DIFFERENTIAL EQUATIONS:

THEORY AND APPLICATIONS

DIFFERENTIAL EQUATIONS: THEORY
AND APPLICATIONS

Maria Catherine Borres

ARCLER
P r e s s

www.arclerpress.com
Differential Equations: Theory and Applications
Maria Catherine Borres

Arcler Press
2010 Winston Park Drive,
2nd Floor
Oakville, ON L6H 5R7
Canada
www.arclerpress.com
Tel: 001-289-291-7705
        001-905-616-2116
Fax: 001-289-291-7601
Email: [email protected]

e-book Edition 2019

ISBN: 978-1-77361-596-7 (e-book)

This book contains information obtained from highly regarded resources. Reprinted material
sources are indicated and copyright remains with the original owners. Copyright for images and
other graphics remains with the original owners as indicated. A Wide variety of references are
listed. Reasonable efforts have been made to publish reliable data. Authors or Editors or Publish-
ers are not responsible for the accuracy of the information in the published chapters or conse-
quences of their use. The publisher assumes no responsibility for any damage or grievance to the
persons or property arising out of the use of any materials, instructions, methods or thoughts in
the book. The authors or editors and the publisher have attempted to trace the copyright holders
of all material reproduced in this publication and apologize to copyright holders if permission has
not been obtained. If any copyright holder has not been acknowledged, please write to us so we
may rectify.

Notice: Registered trademark of products or corporate names are used only for explanation and
identification without intent of infringement.

© 2019 Arcler Press

ISBN: 978-1-77361-403-8 (Hardcover)

Arcler Press publishes wide variety of books and eBooks. For more information about
Arcler Press and its products, visit our website at www.arclerpress.com
ABOUT THE AUTHOR

Catherine is currently taking up Master of Arts in Education Major in

Mathematics in Philippine Normal University – Manila. She is currently
working as a Content Developer for Mathematics at the Affordable Private
Education Center (APEC Schools).
TABLE OF CONTENTS

Preface...........................................................................................................xi

Chapter 1 Basic Concepts of Differential Equations.................................................... 1

1.1. Introduction......................................................................................... 2
1.2. The Bernoulli Equation....................................................................... 31
1.3. Differential Equations of Higher Order............................................... 50
1.4. The Wronskian................................................................................... 85

Chapter 2 Fundamental Concepts of Partial Differential Equations........................ 111

2.1. Introduction..................................................................................... 112
2.2. Classification of Second Order PDE ................................................ 112
2.3. Summary and Discussion................................................................. 141
2.4. Classification of Second Order PDE ................................................ 141

Chapter 3 Application of Differential Equations In Mechanics............................... 143

3.1. Introduction..................................................................................... 144
3.2. Projectile Motion ............................................................................ 161
3.3. Summary and Discussion................................................................. 186

Chapter 4 Elliptic Differential Equation.................................................................. 191

4.1. Introduction..................................................................................... 192
4.2. Boundary Value Problem (BVPs)...................................................... 195
4.3. Some Important Mathematical Tools................................................ 197
4.4. Properties Of Harmonic Functions................................................... 199
4.5. Separation Of Variables ................................................................... 210
4.6. Dirichlet Problem For A Rectangle................................................... 212
4.7. The Neumann Problem For A Rectangle .......................................... 215
4.8. Interior Dirichlet Problem For A Circle ............................................ 217
 4.9. Exterior Dirichlet Problem For A Circle ........................................... 222
4.10. Interior Neumann Problem For A Circle......................................... 227
4.11. Solution Of Laplace Equation In Cylindrical Coordinates .............. 229
4.12. Solution Of Laplace Equation In Spherical Coordinates................. 238
4.13. Miscellaneous Example................................................................. 247
4.14. Summary And Discussions............................................................. 276

Chapter 5 Hyperbolic Differential Equation........................................................... 279

5.1 Introduction...................................................................................... 280
5.2. Solution Of One-Dimensional Wave Equation by
Canonical Reduction .................................................................... 284
5.3. The Initial Value Problem; D’alembert’s Solution.............................. 288
5.4. Summary And Discussion................................................................ 297

Chapter 6 Parabolic Differential Equations............................................................. 301

6.1. Introduction..................................................................................... 302
6.2. Boundary Conditions....................................................................... 304
6.3. Elementary Solutions Of The Diffusion Equation ............................. 305
6.4. Dirac Delta Function....................................................................... 310
6.5. Separation Of Variables Method....................................................... 316
6.6. Maximum-Minimum Principle and Consequences........................... 340
6.7. Miscellaneous Example. ................................................................. 343
6.8. Boundary Conditions....................................................................... 352

Chapter 7 Laplace Transform Methods................................................................... 357

7.1. Introduction..................................................................................... 358
7.2. Transform Of Some Elementary Functions........................................ 362
7.3. Properties Of Laplace Transform...................................................... 364
7.4. Transform Of A Periodic Function.................................................... 370
7.5. Transform Of Error Function............................................................. 372
7.6. Transform Of Bessel’s Function........................................................ 374
7.7. Transform Of Dirac Delta Function.................................................. 376
7.8. Convolution Theorem (Faltung Theorem).......................................... 382

Chapter 8 Green’s Function.................................................................................... 389

8.1. Introduction..................................................................................... 390
8.2. The Eigenfunction Method............................................................... 403

viii
8.3. Summary and Discussion................................................................. 406
References.............................................................................................. 408

Index...................................................................................................... 409

ix
PREFACE

In many respects this book is perfect book for beginners. It combines a light-
hearted approach, well rounded explanations and plenty of practice opportunities.
It makes an ideal companion for those students who are commencing a course
in differential equation at undergraduate or at graduate level.
The study of differential equations is a wide region in pure and applied
mathematics, physics, and engineering. All of these disciplines are anxious with
the properties of differential equations of different types. Pure mathematics
emphasis on the life and incomparability of solutions, while applied number
emphasizes the rigorous occasion of the way for approximating solutions.
Differential equations is an important legislature in modeling virtually every
physical, technical, or biological process, from celestial motion, to bridge
design, to interactions between neurons. Differential equations such as those
used to solve real-life problems may not necessarily be directly solvable, i.e. do
not have closed example solutions. Instead, solution tins be approximated using
numerical methods.
Many fundamental traditions of physics and courtesy tins be formulated as
differential equations. In biology and economics, differential equations are
used to patterns the allotment of complex systems. The mathematical opinion
of differential equations first developed together with the adeptness where the
equations had originated and where the results found application. However,
diverse problems, sometimes originating in quite distinct scientific fields,
may give rise to identical differential equations. Whenever this happens,
mathematical opinion seat the equations tins be viewed as a unifying outlook
sitting diverse phenomena. As an example, consider assembly of enlightenment
and sound in the atmosphere, and of shaft on the surface of a pond.
 CHAPTER

1
BASIC CONCEPTS OF
DIFFERENTIAL EQUATIONS

CONTENTS
1.1. Introduction......................................................................................... 2
1.2. The Bernoulli Equation....................................................................... 31
1.3. Differential Equations of Higher Order............................................... 50
1.4. The Wronskian................................................................................... 85
2 Differential Equations: Theory and Applications

1.1. INTRODUCTION
Mathematical models for real world phenomena often take the form of
equations relating various quantities and their rates of change (derivatives),
For example, the motion of a particle involves the distances covered in

ds
time t and velocity v or acceleration a. Now the rate of change of s
dt
dv
with respect to t is the velocity v and rate of change of velocity with
respect to t is the acceleration a . A particle moving in adtstraight line has an
equation of motion as s = f ( t ) , where t is in seconds and s is in meters.
Its velocity satisfies the equation
ds
v= = 4t 2 + 5t − 3
dt
This leads us to the definition of a differential equation (D.E).
A differential equation is a mathematical equation that relates some
function with its derivatives. In applications, the functions typically represent
physical quantities, the by – product represent their rates of change, and the
equation defines an association between the two. Because such relations are
exceptionally common, differential equations amusement a prominent role
in many disciplines including engineering, physics, economics, and biology.
In pure mathematics, differential equations are studied from several
different perspectives, mostly concerned with their solutions-the set of
functions that satisfy the equation. Only the simplest differential equations
are solvable by explicit formulas; however, some ownership of solutions
of a given differential equation may be determined without discovery their
exact form.

1.1.1. Differential Equation and their Classification

An equation involving one dependent variable and its derivatives with
respect to one or more independent variables is called differential equation.
For example
dy
+ y cos x =
sin x
dx
 Basic Concepts of Differential Equations 3

2
d2y  dy 
2
+ xy   = 0
dx  dx 
3
  dy 2  2 d 2 y
1+    =
  dx   dx 2

1.1.2. Ordinary Differential Equation (O. D. E)

A differential equation, in which ordinary derivatives of the dependent
variable with respect to a single independent variable occur, is called an
ordinary differential equation (O.D.E). Above equations (i) , (ii) and (iii) are
examples of Ordinary Differential Equation.

Partial Differential Equation

A differential equation involving partial derivatives of the dependent
variable with respect to more than one independent variable is called a
partial differential equation.
∂z ∂z
x + y =
nx
∂x ∂y

∂ 2u ∂ 2u ∂ 2u
+ 2 + 2 = 0
∂x 2 ∂y ∂z

Order of Differential Equation

The order of a differential is the order of the highest derivative that occurs
in the equation.

Degree of a Differential Equation

The degree of a differential equation is the greatest exponent of the highest
order derivative that appears in the equation. (The dependent variable and its
derivatives should be expressed in a form free from radicals and fractions).
dy
+ y cos x =
sin x
• dx (order 1, degree 1)
2
d2y  dy 
+ xy   = 0
• dx 2  dx  (order 2, degree 1)
4 Differential Equations: Theory and Applications

3
  dy 2  2 d 2 y
1+    =
•   dx   dx 2 (order 2, degree 2)
2
exponent of d y
is 2 after removing the radical by squaring both sides of
dx 2
the equation
∂z ∂z
x + y = nx
• ∂x ∂y (order 1, degree 1)
2 2 2
∂u ∂u ∂u
+ + 2=0
• ∂x 2 ∂y 2 ∂z (order 2, degree 1)
In this chapter, we shall study ordinary differential equation only.
Recall that a function T : U → V , where U ,V , are vector spaces over the
same field is called linear if, for α , β ∈ F , x, y ∈U ,
T (α x + β y=) α T ( x ) + β T ( y ) .

1.1.3. Linear Differential Equation

 dy d 2 y dny 
F x , y , , 2 ,…, n  =0
An ordinary differential equation  dx dx
to be
dx 
is said
2

linear if F is a linear function of the variables x , dy

n
,
dx dx
d y
,…,
d y
dx
. A similar 2 n

definition applies to partial differential equations. Thus, the general linear

ordinary differential equation of order n is
dny d n −1 y dy
a0 ( x ) + a1 ( x ) +… + an −1 ( x ) + an ( x ) y = F ( x )
dx n
dx n −1 dx

a0 ( x )
Where is not identically zero.
d3y 2
x d y dy
+ 2 e +y = x3
The equation dx 3
dx 2
dx is not linear because of the term
dy
y .
dx
It should be carefully noted that in a linear ordinary differential equation
• The dependent variable y and its derivative are all of degree one.
• No product of y or any of its derivative appear.
• No transcendental function of y and / or its derivative occur.
A differential equation that is not linear is called a nonlinear differential
 Basic Concepts of Differential Equations 5

equation. Differential equations occur in the mathematical formulation of

many problems in science and engineering. Some such problems are:
• Determining the motion of projectile, rocket, satellite or planet.
• Finding the charge or current in an electric circuit.
• Study of chemical reactions.
• Determination of curves with given geometrical properties.

Solution of a Differential Equation

A solution (or integral) of a differential equation is a relation between
the variables, not containing derivatives, such that this relation and the
derivatives obtained from it satisfy the given differential equation identically.
dy
= −µy −µx
For example, the equation dx has a solution y = c.e , where c is
an arbitrary constant.
d2y
+y= 0
The differential equation dx 2 has solution
= y A= cos x, y B sin x and = y A cos x + B sin x, where Aand B
are arbitrary constants.
A solution of a differential equation which contains as many arbitrary
constants as the order of the equation is called a general solution (or integral)
of the differential equation. A solution obtained from the general solution by
giving particular values to the constants is called a particular solution. The
graph of a particular integral is called an integral curve of the differential
equation.

Formation of a Differential Equation

f ( x , y , c1 , c2 , … , cn ) =
0
Given a relation (1)
c , c , …, cn
between variables x, y and containing n constants 1 2 it is always
possible to form a differential equation of order n such that the given relation
(1) is the general solution of the equation. This is done by differentiating (1)
n times thereby obtaining equations and then eliminating the n constants
from the original relation and n derived equations. The method is illustrated
by means of examples.
Example: The equation y= x + a (1)
 6 Differential Equations: Theory and Applications

Represents a family of parallel straight lines for different values of a.

Elimination of one constant ‘a’ requires two equations. The second equation
is obtained by differentiating (1).
dy
=1
Thus dx is the differential equation of the relation (1), with a
eliminated.
Example: From the differential equation by eliminating the two constants
A and B from the relation = y A sin x + B cos x. (1)
Solution: It is clear that three equations are required to eliminate two
unknowns A and B. We obtain two other needed equations by successive
differentiation of (1). Thus from (1), we have
dy
= A cos x − B sin x
dx (2)
d2y
− A sin x − B cos x =
= −y
dx 2 , using (1) (3)
2
d y
2
+ y= 0
So, dx (4)
is the required differential equation and (1) is its general solution.
Example: Find the differential equation of all parabolas whose axes are
parallel to the y − axis.
Solution: General equation of a parabola whose axis is parallel to the
y − axis is

y = ax 2 + bx + c. (1)

In order to obtain its differential equation, we have to eliminate a, b, c

from (1). For that, we need three more equations. Differentiating (1)
successively, we have
dy
= 2ax + b
dx

d2y
= 2a,
dx 2
 Basic Concepts of Differential Equations 7

d3y
= 0.
dx 3

The last equation does not contain any of the constants a, b and c. Thus,
d3y
=0
dx 3 is the differential equation of all parabolas whose axes are parallel
to the y − axis.

Initial and Boundary Condition

We have observed that general solution of a differential equation contains
the same number of arbitrary constants, as is the order of the equation.
Sometimes we need to find the solutions of differential equations subject to
supplementary conditions. Two types of conditions will be often encountered.

Initial Condition
It is often required to find the solution of a differential equation subject to
certain conditions. If the conditions relate to one value of the independent
variable such as
y = y0 at x = x0 (written as y ( x0 ) = y0 ) and
dy
= y′ ( x0 ) x = x0 , where x0 belongs to some interval ]α , β [ then
dx at
they are called initial conditions (or one- point boundary conditions) and 0
x
is called the initial point. An initial value problem consists of a differential
equation (of any order) together with a collection of initial conditions that
must be satisfied by the solution of the differential equation and derivatives
at the initial point.

1.1.4. Boundary Conditions

The problem of finding the solution of a differential equation such that all
the associated constraints relate to two different values of the independent
variable is called a two-point boundary value problem (or simply a boundary
value problem). The associated supplementary boundary conditions are
called two-point boundary conditions.
dy
= 2x
Example: Solve dx (1)
 8 Differential Equations: Theory and Applications

y (1) = 4.
Such that
2
Solution: This is an initial value problem. We note that y= x + c, c
y (1) = 4,
being arbitrary constant, is the general solution of (1). Since we
have
y (1)= 12 + c.

Therefore, 4 =
1+ c 3.
or c =
y x 2 + 3 is the solution of the initial value problem (1).
Thus, =
Note that the general solution represents a family of parabolas for
different values of c. The solution =y x 2 + 3 is a particular member of the

family that passes through

(1 ,4 ).
d2y
2
0
+y=
Example: Solve dx (1)
π  π  1
= y 2, y '  
= .
Subject to the conditions  4  4 2
Solution: Since both the conditions relate to one value of x, namely
π
x= ,
4 this is an initial value problem. We have already noted in previous
example that
= y A sin x + B cos x (2)
is the general solution of (1). Differentiating (2) w. r. t x, we get
=y′ A cos x − B sin x. (3)
By the given conditions, we have respectively from (2) and (3)
π  π π A B
y =2=
A sin + B cos =+
4  4 4 2 2

π  1 π π A B
y′   == A cos − B sin =− .
4  2 4 4 2 2
2 and A − B =
Hence A + B = 1
 Basic Concepts of Differential Equations 9

Solving for Aand B, we find that

3 1
=A = ,B
2 2

With these values of Aand B, the particular solution of (1) is

3 1
=y sin x + cos x
2 2

Example: Verify that 1 y = c cos x

and 2 y = c sin x
are solutions of
2
d y
+ y= 0.
dx 2 Find a particular solution of the equation satisfying the
boundary conditions
π 
y ( 0 ) 1=
= , y   2.
2

Solution: We have
y = c1 cos x.
Differentiating twice, we get
dy
= −c1 sin x
dx
d2y d2y
= c1 cos x = − y +y=0.
dx 2 or dx
2

d2y
y = c1 cos x is a solution of +y=0
Thus, dx 2 (1)

Similarly, it can be checked that

y = c2 sin x is also a solution of (1)
General solution of (1) is
=y c1 cos x + c2 sin x (2)
Applying the boundary conditions, we obtain from (2)
y ( 0 ) = 1 = c1 + 0
and
π 
y  = 2= 0 + c2 .
2
 10 Differential Equations: Theory and Applications

Thus 1
c =1and 2
c = 2. Hence the particular solution of (1) satisfying
the given conditions is
=y cos x + 2sin x
Example: Solve the boundary value problem
2
d y
= ( 0 ) 1, y=
+ y 0, y= (π ) 5.
dx 2
Solution: Applying the boundary conditions to the general solution
=y A sin x + B cos x
of the given equation, we have
y ( 0 ) = 1= B

y (π ) = 5 = − B

Thus, we obtain two values of B and we are unable to determine any

definite value of A. Hence, the boundary value problem has no solution. It
follows that a boundary value problem need not always have a solution.
In above section we classified the differential equations and saw how a
differential equation can be formed by eliminating constants from a given
functional equation. The problem of finding general solution (integral) of
a given differential may or may not exist. Even if the integral of a given
equation exists, it may not be easy to find. We shall only discuss methods of
solutions of special types of differential equations.

Separable Equations
Definition. A differential equation of the type
F ( x ) G ( x ) dx + f ( x ) g ( y ) dy =
0
(1)
Is called an equation with separable variables or simply a separable
equation. Equation (1) may be written as
F ( x) g ( y)
dx + 0
dy =
f ( x) G ( y)

which can be easily integrated.

 Basic Concepts of Differential Equations 11

dy x 2
=
Example: Solve dx y (1)
Solution: Equation (1) can be written as
y dy = x 2 dx.
Integrating both the sides, we get
y 2 x3
= + c1
2 3 or
y 2 2x2 + c
3=
which is the required solution of (1).
dy 1
= .
Example: Solve dx x tan y
Solution: The given equation is separable and can be written as
dx
= tan y dy
x
Integrating, we have
ln x =
− ln cos y + c
or
ln x + ln cos y =
c

ln x cosy = c
i.e., or
x cos y= e=c a ( say )
x cos y = a because a > 0.

Example: Solve
( )
x sin y dx + x 2 +1 cos y dy =
0
(1)

Solution: Dividing (1) by

(x 2
)
+1 sin y
, we have
x
2
dx + cot y dy =
0
x +1
Which is a separable equation. Therefore,
 12 Differential Equations: Theory and Applications

x dx
∫ + ∫ cot y dy= c1= ln c2
x2 + 1
1
2
(
ln x 2 +1 + ln sin y = )
ln c2
i.e.,

or
( )
ln x 2 +1 + 2 ln sin y =2 ln c2 .

( sin y ) ln sin 2 y
2 2
2=
ln sin y ln
= sin y ln =
Now, we can write and
2 2 2
2 ln c2 = ln c2 = ln c = ln c, where c = c > 0
2 2

2 2
Hence ln( x + 1) sin y =
c is the required solution.
Example: Solve the initial value problem
dy 2x
= , y ( 0 ) = −2.
dx y + x 2 y
Solution: We have
dy 2x
=
dx y 1 + x 2( ) or
2x
y dy = dx
1+ x 2
integrating, we obtain
y2
= ln(1 + x 2 ) + ln c,
2 c being a constant.
y2
= ln c 1 + x 2
2
(( ))

( )
2
=y 2 ln( c 2 1 + x 2 )
(i)
y ( 0 ) = −2
Now setting into (1), we have
2 2 4
=4 ln=
c or c e .

So (i) becomes
 Basic Concepts of Differential Equations 13

( ( ))
2
=y 2 ln e 4 1 + x 2

which is the required solution.

1.1.4. Homogenous Equations

f ( x, y )
Definition. A function is called homogeneous of degree n if
f ( tx , ty ) = t n f ( x, y ) ,

x10 + y10 x

xy , 2 2
and sin  
where t is a nonzero real number. Thus x + y  y  are
homogeneous function of degree 1,8 and 0 respectively.
A first order differential equation
dy
= f ( x, y )
dx (1)
is said to be homogeneous function of any degree. If (1) is written in the form
M ( x, y ) dx + N ( x, y ) dy =
0
. Then it is called homogeneous functions of
the same degree.
y
1+
dy x +y 1 x
= ln x − ln y + = ln +
dx x− y y y
1−
The equation x x is homogenous, but
2
dy y 3 + 2 xy  y  y
= = 2
y   + 2   is not homogeneous.
dx x x x
dy  y
=g 
Theorem. A homogeneous equation dx  x  (1)
can be transformed into a separable equation ( in the variables v and x) by
the substitution y = vx.
dy dv
Proof: Put y = vx into (1). Then = v+ x and (1) becomes
dx dx
xdv
v+ = g (v)
dx
dv
v − g (v) + x 0
=
or dx
 14 Differential Equations: Theory and Applications

 v − g ( v )  dx + x dv =
0.
or 
This equation is separable and can be solved as in the previous section.
dy x3 + y 3
= 2 2
Example: Solve dx x y + xy
3
 y
1+ 
dy  x
= 2
dx  y  y
  +  
Solution: We have  x x

So that the equation is homogeneous. Setting y = vx into (1), we obtain

dv 1 + v3 1+ v3 v 2 − v +1
v+ x = = =
dx v + v 2 v ( 1 + v ) v

dv v 2 – v +1 1− v
=x = –v
dx v v
v dx
dv =
or 1 − v x
 1  dx
 −1 +  dv =
or  1−v x
integrating, we get
− v − ln 1 –=
v ln x + ln c

ln cx ( 1 – v ) + v =0
or
y
v by
Replacing x in the above equation, we have
y
ln c ( x – y ) + 0
=
x

which is the required solution.

Example: Solve the initial value problem

(x 2
)
+ 3 y 2 dx − 2 xy dy =
0
,
y ( 2 ) = 6.
 Basic Concepts of Differential Equations 15

2
 y
3 2 3   +1
dy x + 3 y x
= =
dx 2 xy  y
2 
Solution: Here  x  (1)
Which shows that the equation is homogeneous. Putting y = vx into (1),
we have
dv 3v 2 + 1
v+ x =
dx 2v

dv 3v 2 +1 v 2 +1
.
x= = −v
dx 2v 2v

2v dx
2
dv =
or 1 + v x
on integrating, we get
( )
ln 1 + v 2 = ln x + ln c

1 + v2 =
cx
or
y
v by ,
replacing x we obtain
y2
1+ 2 = cx
x
x2 + y 2 =cx x 2
or
Now if x ≥ 0, we can write this as
x2 + y 2 =
cx3
2 3 2
or y= cx − x
± cx 3 − x 2
or y =
Applying the initial condition, we get
y ( 2) =
6=
± 8c − 4
16 Differential Equations: Theory and Applications

or 8=
c − 4 36 =
i.e., c 5
=
Hence y 5 x3 − x 2 is the required solution. We take the plus sign in
y ( 2)
the radical since is positive.

1.1.5. Differential Equation Reducible to Homogeneous Form

Some differential equations are not homogeneous. But they can be reduced
to homogeneous form.

The differential equation

( a1 x + b1 y + c1 ) dx + ( a2 x + b2 y + c2 ) dy =
0
is not homogeneous. But it can be reduced to a homogeneous form as
illustrated below:
a1 b
≠ 1
Case 1. If a2 b2 , then make the transformations
x =X + h, y =Y + k.

The given equation becomes

( a1 X + b1Y + a1h + b1k + c1 ) dX + ( a2 X + b2Y + a2 h + b2 k + c2 ) dY =
0
(1)
Let h and k be the solution of the system of equations
a1h + b1k + c1 = 0

0
a2 h + b2 k + c2 =

Then for these values of h and k , (1) reduces to the homogenous

equation
( a1 X + b1Y ) dX + ( a2 X + b2Y ) dY =
0

in the variables X and Y .

a1 b1
= ,
Case II. If a2 b2 then put= z a1 x + b1 y and the given equation will
reduce to a separable equation in the variable x and z.
dy 2 y – x + 5
=
Example: Solve dx 2 x – y − 4 (1)
 Basic Concepts of Differential Equations 17

Solution: Putting x =X + h, y =+
Y k
So the given equation becomes
dY 2Y – X + 2 K − h + 5
=
dX 2 X –Y + 2hX k − 4 (2)
The solution of the system of equations
−h + 2k + 5 =0 

2h − k − 4 = 0  is h = 1, k = −2.

For these values of h and k , (2) reduce to

dY 2Y − X
=
dX 2 X − Y (3)
which is homogeneous.
Putting Y = vX into (3), we have
dv 2v −1
v+ X =
dX 2 − v which is separable.
 2−v  dX
 2  dv = .
Therefore,  v − 1  X
Integrating, we obtain
v −1 1
ln − ln v 2 =
−1 ln X + ln c0
v +1 2
2
v −1 2 2
ln = ln X + ln c0 + ln v 2 −1
or v+ 1
2

ln
=
v −1
v +1
2
ln c0 X . v 2 −1 { }
2
v −1 2
c c02
= c X . v −1 . v +1 , where=
v +1
2 3
= v –1 c X . v +1
or
18 Differential Equations: Theory and Applications

Y
v by ,
Replacing X this becomes
3
Y−X 2 Y+X
=c X .
X X
3
Y –X
= c Y+X
or (4)
But X= x –1 , Y = y + 2 . Hence (4) takes the form
3
y – x + 3= c y + x +1

which is the required solution.

Example: Solve
( 2 x + y +1 ) dx + ( 4 x + 2 y − 1 ) dy =0 (1)
a1 2 b1 1
= = = ,
Solution: Since 2 a 4 b2 2 we put 2 x + y = z. then (1) becomes

( z +1 ) dx + ( 2 z –1 )( dz − 2dx ) =
0

or
( z +1 – 4 z +2 ) dx + ( 2 z –1 ) dz =0
3 ( 1 – z ) dx + ( 2 z –1 ) dz =0
i.e., which is separable.
Dividing by 1 − z , we have
2 z −1
3 dx + 0
dz =
1− z
 1 
3.dx +  −2 +  dz =0
or  1− z 
integrating, we obtain
3 x − 2 z – ln 1 – z =
c0

Replacing z by 2 x + y, it reduces to
3 x − 2 ( 2 x + y ) − ln 1 − 2 x − y =
c0

– x − 2 y − ln 2 x + y −1 =c0
or
 Basic Concepts of Differential Equations 19

ln 2 x + y −1 + x + 2 y =c, where c =−c0

or
is the required solution.

Exact Equation
M ( x, y ) dx + N ( x, y ) dy
Definition. The expression (1)
is called an exact differential if there exists a continuously differentiable
f ( x, y )
function of two real variables x and y such that the expression
equals the total differential df . We know from calculus that
∂f ∂f
=
df dx + dy.
∂x ∂y
Thus, if (1) is exact then
∂f ∂f
M ( x , y=
) = f x and N ( x , =
y ) = fy.
∂x ∂y
If (1) is an exact differential then the differential equation
M ( x , y ) dx + N ( x , y ) dy =
0

is called an exact equation.

M ( x , y ) dx + N ( x , y ) dy =
0
Theorem. The differential equation (1)
∂M ∂N
= ,
is an exact equation8.U&hKŽ§ if and only if ∂y ∂ x where the func-
M ( x, y ) N ( x, y )
tions and have continuous first order partial derivatives.
Proof. Suppose that the equation (1) is exact so that M dx + Mdy is an exact
f ( x, y )
differential. By definition, there exists a function such that
∂f ∂f
M ( x, =
y ) = f x and N ( x, = y) = fy
∂x ∂y

∂M ∂2 f
M=
y = f=
xy
Then, ∂y ∂y ∂x
∂N ∂2 f
N=x = f=
yx
∂x ∂x ∂y
20 Differential Equations: Theory and Applications

Since M and N possess continuous first order partial derivatives, we

∂M ∂N
=
f xy = f yx
have and, therefore ∂y ∂x as desired.

Solution of an Exact Equation

M ( x , y ) dx + N ( x , y ) dy =
0
If (1)
f ( x, y )
is an exact equation, then there exists a function such that
∂f ∂f
df = dx + dy = M ( x, y ) dx + N ( x, y ) dy
∂x ∂y
∂f ∂f
= M= and N
Therefore ∂x ∂y .
∂f
=M
Integrating ∂x with respect to x, we have
f ( x, y ) =
∫ M dx + h ( y ) .
(2)
h( y)
The constant of integration is an arbitrary function of y since it
must vanish under differentiation w.r.t x.
Differentiating (2) w.r.t. y , we get
∂f ∂
=
∂y ∂y
( )
∫ Mdx + h′ ( y )

∂f ∂
∂ y
=N =
∂ y
( )
∫ Mdx + h′ ( y )
i.e.,
∂
= h′ ( y ) N −
∂ y
(∫ Mdx )
so
f ( x, y ) = c
integrating the above equation w.r.t y, we obtain h and hence
is the required solution of (1).

Example: Solve
( 3x 2
) ( )
y + 2 dx + x3 + y dy =
0
(1)
2 3
M 3 x y + 2 and N= x + y
Solution: Here=
 Basic Concepts of Differential Equations 21

∂M ∂N
= 3= x2 , 3x 2
∂y ∂x

∂M ∂N
=
Thus ∂ y ∂x and so the equation is exact.
To find the solution of (1), we note that the left hand side of the equation
f ( x, y )
is an exact differential. Therefore, there exists a function such that
∂f
= 3x 2 y + 2
∂x (2)
∂f
= x3 + y
And ∂y (3)
Integrating (2) r.t x , we have
w.
f ( x, y ) = x 3 y + 2 x + h ( y ) ,

h( y)
where is the constant of integration. Differentiating the above equation
w.r.t y and using (3), we obtain
∂f
x3 h′ ( y ) =+
=+ x3 y
∂y
h′ ( y ) = y.
or
Integrating, we have
y2
h( y) =
2

y2
f ( x, y ) = x 3 y + 2 x +
Thus, 2
Hence the general solution of (1) is
y2
x3 y + 2 x + =
c
2
 22 Differential Equations: Theory and Applications

Alternative Method
Integrating (2) and (3) w.r.t x and y respectively, we have
f ( x, y ) = x 3 y + 2 y + h ( y )

y2
f ( x, y ) = x 3 y + + g ( x).
and 2
y2
h( y) =
Thus, 2 and g ( x ) = 2 x .
So, the general solution is
y2
x3 y + 2 x + c.
=
2
Example: Solve the initial value problem
( 2 y sin x cos x + y 2
) ( )
sin x dx + sin 2 x − 2 y cos x dy = 0, y ( 0 ) = 3
2
=
Solution: Here M 2 y sin x cos x + y sin x
2
=
and N sin x − 2 y cos x
∂M
= 2sin x cos x + 2 y sin x
∂y

∂N
= 2sin x cos x + 2 y sin x
∂x

∂M ∂N
=
Thus ∂y ∂x showing that the given equation is exact.
f ( x, y )
Hence there exists a function such that
∂f
= 2 y sin x cos x + y 2 sin x
∂x . (1)
∂f
= sin 2 x − 2 y cos x
∂y (2)
Integrating (1) w.r.t . x . , we have
erentiating this equation w.r.t y and using (2), we get
 Basic Concepts of Differential Equations 23

sin 2 x − 2 y cos x + h' ( y ) = sin 2 x − 2 y cos x

= h′ ( y ) 0=
and so h ( y ) c1.
i.e.,
Hence the general solution of the given equation is
f ( x, y ) = c2

i.e.,
y sin 2 x − y 2 cos x + c1 =
c2

y.sin 2 x − y 2 .cos x = c2 − c1 = c

Applying the initial condition that when=x 0,=y 3, we have −9 =c .

2 2
Hence y cos x − y sin x =
9 is the required solution.

1.1.6. Integrating Factors

M ( x, y ) dx + N ( x, y ) dy =
0
Definition. If the differential equation (1)
µ ( x, y )
is not exact but when it is multiplied by a function and the resulting
µ ( x , y ) M ( x , y ) dx + µ ( x , y ) N ( x , y ) dy = 0
equation is exact, then
µ(x ,y)
is called an integrating factor (I.F) of the differential equation (1).
The number of integrating factors of an equation may be infinite.
We list below (without proofs) some rules to find the integrating factors
of equations of special types.
M ( x , y ) dx + N ( x , y ) dy =
0
Rule 1. If (1)
M y − Nx
= P,
is not exact and N where P is a function of x only then (1)
µ ( x) µ ( x)
has an integrating factor which also depends on x. is solution
of the differential equation
dµ
= Pµ
dx

i.e.,
(=
x) exp ∫ P dx
24 Differential Equations: Theory and Applications

∂M ∂N
= My = , Nx .
Note that ∂y ∂x
Nx − M y
= Q,
Rule II. If M where Q is a function of y only, then
0 has an integrating factor
the differential equation Mdx + Ndy =
µ (=
y ) exp ∫ Q dy.

0 (1)
Rule III. If M dx + Ndy =
1
is homogeneous and xM + yN ≠ 0, then xM + yN is an I.F of (1).
0 is of the form y f ( x, y ) dx + x g ( x, y ) dy =
Rule IV. If M dx + Ndy =
0
1
and xM − yN ≠ 0, then xM − yN is an I.F of (1).
The following differential formulas are useful in the calculation of
certain exact equations:
 y  x dy − y dx
d =
x x2

 x  y dx − x dy
d =
 y y2

d ( xy
= ) x dy + y dx

( )
d x 2 + y 2= 2 ( x dx + y dy )

 x  y dx − x dy
d  ln  =
 y xy

 x  y dx − x dy
d  arc tan  = 2
 y  x + y2

Example: Solve
( )
y dx + x 2 y − x dy =
0
 Basic Concepts of Differential Equations 25

Solution. The equation is not exact. Rearranging the equation, we have

y dx − x dy + x 2 y dy =
0

y dx − x dy
0
+ ydy =
or x2
  y
0.
Now it is an exact equation and may be written as − d  x  + y dy =
Integrating, we have
y y2
− + =
c
x 2
2
or xy − 2 y =cx is the general solution.

Example: Solve
( )
x 2 − 2 x + 2 y 2 dx +2 xy dy =
0
(1)
M y − Nx 4 y − 2 y 1
= =
Solution. Here N 2 xy x
dµ µ
µ ( x, y ) =
Therefore, I.F is the solution of dx x or µ = x is an I.F.
Multiplying the equation by x, we have
(x 3
)
− 2 x 2 + 2 xy 2 dx + 2 x 2 y =
0.

This equation is exact. So we can easily find that its solution is

x 2 x3
4
− + x2 y 2 =
c0 4 3 2 2
4 3 or 3 x − 8 x + 12 x y =c is the solution of (1).
 x 
dx +  − sin y  dy =
0
Example: Solve  y  .
Solution. Here, by rule II,
1
−0
Nx − M y y 1
= =
M 1 y
dy ln y
is a function of y only. Therefore, µ ( y ) = exp ∫ y = e = y is an I.F.
Multiplying the equation by y, we have
26 Differential Equations: Theory and Applications

y dx + ( x − y sin y ) dy =
0

or y dx + x dy − y sin y dy =0
d ( xy ) − y sin y dy =0.
or
Integrating, we get xy + y cos y − sin y =
c which is the required
solution.

Example: Solve
(x 2
) ( )
y − 2 xy 2 dx − x3 − 3 x 2 y dy =
0
(1)
Solution: The equation is homogeneous but not exact. We have
xM + yN = x 3 y − 2 x 2 y 2 − x 3 y + 3 x 2 y 2 = x 2 y 2 ≠ 0

1 1
= 2 2
So, using Rule III, xM + yN x y is an I.F. Therefore, multiplying
1
,
(1) by x y
2 2
we obtain
1 2  x 3
0
 −  dx −  2 −  dy =
 y x  y y

This equation is exact. Integrating, we get

x
− 2 ln x + 3ln y =
c
y

is the required solution.

x
,
Note that, here the term y involving both x and y, has been taken
only once.

Example: Solve
( ) ( )
y x . y + 2 x 2 y 2 + x xy − x 2 y 2 dy =
0
(1)
y f ( xy ) dx + x g ( xy ) dy =
0
Solution: The equation is of the form
2 2 3 3 2 2 3 3 3 3
Now, xM − yN = x y + 2 x y − x y + x y = 3 x y ≠ 0.
1
33
Therefore, x y is an I.F.
 Basic Concepts of Differential Equations 27

1
33
,
Multiplying (1) by x y we have
 1 2   1 1
 2 +  dx +  2 −  dy = 0.
x y x   xy y

This is an exact equation. Integrating, we get

1
− + 2 ln x − ln y =
c
xy

as the required solution.

