Chapter 2

Ordinary Differential
Equations

2.1 Introduction
A differential equation is an equation involving an unknown function and
its derivatives. A differential equation is an ordinary differential equation
(ODE) if the unknown function depends on only one independent variable.
If the unknown function depends on two or more independent variables, the
differential equation is a partial differential equation (PDE).

An ODE can be represented as

 
F x, y, y ′ , y ′′ , . . . , y (n) = 0 (2.1)

where y = y(x) is the sought-for function.

The function y = ϕ(x), which converts (2.1) into an identity, is called the so-
lution of the equation. If the solution is represented implicitly, Φ(x, y) = 0,
then it is called an integral.

The order of a differential equation is the order of the highest derivative

appearing in the equation.

Example 2.1 Check that y = sin x is a solution of the equation y ′′ + y = 0. ◀

The integral
Φ(x, y, c1 , c2 , . . . , cn ) = 0 (2.2)
of the differential equation (2.1), which contains n independent arbitrary
constants c1 , c2 , . . . , cn and is equivalent (in the given region) to equation

1
 Class Notes on ECEG-6201
2.1. INTRODUCTION Analytical & Comp. Methods

(2.1), is called the general solution. By assigning definite values to the con-
stants c1 , c2 , . . . , cn in (2.2), we get particular solutions.

Conversely, if we have a family of curves (2.2) and eliminate the parameters

c1 , c2 , . . . , cn from the system of equations
dΦ dn Φ
Φ = 0, = 0, . . . , n = 0,
dx dx
we, generally speaking, get a differential equation of type (2.1) whose gen-
eral solution in the corresponding region is the relation (2.2).

Example 2.2 Find the differential equation of the family of parabolas y = c1 (x −

c2 )2 . ◀

A differential equation along with subsidiary conditions on the unknown

and its derivatives, all given at the same value of the independent variable,
constitutes an initial-value problem. The subsidiary equations are initial
conditions. If the subsidiary conditions are given at more than one value
of the independent variable, the problem is boundary-value problem and the
conditions are boundary conditions.

For example, the problem y ′′ + 2y ′ = ex , y(π) = 1, y ′ (π) = 2 is an initial

value problem, because the two subsidiary conditions are given at x = π.
While the problem y ′′ + 2y ′ = 2ex , y(0) = 1, y ′ (1) = 1 is a boundary-value
problem because the two subsidiary conditions are given at x = 0 and x = 1.

Example 2.3 Find the curve of the family y = c1 ex + c2 e−2x for which
y(0) = 1, y ′ (0) = −2.

Exercise 2.1 Show that for the given differential equations the indicated relations
are integrals (solutions)
1. (x − 2y)y ′ = 2x − y, x2 − xy + y 2 = c2
2. (x − y + 1)y ′ = 1, y = x + cey
3. (xy − x)y ′′ + xy ′2 + yy ′ − 2y ′ = 0, y = ln(xy)
Exercise 2.2 Form the differential equations of the given families of curves
1. y = cx
2. ln xy = 1 + cy
3. x3 = c(x2 − y 2 )

Murad Ridwan, PhD 2 of 29

School of Electrical & Computer Engineering
AAiT, Addis Ababa University.
 Class Notes on ECEG-6201
2.2. FIRST ORDER DIFFERENTIAL EQUATION Analytical & Comp. Methods

4. y = c1 cos 2x + c2 sin 2x
5. (c1 + c2 x)ex + c3 = y
[ans. y −xy ′ = 0; y = xy ′ ln xy ; 3y 2 −x2 = 2xyy ′ ; y ′′ +4y = 0; y ′′ −2y ′ +y =
0.]

Exercise 2.3 Form the differential equation of all circles in the xy-plane. [ans.
(1 + y ′2 )y ′′′ − 3y ′ y ′′2 = 0]

Exercise 2.4 For the given families of curves find the lines that satisfy the given
initial conditions
1. y = c1 sin(x − c2 ), y(π) = 1, y ′ (π) = 0
2. y = c1 e−x + c2 ex + c3 e2x , y(0) = 0, y ′ (0) = 1, y ′′ (0) = −2
[ans. y = − cos x; y = 61 (−5e−x + 9ex − 4e2x ).] ◀

2.2 First Order Differential Equation

A differential equation of the first order in an unknown function y solved for
the derivatives y ′ , is of the form
y ′ = f (x, y) (2.3)
Taking into account that y ′ = dy/dx, the differential equation (2.3) may be
written in the symmetric form
P (x, y)dx + Q(x, y)dy = 0 (2.4)
where P (x, y) and Q(x, y) are known functions.

2.2.1 Separable Equation

First-order equations with variables separable are of the type
y ′ = f (x)g(y) (2.5)
or
X1 (x)Y1 (y)dx + X2 (x)Y2 (y)dy = 0 (2.6)
Dividing both sides of (2.5) by g(y), multiplying by dx and integrating, we
get Z Z
dy
= f (x)dx + c
g(y)
Similarly, dividing both sides of (2.6) by X2 (x)Y1 (y) and integrating we
obtain Z Z
X1 (x) Y2 (y)
dx + dy = 0
X2 (x) Y1 (y)

Murad Ridwan, PhD 3 of 29

School of Electrical & Computer Engineering
AAiT, Addis Ababa University.
 Class Notes on ECEG-6201
2.2. FIRST ORDER DIFFERENTIAL EQUATION Analytical & Comp. Methods

Example 2.4 Solve y ′ = y/x.

Exercise 2.5 Solve the differential equations

1. tan x sin2 y dx + cos2 x cot y dy = 0
2. xy ′ − y = y 3
3. y − xy ′ = a(1 + x2 y ′ )
4. y ′ tan x = y
p cx
[ans. cot2 y = tan2 x + c; x = cy/ 1 + y 2 ; y =a+ 1+ax ; y = c sin x]

Exercise 2.6 Solve the differential equations by changing the variables

1. y ′ = (x + y)2 (Hint: set u = x + y)
2. (2x − y)dx + (4x − 2y + 3)dy = 0 (Hint: set u = 2x − y)
[ans. arctan(x + y) = x + c; 5x + 10y + c = 3 ln |10x − 5y + 6|.] ◀

2.2.2 Homogeneous Equations

A function f (x, y) is a homogeneous function of degree n if, for any λ, it
obeys
f (λx, λy) = λn f (x, y) (2.7)
The differential equation

P (x, y)dx + Q(x, y)dy = 0

with homogeneous functions P (x, y) and Q(x, y) of equal degree can be

reduced to y
y′ = f
x
by means of the substitution y = xu, where u is a new unknown function. It
is transformed to an equation with variables separable. We can also apply
the substitution x = yu.

y
Example 2.5 Solve y ′ = e x + xy .

