (or an equilibrium) point Xo = (xo , yo ) is stable if for any
given  > 0, there is a δ > 0 such that

d(P(0), Xo ) < δ ⇒ d(P(t), Xo ) < , ∀t > 0

Otherwise, the equilibrium point Xo is called unstable.

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 93 / 111
Stability of Equilibria · · · Cont’d.

Definition
... continued
A singular point Xo is called asymptotically stable if motions (or
trajectories) that start out sufficiently close to Xo not only stay close
to Xo but actually approach Xo as t → ∞

i.e., ∃δ > 0 s.t. d(P(o), Xo ) < δ ⇒ lim d(P(t), Xo ) = 0

t→∞

Definition
A singular point is called:
1) a Center if it is surrounded by closed orbits (paths) corresponding to
periodic motions.
A center is stable but not asymptotically stable.

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 94 / 111
Stability of Equilibria · · · Cont’d.
Definition
. . . . Cont’d.
2) a Focus (or Spiral) if all trajectories around Xo “focus” towards (or
outward) it as t → ∞.
A fucus can be asymptotically stable or unstable.
3) a Node if there are infinitely many trajectories entering (or leaving)
the point Xo .
There are four cases → Proper or Improper nodes with each could be
stable or unstable.
4) a Saddle if all trajectories (paths) approach to Xo in one direction and
move away from it in the other direction.
A saddle is always unstable.
The two straight-line trajectories through the saddle (along which the
flow is attracted and repelled) are called the stable and unstable
manifolds respectively.
Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 95 / 111
Stability of Equilibria · · · Cont’d.

In many practical problems we will be interested in the stability of

equilibrium points
That means, we will study if we take an initial point near to an
equilibrium point Xo = (xo , yo ), does the point (x(t), y (t)) on the
solution curve (trajectory) remain near Xo ?
To study this, we approximate the nonlinear system (equation (31))
by its linear terms in the Taylor series expansion in the nbhd of each
singular point.
First let us see stability in Linear Systems.

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 96 / 111
Stability for Linear Systems

Consider a system of linear equations,

ẋ(t) = Ax(t), (32)

where A is a constant n × n matrix and x(t) is a vector function of n

components.
Then the solution of the system depends on the eigenvalues of the
coefficient matrix A.
If A has n distinct real eigenvalues, λ1 , λ2 , . . . , λn , then the general
solution of (32) is given by,

x(t) = c1 e λ1 t v1 + c2 e λ2 t v2 + · · · + cn e λn t vn , (33)

where the vectors vk is an eigenvector of A corresponding to the

eigenvalue λk for each k = 1, 2, . . . , n.

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 97 / 111
Stability for Linear Systems...

It is clear that x(t) ≡ 0, where x(t0 ) = 0 for initial time t0 , is also a

solution of the system (32) (the trivial solution).
0 is a critical point for the given linear system.
To see the relationship between the 0 solution and the general
solution of the linear system when t increases, consider the limit,
h i
lim x(t) = lim c1 e λ1 t v1 + c2 e λ2 t v2 + · · · + cn e λn t vn
t→∞ t→∞
   
= lim e λ1 t c1 v1 + · · · + lim e λn t cn vn ,
t→∞ t→∞

Since the vectors ck vk are independent of t, for all k = 1, . . . , n.

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 98 / 111
Stability for Linear Systems...

Thus we have the following possibilities about the solutions of the system.
If all the eigenvalues λk ’s are positive real numbers, the limit of the
exponential functions tends to infinity;
if all the eigenvalues are negative real numbers each limit approaches
asymptotically to 0.
Hence the general solution of the system converges to 0 if all the
eigenvalues are negative real numbers and diverges if some of the
eigenvalues are positive.

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: Ordinary November
Differential26,
Equations
2023 99 / 111
Stability for Linear Systems...

If some of the eigenvalues are complex, λk = ak + ibk , for some

k = 1, . . . , n, then the corresponding solution is of the form

xk (t) = e ak t ck1 cos bk t + ck2 sin bk t vk .

Then the convergence of the solution xk (t) depends on the real part of the
eigenvalues. That is,
if Re(λk ) < 0 the solution converges to 0,
otherwise it diverges as the amplitudes increases indefinitely.

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: OrdinaryNovember
Differential
26,Equations
2023 100 / 111
Stability for Linear Systems...

To formulate a general characterization for the stability of the solution of a

linear system, we need the result of the following Lemma.
Lemma
Let α, β ∈ R and k ∈ N ∪ {0}, and let the function f (t) be defined by

f (t) = t k e αt cos βt or f (t) = t k e αt sin βt.

Then
i) lim f (t) = 0 if and only if α < 0.
t→∞
ii) There exists a constant C , such that |f (t)| ≤ C for all t ≥ 0 if and
only if either α < 0 or α = 0 and k = 0.

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: OrdinaryNovember
Differential
26,Equations
2023 101 / 111
Stability for Linear Systems...

Theorem (Stability of Zero Solution for Linear Systems)

A system of first order differential equation

ẋ(t) = Ax(t)

with constant coefficient matrix A, has asymptotically stable zero solution

if and only if Re(λ) < 0 for all eigenvalue λ of A.
If there is an eigenvalue λ, with Re(λ) > 0, then the zero solution is
unstable.
Therefore, the stability of a zero solution depend on the sign of the real
parts of all the eigenvalues of the coefficient matrix A.