Linear Equations
A first order ordinary differential equation (ODE) is linear in the dependent
variable y and the independent variable x if it is or can be written in the
form
dy
+ P ( x) y =
Q ( x),
dx (1)
where P and Q are functions of x.
Solution of a Linear Equation.
dy
+ P( x ) =
Q( x )
The linear equation dx may be written as
 P ( x ) y − Q ( x )  dx + dy = 0
(2)
which is of the form M dx + N 0, where
dy =
P ( x ) y − Q ( x ) and N =
M= 1.

∂M ∂N
= P= ( x ) and 0
Now ∂y ∂x
P ( x) = 0
Thus (2) is not exact unless in which case (1) is separable.
However, an integrating factor (depending only on x) of (2) may be easily
µ ( x) µ ( x),
found. Let be an I.F. of (2). Then multiplying (2) by we get
28 Differential Equations: Theory and Applications

 µ ( x ) P ( x ) y − µ ( x ) Q ( x )  dx + µ ( x ) dy = 0
(3)
Now (3) is an exact equation if and only if
∂ ∂
 µ ( x ) P ( x ) y − µ ( x ) Q ( x )  =  µ ( x ) 
∂y ∂x 

d
µ ( x) P ( x) =  µ ( x ) 
This condition reduces to dx 
dµ
µ P ( x) =
i.e., dx
dµ
= P ( x ) dx,
or µ
Integrating, we obtain
ln µ = ∫ P ( x ) dx

exp[ P ( x ) dx ] > 0.
µ =∫
or
e ∫ P dx or [exp ∫ P ( x ) dx ]
Thus is an I.F. of the linear equation (1).
dy ∫ P dx
e ∫ P dx +e P ( x) y =
Q ( x ) e ∫ P dx
Multiplying (1) by e ∫ P dx
, we have dx
d  ∫ P dx 
 y e dx  = Q ( x ) e ∫ P dx
or dx
Integrating this we obtain the solution of (1) in the form
∫[ Q ( x ) e ∫ P dx ] dx + c
y e ∫ P dx =

dy
( x − 1)
+ 4 ( x − 1) y =+
3 2
x 1
Example: Solve dx (1)
Solution: We write the equation in the standard form
dy 4 x +1
+ y =3
dx x − 1 ( x − 1)
4
P ( x) = .
Here x − 1 Therefore, an I.F. of (1) is
 Basic Concepts of Differential Equations 29

4 dx
] =exp ln ( x − 1)  =( x − 1)
4 4
exp[ ∫
x −1  

Multiplying (1) by this I.F., we get

dy
( x − 1) + 4 ( x − 1) y =x 2 − 1
4 3

dx
d 
±( ℵ1)  2 1
4

or dx  
integrating, we obtain
x3
y ( x − 1) =
4
−x+c
3

which is the required solution.

( x + 2 y ) dy
3

dx
=
y
Example: Solve
dy y
= 3
Solution: We have dx x + 2 y
This equation is clearly not linear in y. But in the first order differential
equation, the roles of x and y are interchangeable in the sense that either
variable may be regarded as dependent variable. Let us regard x as dependent
variable and y as independent variable. The equation may be written as
dx 1
2 y3
− x=
± (1)
  1   1 1
I .F
=. exp  ∫  −  dy=
 exp ln =

which is linear in x with   y   y y .
1
,
Multiplying (1) by y we get
1 dx 1
− 2y
x=
y dy y 2

d x
  =2y
or dx  y
 30 Differential Equations: Theory and Applications

x
= y2 + c =
Integrating, we have y or
x y y2 + c ( ) is the required
solution.
Example: Solve the initial value problem
dr π 
+ r tan θ cos 2 θ , =
= r  1
dθ 4
Solution. The equation is linear in r with
I .F . = exp [ ln sec θ ] =
exp  ∫ tan θ dθ  = sec θ ,

Multiplying the given equation by sec θ , we have

dr
sec θ + r sec θ tan θ =
cos θ
dθ
d
[ r secθ ] = cos θ
dθ
Integrating, we obtain
r sec
= θ sin θ + c
Applying the initial condition, we have
π  π 
1.sec
=   sin   + c
4 4
1 1
c= 2− = ,
or 2 2
1
r sec
= θ sin θ +
Therefore, 2
cos θ
=r sin θ cos θ +
2

or 2r sin 2θ +
= 2 cos θ is the required solution.
 Basic Concepts of Differential Equations 31

1.2. THE BERNOULLI EQUATION

dy
Definition. An equation of the form dx
+ P ( x) y =
Q ( x ) y n (1)
is called the Bernoulli differential equation. This equation is linear if
n = 0 or 1. If n is not zero or 1, then (1) is reducible to a linear equation.
Dividing by y , (1) becomes
n

dy
y−n + P ( x) y1− n =
Q ( x)
dx (2)
1− n
In (2), put v = y then it reduces to
dv
+ (1 –n ) P ( x) v =
(1 –n )Q ( x)
dx
which is linear in v.
Note. Consider the equation
dy
f ′( y) + P ( x) f ( y) =Q ( x)
dx

v = f ( y), dv
Letting this equation becomes dx + P ( x ) v =
Q ( x ) which is linear
in v.
1
dy xy
+ 2
=
xy 2

Example: Solve dx 1 − y (1)

1

Solution: Dividing by y , (1) becomes

2

1 1
− dy x
y 2
+ y2 =
x
dx 1 − x 2 (2)
1

Put y = v
2

1 − 12 dy dv
y =
or 2 dx dx
then (2) reduces to
dv x x
+ .v=
dx 2 1 − v 2
( 2 ) (3)
32 Differential Equations: Theory and Applications

This is linear in v.
 x   1 
1

( ) ( )
−
exp  ∫
I .F . = dx  =exp  − ln 1 − x 2  =−
1 x2 4

 2 1 − x 2 ( )   4 
1

( )
−
1− x 2 4
Multiplying (3) by , we get
1
dv x x
( 1− x )
−
2 4 + 5
. v= 1
dx
(
2 1− x 2 ) 4
2 1− x 2 ( ) 4

d  1
 1  1

( ) ( )
− −
2
 1− x
4
v= −  − 2 x 1− x 2 4

dx   4  
or
Integrating, we have
3
1
1 1− x ( 2
) 4

( )
−
v 1− x 2 4
=
− +c
4 3
4
1
1− x 2
v =c ( 1− x ) 2 4
−
3
or
1 1
1 − x2
( )
−
2
y =− c 1 x
2
− 4

or 3
is the required solution of (1).

1.2.1. Equation Solvable for p

2 2 2
Example: Solve x p + xp − y – y =
0. (1)
Solution: we factorize the left hand side of (1) to obtain
(x 2
p 2 − y 2 += )
( xp – y ) ( xp – y )( xp=
+ y +1 ) 0

Therefore, either
xp – y = 0 (2)

or xp + y + 1 =0 (3)
 Basic Concepts of Differential Equations 33

dy dy dx
=y
x =
(2) gives dx or dx x
or y = cx . (4)
xp =− ( y +1 )
From (3), we have
dy dy dx
x = − ( y +1 ) = −
or dx or y + 1 x
x ( y +1 ) = c.
which yields (5)
Combining (4) and (5), the required solution of (1) is
( cx )( xy + x – c ) =
y – 0.

Example: Solve
( ) ( )
xp 3 − x 2 + x + y p 2 + x 2 + xy + y p – xy =0
(1)
Solution: By inspection, we find that p − 1 is a factor of left hand side
of (1). Thus, the given equation is
( p –1 )  ( )
xp 2 − x 2 + y p + xy  =
0

or
( p –1 )( xp – y )( p – x ) = 0.
Therefore, either
p –1 = 0 (2)

or xp – y = 0 (3)
or p – x = 0 (4)
From (2), we have
dy
= 1 or y = x + c
dx (5)
From (3), we get
dy
=
x y= or y cx
dx (6)
From (4), we obtain
dy
=x
dx
34 Differential Equations: Theory and Applications

x2
y= +c or x 2 + 2 ( y – c ) =0.
or 2 (7)
The general solution of (1) is obtained by combining (5), (6) and (7).

Thus
( (
y – x – c )( y – cx ) x 2 − 2 y + 2c =0 )
is the general solution of (1).

1.2.2. Equations solvable for y

Example: Solve= y p 2 x + p (1)

Solution. Differentiating (1) w.r.t.x, we have

dy dp dp
=p 2 + 2 xp
p= +
dx dx dx
dp
( 2 px + 1) + p 2 − p =
0
or dx
dp − p ( p –1 )
=
dx 2 px +1

dx −2 px − 1
=
dp p ( p − 1)

dx 2 x 1
+ =
−
dp p –1 p ( p –1)
i.e., (2)
which is linear in x.
 2 
dp  =exp ln ( p − 1)  =( p − 1) .
2 2
I .F . =exp  ∫
 p −1   

( p − 1)
2

Multiplying (2) by and integrating, we get

x ( p − 1) = c − p + ln p
2

(3)
c − p + ln p
x= .
( p − 1)
2

Substituting this value of x into (1), we have

 Basic Concepts of Differential Equations 35

c − p + ln p
=y p2. +p
( p − 1)
2

y ( p − 1)= p 2 ( c − p + ln p ) + p
2

(4)
Thus, (3) and (4) constitute the required solution of (1) with p as a
parameter.
p 2 x 4 (1)
Example: Solve y + px =
Solution. From (1), we get
=y p 2 x 4 − px

Differentiating the above equation w.r.t.x, we have

dy dp dp
= p = 4 x3 p 2 + 2 px 4 − p−x
dx dx dx

dp dp
2 p − 4 x3 p 2 − 2 x 4 p +x 0
=
dx dx

( ) (
2 p 1 − 2 px3 + x 1 − 2 px3 ) dp
dx
0
=

(1 − 2 px )  2 p + x dp
2 
=
dx 
0

3
Hence, either 1 − 2 px =0
dp
2p + x = 0
or dx (2)
The equation (2) gives
dp dx
+2 = 0
p x

or ln p + 2 ln x =
ln c
c
2 p=
or px = c or x2
36 Differential Equations: Theory and Applications

c
y c2 −
=
Substituting this value of p into (1), we obtain x or
xy − c 2 x + c =0
which is the required solution.
We have yet to deal with relation (2). If we eliminate p from (1) and
(2), we get
1
y=−
4 x 2 (3)
It is easy to check that (3) is also a solution of (1) and this solution does
not involve any constant.

Equation Solvable for x

3
Example: Solve xp = 1 + p (1)
1
x= +p
Solution: We have p (2)
Differentiating (1) w.r .t . y , we have
dx 1 1 dp dp
= = − 2 +
dy p p dy dx

1  1  dp
= 1 − 2 
or p  p  dy
 1  dp
=1  p− 
or  p  dy
Which is a separable equation. Therefore,
 1
=  p −  dp
dy
 p

p2
c+ y= − ln p
Integrating, we get 2

i.e., 2 y =p 2 − 2 ln p − 2c (3)
Thus, (2) and (3) constitute the solution of (1).
 Basic Concepts of Differential Equations 37

1.2.3. Clairaut’s Equation

y xp + f ( p ) ,
=
Definition. The equation (1)
is known as Clairaut’s equation. Differentiating both sides of (1), w.r.t.x,
we get
p =p + xp′ + f ′ ( p ) . p′

Cancelling like terms, we have

xp′ + p′f ′ ( p ) = 0

p′  x + f ′ ( p )  = 0
or
Since one of the factors must be zero, two different solution arise.
• If p′ = 0, then p = c and substitution of this value into (1) yields
the general solution
y cx + f ( c )
=

x + f ′( p) =0, x = − f ′( p)
• If then and (1) can be rewritten as
y= − pf ( p ) + f ( p )
′

Thus x and y are both expressed as functions of p and we obtain the

parametric equations
x = − f ′( p) 

=y f ( p ) − p f ′ ( p ) 
(2)
of a curve, representing a solution of (1). The solution (2) is called the
singular solution. This solution is not deducible from the general solution
p may be eliminated between the two equations in (2) to get a relation in
x and y involving no constant.
Example: Find the general solution and singular solution of
1
y xp + p 4
=
4 (1)
1
y cx + c 4 .
Solution. The general solution of the equation is =
4
Differentiating (1) w.r.t.x, we have
38 Differential Equations: Theory and Applications

p =p + xp′ + p 3 p′
3
or x = − p (2)
We eliminate p from (1) and (2) and get
1
1 4
3 1
x ( − x )3 +
y= ( – x)3 =
− x ( − x )3
4 4
3 4
or 64 y + 27 x =0 is the singular solution.
x 2 ( y − px ) =
yp 2
Example: Solve (1)
Solution: This is not Clairaut’s equation. Let us write
2
x 2 = u and y = v

Then, 2 x dx = du and 2 y dy = dv.

y dy dv
=
Hence x dx du
dy x dv
=
p =
dx y du
Substituting this value of p into (1), we get
2
2 x 2 dv  x 2  dv 
x y− =  
 y du  y  du 
2
dv  dv 
y 2 − x2 = 
or du  du 
2
dv  dv 
v −u = 
or du  du 
2
dv  dv 
= v u + 
or du  du  (2)
This is Clairaut’s equation. Hence its general solution is
v cu + c 2
=
 Basic Concepts of Differential Equations 39

or =y 2 cx 2 + c is the general solution of (1). Differentiating (2) w.r.t u ,

we get
dv d 2 v dv dv d 2 v
= u 2+ +2
du du du du du 2

d 2v  dv 
2 
u+2 = 0,
or du  du  v cu + c 2 , which is the general solution. If
then =
dv
u+2 =0,
du then
dv
u = −2
du (3)
2
dv  dv 
= v u + 
And du  du  (4)
give the singular solution.
dv
Eliminating du from (3) and (4), we obtain
u2
v= −
4
2
or u + 4v =
0 is the singular solution.
2 2
Replacing u , v by x and y respectively, we have
2
• y= cx 2 + c 2 as the general solution.

• x4 + 4 y 2 =
0 as the general solution.

1.2.4. Envelope
f ( x, y , c ) = 0
Let be a one-parameter family of curves. Suppose all
members of the family of curves are drawn for various values of the
parameter c arranged in order of magnitude. The two curves that correspond
to two consecutive values of c will be designated as neighboring curves. The
40 Differential Equations: Theory and Applications

locus of the ultimate points of intersection of neighboring curves is called

f ( x, y, c ) = 0.
the envelope of the family
Let A, B , C represents three neighboring intersecting members of the
family. Let P be the point of intersection of Aand B and Q be the point of
intersection of B and C . By definition P and Q are points on the envelope.
Thus the curve B and the envelope have two contiguous points common,
and therefore, have ultimately a common tangent. Hence B and the envelope
touch each other. In the same way, we may show that the envelope touches
any other member of the family. Thus, a curve E which, at each of its points,
touches someone of the curves of the family is the envelope of the family.
For the family of circles with centre on the x − axis and radius 1 , that
( x − 1)
2
+ y2 =
1,
is, the family of circles with equation
The pair of lines y = ±1 is the envelope. It is easy to see that the lines
y = ±1 touch each member of the family of the circles.

Equation of the Envelope

f ( x, y, c ) = 0,
To find an equation of the envelope of the family consider
two neighboring curves and
f ( x, y, c ) = 0 

f ( x, y , c + h ) =
0 
(1)
Find the intersection of these two curves and let h → 0. The point
of intersection then must approach the point of contact of the curve
f ( x, y , c ) = 0
with the envelope. At the point of intersection of the equation
f ( x, y , c + h ) − f ( x, y , c )
=0
h (2)
is true as well as (1). Letting h → 0, we have from (2)
f c ( x, y , c ) = 0
(3)
and from (1)
f ( x, y , c ) = 0
(4)
 Basic Concepts of Differential Equations 41

If we eliminate c from (3) and (4), we obtain an equation of the envelope.

f ( x, y , c ) = 0
The eliminant is called the c - discriminant of the family .
The c-discriminant may contain loci other than the envelope.
( x − c)
2
− 2y =
0
Example: The family of parabolas has the x-axis as
envelope.

Singular Solution
Let
f ( x, y , p ) = 0
(1)
be a nonlinear first order differential equation in which the left-hand side is
a polynomial in p. The general solution of this differential equation will be
a one parameter family
f ( x, y , c ) = 0
(2)
The envelope E of (2) is a curve which, at each of its points, touches
some one member of the family (2). At a point of contact P of the envelope
and a member of (2), the values of x, y, p are the same. But the values of
x, y, p for the curve at P satisfy (1). Thus, the envelope of family (2) is a
solution of the differential equation (1).
2
0.
Example: Consider p − xp + y =
This is Clairaut’s equation and its general solution (replacing simply p
by c) is
y cx + c 2 (1)
=
Now we find the envelope of the family (1)
f ( x, y, c ) ≡ y − cx + c 2 =0
(2)
∂f
f c ≡ =− x + 2c =0
∂c (3)
Substituting the value of c from (3) into (2), we have
2
x x
0
y − x  +   =
2 2
42 Differential Equations: Theory and Applications

x2 x2
y− + 0
=
i.e., 2 4
2
or 4y = x .
(4) is an equation of the envelope of the family (1).
We check whether (4) is a solution of the given differential equation.
dy
= 4 2=
x or 2 p x.
Differentiating (4) w.r .t . x , we get dx (5)
2
Substituting (4) and (5) into p − px + y, we have
x2 x2 x2
− + 0
=
4 2 4
Thus (4) is a solution of the given differential equation. This solution
does not involve any constant and it cannot be derived from the general
solution by giving particular values to c, such a solution is called a singular
solution (S.S).
f ( x, y , p ) = 0
Definition A solution of a differential equation is called
a singular solution (S.S) if
• It is not derived from the general solution by giving any particular
value to the arbitrary constants
• At each of its points, it is tangent to some member of the one
parameter family of curves represented by the general solution.
We have seen that the envelope, if any, of the one parameter family of
curves represented by the general solution of a differential equation is a
singular solution. Thus the c- discriminant equation may contain singular
solution, if any
The singular solution may also be obtained from the differential equation
directly without finding the general solution. Since at ultimate point of
intersection of neighboring curves the p ' s for the intersecting curves
become equal, and thus the locus of the points where p′s have equal roots
will include the envelope. If we eliminate p from
∂f
f (=
x, y, p ) 0=
and 0
∂p
 Basic Concepts of Differential Equations 43

The resulting equation is called the p-discriminant. The p-discriminant

f ( x, y , p ) = 0
represents the locus for each point of which has equal roots.
f ( x, y , p ) = 0
If envelope of the general solution of exists, it will be
contained in the p-discriminant. Thus, the p-discriminant equation may
contain (i) singular solution (ii) Solutions that are not singular and (iii) such
loci that are not solutions at all.
2
Example: Solve xp − 2 yp + 4 x =
0 (1)
yp xp 2 + 4 x
Solution: 2=
4x
2=
y xp +
p

Differentiating w.r.t.x, we get

dp 4 4 x dp
2 p= x + p+ − 2
dx p p dx

4  4 x  dp
p− = x− 2 
or p  p  dx
2
(
p 2 − 4 x p − 4 dp
=
)
p p2 dx

dp dx
=
or p x
or p = cx (2)
Eliminating p from (1) and (2), we obtain
( )
x c 2 x 2 − 2 ycx + 4 x =
0

2 2
or c x − 2 yc + 4 =0 (3)
From (3), we obtain the c - discriminant as
4 y 2 − 16 x 2 =
0 or y 2 = 4 x 2
The p-discriminant is from (1)
4 y 2 = 16 x 2 or y 2 = 4 x 2
 44 Differential Equations: Theory and Applications

2 2
Since the c-discriminant and p-discriminant are the same viz., y = 4 x
and it satisfies the given differential equation, it is the singular solution.
x 2 p 2 + yp ( 2 x + y ) + y 2 =
0
Example: Solve (1)
By making substitution= y u= , xy v and find the singular solutions.
v v
y= u , x= =
Solution: Let y u
dv
u. −v
dy dx du
= 1,=
Thus du du u2
dy dy du u2
=
p = . =
dx du dx u. dv − v
du

Substituting into (1), we have

2
 
v2  u 2  u2
 
u 2  u. dv − v 
+ 2 v + u 2
. (
dv )
u. − v
 du  du
2
 dv   dv 
2
(
v + 2v + u  u
 du
2


)
− v  +  u. − v  =
 du 
0
or
2
2 dv dv  dv  dv
v + 2uv − 2v 2 + u 3 . − u 2 v + u 2   − 2uv + v2 =
0
du du  du  du
2
dv  dv 
=v u + 
du  du 
which is a Clairaut’s equation. Its solution is
v cu + c 2
=

i.e., xy= cy + c 2 (2)

is the solution of (1).
From (2), the c-discriminant is
 Basic Concepts of Differential Equations 45

0 i.e., y ( y + 4 x ) =
y 2 + 4 xy = 0

Therefore, y = 0, y + 4 x = 0 (3)
Differentiating (3), we have p = −4. Substituting p = −4 into the left-
hand member of (1) and using (3), we obtain
16 x 2 − 4 x ( −4 )( −2 x ) + ( −4 x=
) 32 x 2 − 32=
2
x2 0

Thus y + 4 x =
0 is also a singular solution.

Example: Solve
(x 2
)
− 1 p 2 − 2 xyp − x 2 =0
(1)
Moreover, find the singular solution, if any.
Solution: Solving the equation for y, we have

2y =
(x 2
)
−1 p2 − x2
= xp −
p x
− .
xp x p

Differentiating w.r.t.x, we get

xp′ − p p − xp′ dp
2 p = xp′ + p − 2
− 2
, p′ =
x p where dx
 1 x  p 1
p p′  x − + 2  + 2 −
=
or  x p  x p
 1 1   1 1 
p 1 − 2 + 2 = xp′  1 − 2 + 2 
or  x p   x p 
dp dx
= = or p cx.
Therefore, p x
Substituting this value of p into (1), we get
(x 2
)
− 1 c 2 x 2 − 2 x 2 yc − x =0

or
(x 2
)
− 1 c 2 − 2 yc − 1 =0
(2)
as the general solution of (1).
From (2), the c-discriminant is
4 y 2 + 4 x2 −1 =
0 ( )
46 Differential Equations: Theory and Applications

i.e., x2 + y 2 =
1
which is also envelope of the family (2). We also note that p − discriminat
is
(
4x2 y 2 + 4x2 x2 −1 =
0 )
2 2
i.e., x + y = 1
Since the envelope and the p-discriminant are same, the singular solution
is
x2 + y 2 =
1

Example: Find the singular solution of

x p 2 + x 2 yp + a 3 =
3
0 (1)

Solution: The p − discriminant of (1) is

x 4 y 2 − 4= (
x 3 a 3 0 or x 3 y 2 x − =
4a 3 0) (2)
The parts of (2) that satisfy (1) are singular solutions of the given
equation. From (2), we have
x 0 or y 2 x − 4a=
= 3
0 (3)
We rewrite (1) as
2
dx  dx 
x + x y + a3   =
3 2
0
dy  dy  (4)
Clearly, x = 0 satisfies (4).
Thus x = 0 is a singular solution of (1).
2 3
Differentiating y x − 4a =0 w.r.t x, we have
2 xyp
= + y 2 0 or y ( 2 xp
= + y) 0

y
p=− .
i.e., 2x
Putting this value of p into the left-hand member of (1) and using (3),
we obtain
 Basic Concepts of Differential Equations 47

2
 y  2  y  3 xy 2
x−  + x y−  + a =
 2x   2x 
−
4
+ a3 =
− xy 2 − 4a 3 =
0 ( )
2 3
Thus xy − 4a =0 and its derivative satisfy (1) and so it is a solution

of (1). Hence =x 0 and xy 2 − =

4a 3 0 are singular solutions of (1).

1.2.5. The Ricatti Equation

Definition. We have already studied first order linear differential equation
y′ + P ( x ) y = R ( x )
(1)
q ( x) y2
If we add the term to the left-hand member of (1), we obtain a
nonlinear differential equation
y′ + P ( x ) y + Q ( x ) y 2 =
R ( x)
(2)
(2) is called the Ricatti equation.
Note: In many cases, the solution of (2) cannot be expressed in terms of
elementary functions. However, the Ricatti equation y′ + Py + Qy = R (1)
2

1
y= y1 + ,
can be reduced to a linear equation by the substitution u where y1
is a particular solution of (1) and u is an unknown nonzero function of x .
1
y= y1 +
Proof: Let u be as given. Differentiating w.r.t.x, we have
dy dy1 1 du
=y′ = −
dx dx u 2 dx

Substituting for y and y′ into (1), we get

dy1 1 du  1  2y 1 
− 2 + P  y1 +  + Q  y12 + 1 + 2  =
R
dx u dx  u  u u 
dy1 1  du 
+ Py1 + Qy12 − R − 2  − Pu − 2Qy1u − Q  =
0
or dx u  dx  (2)

Since
y1 is a solution of (1), we have
48 Differential Equations: Theory and Applications

dy1
+ Py1 + Qy12 − R =
0
dx
du
− ( P + 2Qy1 ) u =
Q
and so (2) reduces to dx which is a linear equation. Thus if
a particular solution of (1) is known then its general solution can be found.
dy
−1, y ( 0 ) =
− y2 = 3,
Example: Solve dx given that y1 = 1 is a particular
solution of the given equation.
0, Q =
Solution: Here P = −1, R =
−1.
1
y = 1+ ,
Writing u the given equation reduces to
du
− ( 0 + 2 ( −1)(1) ) =−1
dx
du
+ 2u =
−1,
i.e., dx (1)
which is a linear equation.
∫ 2 dx
I.F. of (1) is e = e2 x .
Multiplying (1) by the I.F., we get
du 2 x
e + 2u e 2 x =
−e 2 x
dx
d
( )
ue 2 x = −e 2 x
or dx
Integrating, we obtain
e2 x
u e 2 x =− ∫ e 2 x dx + c =− +c
2

1 c
u=
− + 2x
or 2 e (2)
1
y = 1+ ,
From u we get by the initial condition,
1 1
y ( 0) =
3=
1+ or u ( 0 ) =
u (0) 2
 Basic Concepts of Differential Equations 49

Hence from (2), we have

1 1
u (0) = = 1
− + c or c =
2 2
1 1 2 – e2 x
u=− + 2x = 2x
Thus 2 e 2e
1 2e 2 x 2 + e2 x
y =1 + =1 + =
Required solution is u 2 − e2 x 2 − e2 x .
dy y
− − x3 y 2 =− x5 ,
Example: Solve dx x (1)
by finding a particular solution.
1
P=− ,Q = − x3 and R =− x5 .
Solution: It is a Ricatti equation with x
1
y1 = x. y= x +
An obvious solution of (1) is Substituting u into (1),
we have
du  1 
(
−  − + 2 − x 3 x  u =− x 3
dx  x 
)
du  1 
+  + 2x4  u = − x3 ,
or dx  x  (2)
which is a linear equation.
1  2 5 2 5
∫  + 2 x 4  dx x x
ln x
=
Its I .F . e = e=
.e xe x  5 5

Multiplying (2) by the I.F. and integrating, we have

2 2 5
x5 x 1 2 x5
ux e 5 = − ∫ x 4 .e 5 = dx + c′ − e 5 + c′
2
2
1 x5
c′ − e 5
u= 2
2
x5
or x e5
Hence the general solution of (1) is
50 Differential Equations: Theory and Applications

 1 5 x5 
2
2 5
x ′
x c + e 
xe 5
 2 
y=
x+ 2
= 2 5
1 x5 1 x
c′ − e 5 c′ − e 5
2 2
2
1 x5
c′ + e 5
y 2
= 2
x 1 5 x5
c′ − e
or 2
2
2 5 x5
y − x e5 x
= = ce 5

or y + x 2c′ is the required solution.

1.3. DIFFERENTIAL EQUATIONS OF HIGHER

ORDER

1.3.1. Introduction
In previous sections, we studied methods of solving special types of linear
and nonlinear differential equations of the first order. In this chapter, we
shall consider systematic methods for the solution of certain classes of
differential equations of order more than one.

1.3.2. Linear Differential Equations

Definition. A linear differential equation of order n in the dependent
variable y and the independent variable x is of the form
dny d n −1 y dy
a0 ( x ) n
+ a1 ( x ) n −1
+…+ an −1 + an ( x ) y = F ( x ) ,
dx dx dx (1)
a0 ( x ) , a1 ( x ) ,…, an −1 ( x ) , an ( x ) and F ( x )
Where are functions of
a ( x)
the independent variable x only and 0 is not identically zero. Using
primes, (1) is also written as
F ( x)
a0 y n + a1 y n −1 +…+ an −1 y′ + an y =
(2)

where
a0 , a1 , …, an −1 , an are real constants.
 Basic Concepts of Differential Equations 51

Equation (1) is with variable coefficients while (2) is with constant

coefficients. In order to solve (2), we shall first consider the equation
a0 y n + a1 y n −1 +… + an −1 y′ + an y =
0
(3)
n
The coefficient of y may be made 1 by dividing throughout by 0 The
a.
differential equation (3) is called homogeneous linear differential equation
of order n. The use of the word homogeneous here is quite different from
F ( x)
the one already mentioned in previous chapter. If is not identically
zero then (2) is called non - homogeneous and (3) is called the associated
homogeneous equation of (2).
y1 ( x ) , y2 ( x ) , …, ym ( x )
Definition. If are m functions of an
c , c , …, cm
independent variable x and 1 2 are constants, then the expression
c1 y1 ( x ) + c2 y2 ( x ) +…+ cm ym ( x )
is called a linear combination of
y1 ( x ) , y2 ( x ) , …, ym ( x ) . y , y , … , ym
We usually write 1 2 instead of
y1 ( x ) , y2 ( x ) ,…, ym ( x ) y , y , … , ym
when it is clear from the context that 1 2
are functions of x.
y1 , y2 , …, ym
As in vector spaces, the m nonzero functions are called
c1 , c2 , …, cm
linearly dependent if and only if, there exist constants at least
one of which is nonzero, such that
c1 y1 + c2 y2 +…+ cm ym = 0.
y1 , y2 , …, ym
The functions are called linear independent if and only if,
c1 y1 + c2 y2 +…+ cm ym = 0
they are not linearly dependent, i.e., if and only if
c1 = c2 = … = cm = 0.
Implies
Now before investigating a solution of (2), we state the following facts:
i. Every homogeneous linear nth − order differential equation
a0 y n + a1 y n −1 +…+ an −1 y′ + an y =
0
has n linearly independent solutions
y1 , y2 , …, yn
y1 , y2 ,…, y _ m
ii. If are n linealy independent solutions of (3), then
52 Differential Equations: Theory and Applications

any linear combination

=yc c1 y1 + c2 y2 +…+ cn yn of y1 , y2 , …, yn
c , c ,…, cm being arbitrary constants.
is the general solution of (3), 1 2
yp y
iii. Let be any particular solution of (2). i.e., p does not contain
y + yp
any constant, then c is the general solution of (2).
(ii) and (iii) can be easily checked by actual substitutions into (3) and (2).
Thus, to find the general solution of (2), we have to find a linearly independent
y y
set of n solutions of (3). So as to determine c and a particular solution p
=y yc + y p
of (2) and then obtain (4)
y
as the general solution of (2). In the general solution (4), c is called the
y
Complementary Function (C.F) and p is called the Particular Integral
(P.I) of (2). These statements are also true for the equation (1) with variable
coefficients.

1.3.3. Homogeneous Linear Equation

dny d n −1 y dy
a0 n
+ a1 n −1
+…+ an −1 + an y, = 0
Consider the equation dx dx dx (1)

where 0 1
a , a ,…, a
n −1 are real constants. To find a solution of (1) we shall try

a judicious guess. The differential equation (1) requires a function y with

the property that if y and its successive derivatives are each multiplied by
a , j= n, n − 1,…,1, 0,
constants j the resulting products are then added; the
sum should equal zero. This can only happen if a function is such that its
various derivatives are constant multiples of itself. The exponential function
y = e mx , m being a constant, has such properties. Here we have
dy d2y dny
= me mx , 2 = m 2 e mx , …, n = m n e mx .
dx dx dx

Substituting into (1) we have

a0 m n e mx + a1m m −1e mx +…+ an −1m e mx + an e mx = 0

or
e mx (a0 m n + a1m n −1e mx +…+ an −1m e mx + an e mx = 0
 Basic Concepts of Differential Equations 53

Since e ≠ 0, we have
mx

a0 m n + a1m m −1 +…+ an −1m + an = 0 (2)

Thus, y = e is a solution of (1) if and only if m is a solution of
mx

(2). Equation (2) is called the characteristics (or auxiliary) equation of the
given differential equation (1). Observe that (2) can be obtained from (1)
m k ( k= n, n − 1, …, 3, 2,1) .
by merely replacing the kth derivative in (1) by
Three cases arise according as the roots of (2) are
i. Real and distinct
ii. Real and repeated
iii. Complex

Case I. Distinct Real Roots

m , m , …, m
Let 1 2 n be n distinct real roots of (2). Then e , e ,…, e
m1 x m2 x mn x

are n distinct solutions of (1). These n solutions are linearly independent.

Hence the general solution of (2) is
= y c1e m1x + c2 e m2 x +…+ cn e mn x , where
c1 , c2 , …, cn are arbitrary constants.

Case II. Repeated Real Roots

d
D≡ ,
In equation (1), writing dx we have

(a D
0
n
)
+ a1 D n −1 +…+ an −1 D + an y = 0  f ( D )  y = 0,
or 

where
( D ) a0 D n + a1D n−1 +…+ an−1D + an
f=

Note that if we write D for m in the characterstic equation (2) then it

f ( D ) = 0. m1 , m2 , …, mn f ( D ) = 0,
is the same as If are the roots of
then (1) may be written as
( D − m1 )( D − m2 )…( D − mn ) y =
0.

Let the root be repeated twice say

m2 = m1.
The part of the general solution of (1) corresponding to the twice repeated
( D − m1 ) y =
2
m 0
root 1 of (2) is solution of
54 Differential Equations: Theory and Applications

i.e., of
( D − m1 )( D − m1 ) y =
0
(3)
Let
( D − m1 ) y = V.

Then (3) becomes

( D − m1 )V =
0

dV
− m1V = 0
or dx (4)
Separating the variables, we have
dV
= m1dx.
V

Therefore, in
V m1 x + k , where k is a constant.
=

V = c2 e m1x , where c2 is a constant.

Replacing V in (4), we obtain

( D − m1 ) y =
c2 e m x 1

dy
− m1 y =
c2 e m1x
dx (5)
= exp ∫ (−m1 dx
Which is a linear equation of order one. Its I.F.
= (
) e − m1x )
Multiplying (5) by e , we get
−m1 x

d
dx
(
ye − m1x = c2 )
y e − m=
1x
c2 x + c1

=
y ( c1 + c2 x ) em x
1

is the part of the general solution corresponding to the

twice repeated
m1. The general solution of (1) is

=
y ( c1 + c2 x ) em x + c3e m x +…+ cn em x
1 3 n
 Basic Concepts of Differential Equations 55

In the same manner, if the characteristic equation (2) has the triple
m1 ,
repeated root the corresponding part of the general solution of (1) is
the solution of
( D − m1 )
3
0
y=

Proceeding as before, we can easily find

(
y = c1 + c2 x + c3 x 2 e m1x )
As the part of the general solution corresponding to this triple repeated
m. m
root 1 If the characteristics equation (2) has the real root 1 occurring
k times then the part of the general solution of (1) corresponding to the
k − fold repeated root m1 is

=
y (c + c x + c x
1 2 3
2
)
+…+ ck x k −1 e m1x

Case III. Complex Roots

Suppose the characteristic equation has the complex number a + ib as a
non- repeated root. Since coefficients of (1) are real, the conjugate complex
number a − ib is also a non- repeated root. The corresponding part of the
general solution is
=y k1e( a + ib ) x + k2 e( a −ib ) x , where k1 and k2 are arbitrary constants.

=y e ax  k1eibx + k2 e − ibx 

= e ax  k1 ( cos bx + i sin bx ) + k2 ( cos bx − i sin bx ) 

= e ax ( k1 + k2 ) cos bx + i ( k1 − k2 ) sin bx 

= e ax [ c1 sin bx + c2 cos bx ]

i ( k1 − k2 ) , c2 =
c1 = k1 + k2
where are two arbitrary constants.
If a + ib and a − ib are conjugate complex roots, each repeated k times,
the the corresponding part of the general solution of (1) may be written as
 56 Differential Equations: Theory and Applications

( ) ( )
y  c1 + c2 x + c3 x 2 +…+ ck x k −1 sin bx + ck +1 + ck + 2 x +…+ c2 k x k −1 cos bx 
=

In the examples given below we use these concepts to solve homogeneous

linear differential equations.

Example: Solve
(D 2
)
+ 4D + 3 y =
0
2
Solution: The characteristic equation is D + 4 D + 3 =0 with roots
D=−1, −3.
= y c1e − x + c2 e −3 x .
Hence the general solution of the given equation is

Example: Solve
(D 3
)
− 5D 2 + 7 D − 3 y =
0.
3 2
Solution. The characteristic equation is D − 5 D + 7 D − 3 = 0 (1)
By inspection, D = 1 is a solution of this equation. The other two roots
( D − 1) ( D 2 − 4 D + 3) =
0
can be found from the quadratic factor of
Hence all roots of (1) are D = 1,1,3.