Exercise 2.7 Solve the differential equations

y
1. y ′ = − 1
x
2. (x − y)y dx + x2 dy = 0
x+y
3. y ′ = −
y

Murad Ridwan, PhD 4 of 29

School of Electrical & Computer Engineering
AAiT, Addis Ababa University.
 Class Notes on ECEG-6201
2.2. FIRST ORDER DIFFERENTIAL EQUATION Analytical & Comp. Methods

p
4. x dy − y dx = x2 + y 2 dx
x + 2y + 1
5. y ′ =
2x + 4y + 3
[ans. y = ln xc ; x = cex/y ; y= c x
x−2; y = 2c x2 − 2c
1
; ln |4x+8y+5|+8y−4x =
c.] ◀

2.2.3 Exact Equations

An exact first-order differential equation is of the form

P (x, y)dx + Q(x, y)dy = 0 (2.8)

for which
∂P ∂Q
= (2.9)
∂y ∂x
In this case P dx + Qdy is an exact differential
∂U ∂U
P dx + Qdy = dU = dx + dy
∂x ∂y
from which
∂U ∂U
P = , Q=
∂x ∂y
∂2U ∂2U
Since ∂x∂y = ∂y∂x , we have
∂P ∂Q
= .
∂y ∂x
If (2.9) holds, the general solution is

U (x, y) = c (∵ P dx + Qdy = dU = 0 ⇒ U (x, y) = c.)

Example 2.6 Find the solution of (3x2 + 6xy 2 )dx + (6x2 y + 4y 3 )dy = 0. ◀

∂Q
If for (2.8), ∂P
∂y ̸= ∂x , the equation (2.8) is inexact. In such cases, (2.8) can
be made exact by an integrating factor µ(x, y) such that

µ(P dx + Qdy) = dU

so that
∂ ∂
(µP ) = (µQ).
∂y ∂x

Murad Ridwan, PhD 5 of 29

School of Electrical & Computer Engineering
AAiT, Addis Ababa University.
 Class Notes on ECEG-6201
2.2. FIRST ORDER DIFFERENTIAL EQUATION Analytical & Comp. Methods

Exercise 2.8 Solve

1. (x + y)dx + (x + 2y)dy = 0
2. (x + y 2 )dx − 2xy dy = 0
1 2
[ans. 2x + xy + y 2 = c; ln |x| − y 2 /x = c.] ◀

2.2.4 Linear Equations

A differential equation of the form
y ′ + P (x)y = Q(x) (2.10)
is called linear. If Q(x) ≡ 0, the equation (2.10) takes the form
y ′ + P (x)y = 0 (2.11)
which is called a homogeneous linear differential differential equation of order
one. In this case, the variables may be separated and we get
R
y = ce− P (x)dx

The inhomogeneous equation (2.10) can be solved by having the integrating

factor R
µ(x) = e P (x)dx (2.12)
so that
µy ′ + P µy = µQ (2.13)
is exact. (2.13) can be written as
Z
d
(µy) = µQ ⇒ µy = µQdx + c
dx
so we have R
µQ(x)dx + c
y= .
µ(x)

Example 2.7 Solve y ′ + x4 y = x4 . ◀

The Bernoulli equation is a first-order equation of the form

y ′ + P (x)y = Q(x)y n (2.14)
where n is a real number. The substitution
z = y 1−n
transforms (2.14) into a linear differential equation in the unknown function
z(x). It is also possible to apply directly the substitution y = ux.

Murad Ridwan, PhD 6 of 29

School of Electrical & Computer Engineering
AAiT, Addis Ababa University.
 Class Notes on ECEG-6201
2.3. SECOND ORDER D.E. Analytical & Comp. Methods

√
Example 2.8 Solve y ′ = x4 y + x y.

Exercise 2.9 Solve the differential equations

y
1. y ′ − x =x
p
2. (1 + y 2 )dx = ( 1 + y 2 sin y − xy)dy
3. 2xyy ′ − y 2 + x = 0.
p
[ans. y = cx + x2 ; x 1 + y 2 + cos y = c; y 2 = x ln xc .] ◀

2.3 Second Order D.E. with Constant Coefficients

A second order homogeneous equation with constant coefficients p and q is
of the form
y ′′ + py ′ + qy = 0 (2.15)
If λ1 and λ2 are the roots of the characteristic equation

λ2 + pλ + q = 0,

then the general solution to (2.15) is written in one of the following three
ways

y = c1 eλ1 x + c2 eλ2 x , if λ1 and λ2 are real and λ1 ̸= λ2 (2.16)

λ1 x
y = e (c1 + c2 x), if λ1 = λ2 (2.17)
αx
y = e (c1 cos βx + c2 sin βx) if λ1 = α + βi and λ2 = α − βi
(2.18)

Exercise 2.10 Verify the above statements. ◀

The solution of the second order inhomogeneous equation with constant

coefficients
y ′′ + py ′ + qy = f (x) (2.19)
may be written in the form
y = yc + yp (2.20)
where yc is the general solution of the corresponding homogeneous equation
(2.15), called the complementary solution, and yp is a particular solution of
(2.19).

The function yp may be found by the method of undetermined coefficients.

But in a general case, the method of variation of parameters can be used.
The method consists in first finding the general solution of the respective

Murad Ridwan, PhD 7 of 29

School of Electrical & Computer Engineering
AAiT, Addis Ababa University.
 Class Notes on ECEG-6201
2.4. SERIES SOLUTION Analytical & Comp. Methods

homogeneous equation, i.e., yc (x). Then, assuming the constants are func-
tions of x, we seek the solution of the inhomogeneous equation (2.19).

Example 2.9 Solve 2y ′′ − y ′ − y = 4xe2x .