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: OrdinaryNovember
Differential
26,Equations
2023 102 / 111
Example
Consider the following the system of differential equations
dx
= 2x − 6y + z,
dt
dy
= 3x − 3y − z,
dt
dz
= 3x − y − 3z.
dt

2 −6 1
The coefficient matrix is then, A =  3 −3 −1  with eigenvalues
3 −1 −3
√ √
λ1 = −2, λ2 = −1 + i 6, and λ3 = −1 − i 6. Then the real parts of the
eigenvalues are −2 and −1,both negative,
  andhence by the above
x(t) 0
theorem the zero solution,  y (t)  ≡  0  is asymptotically stable.
z(t) 0

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: OrdinaryNovember
Differential
26,Equations
2023 103 / 111
Stability - Linearization Technique.
Consider again the system:
dx
= P(x, y )
dt
dy
= Q(x, y )
dt
with critical point (x0 , y0 ).
From the Taylor series (by tangent plane approximation) we have
P(x, y ) ≈ Px (xo , yo )(x − xo ) + Py (xo , yo )(y − yo )
Q(x, y ) ≈ Qx (xo , yo )(x − xo ) + Qy (xo , yo )(y − yo )
Now letting a = Px (xo , yo ), b = Py (xo , yo ), c = Qx (xo , yo ) and
d = Qy (xo , yo ) we have
Ẋ = aX + bY
Ẏ = cX + dY , (34)
where X = x − xo and Y = y − yo .
Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: OrdinaryNovember
Differential
26,Equations
2023 104 / 111
 The above process is called a linearization process.
System (34) can be rewritten as
    
Ẋ a b X
=
Ẏ c d Y

Clearly (0, 0) is a critical point for the linear system (34) [and hence
the point (xo , yo ) is a critical point for the system (31)]
Let
 λ1 and λ2 be the two eigenvalues of the coefficient matrix
a b
.
c d
Then the nature of the critical point (0, 0) of the system (34)
depends upon the nature of the eigenvalues λ1 and λ2 .

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: OrdinaryNovember
Differential
26,Equations
2023 105 / 111
1. If λ1 and λ2 are real, unequal and of the same sign, then the critical
point (0, 0) of the linear system (34) is a node.
If, in addition, both λ1 and λ2 are positive, then the critical point is an
unstable node.
If, both λ1 and λ2 are negative, then the critical point is a stable node.
2. If λ1 and λ2 are real and of opposite sign, then the critical point
(0, 0) of the linear system (34) is a saddle point.
3. If λ1 and λ2 are real and equal, then the critical point (0, 0) of the
linear system (34) is a node.
If, in addition, λ1 = λ2 < 0, then it is a stable node and if
λ1 = λ2 > 0, then it is an unstable node.
If, a = d 6= 0 and b = c = 0, then it is a proper node, otherwise an
improper node.

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: OrdinaryNovember
Differential
26,Equations
2023 106 / 111
4. If λ1 and λ2 are complex conjugates with the real part not zero, then
the equilibrium point (0, 0) of the linear system (34) is a focus or
spiral.
If, in addition, the real part is negative, then the critical point is a
stable focus.
If, the real part is positive then it is an unstable focus.
5. If λ1 and λ2 are pure imaginary, then the equilibrium point (0, 0) of
the linear system (34) is a center.
A center is always stable even though it is not asymptotically stable.

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: OrdinaryNovember
Differential
26,Equations
2023 107 / 111
Remark:

1. In the linearization process it was assumed that

the constants a, b, c, and d are real numbers;
the functions P and Q have continuous first partial derivatives in the
nbhd of the critical points.
The above two requirements will be met, if the Jacobian
∂P ∂P
∂(P, Q) ∂x (xo , yo ) ∂y (xo , yo )
= ∂Q ∂Q 6= 0.
∂(x, y ) (xo ,yo ) ∂x (xo , yo ) ∂y (xo , yo )

2. The constant terms in the linearized system are missing because

P(xo , yo ) = Q(xo , yo ) = 0.
3. The nature of the equilibrium points of the nonlinear system (31) can
be determined from that of the linearized system (34) as in the
following Theorems.

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: OrdinaryNovember
Differential
26,Equations
2023 108 / 111
Theorem (Poincare’s Result)
The classification of all singular points of the non-linear system (31)
correspond in both type and stability with the results obtained by
considering the linearized system (34) except for a center and a proper
node.
In theses exceptional cases
(i) a center of the linearized system could be either a focus or a center
for the nonlinear system.
(ii) a proper node could also be either a spiral or a node for the nonlinear
system (31).
To determine these exceptional cases one requires to study further the
original nonlinear system itself.

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: OrdinaryNovember
Differential
26,Equations
2023 109 / 111
The above procedure (the linearization process) can also be used to solve
a second order non-linear ODE. This can be done by using substitution of
variables y = ẋ, which imply that ẏ = ẍ. This will result in a nonlinear
system of two first order equations.
However, if such an equation has no term in ẋ, we need the following
theorem.
Theorem
If the nonlinear equation ẍ + f (x) = 0 has a singular point in the x ẋ plane
(phase plane), where the linearized system indicates a center or a proper
node, the nonlinear equation also has the same property.

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: OrdinaryNovember
Differential
26,Equations
2023 110 / 111
Example
1. The pair of differential equations
1
ẋ = x − xy
2
ẏ = −2y + xy , x, y ≥ 0,

occur in a study of interacting populations. Find the equilibrium

points and determine their nature.
Ans.: (0, 0) is a saddle equilibrium (not stable)
and the equilibrium point (2, 1/2) is a center (stable).
2. The equation ẍ + ẋ 3 + x = 0 models a harmonic oscillator with cubic
damping - that is, with a damping term proportional to the velocity
cubed. Find the critical point(s) and determine their nature.

Ans.: the point (0, 0) is the only critical point and it is a center for
the nonlinear equation.

Tilahun Abebaw (PhD) (AAU) Advanced Mathematics for Engineers Part II: OrdinaryNovember
Differential
26,Equations
2023 111 / 111