The general solution is

( c1 c2 x ) e x + c3e3 x .
y =+

Example. Solve
(D 3
)
− D2 + D −1 y =
0.
3 2
Solution. The characteristic equation is D − D + D − 1 =0.
By inspection, D = 1 is a root of this equation.

Now,
D 3 − D 2 + D − 1= ( D − 1) ( D 2 + 1)= 0.

Hence the other two roots are D =±i =a + ib with a =0, b =1.
The general solution is
c1e x + e0 x ( c2 sin x + c3 cos x )
y=

c1e x + c2 sin x + c3 cos x.

y=

=
Example: Solve
(
D 2 + D − 12 y 0, where
= )
y ( 2 ) = 2, y′ ( 2 ) 0.
 Basic Concepts of Differential Equations 57

2
Solution. The characteristic equation is D + D − 12 =
0 with roots
D= 3, −4.

Hence the general solution of the given equation is=y c1e3 x + c2 e −4 x (1)

We now determine the co-efficient (constants)

c1 , c2 in (1) from the

initial conditions as follows. Since the initial conditions are given at x = 2,

we rewrite the general solution in the form
=y k1e3( x − 2) + k2 e −4( x − 2)
(2)
= k c=
where 1 1
e , k2 c2 e −8 .
6

Applying initial condition, ( i.e. replacing

c1 , c2 by the new constants
k1 e , k2 e −8 respectively in (1), we get
6

y ( 2 )= 2= k1 + k2
(3)
Differentiating (2) w.r.t.x, we have
dy
y′
= = 3k1e3( x − 2) − 4k2 e −4( x − 2)
dx
Therefore,
y′ ( 2 ) 0 = 3k1 − 4k2
=
(4)
8 6
= k1 = , k2 .
Solving (3) and (4), we obtain 7 7
Hence the solution (2) of the differential equation satisfying the given
conditions is
8 3( x − 2) 6 −4( x − 2)
=y e + e .
7 7

Example. Solve
(D 2
+ 4 D += )
5 0, y (=
0 ) 1, y′ ( 0 ) = 0.
2
Solution. The characteristic equation is D + 4 D + 5 =0 which has
roots D =−2 ±i.
= y e −2 x ( c1 sin x + c2 cos x ) .
The general solution is (1)
 58 Differential Equations: Theory and Applications

Applying the given conditions, we have from (1)

y ( 0 )= 1= c2

Differentiating (1) w.r.t.x, we get

−2 e −2 x ( c1 sin x + c2 cos x ) + e −2 x ( c1 cos x − c2 sin x )
y′ =

Therefore,
y′ ( 0 ) = 0 = −2c2 + c1 , c1 = 2.
giving
c1 and c2
Substituting the values of into (1), the required solution is
=y e −2 x
( 2sin x + cos x ) .

Example. Solve
(D 3
)
− 3D 2 + 4 y= 0, y ( 0=
) 1, y′ ( 0=) −8, y′′ ( 0 ) =
−4.
3 2
Solution. The characteristic equation is D − 3D + 4 =0.
( D + 1) ( D 2 − 4 D + 4 ) =
0.

Therefore, D = −1, 2, 2
y =c1e − x + e 2 x ( c2 + c3 x )
The general solution is (1)
−c1e − x + 2e 2 x ( c2 + c3 x ) + c3e 2 x
y′ =
Now, from (1)
c1e − x + 4e 2 x ( c2 + c3 x ) + 4c3e 2 x
y′′ =

Applying initial conditions, we get from the above three equations

y ( 0 ) = 1 = c1 + c2
(2)
y′ ( 0 ) =−8 =−c1 + 2c2 + c3
(3)
y′′ ( 0 ) =−4 =c1 + 4c2 + 4c3
(4)
Multiplying (3) by -4 and adding to (4), we have
28
= 5c1 − 4c2
(5)
Multiplying (2) by 4 and adding to (5), we obtain
32
9c1 32
= = or c1 .
9
 Basic Concepts of Differential Equations 59

32 23 6
1−
c2 = − and c3 =.
=
Therefore, 9 9 9
The required solution is
32 − x 2 x  23 6  1
=
y
9
e + e  − + x= 
 9 9  9
(
32 e − x − 23 e 2 x + 6 x e 2 x . )

Differential Operators
Let T :V →V , where V is a vector space over a field F , be a linear
transformation (or an operator) from V toV . Linear transformation S ,T , U
defined on V have the following properties.

i. S + T = T + S where ( S + T ) v = S ( v ) + T ( v ) , v ∈ V ,
( S + T ) +U = S + ( T +U )
S ( T + U ) =ST + SU

( ST )U = S (TU )
ii. ST ≠ TS in general.

Here ST : V → V is a linear transformation defined by:

( ST )( v ) = S (Tv )
for all v ∈ V .
Let X be the vector space of all real (or complex) valued real (or complex)
functions possessing a kth order derivative for every= k 1, 2,3, … For each
g ∈ X , y = g ( x ) ∈ R ( or ∈ C ) .D which associates with each y = g ( x )
k

dk y
D k y as k .
, its kth derivative, is a linear operator on X . We write d x Since
the sum and product of the linear operator and scalar multiple of a linear
operator are also linear operators, the linear combination
( D ) a0 D n + a1D n−1 +…+ an−1D + an I
f=
(1)
n −1
Of D , D , …, D, I is also a linear operator. Here, I is the identity
n

= I ( y ) y=
for all y g ( x ) a , a , …, an are
linear operator defined by and 0 1
 60 Differential Equations: Theory and Applications

y = g ( x) f ( D)
scalars. The image of a under is written as
( D ) y a0 D y + a1D
f= n n −1
y +…+ an −1 Dy + an y

dny d n −1 y dy
= a0 n
+ a1 n −1
+…+ an −1 + an y.
dx dx dx
f ( D ) y = 0.
The equation is then a linear differential equation of order
n.
D, D 2 , …, D n and f ( D ) ,
The operator given by (1), are called
d d2
D for , D 2 for 2
differential operators. We have written dx dx and so on
n 2
d dy 2 d y dny
D n for n . = Dy = ,D y D n
y = .
dx Thus dx dx 2 and dx n
= A f (= D ) a0 D n + a1 D n −1 +…+ an −1 D + an I
The expression is called
a differential operator of order n. This may be thought of as an operator
which, when applied to any function y , result in
dny d n − 1y a dy
= ( D ) y a0
Ay f = n
+ a1 n −1
+…+ n −1 + an Iy.
dx dx dx

Nonhomogeneous linear Equations

In this section we discuss the solution of nonhomogeneous linear equation
F ( x)
a0 y n + a1 y n −1 +…+ an −1 y′ + an y =
.

Solution of the equation

f ( D) y = F ( x)
(1)

where
( D ) a0 D + a1D
f= n n −1
+…+ an −1 D + an .

We have already discussed that the general solution of (1) consist of two
parts, namely
i. Complementary Function (C.F)
ii. Particular Solution (P.I)
 Basic Concepts of Differential Equations 61

f ( D) y = 0
The C.F is the solution of the homogeneous equation and
we have described different cases of its solution. Now to find the P . I of (1),
we can write
1
y= F ( x)
f ( D)

1
y= F ( x).
f ( D)
And try to evaluate
f ( D) f ( m ) = 0.
A real number m is said to be a zero of if
f ( D) m1 , m2 ,…, mn .
Suppose has n distinct zeros
1 1
F ( x) = F ( x)
f ( D) ( D − m1 )( D − m2 ) …( D − mn )
Then .
1
F ( x ) = y* .
Suppose D − mn

dy
( D − mn ) y = F ( x)
* − mn y* = F ( x)
Then or dx which is a linear

equation of the first order. Its

= exp ∫ ( −mn ) dx
I .F ( )
= exp ( − mn x ) .

Hence
d
dx
( )
y e − mn x = F ( x ) e − mn x

y* e mn x ∫ F ( x ) e − mn x dx,
=

Omitting the constant of integration, since we are looking for a particular

solution, Therefore
1 1
F ( x) e mn x ∫ e − mn x F ( x ) dx.
f ( D) ( D − m1 )…( D − mn−1 )
1
e mn x ∫ e − mn x F ( x ) dx
Next, we evaluate D – m n −1

As before and continue the process. This is the required P.I of the
equation (1).
 62 Differential Equations: Theory and Applications

Alternatively, we may also write

1 1
F ( x) = F ( x)
f ( D) ( D − m1 )( D − m2 )…( D − mn )
 A1 A2 An 
=  + +…+  F ( x)
 D − m1 D − m2 D − mn 
1
f ( D)
after resolving into partial fractions
= A1 e m1x ∫ e − m1x F ( x ) dx + A2 e m2 x ∫ e − m2 x F ( x ) dx +… An e mn x ∫ e − mn x F ( x ) dx
,
which is the required P.I of the equation (1).

Example: Solve
(D 3
)
−D y =
ex
.
3
0 with roots 0 ,1 , −1
Solution. The characteristic equation is D − D =
C.F =yc =c1 + c2 e x + c3e − x .

ex
yp =
D ( D +1 )( D –1 )
P.I =
1 x
e = e x ∫ e − x .e x dx =x ex ,
Now D − 1 by the method already described
Thus
1 1
ex = xe x
D ( D +1 )( D –1 ) D ( D +1 )

Again,
1 xe x e x
D +1
(
xe =
x
)
e ∫ e .xe dx =−
−x x x

2 4

Hence
1 1  x e x e x  xe x ex xe x 3 x
D ( D +1 )
( )
xe x = 
D 2
−  =∫
4  2
dx − ∫ dx =
4 2 4
− e

Therefore, the general solution of the equation is

 Basic Concepts of Differential Equations 63

xe x 3 x
y =yc + y p =c1 + c2 e x + c3 e − x + − e .
2 4

 3 xe x
=c1 +  c2 −  e x + c3e − x +
 4 2
xe x
=c1 + c2' e x + c3e − x +
2

1.3.4. Alternative Method

f ( D) f ( D),
−1

Consider as the inverse of the operator so that

1
yp = ex
D ( D +1 )( D –1 )

 1 1 1 1 1  x
y p = − + . + . e .
 D 2 1+ D 2 D −1 
1 1
y p =− ∫ e x dx + e − x ∫ e x .e x dx + e x ∫ e − x .e x dx
2 2

1 1 xe x
y p =− e x + e x + xe x = ,
4 2 2
x x
since e already occurs in the C.F and so the terms involving e are omitted.

1.3.5. Working Rules for finding P.I.

1
The method of evaluating F ( x)
involves successive integrations and is
f ( D)
F ( x)
quite unwieldy. We list below short methods for finding the P.I. when
is a function of a particular type.
1
e ax ,
To evaluate f (D)
where f (= D ) a0 D n + a1 D n −1 +…+ an −1 D + an .

We know that D k eax = a k eax , where k is a positive integer.

Hence
 64 Differential Equations: Theory and Applications

f ( D=
) eax (a D
0
n
+ a1 D n −1 +…+ an −1 D + an e ax)
= a0 .D n e ax + a1.D n −1e ax +…+ an −1.D e ax + an .e ax

= a0 .a n e ax + a1.a n −1e ax +…+ an −1. a e ax + an .e ax .

= f ( a ) e ax .
(1)
1
f ( D)
Applying to both sides of (1), we obtain
f ( D) 1
e ax = f ( a ) e ax
f ( D) f ( D)

1
e ax = f ( a ) . e ax
f ( D)

1 e ax
= e ax
, if f ( a ) ≠ 0
f (D) f (a)
i.e.,
f ( a ) = 0, f ( D ) = 0.
if then a is a root of
f ( D ) = 0. ( D – a ) is a factor of
k

Let a be a k-fold root of so that

f ( D ).
Then,
f (=
D) ( D –a ) ∅ ( D ) , where ∅ ( a ) ≠ 0.
k

( D –a ) on e ax p ( x ) , for a polynomial
k
We first check the effect of
p ( x)
in x.
We have
( D – a ) ( eax p ( x ) =) ( )
D e ax p ( x ) − a e ax p ( x )= e ax Dp ( x )

D – a ) ( e Ax p ( x ) ) (=
(= D – a ) ( e ax D p ( x ) ) e ax D 2 p ( x ) .
2
 Basic Concepts of Differential Equations 65

Continuing in this way, we get

( D –a )
k
(e Ax
)
p ( x ) = e ax D k p ( x ) .
(3)
p ( x) = x k
Setting in (3), we are led to
( D –a )
k
(=
xe )k ax
e= D x k !e ax
ax k k

∅ ( D ) ( D – a ) n x k e ax =
k
(
∅ ( D ) k !e ax )
k !∅ ( D ) e ax =
= k !∅ ( a ) e ax ,
by (1)
1
, we obtain
∅( D ) ( D –a )
k

Operating on both sides by

1
x k e ax k !∅ ( a ) e ax
∅( D ) ( D −a )
k

1 x k e ax
e ax =
∅ ( D ) ( D –a )
k
k! ∅( a )
(4)

1.3.6. Principle of Superposition

f (=
D) y (a D 0
n
+ a1 D n −1 +…+ an −1 D + = )
an y F ( x )
(1)

Be a linear differential equation of order n. if

F= ( x ) F1 ( x ) + F2 ( x ) ,
then particular integral of (1) is the sum of particular integrals of
F ( D ) y = F1 ( x )
(2)
F ( D) y F ( x)
and (3)
y1 and y2
Proof: Let be particular integrals of (2) and (3) respectively.
Then
f ( D ) y1 = F1 ( x )

f ( D ) y2 = F2 ( x )
and
 66 Differential Equations: Theory and Applications

Suppose
y= y1 + y2 . Setting this value of y into (1), we obtain

f ( D ) y f ( D )( y1 + y2 )
=

= f ( D ) y1 + f ( D ) y2

= F1 ( x ) + F2 ( x ) = F ( x )

Showing that y is a solution of (1).

Example: Solve
(D 3
) 1 e− x + e2 x .
+ 1 y =+

Solution. The characteristic equation is

1± i 3
3
−1 ,
D + 1 =0 with roots 2
Therefore, C.F is
x
 3 3 
c1e − x + e 2  c2 sin
yc = x + c3 cos x .
 2 2 

The P/I is given by

1
=
yp 3
D +1
(
1+ e− x + e2 x )
1 1 1
= e0 x + 3 e 2 x + e− x
3
D +1 D +1 ( D +1 ) D – D +1
2
( )
e2 x 1 1 1 ax
1+
= + e− x , e ax = e ,
9 3 ( D +1 ) f ( D) f (a)
using
1
2
e − x , with D = −1 when f ( −1) ≠ 0
for D – D +1
e2 x 1 − x
= 1+ + xe ,
9 3
 Basic Concepts of Differential Equations 67

Hence the general solution of the equation is

=
y yc + y p

x
 3 3  e2 x 1 − x
= c1e − x + e 2  c2 sin x + c3 cos x  + 1 + + x. e .
 2 2  9 3

F ( x ) = sin ax or cos ax.

To Find the P.I when
1 1
sin ax and cos ax.
f ( D) f ( D)
Here we have to evaluate
From Euler’s theorem, we have
=
e cos ax + i sin ax
iax

1 1
sin ax and cos ax
f ( D) f ( D)
Thus are respectively imaginary and real
1
eiax .
f ( D)
parts of
If f ( D ) contains only even powers of ( )
D, say f ( D ) = f D 2 ,
it is easy to see
1  sin ax 1  sin ax
=  2 
, provided − a 2 is not a zero of f D 2 . ( )
that
( )
f D 2 cos ax ( )
f −a cos ax

Example: Solve
(D 2
–5 D + 6 y = )
sin 3 x

Solution: The roots of the characteristic equation

2
D − 5D + 6 =0 are 2 and 3.
C.F
= y=
c c1e 2 x + c2 e3 x

1
P.=
I y= 2
sin 3 x
D − 5D + 6
p

1
e3ix
which is imaginary part of ( D – 2 )( D – 3 )
Now
68 Differential Equations: Theory and Applications

1 e3ix
e3ix = ,
( D – 2 )( D – 3 ) ( 3i – 2 )( 3i – 3 )
−1 + 5i 3ix
= e
78

 1 5 
= − + i  ( cos 3 x + i sin 3 x )
 78 78 

1 5
− sin3 x + cos 3 x =
yp
Its imaginary part is 78 78
Hence the required general solution is
1 5
y =c1e 2 x + c2 e3 x − sin 3 x + cos 3 x
78 78
1 1
f ( D) f ( D –a )
e ax F ( x ) = e ax
f ( D)
F ( x).
Theorem. If a is not zero of , then
This replacing of D by D + a is known as exponential shift.
Proof.
We have already shown in previous section that
( D –a )
k
( )
. e ax p ( x ) = e ax D k p ( x )

f ( D)
Using linearity of differential operators, we conclude that, when
is a polynomial (with constant coefficients), then
( )
f ( D – a ) e ax p ( x ) = e ax f ( D ) p ( x )
(1)
f ( D) p ( x) = F ( x)
Suppose , then
1
p ( x) = F ( x)
f ( D)
(2)
From (1) and (2), we have
1
f ( D – a ) e ax F ( x ) = e ax F ( x )
f ( D)
 Basic Concepts of Differential Equations 69

1
,
f ( D –a )
Operating on both sides by we get
1 1
e ax F ( x) = e ax F ( x )
f ( D) f ( D –a )
1 1
e ax F ( x ) = e ax F ( x),
f ( D –a ) f (D)
or
f ( D ).
provided that a is not a zero of

Example: Solve
(D 3
)
+ D2 − 4D − 4 y =
e 2 x cos 3 x.
3 2
Solution. The characteristics equation is D + D − 4 D − 4 =0.
By inspection D = 2 is a root of this equation. The other roots are
D=−2, −1
C.F . = c1e − x + c2 e −2 x + c3e 2 x
yc =

1
.I y=
P= e 2 x cos 3 x
p
( D – 2 )( D + 2 )( D +1 )
1
= e2 x cos 3 x, by the exponential shift
D( D+4 ) ( D+3)

1
= e2 x cos 3 x
( D 3 + 7 D 2 +12 D )
1
= e2 x cos 3 x, putting D 2 = −32
− 9 D – 63 +12 D

1
= e2 x cos 3 x
3 ( D – 21 )

= e2 x
( D + 21) cos 3 x
(
3 D 2 – 441 )
 70 Differential Equations: Theory and Applications

−1
= e2 x . [ −3 sin 3x + 21cos 3x ]
3× 450

e2 x
= ( sin3x – 7 cos 3x )
450
1
cos 3 x,
Alternative Method. Here, for D ( D + 4 )( D + 3 ) we proceed as
follow:
cos 3 x 1
= Re e3ix ,
( D 3 + 7 D 2 +12 D) (
D 3 + 7 D 2 +12 D )
1
= Re
( − 27i − 63 + 36i )
e 3 ix
= Re
9( i − 7)

= Re
( i + 7 ) e3ix
−450

= Re
( i + 7 ) ( cos 3x + i sin 3x)
−450

7 1
=
− cos 3 x + sin 3 x.
450 450

Therefore,
e2 x
=yp ( sin 3x − 7 cos 3x )
450 as before.
y= yc + y p
The general solution is
7 1
= c1e − x + c2 e −2 x + c3e 2 x − cos3 x + sin 3 x.
450 450

e2 x
−x
= c1e + c2 e −2 x 2x
+ c3e + ( sin 3 x – 7 cos 3 x ) .
450
 Basic Concepts of Differential Equations 71

1.3.7. Series Expansion of Polynomials or Expressions

involving Operators.
F ( x)
This method is useful when is a polynomial in x .
f ( D) y = F ( x) F ( x)
Let be such that is a polynomial in x . To
evaluate the particular integral
1
yp = F ( x)
f ( D)
1
f ( D)
It is often useful to expand in a series in D by the Binomial
Theorem for negative exponent so that
y p= ( 1+ a D + a D
1 2
2
+ a3 D 3 +… F ( x ) )
The derivatives on the right will vanish after certain stage since
D n x r 0 if n > r.
=
The following binomial expansions will be useful in this connection:
1
= 1 + x + x 2 + x 3 +….
1– x

1
= 1 − x + x 2 − x3 +….
1+ x

Example: Solve
(D 3
)
− 2 D + 1 y = 2 x3 − 3x 2 + 4 x + 5
3
Solution. The characteristic equation is D − 2 D + 1 =0
( D –1 ) ( D 2 + D –1 ) =
0

−1 + 5 −1 − 5
D =1 , ,
2 2
−1 + 5 –1 − 5
x x
yc =
c1e x + c2 e 2
+ c3 e 2
 72 Differential Equations: Theory and Applications

1
= yp
1− 2 D + D 3
2 x3 − 3 x 2 + 4 x +5 ( )
Next,
1
( )
−1
3
=1 − 2 D − D 3 
Now 1− 2 D + D
( ) ( ) + ( 2D − D )
2 3
1 + 2 D − D3 + 2 D − D3
= 3
+…

1 2 D − D3 + 4 D 2 + 8D3 ,
=+
4
Neglecting D and higher powers of D in view of the degree 3 of the
p ( x)
polynomial in x therefore
(
y p = 1+ 2 D + 4 D 2 + 7 D3 )( 2 x 3
− 3x 2 + 4 x + 5 )
( )
= 2 x3 − 3 x 2 + 4 x + 5 + 2 6 x 2 − 6 x + 4 + 4 ( 12 x − 6 ) + 7 (12 )

= 2 x3 + 9 x 2 + 40 x + 73.

=
y yc + y p
The general solution is
−1 + 5 −1 − 5
x x
= c1e x + c2 e 2
+ c3e 2
+ 2 x3 + 9 x 2 + 40 x +73.

1.3.8. The Method of Undetermined Coefficients

( U. C. Method)
We have studied some special cases in which particular integral can
be evaluated by the inverse operator. Now we consider the method of
undetermined coefficients which can prove simpler in finding the particular
f ( D) y = F ( x) F ( x)
integral of the equation when is
ax
a. An exponential function: e

b. A polynomial:
(
b0 x n + b1 x n −1 +…+ bn )
c. Sinusoidal function : sinax or cos bx
F ( x)
d. The more general case in which is sum of a product of terms
 sin ax 
= F ( x ) e ax b0 x n + b1 x n −1 + …+ bn  (  )
of the above types, such as  cos bx 
 Basic Concepts of Differential Equations 73

yp
The P.I, i.e., will be constructed according to the following table:

F ( x ) is of the form Take y p as

1 .a A. x k
2. ax n ( nis a + veinteger ) xk ( A0 x n + A1 x n −1 +…+ An )
3. a x n e rx ( nis a + veinteger ) x k ( A0 x n + A1 x n −1 +…+ An −1 x + An ) erx

4.Cx n cos ax x k ( A cos ax + B sin ax )

5.Cx n sin ax
6.C.x n . e rx cos ax ( )
x k [ A0 x n + A1 x n −1 +…+ An −1 x + An e rx cos ax

7.C x n e rx sin ax + (B x 0
n
)
+ B1 x n −1 +…+ Bn −1 x + Bn e rx sin ax

In x , k is the smallest nonnegative integer which will ensure that no

k

y
terms in p is already in the C.F.
F ( x) y
If is sum of several terms, write p for each term individually
y
and then add up all of them. The p and its derivatives will be substituted
f ( D) y = F ( x)
into the equation and coefficients of like terms on
the left hand and right-hand sides will be equated to determine the
U .C. A0 , A1 , …, An , B0 , B1 , …, Bn .
The method is illustrated by examples as
follows.
′′ ′ 2 x 3 − 9 x 2 + 6 x
Example: Solve y − 3 y + 2 y = (1)
yc c1e x + c2 e 2 x
=
Solution: C.F. is easily found as
For a particular solution, we assume
(
y p x k Ax 3 + Bx 2 + Cx + D .
= )
yp ,
Since no term of the C.F. is present in we take k = 0 and so
y p = Ax 3 + Bx 2 + Cx + D
74 Differential Equations: Theory and Applications

y 'p = 3 Ax 2 + 2 Bx + C

y"p 6 Ax + 2 B
=

y p , y 'p and y"p

Substituting for into (1), we have
( 6 Ax + 2 B ) − 3 ( 3 Ax 2 + 2 Bx + C ) + 2 ( Ax3 + Bx 2 + Cx + D )
= 2 x3 − 9 x 2 + 6 x
2 Ax 3 + x 2 ( 2 B − 9 A ) + x ( 6 A − 6 B + 2C ) + 2 B − 3C + 2 D
or
= 2 x3 − 9 x 2 + 6 x.

Equating coefficients of like terms, we obtain

x3 : 2 A 2=
Coeff. Of= or A 1.
2
Coeff. Of x : 2 B − 9 A = −9 or 2 B =
9 A − 9 giving B =
0

Coeff. Of x : 6 A − 6 B=
+ 2c 6 =or C 0.
0
Coeff. Of x : 2 B − 3C +=
2D 0 or D 0.
=
y p = x3
So,

The required general solution is

y =c1e x + c2 e 2 x + x 3 .

′′ ′ x 2 e x . (1)
Example: Solve y − 3 y + 2 y =

Solution. The C.F. is

yc c1e x + c2 e 2 x .
=
To construct a particular solution, we use (3) of the table
(
y p x k Ax 2 + Bx + C e x .
= )
Since e is already in C.F ., we take k = 1 so that no terms of C.F. is in
x

yp.
Hence the modified P.I. is
y p = Ax 3 + Bx 2 + Cx e x( )
y 'p = ( Ax 3
) (
+ Bx 2 + Cx e x + 3 Ax 2 + 2 Bx + C e x )
y"p = ( Ax 3
+ Bx 2 + Cx ) e + 2 ( 3 Ax
x 2
)
+ 2 Bx + C e x + ( 6 Ax + 2 B ) e x .
 Basic Concepts of Differential Equations 75

y p , y 'p and y"p

Substituting for into (1), we have
( Ax + Bx + Cx ) e + 2 ( 3 Ax + 2Bx + C ) e + ( 6 Ax + 2B ) e + ( 6 Ax + 2B ) e
3 2 x 2 x x x

− 3 ( Ax + Bx + Cx ) e − 3 ( 3 Ax + 2 Bx + C ) e + 2 ( Ax + Bx + Cx ) e =
3 2 x 2 x 3 2
xe x 3 x

 −3 Ax 2 + ( 6 A − 2 B ) x + C  e x =
x 3e x

Equating coefficients of like terms, we obtain

1
Coeff .of x3 : −3 A =
1 or A =
−
3
Coeff .of x : 6 A − 2 B =
0 or B =
−1

Coeff .of x 0 : C = 0
1
yp = − x 3e x − x 2 e x .
So 3
The required solution is
1
y =c1e x + c2 e 2 x − x3e x − x 2 e x .
3
′′
Example: Solve y + 4 y =
xe x + x sin 2 x (1)
=yc c1 sin 2 x + c2 cos 2 x.
Solution. The C.F., as readily found above, is
′′
For the P.I of (1), we find the P.I. of y + 4 y =
xe x (2)
And y′′ + 4 y =
x sin 2 x (3)
Separately. Their sum will be the P.I of (1) ( by the Principle of
superposition). For a particular solution of (2), we have
=y p x k ( Ax + B ) e x .

yp ,
Since no term of C.F. is in the we must take k = 0. Therefore,
=
yp ( Ax + B ) e x
y 'p =( Ax + B ) e x + Ae x
76 Differential Equations: Theory and Applications

y"p =( Ax + B ) e x + 2 Ae x .

y p , y 'p and y"p

Substituting for into (2), we have
( Ax + B ) e x + 2 Ae x + 4 ( Ax + B ) e x =
xe x

5 Axe x + ( 2 A + 5 B ) e x =
xe x .
or
Equating coefficients of like terms, we obtain
1
Coefficient of xe x : =
5 A 1=
or A
5

2
Coefficient of e x : 2 A + 5 B =
0 or B =
−
25

1 x 2 x
= yp xe − e .
So, P.I of (2) is 5 25
y=
p x ( Cx + D ) sin 2 x + ( Ex + F ) cos 2 x  .
k

Since sin 2 x and cos 2 x are already in the C.F., we must take k = 1, so
that the P.I. is
( ) ( )
y p = Cx 2 + Dx sin 2 x + Ex 2 + Fx cos 2 x

y '
p = ( Cx
+ Dx ) ( 2 cos 2 x ) + ( 2Cx + D ) sin 2 x
2

+ ( Ex + Fx ) ( −2sin 2 x ) + ( 2 E ( x ) + F ) cos 2 x
2

y = ( Cx + Dx ) ( −4sin 2 x ) + ( 2Cx + D )( 2 cos 2 x ) + ( 2Cx + D )( 2 cos 2 x )

"
p
2

+ ( 2Cx + D )( 2 cos 2 x ) + 2C sin 2 x + ( Ex + Fx ) ( −4 cos 2 x )

2

+ ( 2 Ex + F )( −2sin 2 x ) + ( 2 Ex + F )( −2sin 2 x ) + 2 E cos 2 x.

Substituting into (3), we have

( Cx 2
)
+ Dx ( −4sin 2 x ) + 4 ( 2Cx + D ) cos 2 x + 2C sin 2 x + ( Ex + Fx )( −4 cos 2 x )
 Basic Concepts of Differential Equations 77

( ) ( )
−4 ( 2 Ex + F ) sin 2 x + 2 E cos 2 x + 4 Cx 2 + Dx sin 2 x + 4 Ex 2 + Fx cos 2 x =
x sin 2 x.

Equating the coefficients of like terms, we get

1
−8 E =
1 or E =
−
Coeff. Of x sin 2 x : 8

Coeff. Of x cos 2 x=: 8C 0=

or C 0
Coeff. Of sin 2 x : 2C − 4 F
= 0 or F
= 0
1
cos 2 x : 2 E + 4 D
= 0 or D
= .
Coeff. Of 16
Hence P.I of (3) is
1 1
=yp x sin 2 x − x 2 cos 2 x.
16 8 (5)
Sum of (4) and (5) is the required P.I of (I), i.e.,
1 x 2 x 1 1
yp = xe − e + x sin 2 x – x 2 cos 2 x
5 25 16 8 is the particular solution of (1).
Hence the general solution of (1) is
1 2 1 1
y= c1 sin 2 x + c2 cos 2 x + xe x − e x + x sin 2 x − x 2 cos 2 x.
5 25 16 8

1.3.9. The Cauchy-Euler Equation

An equation of the form

n −1
dny n −1 d y dy
a0 x n n
+ a1 x n −1
+…+ an −1 x + an y = F ( x )
dx dx dx

is called Cauchy-Euler (or equidimensional) equation. 0 1

a , a , …, a , a
n −1 n are

real constants. The equation can be reduced to a linear differential equation

=
with constant coefficients by the transformation x e=
t
or t ln x
Then
dy dy dt 1 dy
= = .
dx dt dx x dt
 78 Differential Equations: Theory and Applications

d 2 y d  1 dy  1 d  dy  dy d  1 
= =    + .  
dx 2 dx  x dt  x dx  dt  dt dx  x 

1 d  dy  dt dy  1  1  d 2 y dy 
=  . +  − =   − 
x dt  dt  dx dt  x 2  x 2  dt 2 dt 

d 3 y d  1 d 2 y 1 dy 
=  − 
dx3 dx  x 2 dt 2 x 2 dt 

1 d  d 2 y  d 2 y d  1  1 d  dy  dy d
=  2 + 2 .  2 − 2  + .
2
x dx  dt  dt dx  x  x dx  dt  dt  1 
dx  − 2 
 x 
1 d  d 2 y  dt 2 d 2 y 1 d  dy  dt 2 dy
=  . − −  . +
x 2 dt  dt 2  dx x 3 dt 2 x 2 dt  dt  dx x 3 dt

1 d 3 y 3 d 2 y 2 dy
= − +
x3 dt 3 x3 dt 2 x3 dt
1  d3y d2y dy 
= 3  3
− 3 2
+2 
x  dt dt dt 
d d
= D =
and ∆ , then
If we write dx dt
xD = ∆

x 2 D 2 = ∆ 2 − ∆ = ∆ ( ∆ − 1)

x 3 D 3 =∆ 3 − 3∆ 2 + 2∆ =∆ ( ∆ − 1)( ∆ − 2 )

  

x n D n =∆ ( ∆ − 1)( ∆ − 2 ) …( ∆ − n + 1) .

2 2
Substituting these values of xD, x D , … into (1), we obtain an equation
of nth order with constant coefficients having t as the independendent
variable. This equation can be solved by the previous methods.
 Basic Concepts of Differential Equations 79

d2y dy
x2 2
x3 .
− 2x + 2 y =
Example: Solve dx dx (1)
Solution: Let x = e . Then we have t = ln x
t

xD = ∆.

x 2 D 2 =∆ ( ∆ − 1) .

Substituting into (1), we get

 ∆ ( ∆ − 1) − 2∆ + 2  y = e3t

 ∆ 2 − 3∆ + 2  y = e3t .
(2)
The characteristic equation has roots 2 and 1. Therefore, C.F. of (2) is
yc c1et + c2 e 2t
=

1 1 3t
=yp = e 3t e .
P.I. of equation (2) is ( ∆ − 1)( ∆ − 2 ) 2

1
y =c1et + c2 e 2t + e3t .
The general solution of (2) is 2

Replacing t by
(
ln x or et by x , )
we have
1 3
y =c1 x + c2 x 2 + x
2 as the general solution of (1).
d4y d3y d2y dy
+ 6 x 3 3 + 9 x 2 2 + 3 x + y = (1 + ln x ) .
2
x4 4
Example: Solve dx dx dx dx (1)
Solution: We let x = e so that t = ln x
t

xD = ∆

x 2 D 2 =∆ ( ∆ − 1) = ∆ 2 − ∆.

x 3 D 3 =∆ ( ∆ − 1)( ∆ − 2 ) =∆ 3 − 3∆ 2 + 2∆

x 4 D 4 =∆ ( ∆ − 1)( ∆ − 2 )( ∆ − 3) =∆ 4 − 6∆ 3 + 11∆ 2 − 6∆.

80 Differential Equations: Theory and Applications

Substituting into (1), we get

( ) ( )
 ∆ 4 − 6∆ 3 + 11∆ 2 − 6∆ + 6 ∆ 3 − 3∆ 2 + 2∆ + 9 ∆ 2 − ∆ + 3∆ + 1 y = (1 + t )2
 

(∆ 4
)
+ 2∆ 2 + 1 y = 1 + 2t + t 2
(2)
The roots of the characteristic equation of (2) are ±i, ±i.
yc = ( c1 + c2t ) sin t + ( c3 + c4t ) cos t.
Therefore, C.F. of (2) is
To find a particular integral of (2), we use the method of U .C. Let the
particular integral of (2) be
y p = At 2 + Bt + C

y 'p 2 At + B
=

y"p = 2 A.

Substituting into (2), we obtain

4 A + At 2 + Bt + C =1 + 2t + t 2

At 2 + Bt + ( 4 A + C ) =t 2 + 2t + 1.

Equating coefficients of like terms, we have

1, B =
A= 2, 4 A + C =
1or C =−3.

Therefore, a particular integral of (2) is

y p = t 2 + 2t − 3,

Hence the general solution of (2) is

y= ( c1 + c2t ) sin t + ( c3 + c4t ) cos t + t 2 + 2t − 3 (3)
Replacing t by ln x in (3), we get
y = ( c1 + c2 ln x ) sin ( ln x ) + ( c3 + c4 ln x ) cos(ln x) + ( ln x ) + 2 ln x − 3
2

as the general solution of (1).

 Basic Concepts of Differential Equations 81

Reduction of Order
d2y dy
If one solution of the second order linear equation (1) 0
+ P + Qy =
dx 2 dx
(where P , Q are not necessarily constants and may be functions
of x ) is known, then we can use it to find the general solution of
d2y dy
2
+ P + Qy = F ( x)
dx dx (2)
The procedure, due to D’ Alembert, is known as the method reduction
of order.

Suppose it is known that

y = y1 is a solution of (1). We assume that
y = vy1 (3)
is a solution of (2), where v is a function of x to be determined. From (3),
we get
dy dy dv
= v 1 + y1 ,
dx dx dx

d2y d 2 y1 dv dy1 d 2v
=v 2 +2 + y1 2
dx 2 dx dx dx dx

Substituting into (2), we have

2
d y1 dv dy1 d 2v dy dv
v 2 +2 F ( x)
+ y1 2 + Pv 1 + Py1 + Qvy1 =
dx dx dx dx dx dx

 dv  d y1 
2
d 2 v  dy1 dy
y1 2 +  2 F ( x)
+ Py1  +  2 + P 1 + Q y1  v =
dx  dx  dx  dx dx  (4)
y=y
Since 1 is a solution of (1)
2
d y1 dy
2
+ P 1 + Qy1 = 0
dx dx
and, therefore, equation (4) reduces to
d 2 v  dy1  dv
y1 2 +  2 + Py1  = F ( x)
dx  dx  dx (5)
 82 Differential Equations: Theory and Applications

dv
=u
Setting dx in (5), we obtain
du  dy1 
y1 + 2 + Py1  u =F ( x)
dx  dx 
Which is a linear equation of the first order in u and can be solved
dv
= u, y = vy1. The
for From dx we determine v and hence the solution
method is illustrated by means of example.
d2y
2
+y=csc x.
Example: Solve dx (1)
=yc c1 sin x + c2 cos x.
Solution. The C.F, of (1) is
yc as y1. Let us take y1 to be the value
We may take any special value of
=
y when
of c
c1 1=
and c2 0 . Then assume that y = v sin x (2)
is a solution of (1). From (2), we get
dy dv
= sin x + v cos x
dx dx

d2y d 2v dv
2
=sin x 2
+ 2 cos x − v sin x.
dx dx dx
Substituting into (1), we obtain
d 2v dv
sin x 2
+ 2 cos x − v sin x + v sin x =
csc x
dx dx

d 2v dv
2
+ 2 cot x csc 2 x.
=
dx dx

dv
u= ,
Setting dx the above equation becomes
du
csc 2 x
+ 2 cot x u =
dx (3)
Which is linear equation of first order. An integrating factor of (3) is
 Basic Concepts of Differential Equations 83

exp  ∫ 2 cot xdx

=  exp ln sin 2 =
x  sin 2 x
2
Multiplying (3) by sin x and integrating, we have
u sin 2 x = x,
Neglecting constant of integration since we seek only a particular
solution.
u = x csc 2 x

dv
= x csc 2 x
dx

∫ x csc 2 x dx =
v= − x cot x + ln sin x
(on integrating by parts)
Hence a particular solution of (1) is
v sin x =
y = − x cos x + sin x (ln sin x ).