Exercise 2.11 Solve

1. y ′′ + y = cos x
2. y ′′ − 2y ′ + y = ex + e−x
[ans. c1 cos x + c2 sin x + 12 x sin x; (c1 + c2 x + 12 x2 )e−x + 41 ex .] ◀

2.4 Series Solution of Ordinary Differential Equa-

tions
For a linear differential equation

bn (x)y (n) + bn−1 (x)y (n−1) + . . . + b0 (x)y = R(x) (2.21)

with polynomial coefficients, the point x = x0 is called an ordinary point of

the equation if bn (x0 ) ̸= 0. A singular point of (2.21) is any point x = x1
for which bn (x1 ) = 0.

Example 2.10 List the singular points, if any, of the equations

1. (1 − x2 )y ′′ − 6xy ′ − 4y = 0
2. y ′′ + 2xy ′ + y = sin x
y y
3. y ′′ + 2 y ′ + = 0 ◀
x x

2.4.1 Solution about an Ordinary Point

If x = x0 is an ordinary point of (2.21), then it may be shown that every
solution y(x) of the equation is also analytic at x = x0 . Since every solution
is analytic, y(x) can be represented by a power series
∞
X
y(x) = an (x − x0 )n (2.22)
n=0

Usually we will take x0 = 0. If this is not already the case, then a substitu-
tion x = x − x0 will make it so.

Murad Ridwan, PhD 8 of 29

School of Electrical & Computer Engineering
AAiT, Addis Ababa University.
 Class Notes on ECEG-6201
2.4. SERIES SOLUTION Analytical & Comp. Methods

Example 2.11 Solve the following equations near the ordinary point x = 0
1. y ′ − y = 0
2. y ′ = 2xy
3. y ′′ + 4y = 0
4. (1 − x2 )y ′′ − 6xy ′ − 4y = 0
Exercise 2.12 Find the general solutions near the origin
1. xy ′ − 3y = 6
2. (1 − x2 )y ′ = 2xy
3. y ′′ + 3xy ′ + 3y = 0
2y
4. y ′′ − =0
(1 − x)2
5. (1 − x2 )y ′′ − 2xy ′ + 2y = 0
6. (1 + x2 )y ′′ − 4xy ′ + 6y = 0
Answers
1. y = −2 + a3 x3
2. y = a0 (1 + x2 + x4 + . . .) = a0 /(1 − x2 )
h n 2n
i h i
P∞ P∞ (−3)n x2n+1
3. y = a0 1 + n=1 (−3) n
2 n!
x
+ a1 x + n=1 3·5·7···(2n+1)

1
4. y = a0 1−x + a1 (1 − x)2
5. y = a0 x + a1 (1 − x2 − 13 x4 − 51 x6 − 17 x8 − . . .)
6. y = a0 (1 − 3x2 ) + a1 (1 − 13 x3 )
Exercise 2.13 Find the general solutions about the indicated points
1. y ′′ − 2(x + 3)y ′ − 3y = 0 about x = −3
′′ 2 ′
2. y + (x − 1) y − 4(x − 1)y = 0 about x = 1
Answer
h i
P∞ 3·7·11···(4n−1)(x+3)2n
1. y = a0 1 + n=1 (2n)! +
h P∞ 2n+1
i
5·9·13···(4n+1)(x+3)
a1 (x + 3) + n=1 (2n+1)!
P∞ 4(−1)n (x−1)3n
+ a1 (x − 1) + 14 (x − 1)4
 
2. y = a0 n=0 3n (3n−1)(3n−4)n! ◀

2.4.2 Solution about a Regular Singular Point

Suppose that the point x = x0 is a singular point of the equation

b2 (x)y ′′ + b1 (x)y ′ + b0 (x)y = 0 (2.23)

Murad Ridwan, PhD 9 of 29

School of Electrical & Computer Engineering
AAiT, Addis Ababa University.
 Class Notes on ECEG-6201
2.4. SERIES SOLUTION Analytical & Comp. Methods

with polynomial coefficients. Then b2 (x0 ) = 0, so b2 (x0 ) has a factor (x−x0 )

to some power. Let us put (2.23) into the form
y ′′ + p(x)y ′ + q(x)y = 0 (2.24)
If (x − x0 )p(x) and (x − x0 )2 q(x) are both analytic at x = x0 , we have
the necessary and sufficient conditions to have a finite solution about x0 .
Singular points that have this property are called regular singular points,
whereas any point not satisfying both these criteria is termed an irregular
or essential singularity.

Example 2.12 Discuss the singularities of

1. x4 (x2 + 1)(x − 1)2 y ′′ + 4(x − 1)y ′ + (x + 1)y = 0
2. (1 − x2 )y ′′ − 2xy ′ + l(l + 1) = 0
Example 2.13 Show thatPthe the equation x2 y ′′ + (x2 − x)y ′ + 2y = 0 has no
∞
series solution of the form n=0 an xn at the singular point x = 0. ◀

If x = 0 is a regular singular point of (2.24), then it can be shown that there

exists at least one solution of the form
X∞
r
y=x an xn (2.25)
n=0

where the exponent r may be any real or complex number, and a0 ̸= 0. Such
a series is called a generalized power series or Frobenius series.

Substitute (2.25) in (2.24) and equate the coefficient of the lowest power of x.
This gives a quadratic equation in r, which is known as the indicial equation.

Equate to zero the coefficients of other powers of x to find a1 , a2 , . . . in terms

of a0 . Substitute the values of a1 , a2 , . . . in (2.25) to get the series solution
of (2.24) having a0 as arbitrary constant. Obviously this is not the complete
solution of (2.23), since the complete solution must have two independent
arbitrary constants. The method of complete solution depends on the nature
of roots of the indicial equation.
Case I Distinct roots r1 , r2 not differing by an integer (i.e., r1 − r2 is not
an integer). The complete solution is
y(x) = c1 y|r=r1 + c2 y|r=r2 (2.26)

Case II When the roots are equal r1 = r2 . The complete solution is

∂y(x, r)
y(x) = c1 y|r=r1 + c2 (2.27)
∂r r=r1

Murad Ridwan, PhD 10 of 29

School of Electrical & Computer Engineering
AAiT, Addis Ababa University.
 Class Notes on ECEG-6201
2.4. SERIES SOLUTION Analytical & Comp. Methods

Case III When the roots r1 and r2 (r1 > r2 ) are distinct and differ by an
integer (i.e., r1 − r2 is an integer). The complete solution is
∂[(r − r2 )y(x, r)]
y(x) = c1 y|r=r1 + c2 (2.28)
∂r r=r2

y′ x2 + 1
Example 2.14 Solve y ′′ − + y=0 (Case I)
2x 2x2
Example 2.15 Differentiation of a product function: Suppose that
u = u1 u2 · · · un
each of the u’s being a function of the parameter r. Show that
 ′
u′2 u′n

′ u1
u =u + + ··· +
u1 u2 un
du ′ du′
where u′ = , uk = k (k = 1, 2, . . . , n).
dr dr
Example 2.16 Differentiate with respect to r.
1. y = (ar + b)k
r2 (r + 1)
2. y =
(4r − 1)3 (7r + 2)6
r+n
3. y =
r(r + 1)(r + 2) · · · (r + n − 1)
r3
4. y =
[(r + 2)(r + 3) · · · (r + n + 1)]2
Example 2.17 Solve x2 y ′′ + xy ′ + x2 y = 0 (Case II)

Example 2.18 Solve x2 y ′′ + (x2 − 2x)y ′ + 2y = 0 (Case III).