The general solution of (1) is

y = c1 sin x + c2 cos x − x cos x + sin x(ln sin x ).

d2y dy
( x + 2). 2
− ( 2 x + 5 ) . + 2 y =( x + 1) e x
Example: Solve dx dx (1)
2x
Solution: We note that y = e makes the left-hand side of (1) zero.
Therefore, we put
y = v e2 x

dy  dv 
=  + 2v  e 2 x
dx  dx 

d 2 y  d 2v dv 
2
=  2 + 4 + 4v  e 2 x
dx  dx dx 
Substituting into (1), we obtain
 d 2v dv   dv 
( x + 2)  2
+ 4 + 4v  e 2 x − ( 2 x + 5 )  + 2v  e 2 x + 2v e 2 x =( x + 1) e x .
 dx dx   dx 
 84 Differential Equations: Theory and Applications

d 2v 2 x dv
( x + 2) e +  4 ( x + 2 ) − ( 2 x + 5 )  e 2 x +  4 ( x + 2 ) − 2 ( 2 x + 5 ) + 2  v e 2 x = ( x + 1) e x
dx 2 dx

d 2v dv
( x + 2) 2
+ ( 2 x + 3) =( x + 1) e − x .
dx dx
dv
u= ,
Writing dx the above equation becomes
du
( x + 2) + ( 2 x + 3) u =( x + 1) e − x
dx
du 3 x + 3 x +1 −x
+ u= e
dx x + 2 x + 2 (2)
which is linear equation of first order. An integrating factor of (2) is
 2x + 3    1  
exp  ∫ dx = exp  ∫  2 −  dx
 x+2    x + 2  

e2 x
= exp  2 x − ln ( x + 2=
)

x+2
e2 x
Multiplying (2) by x + 2 , we have
d  e2 x  x +1
u = .e x
dx  x + 2  ( x + 2 )2

e2 x x +1 x
u =
∫ .e dx + c1
x + 2 ( x + 2 )2

ex ex
=∫ dx − ∫ dx + c1
( x + 2)
2
x+2

ex
= + c1
x+2

u =e − x + c1 ( x + 2 ) e −2 x
 Basic Concepts of Differential Equations 85

dv
=e − x + c1 ( x + 2 ) e −2 x .
Therefore, dx
Integrating, we get
1
v= −e − x − c1 ( 2 x + 5 ) e −2 x + c2
4
1
y= −e x − c1 ( 2 x + 5 ) + c2 e 2 x .
ve 2 x =
Hence 4
is the general solution of (1).

Note. It is easy to see that y′′ + Py′ + Qy = 0 is satisfied by y = e ± x .

If 1 ± P + Q = 0 and by y = x if P + Qx = 0. These can be used to find a
solution y′′ + Py′ + Q =
0 by inspection.

1.4. THE WRONSKIAN

y1 , y2 I = [ a ,b]
Definition. If are two differentiable functions of x on
W = W [ y1 , y2 ] ,
then their Wronskian, denoted by is obtained by
y1 y2
W = W [ y1 , y2 ] = y1 y2' − y1' y2 =
y1' y2'

y1 , y2 , y3
Similarly, the Wronskian of three differentiable functions on
I =[ a ,b ]
is defined by
y1 y2 y3
= [ y1 , y2 , y3 ]
W W= y1' y2' y3'
y1" y2" y3"

The definition can be extended in a similar manner for the Wronskian of

I = [ a, b ] .
n differentiable functions on
y1 , y2
Theorem. Let be two solutions of the homogeneous linear equation
d2y dy
2
0,
+ P + Qy =
dx dx (1)
 86 Differential Equations: Theory and Applications

I = [ a, b ] .
Where P, Q are functions of x and are continuous on Then their
W = W [ y1 , y2 ]
Wronskian is either identically zero or is never zero on I .
Proof. We have
=W y1 y2' − y1' y2

Differentiating w.r.t.x, we get

W ′ = y1 y2" + y1' y2' − y1' y2' − y1" y2

= y1 y2" − y1" y2 .

Since
y1 , y2 are solutions of (1), we have

y1" + Py1' + Qy1 =

0 (2)

and
y2" + Py2' + Qy2 =
0 (3)

Multiply (2) by
y2 and (3) by y1 to have an equivalent system

y1" y2 + P y1' y2 + Qy1 y2 =

0
(4)
y1 y2" + P y2' y1 + Qy1 y2 =
0 (5)
Subtracting (4) from (5), we obtain
(y y "
1 2 ) (
− y2 y1" + P y1 y2' − y1' y2 =
0. )
dW
0.
+ PW =
dx
The general solution of the above equation is readily found as
W c exp(− ∫ P dx ).
=

Since the exponential function is never zero, W is identically zero if

c = 0 or is never zero when c ≠ 0.
d2y dy
y ,y 2
+ P 0,
+ Qy =
Theorem. If 1 2 are two solutions of dx dx
=
Then their Wronskian
[ y1 , y2 ] 0 if and only if y1 , y2 are
W W=
linearly dependent.
 Basic Concepts of Differential Equations 87

Proof. If one of the two solutions is identically zero, then the theorem
y1 , y2 y1 , y2
is obviously true. Assume that are both nonzero and let be
linearly dependent. Then
y1 = cy2 , where c is a constant.
y1
=c
y
i.e., 2
d  y1  y2 y1' − y1 y2' dc
 =  = = 0
dx  y2  y22 dx
or
y y ' − y1' y2 =
0
i.e., 1 2
W [ y1 , y2 ] = 0.

d  y1  W
 =  = 0.
W [ y1 , y2 ] = 0, dx  y2  y22
Conversely, if then
y1
= c,
y2 where c is a constant.
y1 cy2 y1 , y2
Thus and so are linearly dependent.
W [ y1 , y2 ] ≠ 0 y1 , y2
Corollary. if and only if are linearly independent.
Above theorems can only be applied to check the linear dependence
or linear independence of differentiable functions which are solutions of
a linear homogeneous O.D.E. Arbitrary differentiable functions exist with
their Wronskian zero but they are linearly independent.

Consider the functions

= 3
y1 x= , y2 x 3
on
]−∞, ∞[ . Here
W ( y1 , y2 ) = 0 y and y2
but 1 are linearly independent.

1.4.1. VARIATION OF PARAMETERS.

d2y dy
2
F ( x).
+ P + Qy =
We found the solution of the equation dx dx (1)
Where P, Q are functions of x, in the previous section by reduction of or-
der of (1). The solution of (1) can be determined by a procedure known as
the method of variation of parameters. This method can be applied even to
equations of higher order.
88 Differential Equations: Theory and Applications

d2y dy
ℵüP Q
Suppose that linearly independent solutions of dx 2 dx (2)
y = y1 ( x ) y = y2 ( x ) .
Are given by and Then the complementary
function of (1) is
=
yc c1 y1 + c2 y2

Where
c1 and c2 are arbitrary constants. We replace the arbitrary

constants 1
c and c2 by unknown functions u1 ( x ) and u2 ( x ) and require that
=y p u1 y1 + u2 y2
(3)
be a particular solution of (1). In order to determine the two functions
u1 and u2 we need two conditions. One condition is that (3) must satisfy (1).
A second condition can be imposed arbitrarily.
Differentiating (3) w.r.t.x, we have
y 'p = (u y + u y ) + u y + u y .
'
1 1
'
2 2
'
1 1 2
'
2

y"p u1" and u2" . To

If we differentiate the above equation, will contain
avoid second derivatives of
u1 and u2 . we put

u1' y1 + u2' y2 =
0. (4)

y 'p u1 y1' + u2 y2'

=
With this condition, we have
y"p u1 y1' + u2 y2'
=
So that
y"p = u1' y1' + u2' y2' + u1 y1" + u2 y2" .
So that
y p , y 'p , y"p
Substituting for into equation (1), we get
(u y + u y + u y + u y ) + P (u y + u y ) + Q (u y + u y ) =
' '
1 1
'
2
'
2
"
1 1 2
"
2 F ( x) '
1 1 2
'
2 1 1 2 2

u ( y + Py + Qy ) + u ( y + Py + Qy ) + u y =
1
"
1
'
1 1 F ( x).
2
"
2
'
2 2
' '
1 2
or
Expressions within the parenthesis are zero since
y1 and y2 are solutions
of (2). Hence
 Basic Concepts of Differential Equations 89

F ( x)
u1' y1' + u2' y2' =
(5)
Taking (4) and (5) together, we have two equations in the two unknowns
u and u2' ;
'
1

u1' y1 + u2' y =
0

F ( x),
u1' y1' + u2' y2' =
and
Solving these, we have
− y2 F ( x ) 
u1' = 
y1 y2' − y1' y2 

' y1 F ( x ) 
u2 =
y1 y2' − y1' y2 
(6)
y ,y W ( y1 , y2 ) = y1 y2' − y1' y2 ≠ 0,
In (6), the Wronskian of 1 2 namely
since
y1 , y2 are linearly independent solutions of (2). Integrating (6), we
u and u2 as
find 1
− y2 F ( x ) − y2 F ( x )
u1 =∫ dx =
∫ dx
y1 y2' − y1' y2 W (7)
y F ( x) y F ( x)
∫ 1'
u2 = '
∫ 1
dx = dx
y1 y2 − y1 y2 W (8)

Where W is the Wronskian of

y1 , y2 .
y
Thus p is completely determined. In numerical problems, instead of
performing the complete process, formulas (7) will be directly applied to
evaluate
u1 and u2 .

d2y dy
2
−2 + y =x e x ln x
Example. Find the general solution of dx dx (1)
Solution. The C.F. of (1) is
=
yc c1e x + c2 x e x
Let
 90 Differential Equations: Theory and Applications

=
y p u1e x + c2 xe x

Here
=y1 e=
x
( x ) x e x ln x
, y2 x e x , F=

( )
W = W ( y1 , y2 ) = y1 y2' − y1' y2 = e x e x + xe x − xe x .e x = e 2 x .

By the formulas of (7) and (8), we have

− y F ( x) − xe x .xe x ln x
∫ 2
u1 = dx =
∫ dx
W e2 x
y F ( x) e x . xe x ln x
∫ 1
u2 = dx =
∫ dx
and W e2 x
x3 x3 x3 x3
u1 =− ∫ x 2 ln xdx =−(ln x). + ∫ dx = − ln x
i.e., 3 3 9 3
x2 x x2 x2
u2 = ∫ x ln x dx = ln x − ∫ dx = − + ln x.
2 2 4 2

 x3 x3  x  x2 x2  x
y=  − ln x  e +  − + ln x  x e
 9 3  4 2
p
 

 x3 x3 x3 x3 
= e x  − ln x − + ln x 
 9 3 4 2 

 5 x3 x3 
= ex  − + ln x  .
 36 6 

The general solution of (1) is

 5 x3 x3 
y = yc + y p = c1e + c2 x e = e  −
x x x
+ ln x  .
 36 6 
Example. Find the general solution of
d2y dy
x 2
2
− x ( x + 2) + ( x + 2) y =
x3
dx dx (1)
 Basic Concepts of Differential Equations 91

Given that y = xe is a solution of the associated homogeneous equation

x

d2y dy
x2 2
− x ( x + 2) + ( x + 2) y =
0
dx dx (2)
Solution. The given equation in the standard form is
d 2 y x + 2 dy x + 2
− + 2 y= x.
dx 2 x dx x
x+2 x+2
P + Qx = − +x 2 = 0, y = x
Since x x is also a solution of (2).
The two solutions, y = x and y = xe x
are linearly independent. The
complementary function of (1) is
y=
c c1 x + c2 xe x .

y= u1 x + u2 xe x
We assume that p
(3)
is a particular solution of (1).
=
Here
y1 x= ( x) x
, y2 xe x , F =

( )
W = W ( y1 , y2 ) = y1 y2' − y1' y2 = x xe x + e x − xe x = x 2 e x

By the formulas of (7) and (8), we have

− y2 F ( x ) y1 F ( x )
u1 = ∫ dx u2 = ∫ dx
W and W
x 2e x x2
u1 = − ∫ 2 x
dx = − x u 2 =
∫ 2 x
dx = −e − x .
i.e., xe and xe
u ,u
Substituting for 1 2 into (3), a particular solution of (1) is
y = yc + y p = c1 x + c2 x e x − x 2 − x.

Example. Explain how the method of variation of parameters can be

applied to find a particular solution of a non-homogeneous linear third order
differential equation whose complementary function is known. Apply the
method to find a particular solution of
92 Differential Equations: Theory and Applications

d 3 y dy
+ cos x.
=
dx 3 dx
Solution. Suppose the C.F. of a linear third order differential equation
d3y d2y dy
3
+ P 2
F ( x)
+ Q + Ry =
dx dx dx (1)

is known to be c 1 1 2 2 3 3y =c y + c y + c y ; y , y , y
1 2 3 being linearly independent

solutions of the associated homogeneous equation of (1).

y =u1 y1 + u2 y2 =u3 y3 , u ,u ,u
We assume that p where 1 2 3 are functions of x.
y 'p = u1 y1' + u3 y3' + u1' y1 + u2' y2 + u3' y3 .

We set
u1' y1 + u2' y2 + u3' y3 =
0.
Then
y =u1 y1' + u2 y2' + u3 y3'
'
p

y"p =u1 y1" + u2 y2" + u3 y3"

y '''p = u1 y1''' + u2 y2''' + u3 y3''' + u1' y1" + u2' y2'' + u3' y3" .

y p , y 'p , y"p , y '''p

Substituting for into equation (1), we have
( u y + u y + u y + u y + u y + u y ) + P (u y + u y + u y )
'''
1 1 2
'''
2
'''
3 3
' ''
1 1
'
2
''
2
' ''
3 3
''
1 1 2
''
2
''
3 3

+Q( u y + u y + u y ) + R( u y + u y + u y ) =
'
1 1 2
'
2 F ( x)
'
3 3 1 1 2 2 3 3

u ( y + Py + Qy + Ry ) + u ( y + Py + Qy + Ry )
1
'''
1
''
1
'
1 1 2
'''
2
''
2
'
2 2

( 2 2 2 ) ( 3 F ( x)
+ u2 y + Py + Qy + Ry2 + u3 y + Py + Qy + Ry3 + u y + u2' y2'' + u3' y3'' =
''' '' ' ''' ''
3
'
3 ) ' ''
1 1
(4)
' ' '
From the system of equations (2), (3) and (4), we find
u , u and u and
1 2 3

on integration we get
u1 , u2 and u3 .
d 3 y dy
3
+ csc x.
=
Now we apply this method to find P.I. of dx dx
 Basic Concepts of Differential Equations 93

The C.F. of the equation is obtained by solving the equation

(
D D +1 = 0.)Thus
c1 + c2 cos x + c3 sin x with
yc =
y p =u1 y1 + u2 y2 + u3 y3 ,
For the particular integral, let
u1 + u2 cos x + u3 sin x.
=
y1 , y2 , y3
Substituting for and their derivatives into equations (2), (3) and
u' ,u' ,u '
(4) above, we have the following system of linear equations in 1 2 3 .
u1' + u2' cos x + u3' sin x =
0

u1' .0 − u2' sin x + u3' cos x =

0

u1' .0 − u2' cos x − u3' sin x =

csc x.

We solve this system by the Gaussian Elimination method. Augmented

matrix of the system is
1 cos x sin x 0 
 
0 − sin x cos x 0 
0 − cos x − sin x csc x 

1 0 0 csc x 
 
0 − sin x cos x 0  by R1 + R3
0 − cos x − sin x csc x 

1 0 0 csc x 
  by ( cos x ) R2
0 − sin x cos x cos 2 x 0 
and ( sin x ) R3
0 − cos x sin x − sin 2 x 1 

1 0 0 csc x 
 2 
0 − sin x cos x cos x 0 
0 0 −1 1 
94 Differential Equations: Theory and Applications

Therefore,
'
1or u3' =
−u = 3 −1, from R3

( − sin x cos x ) u2' + ( cos 2 x ) u3' =

0 R
from 2

or
u2' = − cot x.

u1' = csc x, R1.

from
Hence
∫ csc x dx =
u1 = ln csc x − cot x

and
u3 =∫ − dx =− x.
u1 + u2 cos x + u3 sin x.
yp =

= ln csc x − cot x − ( cos x ) ln sin x − x sin x

is the required particular solution.

1.4.2. Application of Differential Equations

Differential equations can be used in the solution of many problems in
physical, biological and social sciences. In this section, we shall study
such problems which can be formulated in terms of first and second order
differential equations. The solution of the differential equation will lead to
the solution of the given problem.
The following examples illustrate the techniques.
Example. A stone weighting 8 lbs. falls from rest toward the earth from
a great height. As it falls it is acted upon by air resistance that is numerically
1
v ( in pounds )
2 . Where v is the velocity (in feet per second). Find the
velocity and distance fallen at time t.
Solution. We use Newton’s second law F = ma to formulate
mathematical model of this problem. The forces acting on the body are
 Basic Concepts of Differential Equations 95

i.
F1 , the weight of the stone acting downward and hence is positive.
1
ii.
F2 , the air resistance, numerically equal to 2 v, which acts upward
and is therefore negative.
dv
m = F1 + F3
Newton’s second law becomes dt
1 dv 1
=9 − v, taking g =32.
i.e., 4 dt 2
dv
= 32 − 2v
dt

dv
+ 2v =
32
dt

which is a linear equation. We now solve this differential equation. Here

∫ 2 dt
I .F e=
= e 2t .
And so,
d
dt
( )
v e 2t = 32 e 2t

Integrating, we get
2t
16 e 2t + c or v =+
ve = 16 ce −2t

v ( 0 ) = 0.
Now
Hence 0 =
16 + c or c =
−16
=
Therefore,
{
v 16 1 − e −2t } (1)
is the velocity after time t. (1) may be written as
dx
(
= 16 1 − e −2t .
dt
)
(2)

From (2), we have

( )
dx 16 1 − e −2t dt
=

Or x =16t + 8 e −2t + k .
96 Differential Equations: Theory and Applications

x ( 0 ) = 0,
Applying the initial condition we find k = −8.
Hence x =16t + 8 e −2t − 8 is the distance fallen after time t.
Example. A body of constant mass is projected upward from the earth’s
v.
surface with an initial velocity 0 Assuming there is no air resistance, but
taking into consideration the variation of the earth’s gravitational field with
altitude, find the smallest initial velocity for which the body will not return
to the earth (This is the so-called escape velocity).
w( x)
Solution. The general expression for the weight of a body of mass
m is obtained from Newton’s inverse- square law of gravitational attraction.
If R is the radius of the earth and x is the altitude above sea level, then
k
w( x) = ,k
( R + x)
2

being constant. At= x 0,= w mg , hence k = mgR 2

mg R 2
w( x) = .
( R+ x)
2

and The only force acting on the body is its weight

dv − mg R 2
m =
dt ( R + x )2
which acts downward. Thus, the equation of motion is
dv mg R 2
v = − .
( R + x)
2
dt
or
Separating the variables and integrating, we have
1 2 gR 2
= v + c.
2 R+x
=
But
v v= 0 at t 0 i.e., at x = 0.
1 2
= c v0 − gR
Therefore, 2
2 2 2 gR 2
v =v − 2 gR +
0
And so R + x (1)
The escape velocity is found by requiring that v given by (1) remains
positive for all (positive) value of x. Thus we must have
 Basic Concepts of Differential Equations 97

v02 ≥ n 2 g R.

Hence the escape velocity is

=v0 2 gR 6.9miles / sec.taking
= = R 4000 miles

1.4.3. Summary and Discussion

In this chapter, we discussed some elementary aspects of differential
equation. A very important topic in real life problem.
The study of differential equations is a wide sphere in pure and applied
mathematics, physics, and engineering. All of these disciplines are concerned
with the estate of differential equations of various types. Pure sum focuses
on the life and singularity of solutions, while applied tally emphasizes the
rigorous occasion of the methods for approximating solutions. Differential
equations pleasure an important role in modeling virtually every physical,
technical, or biological process, from celestial motion, to bridge design, to
interactions between neurons. Differential equations such as those used to
solve real-life problems may not necessarily be directly solvable, i.e. do not
have closed pattern solutions. Instead, solutions can be approximated using
numerical methods.
Applications. Many fundamental jurisprudences of physics and
chemistry can be formulated as differential equations. In biology and
economics, differential equations are used to mold the behavior of complex
systems. The mathematical opinion of differential equations first developed
together with the sciences where the equations had originated and where
the results found application. However, diverse problems, sometimes
originating in quite distinct scientific fields, may give rise to identical
differential equations.
Whenever this happens, mathematical plan seats the equations can be
viewed as a unifying principle seat diverse phenomena. As an example,
consider dissemination of light and sound in the atmosphere, and of waves
on the surface of a pond. All of them may be described by the same second-
order partial differential equation, the breaker equation, which allows
ourselves to think of brightening and sound as forms of waves, scads like
familiar cascade in the water. Conduction of heat, the theory of which was
developed by Joseph Fourier, is governed by another second-order partial
differential equation, the heat equation. It turns out that many scattering
processes, while seemingly different, are described by the same equation;
98 Differential Equations: Theory and Applications

the Black-Scholes equation in finance is, for instance, related to the heat
equation.
Euler-Lagrange equation in classical mechanics, Hamilton’s equations
in classical mechanics, Radioactive decay in nuclear physics Newton’s
statute of cooling in thermodynamics, The breaker equation , The heat
equation in thermodynamics, Laplace’s equation, which defines harmonic
functions, Poisson’s equation, The geodesic equation, The Navier-Stokes
equations in fluid dynamics, The Diffusion equation in stochastic processes,
The Convection-diffusion equation in fluid dynamics, The Cauchy-Riemann
equations in complex analysis, The Poisson-Boltzmann equation in molecular
dynamics, The shallow water equations and Universal differential equation.
Differential Equation. An equation involving one dependent variable
and its derivatives with respect to one or more independent variables is
called differential equation. For example
dy
+ y cos x =
sin x
dx
Ordinary Differential Equation (O. D. E). A differential equation, in
which ordinary derivatives of the dependent variable with respect to a single
independent variable occur, is called an ordinary differential equation (O.
D. E).
Partial Differential Equation. A differential equation involving
partial derivatives of the dependent variable with respect to more than one
independent variable is called a partial differential equation.
∂z ∂z
x + y =
nx
∂x ∂y
Order of Differential Equation .The order of a differential is the order
of the highest derivative that occurs in the equation.
Degree of a Differential Equation. The degree of a differential equation
is the greatest exponent of the highest order derivative that appears in the
equation. (The dependent variable and its derivatives should be expressed in
a form free from radicals and fractions).
dy
+ y cos x =
sin x
• dx (order 1, degree 1)
 Basic Concepts of Differential Equations 99

2
d2y  dy 
2
+ xy   = 0
• dx  dx  (order 2, degree 1)
3
  dy 2  2 d 2 y
1+    =
  dx   dx 2
• (order 2, degree 2)
Linear Differential Equation. An ordinary differential equation
 dy d 2 y dny 
F x , y , , ,…, n  = 0
 dx dx 2 dx 
is said to be linear if F is a linear
2
dy d y dny
x , , ,…, n
function of the variables dx dx 2 dx
It should be carefully noted that in a linear ordinary differential equation
• The dependent variable y and its derivative are all of degree one.
• No product of y or any of its derivative appear.
• No transcendental function of y and / or its derivative occur.
Nonlinear Differential Equation. A differential equation that is not
linear is called a nonlinear differential equation. Differential equations
occur in the mathematical formulation of many problems in science and
engineering. Some such problems are
• Determining the motion of projectile, rocket, satellite or planet.
• Finding the charge or current in an electric circuit.
• Study of chemical reactions.
• Determination of curves with given geometrical properties.
Solution of a Differential Equation
A solution (or integral) of a differential equation is a relation between
the variables, not containing derivatives, such that this relation and the
derivatives obtained from it satisfy the given differential equation identically.

1.4.4. Initial Condition

It is often required to find the solution of a differential equation subject to
certain conditions. If the conditions relate to one value of the independent
dy
variable such as
y = y0
at
x = x0
(written as
y ( 0 ) 0 and dx = y′ ( x0 )
x = y )
100 Differential Equations: Theory and Applications

at
x = x0 ,
where
x0
belongs to some interval
]α , β [ then they are called
initial conditions (or one- point boundary conditions) and
x0 is called the
initial point.

1.4.5. Boundary Conditions

The problem of finding the solution of a differential equation such that all
the associated constraints relate to two different values of the independent
variable is called a two-point boundary value problem (or simply a boundary
value problem). The associated supplementary boundary conditions are
called two-point boundary conditions.

1.4.6. Separable Equations

Definition. A differential equation of the type
F ( x ) G ( x ) dx + f ( x ) g ( y ) dy =
0
is called an equation with separable
variables or simply a separable equation. Equation (1) may be written as
F ( x) g ( y)
dx + 0
dy =
f ( x) G ( y)

which can be easily integrated.

1.4.7. Homogenous Equations

f ( x, y )
Definition. A function is called homogeneous of degree n
f ( tx , ty ) = t f ( x, y ) ,
n
if Where t is a nonzero real number. Thus
10 10
x + y x
xy , 2 2
and sin  
x + y  y  are homogeneous function of degree 1,8 and
0 respectively.
M ( x, y ) dx + N ( x, y ) dy
Exact Equation. The expression is called
an exact differential if there exists a continuously differentiable function
f ( x, y )
of two real variables x and y such that the expression equals the
∂f ∂f
=df dx + dy.
total differential df . We know from calculus that ∂x ∂y
 Basic Concepts of Differential Equations 101

∂f ∂f
M ( x , y=
) = f x and N ( x , =
y ) = fy.
Thus, if it is exact then ∂x ∂y
Integrating Factors. If the differential equation
M ( x, y ) dx + N ( x, y ) dy = 0
is not exact but when it is multiplied by a
µ ( x, y )
function and the resulting equation
µ ( x , y ) M ( x , y ) dx + µ ( x , y ) N ( x , y ) dy =
0 µ(x ,y)
is exact, then
is called an integrating factor (I.F) of the differential equation. The number
of integrating factors of an equation may be infinite.
We list below (without proofs) some rules to find the integrating factors
of equations of special types.
My − Nx
M ( x , y ) dx + N ( x , y ) dy =
0 = P,
Rule 1. If is not exact and N
µ ( x)
where P is a function of x only then (1) has an integrating factor
µ ( x)
which also depends on x. is solution of the differential equation
dµ
= Pµ
dx
µ (=
x ) exp ∫ P dx
i.e.,
∂M ∂N
= My = , Nx .
Note that ∂y ∂x
Nx − M y
= Q,
Rule II. If M where Q is a function of y only, then the
differential equation
0 has an integrating factor
Mdx + Ndy =
µ (=
y ) exp ∫ Q dy.

0 (1)
Rule III. If M dx + Ndy =
1
is homogeneous and xM + yN ≠ 0, then xM + yN is an I.F of (1).
102 Differential Equations: Theory and Applications

Rule IV. If M dx + Ndy = 0 is of the form y f ( x, y ) dx + x g ( x, y ) dy =

0
1
and xM − yN ≠ 0, then xM − yN is an I.F of (1).
The following differential formulas are useful in the calculation of
certain exact equations:
 y  x dy − y dx
d =
x x2

 x  y dx − x dy
d =
 y y2

d ( xy
= ) x dy + y dx

( )
d x 2 + y 2= 2 ( x dx + y dy )

 x  y dx − x dy
d  ln  =
 y xy

 x  y dx − x dy
d  arc tan  = 2
 y  x + y2

Linear Equations
A first order ordinary differential equation (ODE) is linear in the dependent
variable y and the independent variable x if it is or can be written in the
dy
+ P ( x) y =Q ( x),
form dx where P and Q are functions of x.

The Bernoulli Equation

dy
+ P ( x) y =
Q ( x) yn
Definition. An equation of the form dx (1)
is called the Bernoulli differential equation. This equation is linear if
n = 0 or 1. If n is not zero or 1, then (1) is reducible to a linear equation.
 Basic Concepts of Differential Equations 103

Clairaut’s Equation
y xp + f ( p ) ,
=
Definition. The equation is known as Clairaut’s equation.
Envelope
f ( x, y , c ) = 0
Let be a one-parameter family of curves. Suppose
all members of the family of curves are drawn for various values of the
parameter c arranged in order of magnitude. The two curves that correspond
to two consecutive values of c will be designated as neighboring curves. The
locus of the ultimate points of intersection of neighboring curves is called
f ( x, y, c ) = 0.
the envelope of the family
f ( x, y , p ) = 0
Singular Solution: A solution of a differential equation
is called a singular solution (S.S) if
i. It is not derived from the general solution by giving any par-
ticular value to the arbitrary constants
ii. At each of its points, it is tangent to some member of the
one parameter family of curves represented by the general
solution.

The Ricatti Equation

Definition. We have already studied first order linear differential equation
y′ + P ( x ) y = R ( x )
(1)
q ( x) y2
If we add the term to the left-hand member of (1), we obtain a
nonlinear differential equation
y′ + P ( x ) y + Q ( x ) y 2 =
R ( x)
(2)
(2) is called the Ricatti equation.

1.4.8. Linear Differential Equations

Definition. A linear differential equation of order n in the dependent variable
y and the independent variable x is of the form

dny d n −1 y dy
a0 ( x )n
+ a 1 ( x ) n −1
+…+ an −1 + an ( x ) y = F ( x ) ,
dx dx dx (1)
a ( x ) , a1 ( x ) ,…, an −1 ( x ) , an ( x ) and F ( x )
where 0 are functions of the inde-
104 Differential Equations: Theory and Applications

a ( x)
pendent variable x only and 0 is not identically zero. Using primes,
(1) is also written as
a0 y n + a1 y n −1 +…+ an −1 y′ + an y =F ( x)
(2)
where
a0 , a1 , …, an −1 , an are real constants.

y1 ( x ) , y2 ( x ) , …, ym ( x )
Definition. If are m functions of an
independent variable x and 1 2
c , c , …, cm are constants, then the expression
c1 y1 ( x ) + c2 y2 ( x ) +…+ cm ym ( x )
is called a linear combination of
y1 ( x ) , y2 ( x ) , …, ym ( x ) .

i. Every homogeneous linear nth − order differential equation

a0 y + a1 y n −1 +…+ an −1 y′ + an y =
n
0 has n linearly independent solutions
y1 , y2 , …, yn

ii. If
y1 , y2 ,…, y _ m are n linealy independednt solutions of (3), then

any linear combination c

y c1 y1 + c2 y2 +…+ cn yn of y1 , y2 , …, yn
=

is the general solution of (3), 1 2

c , c ,…, c
m being arbitrary constants.

y y
iii. Let p be any particular solution of (2). i.e., p does not contain
y + yp
any constant, then c is the general solution of (2).

1.4.9. Homogeneous Linear Equation

dny d n −1 y dy
a0 n
+ a1 n −1
+…+ an −1 + an y, = 0
Consider the equation dx dx dx (1)

where
a0 , a1 ,…, an −1 are real constants.

Case I. Distinct Real Roots

Let 1 2
m , m , …, m
n be n distinct real roots of (2). Then e , e ,…, e
m1 x m2 x mn x

are n distinct solutions of (1). These n solutions are linearly independent.

=
Hence the general solution of (2) is
y c1e m1x + c2 e m2 x +…+ cn e mn x , Where
c1 , c2 , …, cn are arbitrary constants.
 Basic Concepts of Differential Equations 105

Case II. Repeated Real Roots

=
y ( c1 + c2 x ) em x + c3e m x +…+ cn em x
1 3 n

In the same manner, if the characteristic equation (2) has the triple
repeated root
m1 , the corresponding part of the general solution of (1) is
the solution of
( D − m1 )
3
0
y=

Proceeding as before, we can easily find

( )
y = c1 + c2 x + c3 x 2 e m1x

Case III. Complex Roots

Suppose the characteristic equation has the complex number a + ib as a
non- repeated root. Since coefficients of (1) are real, the conjugate complex
number a − ib is also a non- repeated root. The corresponding part of the
general solution is
=y k1e( a + ib ) x + k2 e( a −ib ) x , where k1 and k2 are arbitrary constants.
=y e ax  k1eibx + k2 e − ibx 

= e ax  k1 ( cos bx + i sin bx ) + k2 ( cos bx − i sin bx ) 

= e ax ( k1 + k2 ) cos bx + i ( k1 − k2 ) sin bx 

= e ax [ c1 sin bx + c2 cos bx ]

c = i ( k1 − k2 ) , c2 =
k1 + k2
Where 1 are two arbitrary constants.
If a + ib and a − ib are conjugate complex roots, each repeated k times,
the the corresponding part of the general solution of (1) may be written as
( ) ( )
y  c1 + c2 x + c3 x 2 +…+ ck x k −1 sin bx + ck +1 + ck + 2 x +…+ c2 k x k −1 cos bx 
=

Solution of the equation

• Complementary Function (C.F)
• Particular Solution (P.I)
 106 Differential Equations: Theory and Applications

f ( D) y = 0
The C.F is the solution of the homogeneous equation and
we have described different cases of its solution.
Now to find the P.I of (1), we can write
1
y= F ( x)
f ( D)
1
y= F ( x).
f ( D)
and try to evaluate
F ( x ) sin ax or cos ax.
To Find the P.I when
1 1
sin ax and cos ax.
f ( D) f ( D)
Here we have to evaluate
From Euler’s theorem, we have
=
e cos ax + i sin ax
iax

1 1
sin ax and cos ax
f ( D) f ( D)
Thus are respectively imaginary and
1
eiax .
f ( D)
real parts of

If
f ( D)
contains only even powers of
( )
D, say f ( D ) = f D 2 ,
it is
easy to see that
1  sin ax 1  sin ax
=  2  ( )
, provided − a 2 is not a zero of f D 2 .
( )
f D 2 cos ax ( )
f −a cos ax

1.4.10. Series Expansion of Polynomials or Expressions involv-

ing Operators.
F ( x)
This method is useful when is a polynomial in x .
f ( D) y = F ( x) F ( x)
Let be such that is a polynomial in x . To
evaluate the particular integral
 Basic Concepts of Differential Equations 107

1
yp = F ( x)
f ( D)

1
f ( D)
It is often useful to expand in a series in D by the Binomial
Theorem for negative exponent so that
y p= ( 1+ a D + a D
1 2
2
)
+ a3 D 3 +… F ( x )

The derivatives on the right will vanish after certain stage since
D x r 0 if n > r.
= n

The following binomial expansions will be useful in this connection:

1
= 1 + x + x 2 + x 3 +….
1– x
1
= 1 − x + x 2 − x3 +….
1+ x

1.4.11. The Method of Undetermined Coefficients

(U. C. Method)
We have studied some special cases in which particular integral can
be evaluated by the inverse operator. Now we consider the method of
undetermined coefficients which can prove simpler in finding the particular
f ( D) y = F ( x) F ( x)
integral of the equation when is
ax
a. An exponential function: e

b. A polynomial:
(
b0 x n + b1 x n −1 +…+ bn )
c. Sinusoidal function : sinax or cos bx
F ( x)
The more general case in which is sum of a product of terms of
 sin ax 
= (
F ( x ) e ax b0 x n + b1 x n −1 + …+ bn 
cos
)bx

the above types, such as  

The Cauchy-Euler Equation

An equation of the form
 108 Differential Equations: Theory and Applications

n −1
dny n −1 d y dy
a0 x n n
+ a1 x n −1
+…+ an −1 x + an y = F ( x )
dx dx dx

is called Cauchy-Euler(or equidimensional) equation. 0 1

a , a , …, a , a
n −1 n are

real constants. The equation can be reduced to a linear differential equation

=
with constant coefficients by the transformation x e=
t
or t ln x

Reduction of Order
d2y dy
0
+ P + Qy =
If one solution of the second order linear equation dx 2 dx (1)
(where P, Q are not necessarily constants and may be functions
of x ) is known, then we can use it to find the general solution of
d2y dy
2
+ P + Qy = F ( x)
dx dx (2)
The procedure, due to D’ Alembert, is known as the method reduction
of order.

The Wronskian

If
y1 , y2 are two differentiable functions of x on I = [ a , b ] then their
W = W [ y1 , y2 ] ,
Wronskian, denoted by is obtained by
y1 y2
W = W [ y1 , y2 ] = y1 y2' − y1' y2 =
y1' y2'

Similarly, the Wronskian of three differentiable functions

y1 , y2 , y3 on
I =[ a ,b ]
is defined by
y1 y2 y3
= [ y1 , y2 , y3 ]
W W= y1' y2' y3'
y1" y2" y3"

The definition can be extended in a similar manner for the Wronskian of

I = [ a, b ] .
n differentiable functions on
 Basic Concepts of Differential Equations 109

d2y dy
y ,y 2
+ P + Qy = 0,
Theorem. If 1 2 are two solutions of dx dx Then their
=
Wronskian
W W= [ y1 , y2 ] 0 if and only if y1 , y2 are linearly dependent.
W [ y1 , y2 ] ≠ 0 y1 , y2 are linearly independent.
Corollary. if and only if

Variation of Parameters.
d2y dy
2
+ P + Qy = F ( x).
We found the solution of the equation dx dx (1)
where P, Q are functions of x, in the previous section by reduction of
order of (1). The solution of (1) can be determined by a procedure known as
the method of variation of parameters. This method can be applied even to
equations of higher order.
− y2 F ( x ) − y F ( x)
u1 =
∫ ' '
∫ 2
dx = dx
y1 y2 − y1 y2 W
y1 F ( x ) y1 F ( x )
u2 =
∫ dx =
∫ dx
y1 y2' − y1' y2 W

where W is the Wronskian of

y1 , y2 .
 CHAPTER

2
FUNDAMENTAL CONCEPTS OF
PARTIAL DIFFERENTIAL
EQUATIONS

CONTENTS
2.1. Introduction..................................................................................... 112
2.2. Classification of Second Order PDE ................................................ 112
2.3. Summary and Discussion................................................................. 141
2.4. Classification of Second Order PDE ................................................ 141
 112 Differential Equations: Theory and Applications

2.1. INTRODUCTION
Many practical problems in science and engineering , when formulated
mathematically, give partial differential equations (often referred to as PDE
) , In order to understand the physical behavior of the mathematical model,
it is necessary to have some knowledge about the mathematical character
properties, and the solution of the governing PDE, An equation which
involves several independent variables usually denoted by
( x, y, z, t ,…) ,
a dependent function u of these variables , and the partial derivatives of
the dependent function u with respect to the independent variables such as
F ( x , y , z , t … . , u x , u y , u z , ut ,……u xx , u yy , ….u xy , …) = 0
(1.1)
is called a partial differential equation. A few well-known examples are:
( i ) ut= k ( u xx + u yy + u zz )
[linear three-dimensional heat equation]
( ii ) uxx + u yy + uzz = 0
[Laplace equation in three dimension ]
( iii ) u xx + u yy + u zz =
0
[linear three-dimensional wave equation]
( iv=) ut u= u x µ u xx
[one linear one-dimensional Burger equation ]
In all these examples, u is the dependent function and the subscripts
denote b partial differential with respect to these variable.
Definition 1. The order of the partial differential equation is the order of
the highest derivative occurring in the equation. Thus the above examples
are partial differential equation of second order, whereas
ut uu xxx + sin x
=
is an example for third order partial differential equation.

2.2. CLASSIFICATION OF SECOND ORDER PDE

The most general linear second order PDE, with one dependent function u
X ℵ( x1 , x2 , ) , n 1,
on a domain Ù of points is
n n

∑a u x + ∑b u xi + F ( u ) =
ij xi j i G
=i , j 1 =i 1

The classification of PDE depend only on the highest order derivative

present.
 Fundamental Concepts of Partial Differential Equations 113

The classification of PDE is motivated by the classification of the

quadratic equation of the form
Ax 2 + Bxy +Cy 2 + Dx + Ey + F =0 (3)
which is elliptic parabolic or hyperbolic according as the discriminant
B 2 − 4 AC is negative , zero or positive . Thus we have the following
second order linear PDE in two variables x and y
Au xx + Bu xy + Cu yy + Du x + Eu y + Fu =
G
(4)
where the coefficients A, B, C ,… may be functions of x and y, however,
for the sake of simplicity we assume then to be constant . Equation is elliptic,
(x , y )
parabolic of hyperbolic at a point 0 0 according as the discriminant.
B 2 ( x0 , y0 ) − 4 A ( x0 , y0 ) C ( x0 , y0 )
is negative, zero or positive. If this
is true at all points in a domain Ù, then is said to be elliptic, parabolic
in that domain. If the number of independent variables is two or three,
a transformation can always be found to reduce the given PDE to a
canonical form ( also called normal form). In general, when the number of
independent variables is greater than 3, it is not always possible to find such
a transformation excepts in certain special cases. the idea of reducing the
given PDE to a canonical form is that the transformed equation assumes a
simple form so that the subsequent analysis of solving the equation is made
easy.

2.2.1. Canonical Forms

Consider the most general transformation of the independent variables
x and y of to new variables ξ ,η , where
ξ= ξ ( x , y ) ,= η= η ( x , y )
(5)
Such that the function ξ and η are continuously differentiable and the Ja-
cobian
∂ ( ξ ,η ) ξ x ξ y
j= =
∂ ( x, y ) η x η y
(
= ξ xη y − ξ y η x ≠ 0 )
In the domain Ù Eq. 4 holds, using the chain rule of partial differential
. the partial derivative becomes
 114 Differential Equations: Theory and Applications

u x u ξ ξ x + uη η x
=

u y u ξ ξ y + uηη y
=

u xx = u ξξ ξ x2 + 2 u ξη ξ xη x + uηη η x2 + uξ ξ xx + uη η xx

u xy= uξξ ξ xξ y + uξη ( ξ xη y + ξ yη x ) + uηη η xη y + uξ ξ xy + uη η xy

u yy = uξξ ξ y2 + 2 uξη ξ y η y + uηηη y2 + uξ ξ yy + uη η yy

Substituting these expression into the original different equation , we get

Auξξ + Buξη + Cuηη + Duξ + Euη + Fu =
G

where
A=Aξ x2 + B ξ x ξ y + Cξ y2

B= 2 Aξ xη x + B ( ξ xη y + ξ yη x ) + 2 C ξ y η y

C=Aη x2 + Bη xη y + Cη y2

D = Aξ xx + Bξ xy + Cξ yy + Dξ x + Eξ y

= , G G (9)
F F=
It may be noted that the transformed equation has the same form as that
of the original equation under the general transformation
Since the classification of Eq. depends on the coefficient A, B and C ,
we can also rewrite the equation in the form
H ( x , y , u, ux , u y )
Au xx + Bu XY + Cu yy =
(10)
It can be shown easily that under the transformation (5) Eq. (10) takes
one of the following three canonical form :
( i ) uξξ − uηη =
∅ ( ξ ,η , u , uξ , uη )
(11a)
uξη = ∅, (ξ ,η , u , uξ , uη )
or in the hyperbolic case
 Fundamental Concepts of Partial Differential Equations 115

( ii ) uξξ + uηη = ∅ (ξ ,η , u , uξ , uη )
in the elliptic case (11b)
( iii ) uξξ = ∅ (ξ ,η , u, uξ , uη ) (11c)
u = ∅ (ξ ,η , u , uξ , uη )
or ηη in the parabolic case
We shall discuss in detail each of these cases separately .
Using Eq. (9) it can also be verified that

(
B 2 − 4 AC = (ξ xη y − ξ yη x ) B 2 − 4 AC )
2

and therefore we conclude that the transformation of the independent

variables does not modify the type of PDE.

2.2.2. Canonical Form for Hyperbolic Equation

2
Since the discriminant B − 4 AC > 0 for the hyperbolic case, we set
= A 0= and C 0 in Eq. (9) which will gave us the coordinates ξ and η that
reduce the given PDE to a canonical form in which the coefficient of
uξξ , uηη
are zero. Thus we have
A = Aξ x2 + Bξ xξ y + C ξ y2 = 0

C = Aη x2 + Bη xη y + Cη y2 = 0

which, in rewriting , become

2
ξx  ξx 
A  + B 0
 +C =
ξ y  ξ y 
2
η  η 
A x 0
 + B  x  + C =
ηy
  ηy 

Solving these equation for

(ξ x / ξ y ) and (η x / η y )
, we get
ξ x − B + B 2 − 4 AC
=
ξy 2A

η x − B − B 2 − 4 AC
=
ηy 2A
 116 Differential Equations: Theory and Applications

2
The condition B − 4 Ac implies that the slopes of the curves
ξ ( x, y ) C=
= 1 ,η ( x, y ) C2 2
are real . Thus, if B > 4 AC , then at any
point
( x, y ) there exist two real directions given by the two roots (12)
along which the PDE (4) reduces of the canonical form. These are called
characteristics equation. Though there are two solutions for each quadratic,
we have considered only one solution for each. Otherwise we will end up
ξ ( x, y ) = c1 , we have
with the same two coordinates. Along the curve
=d ξ ξ=
x dx ξ y dy 0

Hence,
dy ξ 
= − x 
dx  ξy 
 
η ( x, y ) = c2 ,
Similarly, along the curve we have
dy η 
= − x 
dx  ηy 
 
Integrating Eq. (13) and (14), we obtain the equations of family
= ξ ( x, y ) c=
1 and η ( x, y ) c2 ,
of characteristics which are called the
characteristic of the PDE.(4). Now to obtain the canonical form for the given
PDE, we substitute the expressions of ξ and η into Eq. (8) which reduce
to Eq.(11a) Though there are two solutions for each quadratic , we have
considered only solution for each . Otherwise we will end up with the same
two coordinates .
To make the ideas clearer, let us consider the following examples
3u x +10u xy + 3u yy =
0

Comparing with the standard PDE (4), we have

2
A=3, B =
10, C =3, B − 4 AC =64 > 0. Hence the given equation is a
hyperbolic PDE. The corresponding characteristics are :
dy  ξ   − B + B 2 − 4 AC  1
=
− x  = =
dx  ξy   2A  3
   
 Fundamental Concepts of Partial Differential Equations 117

dy  η   − B + B 2 − 4 AC 
=
− x  =
 =3
dx ηy   2A 
   

To find ξ and η , we first solve for y by integrating the above equations.

Thus. we get
1
y =3 x + c1 , y = x + c2
3
Which give the constant as
y − 3 x, c2 =
c1 = y−x/3
Therefore,
1
ξ =y − 3x =c1 ,η =y, − x =c2
3
These are the characteristics lines for the given hyperbolic equation.
In this example, the characteristics are found to be straight lines in the
( x, y ) − plane along which the initial data, impulse will propagate.
To find the canonical equation, we substitute the expression for
ξ and η into Eq. (9) to get

A = Aξ x2 + Bξ xξ y + Cξ y2 = 3 ( −3) +10 ( −3)(1) + 3 = 0

2

B = 2 Aξ xη x + B (ξ xη y + ξ y η x ) + 2Cξ y η y

 1   1 
= 2 ( 3)( −3)  −  +10 ( −3)(1) +1 −   + 2 ( 3)(1)(1)
 3   3 
 10  100 64
=6 =10  −  + 6 =12 − =−
 3 3 3

C 0,=
= D 0,=
E 0=
F 0
Hence, the required canonical form is
64
= uξη 0=
or uξη 0
3
On integration, we obtain
118 Differential Equations: Theory and Applications

u ( ξ=
,η ) f ( ξ ) + g ( η )

where f and g are arbitrary. Going back to the original variables , the
general solution is
u= ( x, y ) = f ( y −3 x ) + g ( y − x / 3
)

2.2.3. Canonical Form for parabolic Equation

2
For the parabolic equation, the discriminant B −4 AC = 0, which can be true
A = 0 in Eq. ( 9 ) .
if B = 0 and A or C is equal to zero n. Suppose we set first
then we obtain
A = Aξ x2 + Bξ x ξ y + Cξ y2 = 0

ξ  ξ 
A x  + B  x 0
 + C =
 ξy
  ξy 
which gives
ξ x  − B ± B 2 − 4 AC 
= 
ξ y  2A 

Using the condition for parabolic case, we get
ξx B
=−
ξy 2A
(15)
ξ = ξ ( x, y )
Hence, to find the function which satisfies Eq. (15) . we set
dy ξ B
=
− x =
dx ξy 2A
and get the implicit solution
ξ ( x , y ) = C1

In fact, one can verify that A = 0 implies B = 0 as follows :

B= 2 Aξ xη x + B (ξ xη y + ξ y η x ) + 2 C ξ y η x
 Fundamental Concepts of Partial Differential Equations 119

2
Since B − 4 AC =
0 the above relation reduces to
B= 2 Aξ xη x + 2 AC ( ξ xη y + ξ y η x ) + 2C ξ y η x

(
2 Aξ x + Cξ y
= )( Aη x + Cη y )
However,
ξx B 2 AC C
=
− =
− =
−
ξy 2A 2A A

We therefore choose ξ in such a way that both A and B are zero. Then
η can be chosen in any way we like as long as it is not parallel to the ξ
coordinates . In other words. We choose η such that the Jacobian of the
transformation is not zero. Thus we can write the canonical equation for
ξ and η into Eq. ( 8 )
parabolic case by simply substituting which reduces to
either of the forms (11c) .
To illustrate the procedure, we consider the following example:
x 2u xx − 2 xyu xy + y 2u yy =
ex
2 2 2 2 2
The discriminant B − 4 AC = 4 x y − 4 x y = 0, and hence the given
PDE is parabolic everywhere. The characteristics equation is
dy ξ B 2 xy y
=
− x = =− 2 = −
dx ξy 2A 2x x

On integration, we have
And hence ξ = xy will satisfy the characterstics eawuation and we can
chosen η = y . To finds the canonical equation , we substitute the expression
ξ and η into Eq. ( 9 ) to get
for
A = A y 2 + Bxy + cx 2 = x 2 y 2 − 2 x 2 y 2 + y 2 x 2 = 0

B = 0. C = y 2 , D = −2 xy

E 0,=
= F 0,=
G ex
Hence, the transformed equation is
 120 Differential Equations: Theory and Applications

ξ
2
y uηη − 2 xyuξ = e η=x
uηη 2ξ uξ + e
2 η
Or
The canonical form is, therefore
2ξ 1
=
uηη u + 2 eξ /η
2 ξ
η η

2.2.4. Canonical Form for Elliptic Equation

2
Since the discriminant B − 4 AC < 0, for elliptic case, characteristics
equation
dy B − B 2 − 4 AC
−
dx 2A
dy B + B 2 − 4 AC
−
dx 2A

Give us complex conjugate coordinates, say ξ and η . Now , we make

another transformation from
(ξ ,η ) to (α , β ) so that
ξ +η ξ −η
=α = ,β
2 2i
which give us the required canonical equation in the form (11b).
To illustrate the procedure, we consider the following example :
u xx + x 2u yy =
0
2
The discriminant B − 4 AC = −4 x 2 < 0. Hence the given PDE is elliptic
. The characteristics equations are
dy B − B 2 − 4 AC −4 x 2
= =
− =
−ix
dx 2A 2
Integration of these equation yields
x2 x2
iy + =c1 , −iy + =c2
2 2
Hence, we may assume that
 Fundamental Concepts of Partial Differential Equations 121

1 2 1
ξ= x + iy ,η = x 2 − iy
2 2
Now, introducing the second transformation
ξ +η ξ +η
=α = .β
2 2i
We obtain
x2
=α = ,β y
2
The canonical form can now be obtained by computing
A = Aα x2 + β α x α y + cα y2 = x 2

B= 2 Aα x β x + B ( ax β y + α y β x ) + 2c (α y β y )= 0

C = Aβ x2 + Bβ x β y + cβ y2 = x 2

D = Aaxx + Baxy + ca yy + Dax + Ea y = 1

E = Aβ xx + Bβ xy + cβ xy + Dβ x + E β y = 0

F 0,=
= G 0
Thus required canonical equation is
x uaa + x 2uββ + uα =
2
0

u
uαα + uββ =
− α
or 2α
Example 1. Classify and reduce the relation
y2 x2
y 2u xx − 2 xyu xy + x 2u yy = ux + u y
x y
to a canonical form and solve it.
Solution. The discriminant of the given PDE is
B − 4 AC = 4 x 2 y 2 − 4 x 2 y 2 = 0
2
 122 Differential Equations: Theory and Applications

Hence the given equation is of a parabolic type, The characteristics

equation is
dy ξ B −2 xy x
=
− x = =2 = −
dx ξy 2A 2y y

Integration gives
x 2 + y 2 = c1.Therefore,ξ = x 2 + y 2 satisfies the
characteristic equation. The η − coordinates can be chosen arbitrary so that
is not parallel to ξ , i.e. the Jacobian of the transformation is not zero . Thus,
we chosen
ξ=x 2 + y 2 ,η =y2
To find the canonical equation we compute
A Aξ x2 + B ξ xξ y + Gξ y2 = 4 x 2 y 2 − 8 x 2 y 2 + 4 x 2 y 2 = 0

B= 0 , C= 4 x 2 y 2 , D= E= F= G= 0
Hence, the required canonical equation is
2
y 2uηη 0=
4x = or uηη 0

To solve this equation, we integrate functions of ξ . Now , going back to

the original independent variables , the required solution is
( ) (
u y 2 f x2 + y 2 + g x2 + y 2
= )
Example. 2 Reduce the following equation to a canonical form:
(1 + x ) u + ( 1 + y ) u
2
xx
2
yy 0
+ xu x + yu y =

Solution. The discriminant of the given PDE is

B 2 − 4 AC = ( )(
−4 1 + x 2 1 + y 2 < 0 )
Hence the given PDE is an elliptic type. The characteristics equation are

dy B − B 2 − 4 AC −4 1 + x 2 1 + y 2 ( 1+ y2 )( )
= =−= = −i
dx 2A 2 1 + x2 (
1 + x2 )
dy B − B 2 − 4 AC 1+ y2
= = i
dx 2A 1 + x2
 Fundamental Concepts of Partial Differential Equations 123

On integrating, we get
ξ= In ( x + x 2 + 1) – i In ( y + y 2 + 1)= c1

η= In ( x + x 2 + 1) – i In ( y + y 2 + 1)= c2
Introducing the second transformation
ξ +η η −ξ
=α = ,β
2 2i
we obtain

α = In ( x + x 2 + 1)

β = In (y + y 2 + 1)
Then the canonical equation for the given PDE is
0
uaa + uββ =

Example. Reduce the following equation top a canonical form and

hence solve it.
u xx − 2 sin xu xy − cos 2 xu yy − cos xu y =
0

Solution. Computing with the general second order PDE (4) ,we have
1, B=
A= − cos 2 x,
−2sin x, C =
0, E =
D= − cos x, F =
0, G =
0
2 2 2
The discriminant B − 4 AC = 4(sin x + cos x) = 4 > 0. Hence the
given PDE is hyperbolic. The relevant characteristics equations are
dy B − B 2 − 4 AC
= − sin x −1
=
dx 2A
dy B + B 2 − 4 AC
= =−1 − sin x
dx 2A
On integration, we get
y cos x − x + c1 , =
= y cos x + x + c2
Thus, we choose the characteristics line as
 124 Differential Equations: Theory and Applications

ξ =+
x y – cos x =c1 , η =− x + y − cos x =c2
In order to find the canonical equation ,we compute
A = Aξ x2 + Bξ x ξ y + Cξ 2y = 0

B = 2 Aξ xη x + B ( ξ xη y + ξ y η x ) + 2Cξ yη y
= 2 ( sin x +1 )( sinx − 1 ) − 4sin 2 x − 2C ξ yη y

C 0,=
= D 0,=
E 0,=
F 0,=
G 0
Thus, required canonical equation is
uξη = 0

Integrating with respect to ξ , we obtain

uη = f (η )

where f is arbitrary. Integrating once again with respect to η , we have

∫ f (η ) dη + g (ξ )
u=

=u ø ( ç)+ g( î )
or
g (ξ )
where is another arbitrary function. Returning to the old variables
x, y the solution of the Given PDE is
u ( x, y=
) ψ ( y − x − cosx ) g ( y + x − cos x)
Example. Find the characteristics of the equation
u xx + 2u xy + sin 2 ( x ) u yy + u y =
0

when it is of hyperbolic type.

2
Solution. The discriminant B − 4 AC = 4 − 4sin 2 x =
4 cos 2 x . Hence
x ≠ ( 2π − 1 ) π / 2 ,
for all the given PDE is of hyperbolic type. The
characteristics equation are
dy B B 2 4 AC
= = cos
dx 2A
On integration, we get
 Fundamental Concepts of Partial Differential Equations 125

x – sin x + sin x , y =
y= x + sin x + c2
Thus, the characteristics equation are
ξ = y − x + sin x, η = y − x − sin x
Example. Reduce the following equation to a canonical form and hence
solve it.
yu xx + ( x + y ) u xy + xu yy =
0

Solution. The discriminant

B 2 − 4 AC =( x + y ) − 4 xy =( x − y ) > 0
2 2

Hence the given PDE is hyperbolic everywhere except along the

line y = x ; whereas on the line v = x. it is parabolic . when y ≠ x, the
characteristics equations are
dy B  B 2 −4 AC
= =
( x+ y )( x – y )
dx 2A 2y
Therefore,
dy dy x
= 1=
,
dx dx y
On integrating, we obtain
x + c1 , y 2 =
y= x 2 + c2
Hence, the characteristics equation are
ξ =y − x,η =y 2 − x 2
These are straight lines and rectangular hyperbolas. The canonical form
can be obtained by computing
A =Aξ x2 + Bξ xξ y + Cξ y2 =y − x – y + x =0 , B =− 2 ( x − y )
2

0, D=
C= 2( x y ) , F =
0 , E =− 0
G=

Thus, the canonical equation for the given PDE is

−2 ( x − y ) uξη + 2 ( x − y ) uη =
2
0
 126 Differential Equations: Theory and Applications

−2ξ 2 uξη + 2 ( −ξ ) uη =
0
or
∂  ∂u 
ξ=
uξη + uη ξ
=  0
or ∂ξ  ∂η 
Integration yields
∂u
ξ = f (η )
∂η
Again integrating with respect to η , we obtain
1
u= ∫ f ( η ) dη + g ( ξ )
ξ
Hence,
1
u=
y−x
( ) ( )
∫ f y 2 − x2 d y 2 − x2 + g ( y − x )

is then general solution.

Example. Classify and transformation the following equation to a
canonical form:
sin 2 ( x ) u xx + sin ( 2 x ) u xy + cos 2 ( x ) u yy =
x

Solution. The discriminant of the given PDE is

2
B − 4 AC= sin 2 2 x − 4sin 2 x cos 2 x= 0
Hence, the given equation is of parabolic type. The characteristics
equation
dy B
= = cot x
dx 2 A
Integration gives
=y In sin x + c1
Hence, the characteristic equation are :
ξ= y − In sin x,η =y
η is chosen in such a way that the Jacobian of the transformation is nonzero.
Now the canonical from can be obtained by computing
 Fundamental Concepts of Partial Differential Equations 127

A 0=
= , C cos 2 x ,=
, B 0= D 1=
, E 0 ,=
F 0=
,G x

Hence, the canonical equation is

cos 2 ( x ) uηη + uξ =
x

or   ( )
1 −e 2(η – ξ )  uηη = sin −1 eη − ξ − uξ

Example. Show that the equation

2N 1
u xx + ux = utt
x a2
where N and a are constants, is hyperbolic and obtain its canonical form.
Solution. Comparing with the general PDE (4) and replacing y by t ,
we have= A 1,=B 0, C = −1 / a 2 , D = 2 N / x , and E= F= G= 0 . The
discriminant B 2 − 4 AC = 4 / a 2 > 0 . Hence the given PDE is hyperbolic.
The characteristics equation are
dt B  B 2 − 4 AC 4 /a 2 1
= = = 
dx 2A 2 a
Therefore,
dt 1 dt 1
− , =
=
dx a dx a
On integration, we get
x x
− + c1 , t = + c 2
t=
a a
Hence, the characteristic equations are
ξ=
x + at , η =
x − at
The canonical form can be obtained by computing
A = Aξ x2 + Bξ x ξt + Cξt2 = 0 ,

= 2 Aξ xη x + B ( ξ xηt + ξtη x ) + 2Cξt η=

B t 4,

2N 2N
C = 0 , D = Dξ x + Eξt = , E = Dη x + Eηt =
x x
 128 Differential Equations: Theory and Applications

Thus, the canonical equation for the given PDE is

2N
4uξη +
x
( uξ + uη ) =
0

Example. Transform the following differentiate equation to a canonical

form:
u xx + 2u xy + 4u yy + 2u x + 3u y =
0
2
Solution. The discriminant B − 4 AC = −12 < 0. Hence, the given PDE
is elliptic. The characteristics equations are
dy B − B 2 − 4 AC
= =+1 − i 3
dx 2A
dy B + B 2 − 4 AC
= =+1 + i 3
dx 2A
Integration of these equations yields

( )
+ 1 − i 3 x +c1 , y =
y= (
+ 1 + i 3 x + c2 )
Hence, we may take the characteristics equations in the form

( ) (
ξ = y − 1 − i 3 x ,η = y − 1+ i 3 x )
In order to avoid calculations with complex variables , we introduced
the second transformation
ξ +η ξ −η
=a = ,β
2 2i
Therefore.
y − x ,β =
a= ξ +η
The canonical form can now be obtained by computing
A = Aax2 + Bax a y + Ca y2 = 3

= 2 Aax β x + B ( ax β y + a y β x ) + 2Ca y β=
B y 0

C = Aβ x2 + B β x β y + C β y2 = 3
 Fundamental Concepts of Partial Differential Equations 129

D = Aaxx + Baxy + Ca yy + Dax + Ea y = 1

E = Aβ xx + B β xy + C β yy + Dβ x + E β y = 2 3

Thus the required canonical form is

3uaa + 3uββ + ua + 2 3 uβ =
0

1
− ( ua + 2 3uβ =
uaa + uββ = 0
or 3

2.2.5. Adjoint Operators

Let
Lu = ∅ (16)
where L is a differentiate operator given by
dn d n −1
=L a0 ( x ) n + a1 ( x ) n −1 +…+ an ( x )
dx dx
*
One way of introducing the adjoint differential operators L associated
with L is to from the product vLu and integrate it over the interval of
interest . Let
B B

∫vLu=
dx [ .] BA + ∫uL * v dx
A A (17)
*
which is obtained after repeated integration by parts. Here L is the operator
adjoint to L, where the function u and v are completely arbitrary except
*
that Lu and L v should exist.

Example. Let
(
Lu = a ( x ) d 2 u / dx 2 ) + b ( x )( du / dx ) + c ( x ) u ;
*
construct its adjoint L .
Solution. Consider the equation
B B
 d 2u du 
∫AvLu=dx ∫A 
v  a ( x ) + b ( x ) + c ( x ) u  dx
dx 2 dx 
 130 Differential Equations: Theory and Applications

d 2u
B B B
du
∫ ( av )
A
dx 2
dx + ∫ ( bv )
A
dx
dx ∫ ( cv ) u dx
A

d 2u
B B
d
∫A ( av ) dx 2 du = ∫A ( av ) dx u dx
'
( )
B
=  u va  − ∫ ( av ) u ' dx
' B '
A
A

B B
u av  −  u ( av )  + ∫u ( av ) dx
' B ' ''
=
A  A
A

B B
du
∫A ( bv=
) dx  u ( bv )  A − ∫u ( bv ) dx
B '

dx A

B B

∫ ( cv ) u dx = ∫u ( cv ) dx
A A

Therefore,
B B B

∫ vLu dx= u′ ( av ) − u ( av ) u ( bv ) A + ∫u ( av ) − ( bv ) + ( cv ) dx

' '' '

A A

Comparing this equation with Eq. (17), we get

L* v =( av ) − ( bv ) + ( cv ) =av′′ + ( 2a′ − b ) v′ + ( a′′ − b′ + c ) v
'' '

Therefore,
d2 d
L* = a 2
+ ( 2a′ − b ) + ( a′′ − b′ + c )
dx dx
Consider the partial differential equation
L ( u ) = Au xx + 2 Bu xy + Cu xy + Du x + Eu y + Fu= φ

which is valid in a region S of the xy − plane , where A, B, C ,….φ are

functions of x and y .In addition. Linear boundary conditions of the general
form
 Fundamental Concepts of Partial Differential Equations 131

au + β u x =
f

are prescribed over the boundary curve ∂S , the boundary of S , then L

*

is called the adjoint operator. In general, a second order linear partial

differential operator L is denoted by
∂ 2u n n
∂u
L (=
u) ∑ Aij
+ ∑Bi + Cu
∂xi ∂x j i 1 ∂x1
=i, j 1 =
(19)
It’s adjoint operator is defined by
∂2 n n
∂
∑ L* (=
v)
∂xi ∂x j
( Aij v ) − ∑ ( Bi v ) + Cv
i 1 ∂xi
=i, j 1 =
(20)
Aij ò C and Bi ò C1.
2
Here it is assumed that For any pair of functions
u , v ò C 2 , it can be shown that

∂  n  ∂u
n
∂v   n ∂A 
vLu (=
u ) − uL* ( v )
∑  ∑Aij  v  + uv  Bi − ∑
ij
−u  
∂x  j 1  ∂x j
=i 1 = ∂x j  =j 1 ∂x j   (21)
This is known as Lagrange’s identity.
Example. Construct a adjoint to the Laplace operator given by
L ( u=
) uxx + u yy (22)
Solution. Comparing Eq. (22) with the general linear PDE (19) , we
have
= A22 1. From Eq. ( 20 ) the adjoint of (22) is given by
A11 1,=

∂2 ∂2
L* ( v ) = ( ) 2 (v) =
v + vxx =
v yy
∂x 2 ∂y
Therefore,
L* ( u=
) uxx + u yy
Hence, the Laplace operator is a self-adjoint operator
Example. Find the adjoint of the differentiate operator
L ( u=
) uxx + ui (23)
 132 Differential Equations: Theory and Applications

Solution. Comparing Eq. (23) with the general linear PDE (19), we have
A11 1,=
= A22 1.
From Eq (20), the adjoint of (22) is given by
∂2 ∂2
L* ( v ) ( v ) + (v) =
vxx + v yy
∂x 2 ∂y 2
Therefore,
L ( u=
*
) uxx + ut
It may be noted that the diffusion operator is not a self-adjoint operator.

2.2.6. Riemann’s Method

We have noted with interest that a linear second order PDE
L ( u ) = G ( x, y )
2
is classified as hyperbolic if B > 4 AC , and it has two families of real
characteristics curves in the xy − plane whose equations are

Here
(ξ ,η )
are the natural coordinates for the hyperbolic system. In the
=
xy − plane , the curves ξ ( x, y ) c1=
and η ( x, y ) c2
are the characteristics
of the given PDE as shown in Fig 1(a). while in the ξη − plane , the curves
=ξ c=
a and η c2
are families of straight lines parallel to the axes as shown
in Fig 1(b)
A linear second order partial differential equation in two variables , once
classified as a hyperbolic equation, can always be reduce to the canonical
form
∂ 2u
= G ( x, y , u , u x , u y )
∂x∂y
In particular, consider an equation which is already reduced to its
canonical form in the variables x, y :
∂ 2u ∂u ∂u
L ( u )= = a + b + cu= F ( x, y )
∂x∂y ∂x ∂y (24)
 Fundamental Concepts of Partial Differential Equations 133

where L is a linear differential operator and a, b c, F are functions of

x and y only and are differentiable in some domain IR.
v ( x, y )
Let be an arbitrary function having continuous second order
*
partial derivatives. Let us consider the adjoint operator L of L defined by
∂ 2v ∂ ∂
L* ( v ) = − ( av ) − ( bv ) + cv
∂x ∂y ∂x ∂y (25)
Now we introduce
∂v ∂u
M auv – u ,=
= N buv + v
∂y ∂x (26)
Then
M x + N=
y u x ( av ) + u ( av ) x − u x v y − uvxy + u y ( bv ) + u ( bv ) y + v y u x + vu xy

Adding and subtracting cuv, we get

 ∂ 2v ∂ ∂   ∂ 2u ∂u ∂u 
Mx + Ny =
−u  − ( av ) − ( bv ) + cv  + v  + a + b + cu 
 ∂x∂y ∂x ∂y   ∂x∂y ∂x ∂y 
i.e,
vLu − ul *v = M x + N y
(27)
This is knows as Lagrange identity which will be used in the subsequent
*
discussion . The operator L is a self – adjoint if and only if L = L . Now we
shall attempt to solve Cauchy’s problem which is described as follows . Let
L ( u ) = F ( x, y )
(28)
with the condition (Cauchy data)
a curve I n the xy − plane;

This is u , and its normal derivatives are prescribed on a curve à which

is not a characteristics line.
Let à be a smooth initial curve which is also continuous as shown in
Fig 2 since Eq. (24) is in canonical form , x and y are the characteristics
134 Differential Equations: Theory and Applications

coordinates. We also assume that line tangent to Ãis nowhere parallel to the
coordinates axes.
P (ξ ,η )
Let be a point at which the solution to the Cauchy problem is
sought. Let us draw the characteristics PQ and PR through P to meet the
u, ux u y
curve Ãat Q and R. We assume that are prescribed along Ã. Let
∂IR be a closed contour PQRP bounding IR. Since Eq. (28) is already in
canonical form, the characteristics are lines parallel to x and y axes . using
Green’s theorem ,we have
. .
∫ ∫ ( M x + N y ) dxdy= ∮ ( M dy − N dx )
R ∂IR (29)
where ∂IR is the boundary of IR. Applying this theorem to the surface
integral of Eq.(27). We obtain
. .

∫ ( M dy − N dx=) ∫ vL ( u ) − uL ( v ) dx dy

*

∂IR IR (30)
In other words,
. . .

∫ ( Mdy − N dx ) + ∫ ( M dy − N dx ) =
∫ ∫ vL ( v ) − uL ( v )  dx dy *

à RP R

=
Now using the fact that dy 0=
on PQ and dx 0 on PR, we have
. . . .

∫ ( M dy − Ndx ) + ∫ M dy − ∫ N dx =
∫ ∫ v L ( u ) − uL ( v )  dx dy *

à RP PQ IR
(31)
From Eq. (26), we find that
. Q Q

∫=
N dx
PQ
∫buv dx + ∫vux dx
P P

Integrating by parts the second term on the right –hand side and grouping.
the above equation becomes
. Q

∫ N dx = [ nv ] + ∫u ( bv − v ) dx
Q
P x
PQ p
 Fundamental Concepts of Partial Differential Equations 135

Substituting this result into Eq. (31), we obtain

. .

[uv ] p =[uv ]Q + ∫ u ( bv − vx ) dx − ∫ u ( av − v y ) dy
PQ RP

. .
− ∫ ( M dy − N dx ) + ∫ ∫ vL ( u ) − uL* ( v )  dx dy
à IR (32)
v ( x, y;, ξ ,η )
Let us choose to be a solution of the adjoint equation
L* ( v ) = 0
(33)
And at the same time satisfy the following condition :
( i ) vx bv=
= whenu η , i.e , on PQ
(34 a)
( ii ) v y = av
when x = ξ , i.e., on PR (34 b)
( iii=
)v 1 when
= x ξ=
,y η
(34 c)
v ( x, y; ξ ,η )
We call this function as the Riemann function or the
L ( u ) = F , Eq. ( 32 )
Riemann- Green function. Since reduces to
. .
u ] p [uv ]Q − ∫ u ( av − v y ) dy − v ( bu + u x ) dx  + ∫ ∫ ( vF ) dx dy
[=
à IR (35)
This is called the Riemann-Green solution for the Cauchy problem
described by Eq.(28) when
u and u x are prescribed on Ã. Equation (35) can
also be written as
. . .

] p [uv ]Q − Ã∫uv ( a dy − b dx ) + ∫ ( uv y dy − vux dx ) + ∫ ∫ ( vF ) dx dy

[u =
à à IR (36)
This relation gives us the value of at a point p when
u and u x are

prescribed on à . But when

u and u x are prescribed on Ã, we obtain
. . .

[=
u]p [uv ]R − ∫uv ( a dy − b dx ) − ∫ ( uvx dx +vu y dy ) + ∫ ∫ ( vF ) dx dy
à à IR (37)
 136 Differential Equations: Theory and Applications

By adding Eq. (36) and (37), the value of u at a point p when

u and u x

are prescribed on Ã. But when

u and u x are prescribed on Ã, we obtain

. . . .

[u ] p
=
1
2
{ } Ã
1
2 Ã
1
[uv ]Q + [uv ]R − ∫uv ( a dy − b dx ) − ∫u ( vx dx − v y dy ) + ∫v ( ux dx − u y dy ) + ∫ ∫ ( vF ) dx dy
2Ã IR

(38)
Thus, we can see that the solution to the Cauchy problem at a point
(ξ ,η ) depends only on the Cauchy data on Ã. the Knowledge of the
Riemann-Green function therefore enables us to solve Eq. (28) with the
Cauchy data prescribed on a non characteristics curve. Thus, we can see
that the solution to the Cauchy problem at a point
(ξ ,η ) depends only
on the Cauchy data on Ã. the Knowledge of the Riemann-Green function
therefore enables us to solve Eq. (28) with the Cauchy data prescribed on a
non characteristics curve.