Exercise 2.14 Solve the following

1. 9x(1 − x)y ′′ − 12y ′ + 4y = 0
2. x2 y ′′ − xy ′ + y = 0
3. x(1 − x)y ′′ − 3y ′ − y = 0
4. 2x(1 − x)y ′′ + (1 − x)y ′ + 3y = 0
Answers
1. y = A 1 + 13 x + 1·4 2 1·4·7 3
 
3·6 x + 3·6·9 x + · · · + 
Bx7/3 1 + 89 x + 8·11 2 8·11·14 3

10·13 x + 10·13·16 x + · · ·
2. y = Ax + Bx ln x
3. y = Ax(1 + 2x + 3x2 + · · · ) + B[y1 ln x + C(1 + x + x2 + x3 + · · · )]
√  2
3x3 3x4

4. y = A x(1 − x) + B 1 − 3x + 3x 1·3 + 3·5 + 5·7 ◀

Murad Ridwan, PhD 11 of 29

School of Electrical & Computer Engineering
AAiT, Addis Ababa University.
 Class Notes on ECEG-6201
2.5. NUMERICAL SOLUTION Analytical & Comp. Methods

2.5 Numerical Solution of Differential Equations

Most differential equations have no known analytical solution, and even
when one can be found it is often difficult to use. As a result, when so-
lutions are required and an analytical solution either is not known or is
inconvenient to use, it becomes necessary to use methods that produce a
numerical solution directly. However, unlike the general analytical solution
of an initial value problem that can be adapted to any appropriate initial
conditions, a numerical solution is the solution of a specific initial value prob-
lem, so the calculation must be repeated if the initial conditions are changed.

Consider the initial value problem

y ′ = f (x, y), y(x0 ) = y0 (2.29)

We start from y0 = y(x0 ) and proceed stepwise. In the first step we compute
an approximate value y1 of the solution y of (2.29) at x = x1 = x0 + h. In
the second step we compute an approximate value of y2 of the solution at
x = x2 = x0 + 2h, etc. Here h is a fixed number.

In each step the computations are done by the same formula, usually such
formulas are suggested by the Taylor series

h2 ′′
y(x + h) = y(x) + hy ′ (x) + y (x) + · · ·
2!
h2 h3
= y(x) + hf + f ′′ + f ′′′ + · · · (2.30)
2! 3!
where f ′ , f ′′ , f ′′′ , . . . are computed at (x, y(x)).

2.5.1 Euler’s Method

For small h in (2.30), we can approximate

y(x + h) ≈ y(x) + hf, or

yn+1 = yn + hf (xn , yn ), n = 0, 1, 2, . . .

where xn = x0 + nh. This is called Euler’s method. Geometrically it is an

approximation of the curve y(x) by a polygon whose first side is tangent to
the curve at x0 (Fig. 2.1).

Example 2.19 Find y(1.5) for y ′ = 2xy, y(1) = 1 using h = 0.10 and h = 0.05.
Compare with actual values. Verify Figures 2.2 and 2.3 ◀

Murad Ridwan, PhD 12 of 29

School of Electrical & Computer Engineering
AAiT, Addis Ababa University.
 Class Notes on ECEG-6201
2.5. NUMERICAL SOLUTION Analytical & Comp. Methods

Figure 2.1: Euler method.

Figure 2.2: Euler method, h = 0.1.

2.5.2 Improved Euler Method

Taking more terms in (2.30) would produce more accurate numeric results
but the difficulty is that we introduce higher order derivatives. Hence the
general strategy is to avoid the computation of higher derivatives and to
replace it by computing f for one or several suitably chosen auxiliary values
of (x, y).

In the improved Euler method (also called Heun method), in each step we
compute first the auxiliary value
∗
yn+1 = yn + hf (xn , yn ) (2.31)

and the new value

∗ )
f (xn , yn ) + f (xn+1 , yn+1
yn+1 = yn + h (2.32)
2
Geometrically, in the interval from xn to xn + 12 h we approximate the so-
lution by a straight line through (xn , yn ) with slope f (xn , yn ), and then we
∗ ) until x reaches x
continue along the line with slope f (xn+1 , yn+1 n+1 .

Murad Ridwan, PhD 13 of 29

School of Electrical & Computer Engineering
AAiT, Addis Ababa University.
 Class Notes on ECEG-6201
2.5. NUMERICAL SOLUTION Analytical & Comp. Methods

Figure 2.3: Euler method, h = 0.05.

The improved Euler method is a predictor-corrector method, because in each

step we first predict a value by (2.31) and then correct it by (2.32).

Example 2.20 Use the improved Euler’s method to obtain an approximate value
of y(1.5) for the solution
y ′ = 2xy, y(1) = 1
Compare the results for h = 0.1 and h = 0.05. Verify Figures 2.4 and 2.5 ◀

Figure 2.4: Improved Euler method, h = 0.1.

2.5.3 The Runge-Kutta Method

Probably on of the most popular as well as accurate numerical procedures
used in obtaining approximate solution to differential equations is the fourth-
order Runge-Kutta method.