Example. Let
(
Lu = a ( x ) d 2u / dx 2 ) + b ( x )( du / dx ) + c ( x ) u;
*
construct its adjoint L .
Solution. Consider the equation
B B
 d 2u du 
∫vLu=
A
dx ∫A  dx 2 + b ( x ) dx + c ( x ) u  dx
v a ( x )

d 2u
B B B
du
∫ ( av )
A
dx 2
dx + ∫ ( bv ) dx ∫ ( cv ) u dx
A
dx A

d 2u
B B
d
∫ ( av )
A
dx 2
du = ∫ ( av ) ( u ′ ) dx
A
dx

B
= [u ′va ] A − ∫ ( av ) u ′ dx
B '

B B

[u′av ]A − u ( av )  A + ∫u ( av ) 'dx

B ' '
=
A
 Fundamental Concepts of Partial Differential Equations 137

B B
du
( ) ( )  A − ∫u ( bv ) dx
B '
∫A dx 
=
bv dx  u bv 
A

B B

∫ ( cv ) u dx = ∫u ( cv ) dx
A A

Therefore,
B B B

∫ vLu dx= u′ ( av ) − u ( av ) u ( bv ) A + ∫u ( av ) − ( bv ) + ( cv ) dx

' '' '

A A

Comparing this equation with we get

L* v =( av ) − ( bv ) + ( cv ) =av′′ + ( 2a′ − b ) v′ + ( a′′ − b′ + c ) v
'' '

Therefore,
d2 d
L = a 2 + ( 2a′ − b ) + ( a′′ − b′ + c )
*

dx dx

Consider the partial differential equation

L ( u ) = Au xx + 2 Bu xy + Cu xy + Du x + Eu y + Fu= φ

which is valid in a region S of the xy − plane , where A, B, C ,….φ are

functions of x and y .In addition . Linear boundary conditions of the
general form
au + β u x =
f

Are prescribed over the boundary curve ∂S , the boundary of S , then

L* is called the adjoint operator. In general , a second order linear partial
differential operator L is denoted by
∂ 2u
n n
∂u
L (=
u ) ∑ Aij + ∑Bi + Cu
∂xi ∂x j i 1 ∂x1
= i, j 1 =

It’s adjoint operator is defined by
∂2 n n
∂
∑ L* (=
v)
∂xi ∂x j
( Aij v ) − ∑ ( Bi v ) + Cv
i 1 ∂xi
=i, j 1 =

138 Differential Equations: Theory and Applications

Aij ò C 2 and Bi ò C1.

Here it is assumed that For any pair of functions
2
u , v ò C , it can be shown that

∂  n  ∂u n
∂v   n ∂A 
vLu (=
u ) − uL ( v ) ∑  ∑Aij  v
*
 + uv  Bi − ∑
ij
−u  

∂x  j 1  ∂x j ∂x j
=i 1 =  =j 1 ∂x j  
Example. Verify that the Green function for the equation
∂ 2u 2  ∂u ∂u 
+  + = 0
∂x∂y x + y  ∂x ∂y 

x 2 on y x, is given by
u / ∂x 3=
Subject to u = 0 , ∂=
( x + y ) {2 xy + (ξ − η )( x − y ) + 2ξη}
v ( x, y; ξ ,η ) =
(ξ + η )
3

and obtain the solution of the equation in the form

(
u = ( x − y ) 2 x 2 − xy + 2 y 2 )
.
Solution. In the given problem,
∂ 2u 2 ∂u 2 ∂u
L (u ) = + + = 0
∂x∂y x + y ∂x x + y ∂y (39)
Comparing this equation with the standard canonical hyperbolic equation
(24), we have
2
a= b= . C= 0 , F= 0
x+ y
L* ( v ) = 0
Its adjoint equation is where
2
∂v ∂  2v  ∂  2 v 
L* ( v ) = −  −  
∂x∂y ∂x  x + y  ∂y  x + y  (40)
Such that
( i ) L * v = 0 thought the xy − plane
∂v 2
( ii ) = v on PQ, i.e, om y η
∂x x+ y
 Fundamental Concepts of Partial Differential Equations 139

∂v 2
= v on PR, i.e., on x ξ
( ) ∂y
iii x+ y (41)
( iv ) v = 1 at P (ξ ,η )
If v defined by
x+ y
v (=
x , y ; ξ ,η )  2 xy + (ξ − η ( x − y ) + 2ξη 
(ξ + η )
3

(42)
Then

∂v x+ y  2 xy + (ξ − η )( x − y ) + 2ξη 
=  2 y + (ξ − η )  +  
(ξ + η ) (ξ + η )
3 3
∂x  

∂v 1
=  4 xy = 2 x 2 − 2 y (ξ − η ) + 2ξη 
(ξ + η )
3
∂y
or (43)
And
∂ v 4( x + y)
2
=
∂x∂y (ξ + η )3
(44)
∂v 1
=  4 xy = 2 x 2 − 2 y (ξ − η ) + 2ξη 
(ξ + η )
3
∂y
(45)
Using the results described becomes
∂ 2v 2  ∂v ∂v  4v
L* ( v ) −  + +
 ∂x ∂y  ( x + y )
2
∂x∂y x + y

4( x + y) 2
= −  ( )
 4 xy + 2 x 2 + y 2 

(ξ + η ) ( x + y )(ξ + η )
3 3

4( x + y) 4( x + y)
L* ( v ) = − =0
(ξ + η ) (ξ + η )
3 3

or
Hence condition
( i ) is satisfied . Also, on y = η.
∂v 1
=  4 xη + 2η 2 + 2 x (ξ − η ) + 2ξη 
(ξ + η )
3
∂x
140 Differential Equations: Theory and Applications

However,
p
 2uv ∂u 
P
2uv
P
∂v
∫Q  x + y + v ∂x  dx= ∫Q x + y dx + ( uv )Q − Q∫u ∂x dx
P

=
Now, using the condition u 0=
on y x, becomes
. Q
 2uv ∂u 
∫ ∫ vL ( u ) − uL ( v )  dx dy =∫ 
*
− v  dx
IR P
x+ y ∂x 
Q
2uv  2uv ∂u 
P
−∫ dx − ( uv ) p + ( uv )Q + ∫  + v  dx
Q
x+ y R
x+ y ∂x 

2uv
P R
 ∂u 
∫Q x + y dy − ∫P  u ∂x  dy
-
Also, using conditions (ii) –(iv) of the above equation simplifies to
Q
∂u
(u ) p
= ( uv )Q − ∫v dx
R
∂x
Now using the given condition, viz.
∂u
= 3 x 2 on RQ
∂x
We obtain
 2 x  2 x 2 + 2ξη  
Q

( u ) p ( uv )Q − 3 ∫x  
=   dx
2

(ξ + η ) 
3
R 
η
12
=
− 3 ∫
x 5 + x 3 ξη dx ( )
(ξ + η ) ξ
12 1 6 1 4 
=
− ( )
 6 η − ξ + 4 ξη η − ξ 
6 4
( )
(ξ + η )
3

ξ 2 −η 2 
3  (
= 2 ξ 4 + ξ 2η 2 + η 4 ) + 3ξη (ξ 2 + η 4 
(ξ + η )
 Fundamental Concepts of Partial Differential Equations 141

(
= (ξ − η ) 2ξ 2 − ξη + 2η 2 )
Therefore,
(
u ( x, y ) = ( x − y ) 2 x 2 − xy + 2 y 2 )
hence the result.

2.3. SUMMARY AND DISCUSSION

Many practical problems in science and engineering, when formulated
mathematically, give partial differential equations (often referred to as
PDE ). An equation, which involves several independent variables usually,
denoted by
( x, y, z, t ,…) , a dependent function u . of these variables,
and the partial derivatives of the dependent function u with respect to the
independent variables such as
F ( x , y , z , t … . , u x , u y , u z , ut ,……u xx , u yy , ….u xy , …) = 0

is called a partial differential equation. A few well-known examples are:

( i ) ut= k ( u xx + u yy + u zz )
[linear three-dimensional heat equation]
( ii ) uxx + u yy + uzz 0
=
[Laplace equation in three dimension ]
Order of PDE. The order of the partial differential equation is the order of
the highest derivative occurring in the equation. Thus the above examples
are partial differential equation of second order, whereas
ut uu xxx + sin x
=
is an example for third order partial differential equation.

2.4. CLASSIFICATION OF SECOND ORDER PDE

The classification of PDE depend only on the highest order derivative
present. The classification of PDE is motivated by the classification of the
quadratic equation of the form
Ax 2 + Bxy +Cy 2 + Dx + Ey + F =0
which is elliptic parabolic or hyperbolic according as the discriminant
B 2 − 4 AC is negative , zero or positive . Thus we have the following
second order linear PDE in two variables x and y
 142 Differential Equations: Theory and Applications

Au xx + Bu xy + Cu yy + Du x + Eu y + Fu = G

where the coefficients A , B , C ,… may be functions of x and y, however,
for the sake of simplicity we assume then to be constant . Equation is elliptic
(x , y )
, parabolic of hyperbolic at a point 0 0 according as the discriminant.
B 2 ( x0 , y0 ) − 4 A ( x0 , y0 ) C ( x0 , y0 )
is negative , zero or positive. If this
is true at all points in a domain Ù, then is said to be elliptic, parabolic in
that domain . if the number of independent variables is two or three , a
transformation can always be found to reduce the given PDE to a canonical
form ( also called normal form).

Adjoint Operators
Let
Lu = ∅
where L is a differentiate operator given by
dn d n −1
=L a0 ( x ) + a1 ( x ) +…+ an ( x )
dx n dx n −1
*
One way of introducing the adjoint differential operators L associated
with L is to from the product vLu and integrate it over the interval of
interest . Let
B B
dx [ .] + ∫uL
∫vLu=
B
A
*
v dx
A A
*
which is obtained after repeated integration by parts. Here L is the operator
adjoint to L, where the function u and v are completely arbitrary except
*
that Lu and L v should exist.
∂  n  ∂u
n
∂v   n ∂A 
vLu (=
u ) − uL ( v ) ∑  ∑Aij  v
*
 + uv  Bi − ∑
ij
−u  

∂x  j 1  ∂x j ∂x j
=i 1 =  =j 1 ∂x j  
This is known as Lagrange’s identity
 CHAPTER

3
APPLICATION OF
DIFFERENTIAL EQUATIONS IN
MECHANICS

CONTENTS
3.1. Introduction..................................................................................... 144
3.2. Projectile Motion ............................................................................ 161
3.3. Summary and Discussion................................................................. 186
144 Differential Equations: Theory and Applications

3.1. INTRODUCTION
The motion of a particle along a straight line is called rectilinear motion.
Let the particle start from point O along a line. We take the line as x − axis
. Let after time t particle be at point P at a distance x from O. Let r be the
position vector of point p w.r.t. origin O. then

= = xiˆ
r OP

dr dx ˆ
=
v = i
dt dt
v a = a , then
Let v = and
dx
v=
dt

d 2x
a= 2
dt
Also,
dv dv dx dv dv
=
a = = v =v
dt dx dt dx dx

3.1.1. Motion with Constant Acceleration

Let the particle start form O with velocity u at time t = 0 with constant
acceleration a. Let the particle be at distance x from O after time t. Then
dv
a=
dt where a is constant.
Integration will yield
v= at + A

Since velocity v = u at t = 0, therefore, A = u , thus

v = at + u = u + at
 Application of Differential Equations in Mechanics 145

dx
v= ,
Now, since dt therefore eq. (1) gives
dx
= u + at
dt
By integrating
 t2 
x=
ut + a   + B
2
=
Using x 0,=
at t 0, we get, B = 0, hence

at 2
x= ut +
2
dv
a =v ,
Also, dx where a is constant, then
ax = v dv
Integrating
v2
= ac + C
2
u2
,
Since v = u, at x = 0, therefore, C = 2 and hence
v2 − u 2 =
2ax
The equation and are called equation of motion. If the particle moves
with constant retardation the acceleration is taken as negative.

3.1.2. Vertical Motion under Gravity

For a failing body, the acceleration is constant. It is called acceleration due
to gravity and is denoted by ‘g’. In FPS system g = 32 ft / sec .In CGS
2

system g =981 cm / sec and in MKS system g = 9.81 m / sec . If the body
2 2

is projected vertically upward then acceleration g is negative. For falling

bodies equation of motion is
v= u + gt
 146 Differential Equations: Theory and Applications

1 2
x= ut + gt
2
v2 − u 2 =
2 gx
Example
2gh
A particle is projected vertically upward with velocity and
another is let fall from a height h at the same time. Find the height of the
point where they meet each other.
Solution.
Let both particles meet at point P at height x. Then for the first particle
1 2
x= ut − gt
2
u = 2 gh , then
Put
1
=x 2 ght − gt 2
2 (1)
For the second particle

x= ut + gt

1 2
x= ut + gt
2

Put u= 0 , x= h − x, then
1 2
h− x =0+ gt
2 (2)

Using eqs.
(1) and ( 2 ) , we get
u = 2 ght

or t = 2 ght
Thus equation (1) gives
 Application of Differential Equations in Mechanics 147

 h  1  h2  h 3h
x = 2 ght   − g   =h − =
 2 gh
  2  2 gh  4 4

Example
A particle is projected vertically upward. After time t another particle
is sent up from the same point with the same velocity and meets the first
at height h during the downward flight of the first. Find the velocity of the
projection.
Solution
Let u be the velocity of the projection and v be the velocity at height h,
then
v2 − u 2 =
−2 gh

or
=v u 2 − 2 gh
Now, the time taken by the first particle from height h to the highest
point and back to the height h is t, therefore, the time taken from the highest
point is t / 2 .
Velocity at highest point = o
Velocity at height h = v
Here we use the formula
v= u − gt
v 0,=
Put = , t t/2, then
u v=
gt
0= v −
2
From (1) and (2)
gt
u 2 − 2 gh =
2

g 2t 2
u 2 − 2 gh =
4
g 2t 2
u 2 2 gh +
=
4
148 Differential Equations: Theory and Applications

8 gh + g 2t 2
u=
4

3.1.3. Distance Traveled in nth Second

Let 1
x and x b
2 e the distance travelled in the first n and n – 1 seconds
respectively, then from the equation of motion
1
x= ut + at 2
2
1
x=
1 un + an 2
Then 2
1
x2= u ( n − 1) + a ( n − a )
2

And 2

Distance travelled in the nth second =

x1 − x2

a 2  a 2
x1 − x2 = un + n − u ( n − a ) + ( n − 1) 
2  2 
a 2  a 
un +
2  2
(
n −  un − u + n 2 − 2n + 1 

)
=
1
u + a ( 2n − 1)
= 2

3.1.4. Motion with variable Acceleration

( i ) Time Dependent Acceleration
As the acceleration is time dependent, then it is function of time
Therefore we can write:
a = f (t )

dv
= f (t )
or dt
or dv = f ( t ) dt
 Application of Differential Equations in Mechanics 149

Integrating
∫ dv =∫ f ( t ) dt

or=v g (t ) + A
∫ f ( t ) dt ,
where g(t) = and A is constant of integration, which gives the
velocity. Also since
dx
v= ,
dt
Therefore
dx
= g (t ) + A
dt

= dx g ( t ) dt + Adt
or
By integrating, we get
∫ g ( t ) dt + At + B
x=

where B is another constant which gives the distance travelled by the

particle . By using the initial conditions we can find the constants Aand B .
Example 3
Discuss the motion of the particle moving in a straight line if it stars
from rest t = 0 and its acceleration is equal to accost + bsint.
Solution
Give acceleration is
=a accost + b sint
By integrating, we have
v = asint − b cost + A
=
Since v 0= at t 0 , therefore A = b, thus
=C asint – bcost + b
dx
= asint − bcost + b
Also, v = dt
Again integrating
150 Differential Equations: Theory and Applications

−acost – bsint +bt + B

x=
where B is constant of integration.
Since x = 0 , at t = 0, therefore B = a. Thus
−acost – bsint + bt + a
x=
= a (1 − cost ) + b ( t − sint )

(ii) Velocity Dependent Acceleration

Let the acceleration a is function of velocity v, then
a = f (v
dv
a =v , therefore
Since dx
dv
v = dx
f (v)

Integrating
dv
∫v =
∫ dx
f (v)

or =x g (v) + A

dv
g (v) = ∫ v and
f (v)
where A is constant of integration which gives the
distance travelled by the particles.
Again consider a = f(v). since
dv
a= ,
dt
Therefore
dv
= dt or
f (v)
dv
∫ =
∫ dt
f (v)
or
 Application of Differential Equations in Mechanics 151

= t h (v) + B
or
dv
h (v) = ∫ and B is
f (v)
where constant of integration and gives the time .
Example: A particle moves in a straight line with an acceleration kv’ . If
its initial velocity is u, find the velocity and the time spend when the particle
has travelled a distance of x.
Solution.
Given that
a = kv3
dv
a=v ,
Since dx therefore
dv
v = kv3
dx
or v −2 dv = k dx
or ∫ v −2 dv =
kdx
1
or v −= kx + A
1
v = u , x = 0 at t = 0, therefore A = −
Given u
1 1 u
= − kx or c =
Thus v u 1 − kux
dx
v = , then
Again since dt
dx u
=
dt 1 − kux

or ( 1 − kux ) dx =
u dt , thus

∫ ( 1 − kux ) dx =u ∫ dt

 
x − ku  =ut + B
 
152 Differential Equations: Theory and Applications

=
Using x 0,=
at t 0, we get
= B 0. Hence

 x2  x
x − ku   = ut or t = ( 2 − kux )
 2  2u
Example. A Particle starts with a velocity u and move in a straight line.
If the suffers a retardation equal to the square of velocity, find the distance
travelled by the particle in time t.
2
Solution. Given retardation = v , therefore
a = −vt 2
We know that
dv
v = a , there fore
dx
dv
= −v
dx
dv
or = −dx
v
dv
or ∫ =− ∫ dx
v
or In v =− x + A
where a is constant of integration .
Using
A = Inu , thus

u u u
ln   x=
= or e=
x
or v
v v ex
dx
v= ,
Since dt therefore
dx u
=
dt e x
or e x = u dt , then
 Application of Differential Equations in Mechanics 153

e=
x
ut + B
where B is constant of integration .
=
Using x 0=
at t 0, we get B =1 . Hence
in ( ut + 1)
ut + 1 or x =
ex =

(iii) Distance Dependent Acceleration

Let the acceleration be a function of distance x . Then a = f(x).
dv
v = f ( x ) dx
Since dx
∫ f ( x ) dx
or ∫ vdv =

v2
= g ( x) + A
2
∫ f ( x ) dx and A
where g(x) is the constant of integration
± 2g ( x) + 2 A
or v =

We know that
dx
= v, therefore
dt
dx
± 2g ( x) + 2 A
=
dt
dx
dt = ±
2g ( x) + 2 A

Integrating, we get
dx
t =± ∫ +B
2g ( x) + 2 A

where B is constant of integration.

 154 Differential Equations: Theory and Applications

Example
Discuss the motion of the particle moving in a straight line if it starts at
a distance ‘a’ from a point O and move with an acceleration equal to times
its distance from the O.

Solution
Let x be the distance of particle from O, then
Acceleration = a= µ x

Since
or ∫ vdv =µ ∫ xdx
v2 x2
= µ +A
2 2
v 0,=
Given= =
x a at t 0, so that

a2
A = −µ , thus
2
v2  x2 a2 
= µ − 
2  2 2 

± µ x2 − a2
or v =
dx
Take v =µ x 2 − a 2 , since v = , then
dt
dx
∫ µ ∫ dt
=
x − a2
2

x
or cosh −1  =
 µ ∫ dt
a

Using= =
x a at t 0,=
B 0, thus

x
cosh −1   = µ t
a
 Application of Differential Equations in Mechanics 155

or x = a cosh( µ t )
dx
v=− µ x 2 − a 2 , since v = , then
If we take dt
dx
∫ = − µ ∫ dt + C
x − a2
2

x
− µ t +C
cosh −1   =
a
x a=t 0, C= 0, thus
Using =
x
cosh −1   = − µ t
a
( )
or = a cosh − µ t = a cosh ( µt )
3.1.5. Graphical Solution of Rectilinear Problem
y = f ( x)
Consider the graph of the curve in the xy − plane, then

dy
(a) =
dx slope of the tangent at any point
(b ) Area bounded by the curve=y f ( x ) , from
b
= =
x a to =
x bs is ∫ ydx
(c) (i) Velocity- Time Curve
a

Consider the graph of velocity time graph v = v(t), then acceleration = a =

 156 Differential Equations: Theory and Applications

dv
dt
or a = slope of velocity – time curve = tan θ

dx
v= ,
We know that dt thus
dx = vdt
=t t2=t t2

∫ dx = ∫ vdt
or
=t t1=t t1
t = t2

or
t =t
x t = t2 ∫ vdt
1 = t =t1
t = t2

or x ( t1 ) − x ( t2 ) =
∫ vdt t =t1

Change in distance = area bounded by velocity – time curve, t-axis

=t t1=
and t t2
(ii) Space – Time curve
 Application of Differential Equations in Mechanics 157

x = x (t ) ,
Consider the space- time curve as
dx
velocity= v=
dt
velocity = slope of the space-time curve
= tanθ
(ii) Acceleration-time graph
a = a (t ) ,
Consider the acceleration – time graph then, since
dv
a = or dv adt
dt
=t t2=t t2

∫ dv = ∫ adt
or
=t t1=t t1

t = t2
t = t2
v t =t =
1 ∫ adt
or t =t1

t = t2

v ( t1 ) − v ( t2 ) =
∫ adt
or t =t1

Change in velocity = area bounded be the acceleration – time curve,

t − axis t= t1 and t = t2
(iv) Velocity-space curve
158 Differential Equations: Theory and Applications

v = v ( x).
Consider the velocity- space curve Now since
dv
a=v
dx
where
dv
Slope of tangent = dx
Therefore
dv
tanθ =
dx
In triangle PNG
GN GN
tan
= θ =
GP v
From (1) and (2), we have
dv GN dv
= v = GN
dx v or dx
Therefore
Acceleration= a= Lenght of subnormal at P
Example. A particle starts from rest with a constant acceleration a.
when its velocity acquires a certain value v, it moves uniformly and then its
velocity start decreasing with constants retardation 2a till it comes to rest .
 Application of Differential Equations in Mechanics 159

Find the distance traveled by the particle. If the time taken from rest to rest
is t.
Solution

Let 1
t , and t and t
2 3 be the times for accelerated, uniform motion and

retorted motion respectively , then

t = t1 + t2 + t3
v
=
Acceleration = a = slope of OA t1
v
t1 =
or a

v
2a =
Similarly retardation = t3
v
t3 =
or 2a
t = t1 + t2 + t3
Thus
v v
= + t2 +
a 2a

This implies that

3v
t2 = t −
2a
Distance = area under the velocity- time curve
= area of trapezium OABC
160 Differential Equations: Theory and Applications

1
= ( OC + AB )( AD )
2

1
= ( t1 + t2 + t3 + t2 ) v
2

1
= ( t + t2 ) v
2

1 3v  3v 2
=  t + t −  v = vt −
2 2a  4a
Example
A particle moving along a straight-line stars from rest and is accelerated
uniformly till it attains the velocity v. The motion is the retorted and the
particle come to rest after traversing a total distance x. if the acceleration is
f, find the retardation and the total time take y the particle from rest to rest.
Solution

Let
t1 and t2 be the time for acceleration and retorted motion respectively,
then

Acceleration = f
= slope of OA
v v
= f = or t1
Or t1 f
Let g be the retardation, then
g = slope of AB
v
2
= t
 Application of Differential Equations in Mechanics 161

v
t2 =
or g
1
= ( t1 + t2 ) v
Distance = x = area of triangle OAB 2
2x
t1 + t2 =
or v
2x
Thus total time = v
v v
= t1 = and t2 in (1) ,
Put f g then
1 v v  1 1 1  2
x=  + v =  + v
2 f g  2 f g 
2x 1 1
=
2
+
or v f g

1 2 x 1 2 xf − V 2
= 2 − =
or g v f fv 2
fv 2
g=
or 2 xf − v 2

3.2. PROJECTILE MOTION

An object thrown into the space with certain velocity from a gun or dropped
from a moving plane under the action of gravity is called a projectile.
Thus, a projectile move with a constant horizontal velocity and at the same
falls freely under the action of gravity. The path of projectile is called the
trajectory.

3.2.1. Equation of the Trajectory of a projectile

v
Let the particle m be projected from point O with an initial velocity o making
an angle α with the horizontal. Take O as the origin and the horizontal and
vertical lines through O as x − axis and y − axis respectively.
162 Differential Equations: Theory and Applications

p ( x, y )
Suppose that after time t, the particle is at point whose position
vector is r . i.e
r= xiˆ + y ˆj
ÿ
 ˆ + y ˆj
r = xi
¨ ¨ ¨
r x iˆ + y ˆj
=

The gravitational force F acting on the particle at P is

F = mgjˆ
¨
m r = −mgjˆ
¨
ˆ
or r = gj
¨ ¨
x iˆ + y ˆj =
− gjˆ,

Then
¨ ¨
x= 0 and y = −g
Integrating, we get
ÿ
x = A and y =
− gt + B

at t = 0, the particle is at O, therefore

=
at t 0,=x v0 cosα=
, y vo sinα
 Application of Differential Equations in Mechanics 163

Therefore
= o cosα , B
A v= vo sinα
Hence
x =v0 cos α , y =
− gt + vo sinα
which gives the horizontal and vertical component of velocity at any time t.
again integrate
t2
x =( v0 cos α ) t , and y =− g +t + D
2
at=t 0,=
x 0,=
y 0, therforeC
= 0,=
D 0,
Hence,
x = ( v0 cos α ) t

1 2
=y v0 sinα − gt
2
The eqs (1) and (2) are parametric equation of trajectory. To find the
Cartesian equation, eliminate t from the eqs. (1) and (2), we get

which is the Cartesian equation of trajectory . from

( 3) , we have
gx 2 sinα
−x =
−y
2vo cos α
2 2
cosα

sinα  2vo2 cos 2 α  2v 2 cos 2 α

x2 − x  =−y o
cosα  g  g
or
2 2 x v02 sinα cos α vo2 cos 2 α
x − −2 y
or g = g
2
2 v02 sinα cos α  v02 sinα cos α 
x − 2x + 
g  g 
or
2
v 2 cos 2 α  v02 sinα cos α 
= −2 y o + 
g  g 
 164 Differential Equations: Theory and Applications

2
 v02 sinα cos α  2vo2 cos 2 α  vo2 sin 2 α 
 x −  =  y − 
g g 2g 
or   
Now
(i) Vertex
 v0 sinα cos α vo2 sin 2 α 
2

 , 
 g 2g 

is the verte3x of the

(ii) Latus Rectum
2vo2 cos 2 α
4a =
Latus Rectum = g
(iii) Maximum Height
Maximum height attained above x- axis

H = Maximum height attained above x – axis

= y − Coordinate of vertex

vo2 sin 2 α
=
2g
dy
=0
This equation can also be obtained by putting dx in the equation of
trajectory.
(iv) Equation of Directrix
Height of Directrix above x – axis
1
= H + ( latus rectum )
4
 Application of Differential Equations in Mechanics 165

vo2 sin 2 α vo2 cos 2 α

= +
2g 2g
vo2 v2
=
2g
(sin 2 α + cos 2 α= o
2g
)
vo2
y=
Equation of Directrix is 2g
(v) Focus
x – Coordinate of focus = x – coordinate of vertex
v02 sinα cos α
=
g
1
( latus rectum )
y – Coordinate of focus = H - 4
vo2 sin 2 α vo2 cos 2 α
= +
2g 2g
vo2
=
2g
(
sin 2 α + cos 2 α )
vo2
=− cos 2α
2g

Thus, the focus

 v 2 sinα cos α v2   v2 v2 
= 0 , − o cos 2α  =  o sin 2α − o cos 2α 
 g 2g   2g 2g 
( vi ) Timeof Flight
The time taken by the particle to reach the horizontal plane through the
point of projection is ca;;ed its time of flight. In this case put y = 0, in the
equation
t2
=y g + (vo sin α ) t
2
 166 Differential Equations: Theory and Applications

t2
0= − g + (vo sin α ) t
Then 2
sincet ≠ 0, therefore

2vo sinα
t=
or g is the time of flight.
(vii) Range
(a) Horizontal Range
The horizontal range of the projectile is the horizontal distance described
by the particle during the time of flight. Therefore
Range R = (horizontal velocity) (time of flight)
 2v sinα 
vo cos α  o 
 g 
vo2 2 ( sin α cosα ) 2vo2 sin 2α
=
g g
This range can also be obtained by putting y = 0 in the Cartesian equation
of trajectory i.e putting y = 0 in the equation
gx 2
=y xtanα − 2
sec 2 α
2vo

Also, it is notable that this range will remain same if we replace

π
α by α= −α
2
(b) Maximum Range
vo2 sin 2α
Range R = g
dr
=0
For maximum range dα
 Application of Differential Equations in Mechanics 167

2vo2
⇒ cos 2 α =
0
g

⇒ cos 2α =
0
π
⇒ 2α =
2

π
⇒ α=
4

Therefore
vo2
Rmax =
g

3.2.2. Range in inclined plane

Let β the inclination of the inclined plane. Let a particle is projected from
point O with velocity
vo by making an angle α to the horizontal where
P ( x, y )
β < α . Let the projectile meet the inclined plane at then OP = r is
called the range of the inclined plane.

= =
x rcos β , y rsinβ

P ( x, y ) = ( rcos β , rsinβ )

Equation of trajectory is
1 x2
=y tan α − g 2 sec 2 α
2 vo
 168 Differential Equations: Theory and Applications

p ( rcos β , rsinβ )
The point lies on it, then
1 r cos 2 β
r sinβ r β tanα − g 2
=
2 vo cos 2 α

grcos 2 β sinα
= cos β − sinβ
or 2 vo
2
cos 2
α cosα
sinα cos β − cosα sinβ
=
cosα

sin (α − β )
=
cosα

Thus
2vo2 cos 2 α sin (α − β )
=r ×
g cos 2 β cosα

2vo2 cos 2 α sin (α − β )

=
g cos 2 β

For range down the inclined plane, change β to − β .

Maximum Range
Equation (1) can be written as, by using
sin (α + β ) − sin (α − β ) =
2 cos α sinβ

vo2
=r (sin
= ( 2α − β ) 1 , sothat
gcos 2 β

π β π
2α − β = or α = +
2 2 4

Thus
vo2
=rmax ( 1 − sin β )
gcos 2 β
 Application of Differential Equations in Mechanics 169

vo2 ( 1 − sinβ ) vo2

=
g (1 − sin 2 β ) g (1 − sin β )
dr
Or for maximum range put dx 0
cos ( 2α − β )
or =0
π
or 2α − β = 2

Time of Flight
Horizontol distance
Time of flight = Horizontol velocity
x rcos β
= =
vo cos α vo cos α
Using (1), we get
2vo2 cos 2 α sin (α − β ) cos β
Time of flight = g cos 2 β vo cosα

2vo2 sin (α − β )
=
g cos β

For time of flight down the plane, change β to − β .

Example: Show that the speed at a point P of the projectile is the same
as attained by a particle of same mass at P falling freely from the point
vertically above P on the Directrix of the trajectory.
Let vo be The Velocity of projection and v be the velocity at point p ( x, y )
of trajectory, then
v= xi  ˆ + yj
ˆ

v= ( v0 cosα ) iˆ + ( vo sinα − gt ) ˆj
v=
v = ( )
v0 cos 2 α + ( vo sin α − gt )
2
 170 Differential Equations: Theory and Applications

v= vo2 cos 2 α + vo2 sin 2 α − 2 gtvo sinα + g 2t 2

 1 
v 2 − 2 g  vot sinα − gt 2 
=
 2 
Since
1 2
=y vo t sin α − gt
2
Therefore
=
v v02 − 2 gy

P ( x, y )
Let N be the point on the Directrix vertically above . From fig.
= MP + PN
MN
V02
= y + PN
2g
Let v’ be the velocity of the particle in failing freely a distance NP. Using
the equation of motion
v 2 −U 2 =
2 gx
where u is initial velocity and v is final velocity
u 0,=
Here= , v v’, then
x PN=
v '2 − 0 =2 g ( PN )

 v2 
=v '2 2 g  o − y 
 2g 
v′=
2
vo2 − 2 gy

v′
= vo2 − 2 gy
`

From
(1) and ( 2 ) , we have
v′ = v
 Application of Differential Equations in Mechanics 171

3.2.3. Parabola if Safety

As the equation of the projectile is
1 x2
=y x tanα − g 2 sec 2 α
2 v
which shows that for different values of α we have different trajectories.
So, we define that the locus of all the points on different trajectories which
lie at maximum distance from the point of projection O is called the parabola
of safety. All the trajectories from O must touch the parabola of safety at
some point. Thus, it is envelope of the family of all the trajectories from with
same velocity of protestation
vo ,α being regarded as parameter. Equation
of trajectory is
1 x2
=y x tan α − g 2 sec 2 α
2 vo

1 x2
=y x tan α − g 2
2 v
gx 2  gx 2 
tan α − x tan α +  α + 2  =
2
0
2vo2  2vo 

It is quadratic equation in tanα . For the equation of envelope , put

discriminate = 0, i.e
 gx 2   gx 2 
x2 − 4  2   y + 2  = 0
 vo   vo 

Since x ≠ 0, dividing it by x , we get

2
172 Differential Equations: Theory and Applications

 g  gx 2 
− 2  2  y + 2  = 0
v
 o  vo 

2 gy g 2 x 2
− − 2 = 0
vo2 vo

g 2 x 2 2 gy
= +1
vo2 vo2

2 vo2  vo2 
−2  y −
x = 
g 2g 
which is the equation of the parabola of safety having vertex
 vo2 
 0, 
 2g 
2vo2
Latus Rectum = g
Equation of Directrix is
vo2 1
=
y + ( latus rectum )
2g 4

vo2 1  2vo2 
=
y +  
2g 4  g 

vo2 vo2 vo2

y= + =
2g 2g g

x − coordinate of focus = 0
x − coordinate of vertex =
y − coordinate of focus = 0
y − coordinate of vertex =
1
− ( latus rectum )
4
 Application of Differential Equations in Mechanics 173

v 2 1 2vo2
=o − 0
=
2g 4 g

Therefore, focus

3.2.4. Projection with Air Resistance

Let a particle of mass m be projected with an initial velocity

vo making an
angle α with the horizontal as shown in the figure.

Here we restrict ourselves to the simple case of air resistance proportional

to the first power of velocity. We suppose that after some time t the particle
is at P(x, Y). The equations of motion are, with mass canceled,
¨
x = −kx
¨
y=− g − ky
From (1)
dx
= −kx
dt

dx
= −kdt
x
Integrate
In ( x ) =
−kt + c1

x vo cos α , therefore c=
t 0=
= in ( vo cosα )
At 1
 174 Differential Equations: Theory and Applications

−kt + in ( vo cosα )
=
In(x)
x = vo cos α e − kt (3)
v
− o cosα e − kt + c2
x=
k
=t 0=
At , x 0, there fore
vo
c2 = cosα , thus
k
vo v
=x cos α e − kt + o cosα
k k
v
x= ( )
− o 1 − e − kt cosα
k
or (4)
Now, consider equation (2)
dy
=− g − ky
dt
dy
= −dt
or g + ky
Integrate
in ( g + ky ) =
−kt + c3

=t 0,=
At
y vo sina, that implies that
c3 in ( g + kvo sin α )
=

In
( g + ky ) =
−kt + In ( g + kvo sinα )

g + ky
In = −kt
In g + kvo sin α

g + ky = ( g + kvo sin α ) e − kt

g g 
− +  + kvo sinα  e − kt
y=
k k  (5)
Again integrating
 Application of Differential Equations in Mechanics 175

g 1g 
− −  + vo sinα  e − ky + c4
y=
k kk 
at=t 0,=
y 0, this implies that
1g 
=
c4  + vo sinα 
kk 
Thus
g 1g  1g 
− −  + vo sinα  e − kt +  + vo sinα 
y=
k kk  kk 
g 1g 
− t −  + vo sinα  1 − e − kt
y=
k kk 
( )
(6)
The equation of the path is obtained by eliminating t from equation
( 4 ) and ( 6 ) as
xcosα  g  g  xcosα 
y=
−  + vo sinα  + 2 In 1 − 
vo  k  K  vo 
Example
A shell burst on contact with the ground and pieces from it fly in all
directions with all speed up to 80 ft. / sec. Prove that a man 100 ft. is in
5
danger for 2 seconds.
Solution
We know that the range of the particle is
vo2 sin 2α
R=
g
=
Here
R 100
= ft., vo 80 ft / sec., g = 32 ft / sec 2 ., therefore
6400 sin 2α
100 =
32
1
sin 2α =
or 2
176 Differential Equations: Theory and Applications

or α = 15 , 75
o o

For the range 100 ft. there are two directions of projection. Let
t1 and t2
be the times of flight respectively, then
2vo sin15o
t1 =
g
2vo sin75o
t2 =
and g

The man is in danger for time =

t2 − t1

2vo sin15o 2vo sin75o

= −
g g

2vo
=
g
(
sin15o − sin75o )
160   75 + 15   75 − 15  
=  2 cos   sin  
32   2   2 

5
(
5=
2cos 45o sin30o ) 2
sec.

Example Find the range of a rifle bullet when α is the

v
elvation of projection and o is speed. show that, if the rifle is
field with the same elevation and speed from a car travelling with
2vvo sinα
v towards the target, the range will be increase by g
Solution
We know that the range is given by
R = ( horizontal velocity )( time of flight )

 2vo sinα  2v02 sin α cosα

R v=
o cosα  
 g  g
 Application of Differential Equations in Mechanics 177

When the shell is fired from a car moving with speed v towards the
target the horizontal velocity is increased by v.