The fourth-order Runge-Kutta method consists of determining appropirate

constants so that a formula such as
yn+1 = yn + ak1 + bk2 + ck3 + dk4

Murad Ridwan, PhD 14 of 29

School of Electrical & Computer Engineering
AAiT, Addis Ababa University.
 Class Notes on ECEG-6201
2.5. NUMERICAL SOLUTION Analytical & Comp. Methods

Figure 2.5: Improved Euler method, h = 0.05.

agrees with a Taylor expansion out to h4 or the fifth term. The constants
a, b, c, d and the auxiliary quantities k1 , k2 , k3 , k4 are given as

yn+1 = yn + 61 (k1 + 2k2 + 2k3 + k4 )

k1 = hf (xn , yn )
k2 = hf (xn + 12 h, yn + 12 k1 ) (2.33)
1 1
k3 = hf (xn + 2 h, yn + 2 k2 )
k4 = hf (xn + h, yn + k3 )

The method is well suited to the computer because it needs no special start-
ing procedure, makes light demand on storage, and repeatedly uses the same
straight forward computational procedure. It is numerically stable.

Example 2.21 Derive the second order Runge-Kutta method by finding the con-
stants a, b, α, and β such that the formula

yn+1 = yn + ak1 + bk2

where
k1 = hf (xn , yn ), k2 = hf (xn + αh, yn + βk1 )
agrees with a Taylor series expansion out to h3 .

Exercise 2.15 Derive (2.33).

Example 2.22 Use the Runge-Kutta method to obtain an approximation to y(1, 5)

for the solution
y ′ = 2xy, y(1) = 1
use h = 0.1 and verify Figure 2.6. ◀

Murad Ridwan, PhD 15 of 29

School of Electrical & Computer Engineering
AAiT, Addis Ababa University.
 Class Notes on ECEG-6201
2.5. NUMERICAL SOLUTION Analytical & Comp. Methods

Figure 2.6: Runge-Kutta method, h = 0.1.

2.5.4 Milne’s Method

In a one-step method we compute yn+1 using only a single step, namely,
the previous value yn . One-step methods are ‘self-starting’, they need not
help to get going because they obtain y1 from the initial value y0 , etc. The
Euler, Improved Euler and Runge-Kutta are examples of a one-step method.

In contrast, a multi-step method uses in each step values from two or more
previous steps. These methods are motivated by the expectation that the ad-
ditional information will increase accuracy and stability. But to get started,
one needs values, say, y0 , y1 , y2 , y3 in a 4-step method, obtained by Runge-
Kutta or other accurate methods. Thus, multi-step methods are not self-
starting.

Milne’s method is an example of multi-step method. The predictor is

∗
yn+1 = yn−3 + 43 h(2yn′ − yn−1
′ ′
+ 2yn−2 ) for n ≥ 3 (2.34)

where

yn′ = f (xn , yn ), ′
yn−1 = f (xn−1 , yn−1 ), ′
yn−2 = f (xn−2 , yn−2 )

and the corrector is

′
yn+1 = yn−1 + 13 h(yn+1 + 4yn′ + yn−1
′
) (2.35)
′
where yn+1 ∗ ).
= f (xn+1 , yn+1

Exercise 2.16 For the following differential equations, construct a table computing
the indicated values of y(x) using Euler, improved Euler and Runge-Kutta methods.
Use h = 0.1 and compute to four rounded decimal places.
1. y ′ = 2 ln(xy), y(1) = 2.
y(1.1), y(1.2), y(1.3), y(1.4), y(1.5)

Murad Ridwan, PhD 16 of 29

School of Electrical & Computer Engineering
AAiT, Addis Ababa University.
 Class Notes on ECEG-6201
2.6. SPECIAL FUNCTIONS Analytical & Comp. Methods

2. y ′ = sin x2 + cos y 2 , y(0) = 0.

y(0.1), y(0.2), y(0.3), y(0.4), y(0.5)
√
3. y ′ = x + y, y(0.5) = 0.5
y(0.6), y(0.7), y(0.8), y(0.9), y(1.0)
4. y ′ = xy + y 2 , y(1) = 0.
y(1.1), y(1.2), y(1.3), y(1.4), y(1.5)
Exercise 2.17 Use the Euler method to obtain the approximate value of y(0.2)
where y(x) is the solution of

y ′′ + xy ′ + y = 0
y(0) = 1, y ′ (0) = 2.

Exercise 2.18 Use Milne’s method to approximate the value of y(0.4) where

y ′ = 4x − 2y, y(0) = 2

Use the Runge-Kutta formula and h = 0.1 to obtain the values of y1 , y2 and y3 . ◀

2.6 Special Functions

2.6.1 Gamma and Beta Functions
The gamma function denoted by Γ(n) is defined as
Z ∞
Γ(n) = xn−1 e−x dx (2.36)
0

It may be regarded as a generalization of n! for n positive integers. In par-

ticular Γ(1) = 1.

A recursive formula for the gamma functions is

Γ(n + 1) = nΓ(n) (2.37)

In particular if n is a positive integer, then

Γ(n + 1) = n!, n = 1, 2, 3, . . . (2.38)

For this reason Γ(n) is sometimes called the factorial function.

Example 2.23 Verify (2.37) and (2.38). ◀

Murad Ridwan, PhD 17 of 29

School of Electrical & Computer Engineering
AAiT, Addis Ababa University.
 Class Notes on ECEG-6201
2.6. SPECIAL FUNCTIONS Analytical & Comp. Methods

From (2.37), Γ(n) can be determined for all n > 0 when the values for
1 ≤ n < 2 (or any other interval of unit length) are known. For instance,
Γ 11 7


7 7
3 3

4 = Γ 4 + 1 = 4Γ 4 = 4Γ 4

The value of Γ 34 can be obtained from the table of gamma functions.

Figure 2.7: The gamma function.

1

√
Example 2.24 Show that Γ 2 = π.

Example 2.25 Evaluate the following integrals

Z ∞ Z 1
dx
1. x3 e−x dx 4. √
0 0 − ln x
Z ∞
2. x6 e−2x dx Z ∞
0 n
Z ∞ 5. xm e−ax dx, m, n, a posi-
2
3. 3−4z dz 0
tive constants. ◀
0

The beta function is defines as

Z 1
B(m, n) = xm−1 (1 − x)n−1 dx, m > 0, n > 0 (2.39)
0

The beta function is symmetric, i.e.,

B(m, n) = B(n, m) (2.40)
The beta function can be expressed through gamma function
Γ(m)Γ(n)
B(m, n) = (2.41)
Γ(m + n)

Murad Ridwan, PhD 18 of 29

School of Electrical & Computer Engineering
AAiT, Addis Ababa University.
 Class Notes on ECEG-6201
2.6. SPECIAL FUNCTIONS Analytical & Comp. Methods