Horizontal velocity =
vo cosα + v

New range + R’
= ( newhorizontal time of flight )

 2vo sinα 
= ( vo cosα + v )  
 g 

2v02 sin α cosα  2vo v sinα 

'
=R + 
g  g 
 2vo v sinα 
r′ − R =  
Increase in range =  g 
Example
A battleship is steaming ahead with speed v and a gun is mounted on a
battleship so as to point straight backward and is set at an angle of elevation
α if vo is the speed of projection (relative to gun), show that the range is
 2vo sinα 
  ( vo cosα − v )
 g 
Prove also that the angle of elevation for maximum range is
 v + v 2 + 8v 2 
cos −1  o

 4vo 
 
Solution
The ship is moving with speed v away from the target, so the horizontal
velocity of the projectile will be decreased by v.
vo cos α − v
Actual horizontal velocity =
Actual range = R = (horizontal velocity) (time of flight)
 2vo sinα 
( vo cosα − v )  
 g 
178 Differential Equations: Theory and Applications

 2vo sinα 
  ( vo cosα − v )
 g 
Since
 2v sinα 
R  o  ( vo cosα − v )
 g 

vo
(vo sin 2α − 2v sinα )
g

Therefore, for maximum range

dR 2vo
= ( vo cos 2α − v cos
= α) 0
dα g

vo cos 2 α − vcosα =
0
vo (2 cos 2 α − 1) − vcosα =0

vo (2 cos 2 α − 1) − v cosα =0

2v0 cos 2 α − vcosα − v0 =

0

v ± v 2 + 8vo2
cosα =
4Vo

d 2 R 2vo
= ( −2vo sin 2α + vsinα )
dα 2 g

 2v sinα 
=
− o  4vo cosα −
 g 

v + v 2 + 8vo2
cos α =
when 4Vo
dR 2vo sinα
= ( v 2 + 8v02 ) =
−ve
dα g
 Application of Differential Equations in Mechanics 179

For 0 < α < 90, sinα is positive .

Thus, R is maximum when
v + v 2 + 8vo2
cosα =
4Vo

 v + v 2 + 8v 2 
α = cos  −1 o

 4Vo 
or  
 2 2 
−1 v + v + 8vo
α = cos  
 4Vo 
or  
Example
Prove that the speed required to project a particle from a height h to fall
a horizontal distance a from the point of projection is at least

g ( a 2 + h 2 − h)

Solution
Equation of trajectory is
gx 2
=y xtan α − sec 2 α
2vo2

Point
( a, −h ) lies on it, then
ga 2
−h a tan α −
= sec 2 α
2v02
180 Differential Equations: Theory and Applications

ga 2
−h a tan α −
= 2
(1 + tan 2 α )
2v0

ga 2 tan 2 α − 2avo tanα + ga 2 − 2hvo2 ) ≥ 0

Since tanα is real therefore, discriminant ≥ 0, i.e.

(
4a 2 vo4 − 4 ga 2 ga 2 − 2hvo2 ≥ 0 )
(v ) ( )
2
2
0 + gh ≥ g 2 a 2 + h 2

v02 + gh ≥ g a 2 + h 2

v02 ≥ g ( a 2 + h2 − h )
vo ≥ g ( a 2 + h2 − h )
v
Therefore, least value of o is
g ( a 2 + h2 − h )
Example
A particle is launched at an angle a from a cliff of height H above sea
level. If it falls into the sea at a distance D from the base of the cliff , prove
that the maximum height above the sea level is
 Application of Differential Equations in Mechanics 181

Solution
Equation of the trajectory is

Point (D, -H) lies on it, therefore

gD 2
−
= H D tan α − 2
sec 2 α
2vo

gD 2
H + D tan α −
2vo2 cos 2 α

gD 2
vo2 =
2 cos 2 α ( H + D tan α )

Now, h height above x − axis, so that

=

vo2sin 2α
h=
2g

Putting the value of

vo2 ,then

gD 2 sin 2 α
h=
2 cos 2 α ( H + D tanα ) 2 g

D 2 tan 2 α
=
4 ( H + D tanα )

Height above sea level= H + h

D 2 tan 2 α
= H+
4 ( H + D tanα )

Example
A front and a slip are both armed with guns which give their projectile
2 gk and guns in the fort are at height h above the guns
a muzzle velocity
 182 Differential Equations: Theory and Applications

d d
in the ship. If 1 and 2 are the greatest horizontal ranges at which the fort
and ship respectively can engage, prove that
d1 k +h
=
d2 k −h
Solution
Let S be the ship and F be the fort.

β and SF = r
Let ∠XSF =
d 2 is greatest horizontal range for gun in ship, so r is maximum range

on inclined plane with inclination β . Now,

vo2
r=
g ( 1 + sinβ )

vo = 2 gk in the above equation, then

Put
2 gk 2k
=r =
g ( 1 + sinβ ) 1 + sin β
r + r sin β =
2k

Since form the figure h = r sin , therefore

r+h= 2k

Also from right angle triangle SAF, we have

2
r= h 2 + d 22
So that
 Application of Differential Equations in Mechanics 183

( 2k − h )
2
h 2 + d 22 = = 4k 2 − 4kh + h 2

d 22 = 4k 2 − 4kh = 4k ( k − h )
(1)
For gun in fort, change h to – h, then
d12 4k ( K + h )
=
(2)
Dividing (2) by (1), then
d1 k +h
=
d2 k −h
Example
The horizontal rang of a ball for two different projection is same and
equal to R. if 1
h and h
2 are the greatest height in the two parts for which

this is possible prove that

R 2 = 16h1h2
Solution
v
Let R be the range, when α is angle of the projection and o is velocity
of projection , then
2vo2 sinα cosα
R=
g
π
−α,
Let R’ be the range when angle of projectile is 2 then
π  π 
2vo2 sin  − α  cos  − α 
R′ 2  2 
g
2vo2 sinα cosα
R′ =
or g
π
−α
which show that R is the same when angle of projection are α and 2 .
h and h2 be the maximum height in these cases respectively, therefore
Let 1
 184 Differential Equations: Theory and Applications

vo2 sin 2 α
h1 =
2g

π 
vo2 sin 2  − α 
 vo cos α
2 2
=h2 = 2
2g 2g

Hence
vo2 sin 2 α  vo2 cos 2 α 
16h1h2 = 16  
2g  2g 

vo4 sin 2 α cos 2 α

4= R2
g

i.e R 2 = 16 h1h2

Example
Prove that if time of flight of a bullet over a horizontal range R is T
seconds then the inclination of direction to the horizontal is
 gt 2 
tan −1  
 2r 

Solution
We know that
2 sin α
T= o
g

2vo2 sinα cosα

R=
g

gT 2 4vo2 sin 2 α g
= tanα
2R g 4vo sinα cosα
2

Hence
 Application of Differential Equations in Mechanics 185

 gT 2 
α = tan −1  
 2R 
Example
A cricket ball thrown from height of 6 feet at an angle of 30 with the
o

horizontal speed of 60 ft. / sec . is caught by another player at height 2 feet

from the ground. How far apart were the two men?
Solution
Let two men are x feet apart. As the ball is caught at a height of 2 feet
from the ground, so the distance it falls vertically is 6-2 = 4ft. The point
( x, −4 ) lies on the trajectory. Now
1 2
=y ( vo sinα ) t − gt
2
y=
Here
−4 , v0 = 6 − ft / sec., α= 3= 0 , t= ?
1
−4 ==( 60 sin30 ) t − ( 32 ) t 2
2

8t 2 −15t − 2 =0

or (t-2) (8t+1) = 0
1
t = 2 or t = −
8

1
t ≠− ,
Since 8 therefore t = 2.
x = Horzontal distance

= ( Horizontal velocity )( time )

= (=
vo cosα ) t ( 60 cos30 ) ( 2 )
0

x = 60 3 ft.
186 Differential Equations: Theory and Applications

3.3. SUMMARY AND DISCUSSION

Rectilinear Motion. The motion of a particle along a straight line is called
rectilinear motion.

Vertical Motion under Gravity

For a failing body, the acceleration is constant. It is called acceleration due
to gravity and is denoted by ‘g’. In FPS system g = 32 ft / sec .In CGS
2

system g =981 cm / sec and in MKS system g = 9.81 m / sec . If the body
2 2

is projected vertically upward then acceleration g is negative. For falling

bodies equation of motion is
v= u + gt

1 2
x= ut + gt
2

v2 − u 2 =
2 gx

Distance traveled in nth Second

x and x b
Let 1 2 e the distance travelled in the first n and n – 1 seconds
respectively, then from the equation of motion
1
x= ut + at 2
2
1
x=
1 un + an 2
then 2
1
x2= u ( n − 1) + a ( n − a )
2

and 2
x1 − x2
Distance travelled in the nth second =
a 2  a 2
x1 − x2 = un + n − u ( n − a ) + ( n − 1) 
2  2 
a 2  a 
un +
2  2
(
n −  un − u + n 2 − 2n + 1 

)
=
 Application of Differential Equations in Mechanics 187

1
u + a ( 2n − 1)
= 2
Motion with variable Acceleration
(i) Time Dependent Acceleration
(ii) Velocity Dependent Acceleration
(iii) Distance Dependent Acceleration

Projectile
An object thrown into the space with certain velocity from a gun or dropped
from a moving plane under the action of gravity is called a projectile.
Thus, a projectile move with a constant horizontal velocity and at the same
falls freely under the action of gravity. The path of projectile is called the
trajectory.
Horizontal and vertical component of velocity at any time t are
,
are the parametric equation of trajectory.
(i) Vertex
 v sinα cos α vo2 sin 2 α 
2


0
, 
 g 2g 

(ii) Latus Rectum

2vo2 cos 2 α
4a =
Latus Rectum = g
(iii) Maximum Height
vo2 sin 2 α
=
2g
(iv) Equation of Directrix
2
v
y= o
2g
(v) Focus
188 Differential Equations: Theory and Applications

 v 2 sinα cos α v2   v2 v2 
= 0 , − o cos 2α  =  o sin 2α − o cos 2α 
 g 2g   2g 2g 

The time taken by the particle to reach the horizontal plane through the
point of projection is called its time of flight.
2v sinα
t= o
g is the time of flight.
(vii) Range
(a) Horizontal Range
The horizontal range of the projectile is the horizontal distance described
by the particle during the time of flight. Therefore
Range R = (horizontal velocity) (time of flight)
 2v sinα 
vo cos α  o 
 g 
vo2 2 ( sin α cosα ) 2vo2 sin 2α
=
g g
This range can also be obtained by putting y = 0 in the Cartesian equation
of trajectory i.e putting y = 0 in the equation

Also, it is notable that this range will remain same if we replace

π
α by α= −α
2
(b) Maximum Range
vo2 sin 2α
Range R = g
dr
=0
For maximum range dα
 Application of Differential Equations in Mechanics 189

2vo2
⇒ cos 2 α =
0
g
⇒ cos 2α =
0
π
⇒ 2α =
2
π
⇒ α=
4
vo2
Rmax =
Therefore g

Time of Flight
Horizontol distance
Time of flight = Horizontol velocity
x rcos β
= =
vo cos α vo cos α
Using (1), we get
2vo2 cos 2 α sin (α − β ) cos β
Time of flight = g cos 2 β vo cosα

2vo2 sin (α − β )
=
g cos β

For time of flight down the plane, change β to − β .

Parabola of Safety
As the equation of the projectile is
1 x2
=y x tanα − g 2 sec 2 α
2 v
which shows that for different values of α we have different trajectories. So
we define that the locus of all the points on different trajectories which lie
at maximum distance from the point of projection O is called the parabola
of safety.
 CHAPTER

4
ELLIPTIC DIFFERENTIAL
EQUATION

CONTENTS
4.1. Introduction..................................................................................... 192
4.2. Boundary Value Problem (BVPs)...................................................... 195
4.3. Some Important Mathematical Tools................................................ 197
4.4. Properties Of Harmonic Functions................................................... 199
4.5. Separation Of Variables ................................................................... 210
4.6. Dirichlet Problem For A Rectangle................................................... 212
4.7. The Neumann Problem For A Rectangle .......................................... 215
4.8. Interior Dirichlet Problem For A Circle ............................................ 217
4.9. Exterior Dirichlet Problem For A Circle ........................................... 222
4.10. Interior Neumann Problem For A Circle......................................... 227
4.11. Solution Of Laplace Equation In Cylindrical Coordinates .............. 229
4.12. Solution Of Laplace Equation In Spherical Coordinates................. 238
4.13. Miscellaneous Example................................................................. 247
4.14. Summary And Discussions............................................................. 276
192 Differential Equations: Theory and Applications

4.1. INTRODUCTION
In previous topics, we have seen the classification of second order partial
differential equation into elliptic, parabolic and hyperbolic types. In this
chapter we shall consider various properties and technique for solving
Laplace and Poisson equations which are elliptic in nature.
Various physical phenomena are governed by the well-known Laplace
and poison equations. A few of them, frequently encountered in applications
are: study heat condition, magnetic potential, torsion of prismatic shaft,
bending of prismatic beams, distribution of gravitational potential, etc.
In the following two sub-section, we shall give the derivation of Laplace
and Poisson equations in relation to the most frequently occurring physical
situation , namely the gravitational potential.

4.1.1. Derivation of Laplace Equation

m and m1
Consider two particles of masses situated at Q and P separated
by a distance r as shown in . According to Newton’s universal law of
gravitational, the magnitude of the force, proportional to the product of their
masses and inversely proportional to the square b of the distance. Between
is given
mm1
F =G
r 2 (1)

where G is the gravitational constant. It r represent the vector PQ,
assuming unit mass at Q and G = 1, the force at Q due to the mass at p
is given by
mr m 
F= − 13 = ∇ 1 
r  r  (2)
which is called the intensity of the gravitational force. Suppose a particle of
m
unit mass moves under the attraction of a particle of mass 1 at P infinity
up to Q; then the work done by the force F is
r r
m  m1
∫∞F .dr =
∫∞∇  r1  .dr =
r (3)
 Elliptic Differential Equation 193

This is defined as the potential V at Q due to a particle at P and is

denoted by
m1
V=
r (4)

Figure 1

From Eq. 2, the intensity of the force at P is

F = −∇V (5)

Now, if we consider a system of particles of masses

m1 , m2 , …., mn

which are at distances

r1 , r2 , …, rn respectively, then the force of attraction

per unit mass at Q due to the system is

n n
mi m
=i 1 =
∑∇
=
ri
= ∇∑ i
i 1 ri (6)

The work done by the force acting on the particle is

r n
mi
∫F .dr = ∑
i =1 ri
= −V
∞ (7)
Therefore,
mi n 2 mi
n
∇2 V = − ∇2 ∑ = ∑∇ = 0, ri ≠ 0
ri i 1
=i 1 = ri (8)
where
194 Differential Equations: Theory and Applications

2 ∂2 ∂2 ∂2
∇= div ∇= + +
∂x 2 ∂y 2 ∂z 2 is called the Laplace operator.
In the case of continuous distribution of matter of density p in a volume
r , we have
.
p (ξ ,η , ζ )
V ( x, y, z ) =∫∫ ∫ dr
r
r (9)

where
{
r = ( x − ξ ) + ( y −η ) + ( z − ζ )
2 2
}
2 1/2

and Q is outside the body. It

can be verified that
∇ 2V = 0 (10)
which is called the Laplace equationA2I.

4.1.2. Derivation of Poisson Equation

m , m , … mn .
Consider a closed surface S consisiting of particle of masses 1 2
n

∑ mi = M
Q
Let be any point on S . Let i =1 be the total mass inside S , and let
g1 , g 2 , …, g n be the gravity field at Q due to the presence of m1 , m2 …, mn
n

∑ gt = g ,
S
respectively within . Also, let i =1 the entire gravity field at Q.
Then, according to Gauss law, we have
.
∫ ∫ g.dS =
− 4π GM
S (11)
.
M =∫ ∫ ∫ pdτ , p
where r is the mass density function and τ is the volume
in which the masses are distributed through. Since the gravity field is
conservative. We have
g = ∇V (12)

where V is a scalar potential. But the Gauss divergence theorem states that
 Elliptic Differential Equation 195

. .
∫ ∫g .dS = ∫ ∫ ∫ ∇.g dτ
S τ (13)
Also. Eq. (11) gives
. .
∫ ∫g ô. dS =−4π G ∫ ∫ ∫ pd
S τ (14)
Combining Eq. (13) and (14), we have
.
∫ ∫ ∫ (∇.g + 4π Gp ) dτ =0
τ

Implying
∇.g =−4π Gp =∇. ∇ v
Therefore,
∇2 V =− 4π Gp (15)

This equation is known as poisson’s equation .

4.2. BOUNDARY VALUE PROBLEM (BVPS)

The function V , whose analytical form we seek for the problems stated
in section in addition to satisfying the Laplace and Poisson equation in a
3
bounded region IR in R , should also satisfy certain boiundary conditions
on the boundary ∂IR of this region Such problems are referred to as
boundary value problems (BVPs) for Laplace and Poisson equations. We
shall denote the set of all boundary points of IR by ∂IR . By the closure of
IR, we mean the set of all interior points of IR together with its boundary
points and is denoted by IR, we mean the set of all interior point together
IR IRU ∂IR.
Symbolically, =

If a function then all its derivative of order n

are continuous.
0
If it belong to c , then we mean f is continuous.
196 Differential Equations: Theory and Applications

There are mainly three types of boundary value problems for Laplace
equation. If and is specified on the boundary ∂IR of some finite region
IR , the problem of determining a function ψ ( x, y, z ) suchthat ∇ ψ =
2
0

within IR and satisfying = ψ f on ∂IR is called the boundary value

problem of first kind, or the Dirichlet problem. For example, finding the
steady state temperature within the region IR when no heat sources or sinks
are present and when the temperature is prescribed on the boundary ∂IR, is
Dirichlet problem. Another example would be to find the potential inside the
region IR when the potential is specified on the boundary ∂IR.These two
examples correspond to the interior Dirichlet problem.

Similarly, if and is prescribed on the boundary ∂IR of a finite

ψ ( x, y , z )
simply connected region IR, determinanting function which
satisfies ∇ ψ =
2
0 outside IR and is such that ψ = f on ∂IR, is called an
exterior Dirichlet problem. For example, determination of the distribution
of the potential outside a body whose surface potential is prescribed , is
an exterior Dirichlet problem. The second type of BVP is associated with
ψ ( x, y , z )
Von Neumann. The problem is to determine the function
that ∇ ψ 0 within IR while ∂ψ∂n. is specified at every point of
2
=
so
∂IR, where ∂ψ / ∂n denotes the normal derivative of the field variable ψ .
This problem is called the Neumann problem. If ψ is the temperature,
∂ψ / ∂n is the heat flux representing the amount of heat crossing per unit
volume per unit time along the normal direction, which is zero when insulated
. The third type of BVP is concerned with the determination of the function
ψ ( x, y , z )
such that ∇ ψ =
2
0 within IR, while a boundary condition of
the form ∂ψ / ∂n + hψ = f , where h ≥ 0 is specified at every point of
∂IR. This is called a mixed BVP or Churchill’s problem. If we assume
Newton’s law of cooling, the heat lost is hψ , where ψ is the temperature
difference from the surrounding medium and h > 0 is a constant depending
on the medium. The heat f supplied at a point of the boundary is partly
conducted into the medium and partly lost by radiation to the surrounding.
Equating these amounts, we get the third boundary condition. The third
 Elliptic Differential Equation 197

type of BVP is concerned with the determination of the function ψ ( x, y, z )

such that ∇ ψ =
2
0 within IR, while a boundary condition of the form
∂ψ / ∂n + hψ = f , where h ≥ 0 is specified at every point of ∂IR. This
is called a mixed BVP or Churchill’s problem. If we assume Newton’s law
of cooling, the heat lost is hψ , where ψ is the temperature difference from
the surrounding medium and h > 0 is a constant depending on the medium.
The heat f supplied at a point of the boundary is partly conducted into
the medium and partly lost by radiation to the surrounding. Equating these
amounts, we get the third boundary condition

4.3. SOME IMPORTANT MATHEMATICAL TOOLS

Among the mathematical tools, we employ in deriving many important
results. The Gauss divergence theorem plays a vital role, which can be stated
as follows : Let ∂IR be a closed surface in the xyz − space and IR denote
the bounded region enclosed by ∂IR in which F is a vector belonging to
c1 in IR and continuous on IR. Then
. . .

∫∫ F . nˆ dS = ∫ ∫ ∫ ∇.F dV
∂ IR IR

where dV is an element of volume, dS is an element of surface area . and

n̂ the outward drawn normal.
Green’s identities which follow form divergence theorem are also useful
and they can be derived as follows : Let F = f φ , where f is a vector function
of position and φ is a scalar function of position. Then
. . .
∫ ∫ ∫ ∇. ( f φ ) dV =∫ ∫ nˆ. f φ dS
IR ∂ IR

Using the vector identity

∇. ( f φ ) = f . ∇φ + φ∇. f

we have
198 Differential Equations: Theory and Applications

. . . .
∫ ∫ ∫ f=
IR
.∇.φ dV ∫ ∫ n. f φ dS ∫ ∫ ∫ φ ∇. f dV
∂ IR IR

If we choose f = ∇ψ , the above equation yields

. . . .
∫ ∫∇φ .∇ψ=
dV ∫∫φ nˆ.∇ψ dS − ∫ ∫ ∫ φ ∇ ψ dV
2

IR ∂ IR IR
( 17 )
Noting that nˆ.∇ψ is trhe derivative of ψ in the direction of nˆ , we
introduced the notation
nˆ .∇ψ =
∂ψ / ∂n
Into Eq. (17) to get

( 18 a )
This equation is known as Green’s first identity. Of course, it is assumed
that both φ and ψ possess continuous second derivatives.
Interchanging the role of φ and ψ , we obtain form relation (18 a) the
equation

( 18 b )
Now, subtracting Eq. (18b) from Eq. (18) we get

( 19 )
This is known as Green’ second identity. If we set φ = ψ in Eq. (18a)
we get
. . . .
∂φ
∫ ∫ ∫ (=
∇φ ) dV
2

IR
∫ ∫φ
∂ IR
∂n
dS − ∫ ∫ ∫ φ∇ 2 φ dV
IR ( 20 )
which is a special case of Green’s first identity.
 Elliptic Differential Equation 199

4.4. PROPERTIES OF HARMONIC FUNCTIONS

Solutions of Laplace equation are called harmonic functions which possess
a number of interesting properties, and they are presented in the following
theorems.
Theorems. If a harmonic function vanishes everywhere on the boundary
, then it is identically zero everywhere.

Proof. If φ is a harmonic function, then ∇ φ =

2
0 in IR . Also, if φ = 0

on ∂IR, we shall show that= φ 0 in= IR IRU ∂IR. Recalling Green’s first

identity, i.e., Eq.

( 20 )
. . . .
∂φ
∫ ∫ ∫ ( ∇=
φ ) dV
2
∫∫ φ ∂n
dS − ∫ ∫ ∫ φ ∇ 2 φ dV
IR ∂ IR IR
And using the above facts we have, at once, the relation
.
∫ ∫ ∫ ( ∇φ ) dV =0
2

IR

( ∇φ )
2

Since is positive, it follows that the integral will be satisfied only

if ∇ φ =0. This implies that φ is a constant in IR . Since φ is continuous
in IR and φ is zero on ∂IR, it follows that φ = 0 in IR.
Theorem. If φ is a harmonic function in IR and ∂φ / ∂n =0 on ∂IR,
then φ is a constant in IR .
Proof. Using Green’s first identity and the data of theorem, we arrive at
.
∫ ∫ ∫ ( ∇φ ) dV =0
2

IR

Implying ∇φ =0, i.e.,φ is a constant in IR. Since the value of φ is

not known on the boundary ∂IR while ∂φ / ∂n =0, it is implied that φ is a
constant on ∂IR and hence on IR .
Theorem. If the Dirichlet problem for a bounded region has a solution,
then it is unique .
200 Differential Equations: Theory and Applications

Proof. If
φ1 and φ2 are two solutions of the interior Dirichlet problem,
then
∇ 2φ1 = 0 in IR; φ1 = f on ∂IR

∇ 2φ2= 0 in IR; φ2 = f on ∂IR

ψ= φ1 − φ2 . Then
Let
∇ 2φ =∇ 2φ1 −∇ 2φ2 = 0 in IR;

ψ = φ1 − φ2 = f − f = 0 on ∂IR;

Therefore,
∇ ψ = 0 in IR, ψ = 0 on ∂IR
2

φ = φ2 . Hence
Now using, we obtain φ = 0 on IR, which implies that 1
the solution of the Dirichlet problem is unique.
Theorem. If the Neumann problem for a bounded region has a solution,
then it is either unique or it differ from one another by a constant only.

Proof. Let
φ1 and φ2 be two distinct solution of the Neumann problem.
Then we have
∂φ1
∇ 2φ1 = 0 in IR; = f on ∂IR,
∂n
∂φ2
∇ 2φ2 = 0 in IR; = f on ∂IR,
∂n
ψ= φ1 − φ2 . Then
Let
∇ 2ψ = ∇ 2φ1 −∇ 2φ2 = 0 in IR

∂φ ∂φ1 ∂φ2
= − = 0 on ∂IR
∂n ∂n ∂n
IR,i.e.,φ1 − φ2 =
Hence from theorem ψ is a constant on constant.
Therefore, the solution of the Neumann problem is not unique. Thus, the
 Elliptic Differential Equation 201

solutions of a certain Neumann problem can differ from one another by a

constant only.

4.4.1. The Spherical Mean

p ( x, y , z )
Let IR be a region bounded by ∂IR and let be any point in IR.
S ( p, r )
Also, let represent a sphere with centre at P and radius r such that
it lies entirely within the domain IR as depicted in. Let u be a continuous
function in IR. Then the spherical mean of u denote byu u is defined as
. .
1
u (r ) = ∫ ∫ u ( Q ) dS
4π r 2 S ( p ,r )

( 21 )

Figure 2

Q (ξ ,η , ζ ) S ( p, r )
where is any variable point on the surface of the sphere
and dS is the surface element of integration. For a fixed radius r , the value
u (r )
is the average of the value of u taken over the spehere S 9 p, r ) , and
hence it is called the spherical mean. Taking the origin at p, in terms of
spherical polar coordinates, we have
ξ= x + r sin θ cosφ
202 Differential Equations: Theory and Applications

η= y + r sin θ sinφ

ζ = z + r cos θ
Then the spherical mean can be written as
2π π
1
u ( r=) u ( x + r sin θ cos φ , y + r sinθ sinφ , z + r cosθ ) r 2 sin θ dθ dφ
4π r 2 =θ ∫0=θ ∫0

S ( p, r ) , u too
Also since u is a continuous on is a continuous function
of r on some interval 0 < r ≤ R, which can be verified as follows.
1
2π π
u ( Q ) 2π π
u (r ) u ( Q ) sinθ dθ dφ = sin θ d θ dφ u ( Q )
4π r 2 =φ∫0=θ ∫0 4π ∫0 ∫0
=

Now, taking the limit as r → 0, Q → p, we have

Ltr →0 u ( r ) = u ( p ) ( 22 )

Hence, u is continuous in 0 ≤ r ≤ R.

4.4.2. Mean Value Theorem for harmonic Functions

p ( x, y , z )
Theorem. Let u be harmonic in a region IR. Also, let be a
S ( p, r )
given point in IR and be a sphere with centre at p such that
S ( P, R )
is completely contained in the domain of harmonicas of u. Then
.
1
( p ) u=
u= (r ) ∬ u ( Q ) dS
4π r 2 s( p ,r )

u (r )
Proof. Since u is harmonic in IR, it spherical mean is continuous
in IR and is given by
. 2π π
1 1
u (r ) = 2 ∬ ( )
u Q dS ∫ ∫ u (ξ ,η , ζ ) r
2
sinθ dθ \ dφ
4π r s( p ,r ) 4π r 2 0 0

Therefore,
 Elliptic Differential Equation 203

du ( r ) ππ
=
dr ∫ ∫ ( uξ ξ
0 0
r + uηηr + uζ ζ r ) sinθ dθ dφ

2π π
1
=
4π r 2 ∫ ∫ ( uξ
0 0
sinθ cosφ +uτ sinθ sinφ + uζ cosθ ) sin θ d θ dφ
( 23 )
Nothing that sin θ cosφ ,sin θ sin φ and cos θ are the direction cosines of
nˆ on S ( p , r ) ,
the normal
∇u= iuξ + juη + kuξ , nˆ= ( in1 , jn2 , kn3 ) ,
The expression within the parenthesis of the integrand of eq. (23) can be
written as ∇. nˆ . Thus
du ( r ) 1
.

=
dr 4π r 2
∬ ∇u. nr
ˆ 2
sin θ dθ dφ
s( p ,r )

.
1
=
4π r 2
∬∇u. nˆ dS
s( p ,r )

.
1
= ∬ ∇.∇u dV ( by divergencetheorem )
4π r 2 V ( p ,r )
.
1
= 2 ∬
= 0 ( sinceu is harmonic )
∇ 2u dV
4π r V ( p ,r )

du
= 0,
Therefore, dr implying u is constant.
Now the continuity of u at r = 0 gives, from Eq. (22), the relation
.
1
u (r ) −u ( p ) = ∬ u (Q ) d S
4π r 2 V ( p ,r )

( 24 )

4.4.3. Maximum-Minimum Principal and Consequences

Theorem. Let IR be a region bounded ∂IR. Also. Let u be a function
which is continuous in a closed region IR and satisfies the Laplace equation
204 Differential Equations: Theory and Applications

∇2 u =0 in the interior of IR . Further, if u is not constant everywhere on

IR , then the maximum and minimum values of u must occur only on the
boundary ∂IR .
Proof. Suppose u is a harmonic function but not constant everywhere
on IR. if possible , let u attain its maximum value M at some interior
point p in IR.Since M is the maximum of u which is not a constant , there
S ( p, r )
should exist a spehere about p such that some of the values of u
S ( p ,r ) ,
on must be less than M . But by the mean value property, the
u on S ( p , r ) ,
value of u at P is the average of the values of and hence it
is less than M . This contradicts the assumption that u = M at p. Thus x u
S ( p, r ) .
must be constant over the entire sphere
Let Q be any other point inside IR which can be connected to P by
an arc lying entirely within the domain IR. By convergence this arc with
spheres and suing the Henie-Borel Theorem to choose a finite number of
covering spheres and repeating the argument given above , we can arrive
at the conclusion that u will have the same constant value at Q as at P
. Thus u cannot attain a maximum value at any point inside the region
IR. Therefore, u can attain its maximum value only on the boundary ∂IR.
By convergence this arc with spheres and suing the Henie-Borel Theorem
to choose a finite number of covering spheres and repeating the argument
given above , we can arrive at the conclusion that u will have the same
constant value at Q as at P . Thus u cannot attain a maximum value at any
point inside the region IR. Therefore, u can attain its maximum value only
on the boundary ∂IR.
Some important consequence of the maximum-minimum principle are
given in the following theorems.
Theorem.

Proof. Let
u1 and u2 be two solutions of the Dirichlet problem and let
f1 and f 2 be a given continuous functions on the boundary ∂IR such that

∇ 2u1= 0 in IR ; u1 = f1 on ∂IR ,
 Elliptic Differential Equation 205

∇ 2 u2 = 0 in IR ; u 2= f 2 on ∂IR ,

Let
u= u1 − u2 . Then,

∇ 2 u =∇ 2 u1 −∇ 2 u 2 = 0 in IR ; u =
f1 − f 2 on ∂IR
Hence, u is a solution of the Dirichlet problem with boundary condition
u= f1 − f 2 on ∂IR. By the maximum-minimum principle , u attains the

maximum and minimum values on ∂IR. Thus at any interior point in IR,
we shall have , for a given ε > 0,
− ε < u min ≤ u ≤ u max < ε

Therefore,
u < ε . in IR, implying u1 − u2 < ε on IR

Thus, small changes in the initial data bring about an arbitrary small
change in the solution. This completes the proof of the theorem.

Theorem. Let
{ fn }
be a sequence of the functions, each of which is
{f }
continuous on IR and harmonic on IR. if the sequence n convergence
uniformly on ∂IR, then it converges uniformly on IR .
{f }
Proof. Since the sequence n converges uniformly on ∂IR, for a given
ε > 0 , we find an integer N such that
f n − f m ε for n, m N

Hence, form stability theorem, for all n, m > N it follows immediately

that
f n − f m < ε In IR

Therefore,
{ f n } convergence uniformly on IR.
Example. Show that if the two-dimensional Laplace equationc¤X
∇2 u = 0 is transformed by introducing plane polar coordinates r , θ defined
=
by the relations x r=cos θ , y r sin θ , it takes the form
206 Differential Equations: Theory and Applications

∂ 2u 1 ∂u 1 ∂ 2u
+ + 0
=
∂ r 2 r ∂r r 2 ∂ θ 2
Solution. In many practical problems, it is necessary to write the
Laplace equation‑« X in the various coordinates systems. For instance,
if the boundary of the region ∂IR is circle, then it is natural to use polar
=
coordinates defined cosθ , y r sin θ . Therefore,
by x r=
 y
x2 + y 2 , θ =
r2 = tan −1   .
x
sin θ cosθ
rx = cosθ , ry = sinθ , θ x = - θ
r , y= r
Since
 sinθ 
u =u ( r , θ ) u x =ur rx + uθ θ x = ur cos θ − uθ 
 r 
Similarly,
 cosθ 
uy =ur ry + uθ θ y = ur sinθ + uθ 
 r 
Now for the second order derivatives,
 sinθ   sinθ   sinθ 
u xx =( ux )x =( ux )r rx + ( ux )θ θ x = ur cosθ −uθ  cos θ +  ur cos θ − uθ  − 
 r r  r θ  r 

Therefore,
 sinθ sinθ 
u xx=  urr cos θ − uθ r + uθ 2  cosθ
 r r 
 sinθ cosθ   sinθ 
+  urθ cosθ − ur sinθ − uθθ − uθ  − 
 r r  r  (25)
Similarly, we can show that
 cosθ cosθ 
u yy =  urr sin θ + urθ − uθ 2  sinθ
 r r 
 cosθ sinθ   cosθ 
+  urθ sinθ + ur cosθ + uθθ − uθ  
 r r  r  ( 26 )
 Elliptic Differential Equation 207

By adding Eq. (25) and (26) and equating to zero, we get

1 1
u xx + u yy = urr + ur + uθθ = 0
r r2 ( 27 )
which is the Laplace equation in polar coordinates. One can observe that the
Laplace equation in the Cartesian coordinates has constant coefficients only,
whereas in polar coordinates , it has variable coefficients.
Example. Show that in cylindrical coordinates r , θ , z defined by the
=
relations cosθ , y r=
x r= sin θ , z z , the Laplace equation ∇ 2u =
0 takes
the form
∂ 2u 1 ∂u 1 ∂ 2u ∂ 2u
+ + + 0
=
∂ r 2 r ∂r r 2 ∂θ ∂z 2
Solution. The Laplace equation in Cartesian coordinates is
∂ 2u ∂ 2u ∂ 2u
∇2 u = + + =0
∂x 2 ∂y 2 ∂z 2
The relations between Cartesian and cylindrical coordinates gives
r2 =x 2 + y 2 ,θ =tan −1 ( y / x ) , z =z

Since
u = u ( r ,θ , z )
u x = u x rx + uθ θ x + u z z x = ur cos θ − u  sin θ 
θ 
 r 

 cos θ 
u y = ur ry + uθ θ y + u z z y = ur sin θ − uθ  
 r 
u z = ur rz + uθ θ z + u z= u z
For the second order derivatives, we find
( ux )x =
u xx = ( ux )r rx + ( ux )θ θ x + ( ux ) z z x
  sinθ    sinθ    sinθ 
= ur cosθ − uθ    cos θ + ur cosθ − uθ   − 
  r  r   r  θ  r 

 sinθ sinθ 
=  urr cosθ − urθ + uθ 2  cosθ
 r r 
 208 Differential Equations: Theory and Applications

 sinθ cosθ  sinθ 

+  urθ cosθ −ur sinθ −uθθ − uθ  − 
 r r  r  (28)
Similarly
( u y ) y =( u y ) ry + ( u y ) θ y + ( u y ) z y
u yy =
r θ z

 cosθ cosθ 
=  ur sinθ + uθ − uθ 2  sin θ
 r r 
 cosθ sinθ  cos θ 
+  urθ sinθ + +ur cosθ + uθθ − uθ  
 r r  r  (29)
u zz = u zz (30)
Adding Eq. (28)-(30), we obtain
1 1
∇ 2u = urr + ur + 2 uθθ + u zz
r r (31)
Example. Show that in spherical polar coordinates r , θ , φ defines by
relations x r sin
the = = θ cos φ , y r=
sinθ sin φ , z r cos θ , , the Laplace
2
0 takes the form
equations ∇ u =
∂  2 ∂u  1 ∂  ∂u  1 ∂ 2u
 r  +  sin θ  + 0
=
∂r  ∂r  sinθ ∂θ  ∂θ  sin 2 θ ∂φ 2
Solution. In Cartesian coordinates , the Laplace equation is
2
∇ u = u xx + u yy + u xx + u zz = 0

In spherical coordinates,
z
u = u ( r , θ , φ ) , r 2 = x 2 + y 2 + z 2 , cos θ = , tan φ = y / x
r .
It can be easily verified that
cosθ cosφ cosθ sinφ sin θ
θx = , θy = ,θ z = −
r r r
sinφ cosφ
=φx = ,φ y = φz 0
r sinφ r sinφ
 Elliptic Differential Equation 209

Now,
cos θ cos φ sinφ
u x = u x rx + uθ θ x + uφ φx = ur sin θ cosφ + uθ − uφ
r rsinφ

cos θ sinφ uφ sinφ

u y = u y ry + uθ θ y + uφφ y = ur sin θ sin φ + uθ +
r r sinφ
 sinθ 
u z = ur rz + uθ θ y + uφ φ y = ur cos θ + uθ  − 
 r 
For the second order derivatives,
u xx = ( u x ) r rx + ( u x ) θ θ x + ( u x )φ φx

 cosθ cosφ sinφ 

=  ur sin θ cos φ + uθ − uφ  (sin θ cosφ )
 r r sin θ  r

 cosθ cosφ sinφ   sin θ cos φ 

=  ur sin θ cos φ + uθ − uφ   
 r r sin θ  θ  r 
 cosθ cos φ sinφ   sin θ 
=  ur sin θ cos φ + uθ − uφ  − 
 r r sin θ  φ  rsin θ 

cos 2 θ cos 2 φ sin 2 φ

(
= sin θ cos φ
2 2
) u rr
+
r2
uθθ + 2 2 uφφ
r sin θ
 2sin θ cos θ cos 2 φ   2sinφ cos φ 
+ urθ   + urφ  − 
 r   r 

 2cos θ cos φ sinφ   cos 2 θ cos 2 φ sin 2 φ 

+ uθφ  −  + ur  − + 
 r 2 sin θ   r r 

 sin φ cos φ cos 2 φ cos φ sinφ sin φ cos φ 

+ uφ  + + 2 2 
 r2 r 2 sin 2 θ r sin θ 
(32)
u yy =( u y ) ry + ( u y ) θ θ y + ( u y ) φ y
r φ

 cosθ sin φ cos φ 

=  ur sin θ sin φ + uθ + uφ ,
 r r sin θ 
 210 Differential Equations: Theory and Applications

 cos θ sin φ cos φ  cos θ sin φ

+  ur sin θ sin φ + uθ + uφ 
 r sin θ θ r

 cos θ sin φ cos φ  cos φ

+  ur sin θ sin φ + uθ + uφ 
 r sin θ φ r sin θ

cos 2 θ sin 2 φ cos 2 φ

( )
= sin 2 θ sin 2 φ urr +
r2
uθθ +
r 2 sin 2 θ
uφφ

 2sin θ cos θ sin 2 φ   2 cos φ sin φ 

=
+ u rθ   + urφ  
 r   r 

 2 cos θ cos φ sin φ   cos 2 θ sin 2 φ cos 2 φ 

+uθφ  +
 ru + 
 r 2 sin θ   r r 
 sin φ cos φ sin φ cos φ cos 2 θ sin φ cos φ 
+ uθ  − − 2 2 − 
 r2 r sin θ r 2 sin θ 
Similarly,
u zz = ( u z )r rz + ( u z )θ θ z + ( uz )φ φz

 sinθ   sinθ  sin θ

=  ur cos θ − uθ  ( cos θ ) +  ur cos θ − uθ  —
 r r  r θ r
2sin θ cosθ sin 2 θ sin 2θ cosθ sin θ
= urr cos 2 θ − urθ + uθθ 2
+ u + uθ
r2
r
r r r (34)
Adding Eq. (32) - (34), we obtain
1 1 2 cos θ
∇ 2u = urr + u + 2 2 uφφ + ur + 2
2 θθ
uθ = 0
r r sin θ r r sinθ
which can be rewritten as
∂  2 ∂u  1 ∂  ∂u  1 ∂ 2u
∇ 2u
=  r  +  sin θ  + = 0
∂r  ∂r  sin θ ∂θ  ∂θ  sin 2 θ ∂φ 2 (35)

4.5. SEPARATION OF VARIABLES

The method of Separation of variables is applicable to a large number of
classified linear homogeneous equations. The choice of the coordinate
 Elliptic Differential Equation 211

system in general depends on the shape of the boundary For example,

consider a two –dimensional Laplace equation in Cartesian coordinates .
∇ 2u = u xx + u yy = 0
(36)
We assume the solution in the form
u ( x, y ) = X ( x ) Y ( y )
(37)
Subtracting in Eq. (36), we get
X ′′Y + Y ′′X =0
i.e.
X ′′ Y ′′
= = k
X Y
where k is separation constant. Three cases arise.
2
Case I Let k = p , p is real. Then
d2X 2 d 2Y
2
=
p X 0 and 2
p 2Y 0
+=
dx dy
whose solution is given by
X c1e px + c2 e − px
=
and
=Y c3 cos py +c4 sin py
Thus the solution is
u ( x, y ) = ( )
c1e px + c2 e − px ( c3 cos py +c4 sin py )
(38)
Case II Let k = 0. Then
d2X d 2Y
= 0=
and 2 0
dx 2 dy
Integrating twice, we get
=
X c5 x + c6
and
=
Y c7 y + c8
 212 Differential Equations: Theory and Applications

The solution is therefore,

u ( x, y ) =
( c5 x + c6 )( c7 y + c8 ) (39)
2
Case III let k = − p . proceeding as in Case I, we obtain
=X c9 cos px + c10 sin px
py − py
=
Y c11e + c12e
Hence, the solution in this case is
u ( x, y ) = (
( c9 cos px + c10 sin px ) c11e + c12e
py − py

) (40)
c ( i = 1, 2,..,12 )
In all these cases, 1 refer to integration constants, which
are calculated by using the boundary conditions.