Example 2.26 Verify (2.40) and (2.41). ◀

Many integrals can be expressed through beta and gamma functions. Two
of special interest are
Z π/2
1
sin2m−1 θ cos2n−1 θ dθ = B(m, n)
0 2
∞
xp−1
Z
π
dx = Γ(p)Γ(1 − p) = , 0<p<1
0 1+x sin πp

Exercise 2.19 Show that

√
1. Γ − 12 = −2 π

8 √
2. Γ − 52 = − 15


π
Z 1
(−1)n n!
3. xm (ln x)n dx = , n is positive integer and m > −1
0 (m + 1)n+1
Z ∞ √
−k2 x2 π
4. e dx =
−∞ k
Z ∞ c
x Γ(c + 1)
5. x
dx = ,c > 1
0 c (ln c)c+1
Z π/2 √
π
6. tan θ dθ = √
0 2
7. yB(x + 1, y) = xB(x, y + 1)
Z a
8. xn−1 (a − x)m−1 dx = am+n−1 B(m, n)
0
Z b
9. (x−a)m (b−x)n dx = (b−a)m+n+1 B(m+1, n+1) [Hint: put x = a+(b−a)z]
a
Z π/2  
p 1
q p+1 q+1
10. sin x cos x dx = B ,
0 2 2 2
11. B(m, n) = B(m + 1, n) + B(m, n + 1)
Z ∞ Z 1 p−1
y q−1 x + xq−1
12. dy = dx = B(p, q) ◀
0 (1 + y)p+q 0 (1 + x)p+q

2.6.2 Legendre Functions

The differential equation

(1 − x2 )y ′′ − 2xy ′ + n(n + 1)y = 0 (2.42)

Murad Ridwan, PhD 19 of 29

School of Electrical & Computer Engineering
AAiT, Addis Ababa University.
 Class Notes on ECEG-6201
2.6. SPECIAL FUNCTIONS Analytical & Comp. Methods

when n is real number, is called Legendre’s differential equation. This equa-

tion is of considerable importance, particularly in boundary value problems
involving spherical configurations. Any solution of (2.42) is called Legendre
function.

Dividing (2.42) by 1−x2 we see that the coefficients of the resulting equation
are
P∞analyticm at x = 0, so that we may apply the power series method y =
m=0 am x . The resulting solution becomes

y(x) = Ay1 (x) + By2 (x) (2.43)

where
n(n + 1) 2 (n − 2)n(n + 1)(n + 3) 4
y1 (x) = 1 − x + x − +···
2! 4!
(n − 1)(n + 2) 3 (n − 3)(n − 1)n(n + 2)(n + 4) 5
y2 (x) = x − x + x − +···
3! 5!
These series converge for |x| < 1.

Exercise 2.20 Verify (2.43). ◀

In many applications, the parameter n in Legendre’s equation is non-negative

integer. In this case the series has only finite terms. It is customary to
choose am = 1 when m = 0. Then the resulting solution of (2.42) is called
the Legendre polynomial of degree n and is denoted by Pn (x) which is given
by
M
X (2n − 2m)!
Pn (x) = (−1)m xn−2m (2.44)
2n m!(n − m)!(n − 2m)!
m=0
(2n)! n (2n − 2)!
= n 2
x − n xn−2 + − · · ·
2 (n!) 2 1!(n − 1)!(n − 2)!

where M = n/2 or (n − 1)/2, whichever is an integer.

Exercise 2.21 Verify (2.44). ◀

Murad Ridwan, PhD 20 of 29

School of Electrical & Computer Engineering
AAiT, Addis Ababa University.
 Class Notes on ECEG-6201
2.6. SPECIAL FUNCTIONS Analytical & Comp. Methods

In particular
P0 (x) = 1
P1 (x) = x
1 2
P2 (x) = 2 (3x − 1)
1 3
P3 (x) = 2 (5x − 3x)
1 4 2
P4 (x) = 8 (35x − 30x + 3)
1 5 3
P5 (x) = 8 (63x − 70x + 15x)
1 6 4 2
P6 (x) = 16 (231x − 315x + 105x − 5)
1 7 5 3
P7 (x) = 16 (429x − 693x + 315x − 35x)

Figure 2.8: Legendre polynomials.

Rodrigues’ formula: The Legendary polynomials can be obtained from

Rodrigues’ formula
1 dn  2
(x − 1)n

Pn (x) = n n
(2.45)
2 n! dx

Exercise 2.22 Verify (2.45) by applying the binomial theorem to (x2 − 1)n ,
differentiating n times term by term, and comparing the result with (2.44).

Example 2.27 Express f (x) = x3 − 5x2 + x + 2 in terms of the Legendre

polynomials.

Murad Ridwan, PhD 21 of 29

School of Electrical & Computer Engineering
AAiT, Addis Ababa University.
 Class Notes on ECEG-6201
2.6. SPECIAL FUNCTIONS Analytical & Comp. Methods

Generating function: It can be shown that Pn (x) is the coefficient of un

in the expansion of (1 − 2xu + u2 )−1/2 in ascending powers of u, i.e.,
∞
1 X
√ = Pn (x)un (2.46)
1 − 2xu + u2
n=0

Exercise 2.23 Verify (2.46). ◀

Orthogonality: The Legendre polynomials are orthogonal over [−1, 1], i.e.,
Z 1 
0, if m ̸= n;
Pm (x)Pn (x)dx = 2 (2.47)
−1 2n+1 , if m = n.

Exercise 2.24 Prove (2.47). ◀

The orthogonality of Legendre polynomials enable us to expand a function

f (x), defined from x = −1 to x = 1, in a series of Legendre polynomials.

Let
∞
X
f (x) = an Pn (x) (2.48)
n=0

To determine an , multiply both sides of (2.48) by Pm (x) and integrating

w.r.t. x from −1 to 1, we have
Z 1 ∞
X Z 1
f (x)Pm (x)dx = an Pn (x)Pm (x)
−1 n=0 −1
∞ 
X 2 1, if m = n;
= an δmn , where δmn =
2n + 1 0, otherwise.
n=0
2
= am
2m + 1
which gives us Z 1
1
an = (n + 2) f (x)Pn (x)dx
−1

The expansion of f (x) given by (2.48) is known as Fourier-Legendre series.

Exercise 2.25 Using (2.46) show that

1. Pn (1) = 1

Murad Ridwan, PhD 22 of 29

School of Electrical & Computer Engineering
AAiT, Addis Ababa University.
 Class Notes on ECEG-6201
2.6. SPECIAL FUNCTIONS Analytical & Comp. Methods

2. Pn (−x) = (−1)n Pn (x)

2n!
3. P2n (0) = (−1)n
22n (n!)2
4. P2n+1 (0) = 0
Exercise 2.26 Verify the following recurrence relations
1. (2n + 1)xPn (x) = (n + 1)Pn+1 (x) + nPn−1 (x)
2. nPn (x) = xPn′ (x) − Pn−1
′
(x)
′ ′
3. (2n + 1)Pn (x) = Pn+1 (x) − nPn−1 (x)
4. Pn′ (x) = xPn−1
′
(x) + nPn−1 (x)
5. (1 − x2 )Pn′ (x) = n [Pn−1 (x) − xPn (x)]
6. (1 − x2 )Pn′ (x) = (n + 1) [xPn (x) − Pn+1 (x)]
Exercise 2.27 Show that
1. x3 = 52 P3 (x) + 35 P1 (x)
2. Pn (−1) = (−1)n
3. Pn′ (1) = 12 n(n + 1)
4. Pn′ (−1) = 12 (−1)n−1 n(n + 1)
Exercise 2.28 Prove that
Z 1
1. Pn (x)dx = 0 if n ̸= 0
−1
Z 1
2. P0 (x)dx = 2
−1
Z 1
3. xm Pn (x)dx = 0 if m, n are positive integers and m < n
−1
Z 1
2n
4. xPn (x)Pn−1 dx =
−1 4n2 − 1
Z 1
1
5. Pn (x)dx = Pn−1 (0)
0 n+1
Exercise 2.29 Obtain the Fourier-Legendre expansion of f (x) when

0, −1 < x < 0;
1. f (x) =
1, 0 < x < 1.

0, −1 < x ≤ 0;
2. f (x) =
x, 0 < x < 1.
Answer
1
1. 2 P0 (x) + 34 P1 (x) − 7
16 P3 (x) + ···
1 1 5 3
2. 4 P0 (x) + 2 P1 (x) + 16 P2 (x) − 32 P4 (x) + ··· ◀

Murad Ridwan, PhD 23 of 29

School of Electrical & Computer Engineering
AAiT, Addis Ababa University.
 Class Notes on ECEG-6201
2.6. SPECIAL FUNCTIONS Analytical & Comp. Methods

2.6.3 Bessel’s Function

The differential equation
x2 y ′′ + xy ′ + (x2 − n2 )y = 0 (2.49)
is called Bessel’s equation of order n and its particular solutions are called
Bessel’s function of order n.

Since x = 0 is a regular singular point of (2.49), we can use Frobenius series

to obtain the solution. The two roots of the resulting indicial equation are
r = n (≥ 0) and r = −n. Based on the value of r we have different solutions:
Case I When n ̸= 0 or when n is not an integer.

The solution is
y(x) = c1 Jn (x) + c2 J−n (x) (2.50)
where Jn (x), called the Bessel function of the first kind of order n, is
∞
X (−1)m  x n+2m
Jn (x) = (2.51)
m!Γ(n + m + 1) 2
m=0

and J−n (x), called the Bessel function of the first kind of order −n, is
∞
X (−1)m  x −n+2m
J−n (x) = (2.52)
m!Γ(−n + m + 1) 2
m=0

Case II When n = 0.

The Bessel’s equation takes the form

xy ′′ + y ′ + xy = 0
This is called Bessel’s equation of order zero. The complete solution
is
y(x) = c1 J0 (x) + c2 Y0 (x) (2.53)
where J0 (x) is Bessel function of the first kind of order zero. From
(2.51),
∞
X (−1)m  x 2m
J0 (x) = , [since Γ(m + 1) = m!] (2.54)
(m!)2 2
m=0

Y0 (x) is called Bessel function of the second kind of order zero or

Neumann function.
∞
(−1)m+1
 
X 1 1 1
Y0 (x) = J0 (x) + 1 + + + · · · + (2.55)
(m!)2 2 3 m
m=1

Murad Ridwan, PhD 24 of 29

School of Electrical & Computer Engineering
AAiT, Addis Ababa University.
 Class Notes on ECEG-6201
2.6. SPECIAL FUNCTIONS Analytical & Comp. Methods

Case III When n is an integer.

The two functions Jn (x) and J−n (x) are not independent but are con-
nected by the relation

J−n (x) = (−1)n Jn (x)

A solution can be obtained by letting y = u(x)Jn (x) in (2.49). Solving

for u(x), we obtain
Z
dx
u(x) = c2 + c1
xJn2 (x)
The complete solution becomes
 Z 
dx
y = c2 + c1 Jn (x)
xJn2 (x)
or
y(x) = c1 Jn (x) + c2 Yn (x) (2.56)
where Z
dx
Yn (x) = Jn (x) (2.57)
xJn2 (x)
The function Yn (x) is called the Bessel function of the second kind of
order n or Neumann function of order n.

Figure 2.9: The first three integer order Bessel functions of the first kind.

Murad Ridwan, PhD 25 of 29

School of Electrical & Computer Engineering
AAiT, Addis Ababa University.
 Class Notes on ECEG-6201
2.6. SPECIAL FUNCTIONS Analytical & Comp. Methods

Exercise 2.30 Derive (2.50), (2.53) and (2.56). ◀

Generating function: Bessel functions of various orders can be derived as

coefficients of various power in the expansion
hx i ∞
X
exp (t − 1
t) = tn Jn (x) (2.58)
2 n=−∞

Exercise 2.31 Verify (2.58). ◀

Figure 2.10: The first three integer order Bessel functions of the second kind.

Orthogonality: If α and β are the roots Jn (x) = 0, then

Z 1 
0, if α ̸= β;
xJn (αx)Jn (βx)dx = 1 2 (2.59)
0 J
2 n+1 (α), if α = β.

From the orthogonality property of Bessel functions, we can expand a func-

tion f (x) in fourier-Bessel series in the range 0 to a. Let
∞
X
f (x) = ci Jn (λi x) = c1 Jn (λ1 x) + c2 Jn (λ2 x) + . . . (2.60)
i=1

where λ1 , λ2 , . . . are the roots of the equation Jn (λa) = 0.

Murad Ridwan, PhD 26 of 29

School of Electrical & Computer Engineering
AAiT, Addis Ababa University.
 Class Notes on ECEG-6201
2.6. SPECIAL FUNCTIONS Analytical & Comp. Methods

To determine ci , we multiply both sides of (2.60) by xJn (λi x) and integrate

w.r.t. x between the limits 0 to a. From the orthogonality property, all
integrals on the right hand side will vanish except the one containing ci and
we have Z a
2
ci = 2 2 xf (x)Jn (λi x)dx (2.61)
a Jn+1 (λi a) 0

Exercise 2.32 Verify (2.59) and (2.61).

Exercise 2.33 Show that the Fourier-Bessel series in J2 (λi x) for f (x) = x2 (0 <
∞
X J2 (λi x)
x < a) where λi a are positive roots of J2 (x) = 0, is x2 = 2a2 . ◀
i=1
aλ i J3 (λi a)

Example 2.28 Show that

q
2
1. J1/2 (x) = πx sin x
d
2. dx [xn Jn (x)] = xn Jn−1 (x)
Exercise 2.34 Show that
q
2
1. J−1/2 (x) = πx cos x

2. d
dx [x−n Jn (x)] = −x−n Jn+1 (x)
3. Jn′ (x) + nx Jn (x) = Jn−1 (x)
x
4. Jn (x) = 2n [Jn−1 (x) + Jn+1 (x)]
q h i
2 3−x2 3
5. J5/2 (x) = πx x2 sin x −
cos x 2

6. J4 (x) = x483 − x8 J1 (x) + 1 − x242 J0 (x)

Z
2
7. J3 (x)dx = −J2 (x) − J2 (x)
x
Z
1
8. xJ02 (x)dx = x2 [J02 (x) + J12 (x)] + c, c− a constant ◀
2

A number of second order differential equations with variable coefficients can

be reduced to Bessel’s equation by a suitable transformation and, hence, can
be solved in terms of Bessel functions.

Consider the differential equation

x2 y ′′ + (1 − 2α)xy ′ + [β 2 γ 2 x2γ + (α2 − n2 γ 2 )]y = 0 (2.62)
where α, β, γ and n are constants. Putting X = βxγ and Y = x−α y, equa-
tion (2.62) reduces to
X 2 Y ′′ + XY ′ + (X 2 − n2 )Y = 0 (2.63)

Murad Ridwan, PhD 27 of 29

School of Electrical & Computer Engineering
AAiT, Addis Ababa University.
 Class Notes on ECEG-6201
2.6. SPECIAL FUNCTIONS Analytical & Comp. Methods

which is Bessel’s equation. The solution is


c1 Jn (X) + c2 Yn (X), for n- integer;
Y (X) = (2.64)
c1 Jn (X) + c2 J−n (X), for n- noninteger.

In terms of the original variables x and y, the solution becomes

 α
x [c1 Jn (βxγ ) + c2 Yn (βxγ )], for n- integer;
y(x) = (2.65)
xα [c1 Jn (βxγ ) + c2 J−n (βxγ )], for n- noninteger.

Example 2.29 Solve

1. y ′′ − x2 y ′ + 4(x2 − 1
x2 )y =0
′′ ′
2. xy − 3y + xy = 0
Exercise 2.35 Find the solutions of the following differential equations in terms
of Bessel functions.
1. xy ′′ + y = 0
2. y ′′ + x1 y ′ + 3 − 1


4x2 y =0
′′ ′ 1
3. xy + y + 4y =0
Answer :
√ √ √
1. y(x) = x [c1 J1 (2 x) + c2 Y1 x)]
√ √
2. y(x) = c1 J1/2 ( 3x) + c2 J−1/2 ( 3x)
√ √
3. y(x) = c1 J0 ( x) + c2 Y0 ( x) ◀

2.6.4 Hypergeometric Function

The differential equation

x(1 − x)y ′′ + [c − (a + b + 1)x]y ′ − aby = 0 (2.66)

where a, b, c are real constants, is called the (Gauss’s) hypergeometric differ-

ential equation. It has regular singular points at x = 0, 1. The corresponding
indicial equation has roots r1 = 0 and r2 = 1 − c.

For r1 = 0, the Frobenius method yields

ab x a(a + 1)b(b + 1) x2
y1 (x) = F (a, b, c; x) ≡ 1 + + + · · · (2.67)
c 1! c(c + 1) 2!
∞
Γ(c) X Γ(a + n)Γ(b + n) xn
=
Γ(a)Γ(b) Γ(c + n) n!
n=0

Murad Ridwan, PhD 28 of 29

School of Electrical & Computer Engineering
AAiT, Addis Ababa University.
 Class Notes on ECEG-6201
2.6. SPECIAL FUNCTIONS Analytical & Comp. Methods

where c ̸= 0, −1, −2, . . .. The function F (a, b, c; x) is called the hypergeomet-

ric function or hypergeometric series. The series converges for |x| < 1.

For r2 = 1 − c, the Frobenius method yields (where c ̸= 2, 3, 4, . . .)


1−c (a − c + 1)(b − c + 1) x
y2 (x) = x 1+ +
(−c + 2) 1!
(a − c + 1)(a − c + 2)(b − c + 1)(b − c + 2) x2

+ · · · (2.68)
(−c + 2)(−c + 3) 2!
= x1−c F (a − c + 1, b − c + 1, 2 − c; x)

The complete solution is then

y(x) = c1 y1 (x) + c2 y2 (x) (2.69)

Exercise 2.36 Derive (2.67) and (2.68). ◀

The general nature of the hypergeometric equation allows us to write a large

number of elementary functions in terms of the hypergeometric function.

Exercise 2.37 Show that

1
1. 1−x = 1 + x + x2 + . . . = F (1, 1, 1; x) = F (1, b, b; x) = F (a, 1, a; x)
2. If a or b is a negative integer, (2.67) reduces to a polynomial.
3. (1 + x)n = F (−n, b, b; −x)
4. (1 − x)n = 1 − nxF (1 − n, 1, 1; x)
5. tan−1 x = xF ( 21 , 1, 32 ; −x2 )
6. ln(1 + x) = xF (1, 1, 2; −x)
7. F (a, b, c; x) = (1 − x)c−a−b F (c − a, c − b, c; x)
8. F ′ (a, b, c; x) = ab
c F (a + 1, b + 1, c + 1; x) ◀

Murad Ridwan, PhD 29 of 29

School of Electrical & Computer Engineering
AAiT, Addis Ababa University.