4.6. DIRICHLET PROBLEM FOR A RECTANGLE

The Dirichlet problem for a rectangle is defined as follows:
2
PDE: ∇= u 0 , 0 ≤ x ≤ a ,0 ≤ y ≤ b

BCs:
( x, b ) u=
u= ( a , y ) 0, u=
( 0 , y ) 0 , u=
( x ,0 ) f ( x ) (41)
This is an interior Dirichlet problem. The general solution of the
governing PDE, using the method of variables separable, is discussed in
previous section. The various solution of the Laplace equation are given by
Eq. (38-40). Of three solutions, we have to choose that solution which is
consistent with the physical nature of the problem and the given boundary
conditions as despaired .

Figure 3: Dirichlet Boundary Conditions

 Elliptic Differential Equation 213

Consider the solution given by Eq. (38)

u ( x, y ) = ( )
c1e px + c 2e px ( c3 cos py +c4 sin py )

u ( 0 , y ) = 0, we get
Using the boundary condition :
( c1 + c2 )( c3 cos py + c4 sin py ) =
0

which means that either

c1 + c2 0 or c3 cos py +=
= c4 sin py 0.
But
c3 cos py + c4
sin py ≠ 0;
Therefore
0
c1 + c2 =
(42)
u ( a , y ) = 0 , Eq. ( 38 )
Again, using the BC; gives
(c1eap
+ c2e − ap ) ( c cos py +c
3 4 sin py ) =
0

Implying thereby
0
c1e + c2 e − ap =
ap
(43)
c ,c ,
To determine the constants 1 2 we have to solve Eq. (42) and ( 43),
being homogeneous, the determinant
11
=0
e e − ap
ap

For the exercise of non-trivial solution, which is not the case. Hence
±( , ) = 0
only trivial solution is possible.
u ( x, y ) =
( c5 x + c6 )( c7 y + c8 ) ,
If we consider the solution given by Eq. (39)
the boundary conditions:
u=( 0, y ) u=( a, y ) 0 again yields a trivial
solution. Hence, the possible solutions given by Eq. (38) and (39) are ruled
out. Therefore, the only possible solutions given by Eq. (38) and (39) are
rule out Therefore, the only possible solutions obtained from Eq.(40) is
 214 Differential Equations: Theory and Applications

u( x , y ) = (
( c9 cos px + c10 sin px ) c11e py + c12 e− py )
u ( 0, y ) = 0, c9 = 0. Also, the other BC:
Using the BC : we get
u ( a, y ) = 0
yeilds
( )
c10 sin pa c11e py + c12 e − py =
0

For non trivial solution, 10 cannot be zero , implying sin pa = 0 , which

c
n π or p n=
is possible if pa = = π / a., n 1 , 2 , 3, ... therefore, the possible
non-trivial solution after using the superposition principle is
∞
nπ x
=u ( x , y) ∑ sin  an exp ( nπ y / a ) + bn exp ( −nπ y / a ) 
n =1 a  (44)
u ( x ,b ) = 0 ,
Now, using the BC : we get
nπ x
sin  an exp ( nπ b / a ) +bn exp ( −nπ b / a )  =
0
a 
Implying thereby
an exp ( n π b / a ) + bn exp ( −nπ b / a ) =
0

which gives
exp ( nπ b / a )
bn = −an
exp ( − nπ b / a ) =
, n 1 , 2 ,…, ∞
The solution (44) now becomes
∞
2an sin ( nπ x / a )  exp {nπ ( y − b ) / a} − exp {−nπ ( y − b ) / a} 
u(x , y ) =∑  
n =1 exp ( − nπ b / a )  2 
∞
2an
∑ exp ( −nπ b / a ) sin ( nπ x / a ) sin h {nπ ( y − b ) / a}
n =1

  nπ x  
2an / exp    = An . Then the solution can be written in the
Let   a 
form
 Elliptic Differential Equation 215

∞
u ( x , y)
= ∑ A sin ( nπ x / a ) sin h {−nπ ( y − b ) / a}
n =1
n

Finally , using the non-homogenous boundary condition:

u ( x, y ) = f ( x ) ,
we get
∞

∑ A sin ( nπ x / a ) sin h ( −nπ x / b ) =

n =1
n f ( x)

which is half-range Fourier series. Therefore ,

2
a
An sin h ( −nπ x / b ) = f ( x ) sin ( nπ x / b ) dx
a ∫0
Thus, required solution for the given Dirichlet problem is
∞
u ( x, y ) = ∑An sin ( nπ x / b ) sin h nπ y − b / a
n =1
{ ( ) } (47)
where
2 1
a
An =
a sin h ( −nπ x / b ) ∫ f ( x ) sin ( nπ x / b ) dx
0

4.7. THE NEUMANN PROBLEM FOR A

RECTANGLE
The Neumann problem for a rectangle is defined as follows
PDE : ∇ 2 u= 0, 0 ≤ x ≤ a, 0 ≤ y ≤ b

x ( 0, y )
BCs : u= x ( a, y )
u= y ( x, y )
0, u= y ( x, b )
0 , u= f ( x)
(48)
The general solution of the Laplace equation using the method of
variables separated is found to be
u ( x , y) = (
( c1 cos px +c2 sin px ) c3e py + c4e− py )
u x ( a, y ) = 0
The BC : gives
0= (
−c2 p c3e py + c4 e − py )
 216 Differential Equations: Theory and Applications

Implying
c2 = 0. Therefore,

(
u ( x , y ) c1 cos px c3e py + c4 e − py
= ) (49)
u x ( a, y ) = 0
The BC: gives
(
0 c1 p sin pa c3e py + c4 e − py )
For non-trivial solution,
c1 ≠ 0, implying
nπ
sin =
pa 0, = π,
pa n= (π 0,1, 2,…)
=
a
Thus the possible solution is
nπ x
=u (=
x ,y) cos
a
(
Ae nπ y / a + Be − nπ y / a )
(50)
u y ( x, 0 ) = 0,
Now, using the BC : we get
nπ x  nπ nπ 
0 cos A −B 
a  a a 

Implying B = A. Thus, the solution is

nπ x   nπ y   nπ y  
u ( x , y ) A cos
=  exp   + exp  − 
a   a   a 
nπ x nπ y
2 A cos cosh
a a

Using the superposition principle and defining

2 A = An , we get
∞
nπ x nπ y
u = ∑An cos cosh
n =0 a a
u y ( x, b ) = f ( x ) ,
Finally, using the BC: we get
∞
nπ x nπ nπ b
f ( x ) = ∑An cos sin h
n =1 a a a
which is the half-range Fourier cosine series. Therefore,
 Elliptic Differential Equation 217

nπ nπ b 2 nπ x
a
An sinh = ∫ f ( x ) cos dx
a a a0 a
Hence, the required solution is
∞
nπ x nπ y
u A0 + ∑ An cos
= cosh
n =1 a a (52)
where
2 1 nπ x
a
An = ∫ f ( x ) cos dx
nπ sinh ( nπ b / a ) 0 a

4.8. INTERIOR DIRICHLET PROBLEM FOR A

CIRCLE
The Dirichlet problem for the circle is defined as follows:

PDE: ∇ u= 0, 0 ≤ r ≤ a, 0 ≤ 0 ≤ 2π
2

u ( a=
, θ ) f ( θ ) , 0 ≤ θ ≤ 2π
BC:
f (θ )
where is a continuous function on ∂IR. The task is to find the value
of u at any point in the interior of the circle IR in terms of its value on ∂IR
such that u is single valued and continuous on IR.
In view of circular geometry , it us natural to choose polar coordinates
to solve this problem and then use the variables separable method. The
requirement of single-valuedness of u in IR implies the periodicity
condition, i.e.,
u ( r ,θ =
+ 2 π ) u ( r , θ ) , 0 ≤ r ≤ a,
(54)
2
0 which in polar coordinates can be written as
From Eq. (27) , ∇ u =
1 1
0
urr + ur + 2 u θθ =
r r
u ( r ,θ ) R ( r ) H (θ ) ,
If the above equations reduces to
 218 Differential Equations: Theory and Applications

1 1
R′′H + R′H + 2 RH ′′ =
0
r r
This equation can be rewritten as
r 2 R′′ + rR′ H ′′
= −= k
R H (55)
which means that a function of r is equal to a function of θ and, therefore,
each must be equal to a constant k ( a separation constant ).
Case I Let k = λ . Then
2

r 2 R '' + rR ' − λ 2 R =
0 (56)

which is Euler type of equation and can be solved by setting r = e . Its

z

solution is
R=c1eλ z + c2 e – λ z =
c1r λ + c2 r – λ
Also,
H ′′ + λ 2 H = 0
whose solution is
=H c3 cos λθ + c4 sin λθ
Therefore,
u ( r ,θ ) = (
c1r λ + c2 r – λ ) ( c cos λθ + c
3 4 sin λθ )
(57)
Case II Let k = −λ 2 . Then
r 2 R′′ + rR′ + λ 2 R
= 0 , H ′′ − λ 2 H= 0
Their respective solutions are
=R c1 cos ( λ in r ) + c2 sin ( λ in r )

H c3 e λθ + c4 e − λθ
=
Thus
u ( r ,θ ) (
c1 cos ( λ in r ) c2 sin ( λ in r )  c3 e λθ +c4 e – λθ ) (58)
Case III Let k = 0 , then we have
 Elliptic Differential Equation 219

r R '' + R ' =
0
R' ( r ) = V ( r ) ,
Setting we obtain
dV dV dr
r +V =0, + = 0
dr i.e. , V r

Integrating , we get in
Vr in c . Therefore,
c1 dR
V= =
r dr
On integration ,
=R c1 In r + c2
Also,
H ′′ = 0
After integrating twice , we get
H c3θ + c4
=
Thus,
u ( r ,θ ) =( c1 in r + c2 )( c3θ + c4 )
(59)
Now, for the interior problem, r = o is a point in the domain IR and
since in r = 0 is a point in the domain IR and since In r is not defined
at r = 0, the solutions (58) and (59) are not acceptable. Thus the required
solution is obtained from Eq. (57). The periodicity condition in θ implies
c3 cos λ + c4 sin ( λ (θ + 2π ) ) + c4 sin ( λ (θ + 2 π ) )

i.e.,
c3 cos λθ − cos ( λθ + 2λπ )  + c4 [sin ( λ (θ + 2π ) )] =
0

or
2sin λ π c3 sin ( λθ + λπ ) − c4 cos ( λθ + λπ )  =
0

Implying sin
λπ 0 ,=
= λπ nπ = , λ n= ( n 0 ,1 , 2 , ,…) . Using the
principle of superposition and renaming the constants, the acceptable general
solution can be written as
 220 Differential Equations: Theory and Applications

∞
u ( r ,θ ) = ( )
∑ cn r n + dn r −9 ( an cos nθ + bn sin nθ )
n =0 (60)

At r = o, the solution should be finite, which requires

d n = 0. Thus the
appropriate solution assumes the form
∞
u ( r ,θ )
= ∑r ( A cos nθ + B
n
n n sin nθ )
n =0

For n − 0, let the constant A0 be A0 / 2. Then the solution is

A0 ∞ n
u ( r ,θ ) = + ∑ r ( a n cos n θ + Bn sin nθ )
2 n =1 (61)

which is a full-range Fourier series. Now we have to determine

An and Bn
u ( a, θ ) = f (θ )
so that the BC: is satisfied.
i.e.,
∞
=f (θ ) ∑a n
( An cos n θ + Bn sin nθ )
n =1

Hence,
2π
1
A0 =
π ∫ f (θ ) dθ
0

2π
a n An = 1π ∫ f (θ ) cos θ dθ
0 (62)
2π
1
a= ∫ f (θ ) sin nθ d=
θ ,n 1 , 2 , ,…
n
Bn
π 0

In Eq. we replace the dummy variables θ by φ to distinguish this variable

from the current variables θ in Eq. Substituting Eq. into Eq., we obtain the
relation
2π ∞  n 2π
1 r cos nθ
u ( r ,θ )
= ∫0 f ( φ ) d φ + ∑  n ∫ cos ( nφ ) f (φ ) dφ
2π n =1  a π 0
 Elliptic Differential Equation 221

2π
r n sin nθ 
+ n ∫0 sin ( nφ ) f ( φ ) d φ 
a π 
Interchanging the order of summation and integration, we get
2π 2π n
1 1 ∞
r
u ( r ,θ ) = ∫ f (φ ) d φ + ∫0 f ( φ ) ∑   {cos nφ cos nθ + sin nφ sin nθ } dφ
2π 0 π n =1  a 

1
2π
 1 ∞  r n 
=∫ f (φ )  + ∑   cos n (φ − θ )  dφ
π 0  2 n =1  a   (63)
∞ 2
r
c ∑   cos n (φ − θ )
n =1  a 

∞ n
r
s ∑   cos n (φ − θ )
n =1  a 

So that
∞ n
 r i(φ −θ ) 
∑
c + is =  e
n =1  a


r
r < a,   < 1 and e1(φ −θ ≤ 1,
Since a

(r / a) r i (φ −θ )
n
 r  i(φ −θ ) 
∞
=c + is ∑ =  e 
n =1  a     r  i(φ −θ ) 
1 −  a  e 
   

  e
 
a
{
 r  i(φ −θ )
(
− r 2 / a2 )}
=
  r  i(φ −θ )    r  i(φ −θ ) 
1 −  a  e  1 −  a  e 
      

Equating the real part on both sides, we get

222 Differential Equations: Theory and Applications

 r   r 2 
  cos (φ − θ ) −  2  
 a   a  a2 − r 2
c =
  2r  2  2  a 2 − 2ar cos (φ − θ ) + r 2 
1 −
  a  cos ( φ − θ ) + r 2
/ a ) 
   
Thus , the required solution takes the form

u ( r ,θ ) =
1
2π
(a 2
)
− r 2 f (φ )
2π ∫ a 2
− 2ar cos (φ − θ ) + r 2 
dφ
0  (64)
This is known as Poisson’s integral formula for a circle, which gives a
unique solution for the Dirichlet problem. The solution can be interpreted
physically in many ways: it can be thought of as finding the potential
u ( r , θ ) as f (φ )
a weighted average of the boundary potential weighted by
the Poisson kernel p, given by
a2 − r 2
p=
 a 2 − 2ar cos (φ − θ ) + r 2 

u ( r ,θ )
It can also be thought of as a steady temperature distribution
in a circular disc, when the temperature u on its boundary ∂IR is given by
u = f (φ )
which is independent of time.

4.9. EXTERIOR DIRICHLET PROBLEM FOR A

CIRCLE
The exterior Dirichlet problem is described by
PDE : ∇ 2 u =
0
BC : u ( a, θ ) = f (θ )
(65)
u must be bounded as r → ∞
By the method of separation of variables, the general solution of
2
0 in polar coordinates can be written as
∇ u=
∞
u ( r ,θ ) = (
∑ cn r n + dn r − n )( an cos nθ + bn sin nθ )
n =0
 Elliptic Differential Equation 223

c = 0.
Now as , r → ∞ , we require u tobe bounded, and therefore, n
After adjusting the constants, the general solution now reads
∞
u ( r ,θ )
= ∑( A n cos n θ + Bn sin nθ )
n =0

With no loss of generality, it can also be written as

A0 ∞ − n
u ( r ,θ ) = + ∑r ( An cos nθ + Bn sin nθ )
2 n =1 (66)
u ( a,θ ) = f (θ ) ,
Using the BC: we obtain
∞
A0
f (θ ) = + ∑a – n ( An cos n θ + Bn sin nθ )
2 n =1
f (θ ) ,
This is a full range Fourier series in where
2π
1
A0 =
π ∫ f ( θ ) dθ
0

2π
1
∫ f ( θ ) cos nθ dθ
−n
a An =
π 0 (67)
2π
1
a − n Bn =
π ∫ f ( θ ) sin nθ dθ
0

In Eq. (67) we replace the dummy variables θ by φ so as to distinguish

it from the current variable θ . We then introduced the changed variables
into solution (66) which becomes
2π ∞  −n 2π
1 r an
u ( r ,θ )
= ∫ f (φ )d φ + ∑  cos n θ ∫ cos ( nφ ) f ( φ ) d φ
2π 0 n =1  π 0

2π
r −n an 
sin nθ ∫0 sin ( nθ ) f ( φ ) d φ 
+ π  or
 224 Differential Equations: Theory and Applications

1
2π
 1 ∞  a n 
u ( r , θ ) =∫ f (φ )  + ∑   cos n (φ − θ )  dφ
π 0  2 n =1  r   (68)
Let
∞ n
a
C ∑   cos n (φ − θ )
n =1  r 

∞ n
a 
S = ∑ 
n =1  n  sin
n (φ − θ )

Then,
n
 a  i ( φ −θ ) 
∞

∑
C + iS =   e
n =1  r 


Since
a
< 1 , e i (φ − θ ) ≤ 1
r
we have
 a   i(φ −θ ) 
a e i (φ −θ )   e 
=C + iS = r
r   a  i (φ – θ )    a  i (φ – θ )    a  – i (φ – θ ) 
1 r  e  1 –  r  e  1 −  r e 
          
Hence,
 a   a2  
   cos ( φ − θ ) −  2  
 r   r 
C= 
  2a   a2 
1
 −  cos ( φ − θ ) +  2 
  r   r 
Thus the quantity in the square bracket on the right-hand side of Eq. (68)
becomes
 Elliptic Differential Equation 225

 a   a2  
  cos ( φ –θ )  2 
1  r   r  r2 +a2
+ =
2   2a  2  2  r 2 − 2ar cos ( φ −θ ) + a 2 
2 1−   cos ( φ – θ ) + a 2
(
/ r  )
  r  

Therefore, the solution of the exterior Dirichlet problem reduce to that

of an integral equation of the form

u ( r ,θ ) =
1
2π
(r 2
)
− a 2 f (φ )
2π ∫
0  r − 2ar cos ( φ − θ ) + a 2 
2
dφ
(69)
Example. Find the steady state temperature distribution in a semi-
circular plate of radius a. insulated on both the faces with its curved
U0
boundary kept at a constant temperature and its bounding diameter kept
at zero temperature .
Solution The Governing heat flow equation
ut = ∇ 2 u

0,
In the steady state, the temperature is independent of time ; hence
and the temperature satisfies the Laplace equation . The problem can now be
stated as follows : To solve
1 1
PDE : ∇ 2 u ( u , θ ) = u rr + + uθ θ = 0
ur r 2

= u ( a, θ ) U 0 =
, u ( r , 0 ) 0,=
u ( r,π ) 0
BCs:
The acceptable general solution is
u ( r ,θ ) = (
cr λ + Dr − λ ) ( Acos λθ + B sin λθ ) (70)
u ( r , 0 ) = 0, u ( r,π ) = 0
From the BC : we get A = 0; however, the BC :
also gives
(
B sin λπ cr λ + Dr λ − =
0 )
 226 Differential Equations: Theory and Applications

Figure 4

Implying either B = 0 or sin = λπ 0= . B 0 gives a trivial solution. For

a non-trivial solution. We λ = n . Hence the possible solution is
= (
u ( r ,θ ) B sin nθ Cr λ + Dr − λ ) (71)
In eq. (71), we observe that as r → 0, the term r → ∞. But the solution
−λ

=
should be finite at r 0=and so D 0 . Then after adjusting the constants, it
follows form the superposition principle that
∞
u ( r ,θ ) = ∑ B n a n sin nθ
n =1

which is a half –range Fourier sine series . Therefore,

π  4U 0
2  , for
= n 1 , 3 ,…
= =∫ U 0 sin n θ d θ  nπ
n
Bn a
π 0  0 , for n = 2 , 4. , , , , ,

Hence,
4U 0
=
Bn , n 1 , 3 ,…
=
nπ a n

With these values of

Bn , the required solution is
n
4U 0
1 r∞
u ( r ,θ ) = ∑
π π n  a  sin nθ
 Elliptic Differential Equation 227

4.10. INTERIOR NEUMANN PROBLEM FOR A

CIRCLE
The interior Neumann problem for a circle is described by
PDE : ∇ 2 u =
0, 0 ≤ r < a ; 0 ≤θ 2π (72)
∂u ∂u ( a , r )
=
BC : = g= (θ ) ,r a
∂n ∂r
Following the method of separation of variables, the general solution
2
0 in polar coordinates is given by
(60) of equation ∇ u =
∞
u ( r ,θ ) = (
∑ cn r n + d n r − n )( a n cosnθ + bn sin nθ )
n=0

d n = 0.
At r = 0, the solution should be finite and , therefore, Hence ,
after adjusting the constant the general solution becomes
∞
u ( r , θ ) = ∑r n ( An cos nθ )
n=0

With no loss of generally , this equation can be written as

∞
u ( r ,θ )
= ∑r n
( An cosn θ + Bn sin nθ )
n=0 (73)
0
∂u
∑ nr n −1
( An cos nθ + Bn sin nθ )
∂r n =1

Using the BC:

∂u
( a ,θ ) = g ( θ )
∂r
we get
∞
g (θ ) ∑ na ( A cos nθ
n –1
n + Bn sinθ )
n =1

g (θ )
which is a full-range. Fourier series in , where
2π
1
na n −1 An =
π ∫ g ( θ ) cos nθ
0
dθ
 228 Differential Equations: Theory and Applications

2π
1
na n − 1 Bn =
π ∫ g ( θ ) sin nθ
0
dθ

Here, we replace the dummy variables θ by φ to distinguish form the

θ in Eq ( 74 ) .
current variables now introducing Eq. (75) into Eq.(73), we
obtain
2π
A0 ∞ rn
u ( r , θ=
) +∑ ∫ g ( φ ) ( cos nφ cos nθ sin nφ sin nθ )dφ
2 n = 1 nπ a n − 1 0

or
2π ∞ n
A0 r 
u ( r , θ=
) + ∫ g (φ ) ∑   1/ nπ cos n (φ –θ ) dφ
2 0 n = 1 a 
(76)
or this solution can also be expressed in an alternative integral form as
follows: Let
n
r a
C=
∑  cos n (φ − θ )
 a  nπ
n
r a
S=
∑  sin n (φ − θ )
 a  nπ
Therefore,
n n
 r  a in (φ −θ ) a ∞  r i (φ −θ )  1
C + iS =
∑  e =∑ e
 a  nπ π n =1  a  n
  r i (φ −θ )   r i (φ −θ )   r i (φ −θ )  
  e   e   e  
a a  + a  + a  +…
= 
π 1 2 3 
 

or
a  r  a  r r 
− In 1 − cos (φ − θ ) − i sin (φ − θ ) 
C + iS = In 1 − ei (φ −θ )  =
π  a  π  a a  (77)
 Elliptic Differential Equation 229

To get the real part of In z , we may note that

= =
W In z or z e w
i.e.,
x + iy= eu +iv= eu cos v + ieu sin v.
Therefore,
=x e=
u
cos v, y eu sin v
2
e 2u = x 2 + y 2 = z

u = In z .
i.e.,
Therefore ,
2 2
a  r  r 
− In 1 − cos (φ − θ )  +  sin (φ − θ ) 
C=
π  a  a 

a a 2 − 2ar cos (φ − θ ) + r 2
= − In
π a2
Thus the required solution is

A0 a
2π
a 2 − 2a r cos ( φ − θ ) + r 2
u ( r , θ=) − ∫ In g (φ ) d φ
2 π a2
0 (78)
which is again an integral equation.

4.11. SOLUTION OF LAPLACE EQUATION IN

CYLINDRICAL COORDINATES
The Laplace equation in cylindrical coordinates assumes the following form:
1 1
∇ 2 u = urr + ur + 2 uθθ + u zz = 0
r r (79)
We now seek a separable solution of the form
u ( r ,θ , z ) = F ( r ,θ ) Z ( z )
(80)
Subtracting Eq. (80) into Eq. (79), we get
 230 Differential Equations: Theory and Applications

∂2 F 1 ∂F 1 d 2F d 2Z
Z + Z + Z + F 0
=
∂r 2 r ∂r r 2 dθ dz 2
 ∂2 F 1 ∂ F 1 d 2F  1 d2 Z 1
 2 + +  =− = k ( say )
 ∂r r ∂ r r 2 dθ 2  F d z2 Z
or
where k is a separation constant. Therefore , either
d 2Z
0
+ kZ = ( 81)
d z2
or
∂2 F 1 1 ∂2 F
+ ∂ F \ ∂r + − KF =0 ( 82 )
∂r2 r r 2 ∂θ 2
If k is real and positive , the solution of Eq. (81) is
=Z c1 cos kz + c2 e − kz

If k is equal to zero, the solution of Eq. (81) is

=
Z c1 z + c2
From physical considerations, one would except a solution which decays
with increasing z and, therefore, the solution corresponding to negative
k is acceptable. Let k = −λ 2 . Then
=Z c1 e λ z + c2 e – λ z ( 83 )
Equation (82) now become
∂ 2 F 1 ∂F 1 ∂ 2 F
+ + 2 + λ2F =
0
∂r 2
r ∂r r ∂θ 2

F ( r , θ ) = f ( r ) H (θ )
Let .Substituting into the above equations we get
1 ' 1
f '' H + f H + 2 f H '' + λ 2 f H =
0
r r
or
1 H ''
( r 2 f '' + rf ' + λ 2 r 2 ) f
=
−
H
K '' ( say )
=
 Elliptic Differential Equation 231

Form physical consideration, we except the solution to be periodic in

θ , which can be obtained when k ′ is positive and k ' = n 2 . therefore , the
acceptable solution will be
=H c3 cos n θ + c4 sin nθ ( 84 )
' 2
When k = n , we will also have
d2 f d f
r2
dr 2
+r
dr
(
+ λ 2 r 2 − n2 f =0 ) ( 85 )

which is a Bessel ‘s equation whose general solution is given by
=f Ajn ( λ r ) + BY n ( λ r ) ( 86 )

Jn ( λr ) Y n ( λr )
Here , and are the nth order Bessel function of first
Y ( λr ) → − ∞ r → 0 ,Yn ( λ r )
and second kind, respectively. Since n as
becomes unbounded at r = 0. Continuity of the solution demands B = 0 .
2
0 is
Hence the most general and acceptable solution of ∇ u =
u ( r ,θ , r ) = (
J n ( λ r ) c1eλ z + c2 e – λ z ) ( c cos nθ + c
3 4 sin nθ ) ( 87 )
Example. A homogeneous thermally conducting cylinder occupies
the region 0 ≤ r ≤ a, 0 ≤ θ ≤ 2π , 0 ≤ z ≤ h , where r ,θ , z are cylindrical
coordinates . The top z = h and the lateral surface r = a are held at 0 ,
o

0
while the base z = 0 is held at 100 . Assuiming that there are no sources of
heat generation within the cylinder, find the steady –temperature distribution
within the cylinder.
Solution. The temperature u must be a single valued continuous
function. The steady state temperature satisfies the Laplace equation ‡E
inside the cylinder. To compute the temperature distribution inside the
cylinder , we have to solve the following BVP:
2
0
PDE: ∇ u =
= : u 0= on z h ,
o
BCs
=u 0=
o
on r a ,
=u 100
= o
on z 0
 232 Differential Equations: Theory and Applications

The general solution of the Laplace equation in cylindrical coordinates

is
(
r ( r , θ , z ) = J n ( λ r )( c1 cosnθ +c2 sin nθ ) c3 e λ z + c 4 e – λ z )
0
Since the face z = 0 is maintained at 100 and since the other and lateral
0
surface of the cylinder are maintained at 0 , the temperature at any point
inside the cylinder is obviously independent of θ . This is possible only
when n = 0 in the general solution .Thus,
(
u ( r , z ) j 0 ( λ r ) Ae λ z + Be − λ z
= )
: u 0=
Using the BC= on z h, we get

= (
0 J 0 ( λ r ) Ae λ h + Be − λ h )
λ −λh
Implying thereby Ae + Be 0 , from which
=
Aeλ h
B= −
e −λh
Therefore, the solution is
J0 ( λr ) A
=u( r ,z ) −λh
 λ ( z −h ) − e –λ ( z − h ) 
e 
e

or
( r , z ) J 0 ( λ r ) A1 sinh λ ( z − h )
= u=

A = 2 A e −λh
where 1 / , Now using the BC : u = 0 on r = a, we have
0 A1 J 0 ( λ a ) sinh λ ( z − h )

j0 ( λ a ) = 0 ,
Implying which has infinitely many positive roots.
Denoting them by
ξ n , we have ξ n = λ a , and therefore,
ξn
λ=
a
Thus the solution is
 Elliptic Differential Equation 233

ξ r  ξ 
u( r ,z ) =
A1 j0  n  sinh  n ( z − h )  , n =
1 , 2 ,…
 a  a 
Using the principle of superposition , we have
∞
ξr ξ 
u( r ,z ) ∑A n j0  n  sinh  n ( z − h ) 
n =1  a  a 
0
BC : u 100
The= = on z 0 gives
∞
ξ r  ξ 
100 ∑A n j0  n  sin h  n ( z − h ) 
n =1  a  a 
BC : u 100
The = = on z 0 gives
∞
 ξ h ξ r 
=100 ∑A
n =1
n j0  − n  j0  n 
 a   a 

which is a Fourier-Bessel series. Multiplying both sides with

rJ 0 (ξ m r / a )
and integrating, we get
a
ξ r  ∞
 ξh ξ r  ξ 
a
100 ∫ r =
j0  m  dr ∑ An sin h  − n  ∫ rjo  n  j0  mr  dr
 a 0  a   a 
0  a  n =1

Using the orthogonality property of Bessel’ s function , namely,

a  0, if i ≠ j
 2
∫0 xJ n ( ai x ) J n ( a j x ) dx =  a j 2 ( α ) , if i = j
 n +1
 2
i

ai , a j jn ( x ) = 0 ,
Where are the zeros at we have
1
 ξn r   ξn h  a 2
∞ 2
100 ∫ r j0=
 a  dr ∑ An sinh  a  2 j1 ( ξ n )
−
0   n =1  
Therefore,
200  ξn r
a

An =
 ξh ∫ rJ 0  a

 dr

a 2 sin h  − n  j12 ( ξ n )0
 a 
234 Differential Equations: Theory and Applications

Setting
ξn r a
= , dr
x= dx
a ξn

The relation for

An can also be written as
ξn
200
An =
 ξ h ∫ x j ( x ) dx
0
ξ n2 sinh  − n  j12 ( ξ n ) 0
 a 

Using the recurrence relation

d
x n jn –1 ( x ) =  x n jn ( x )  ,
dx 

For n =1 , we get
∫ x j0 ( x ) dx =
x j1 ( x )

Now,
An can be written as
ξ
 
 200 x j1 ( x )  200
An = 
 ξ 2 sinh  − ξ n h  j 2 ξ  ξ n sin h \ a ) j1 ( ξ )
 n  a  1 ( n )
  0
Hence, the required temperature distribution inside the cylinder is
 r   ξ  
∞
j0  ξ n  sin h   n  ( z − h ) 
 a  a  
u ( r , z ) = 200 ∑
n =1 ξ n sin h ( − ξ n h / a ) j1 (ξ n )

where
xn are the positive zeros of j 0 ( ξ ) .
Example . Find the potential u inside
the cylinder
0 ≤ r ≤ a, 0 ≤ θ ≤ 2π , 0 ≤ z ≤ h, if the potential on the top z = h and on the
lateral surface r = a is held at zero, while on the base z = 0, the potential
, 0 ) V0 (1 − r 2 / a 2 ) ,
u ( r ,θ =
where 0 is a constant r , θ , z are
V
is given by
cylindrical polar coordinates .
 Elliptic Differential Equation 235

Solution. The potential u must be a single-valued continuous function

and satisfy the Laplace equation inside the cylinder. To compute the potential
inside the cylinder, we have to solve the following BVP:
2
0
PDE: ∇ u =
=
BCs: u 0=
on z h,
=u 0=
on r a,

 r2 
V0 1 − 2  on z =
u= 0
 a 
In cylindrical coordinates , the general solution of the Laplace
equation\ /\ is
(
u ( r , θ , z ) = J n ( λ r )( c1 cos nθ + c2 sin nθ ) c3eλ z + c4 e − λ z )
 r2 
V0 1 − 2 
Since the face z = 0 has potential  a  which is purely, a
function of r and is independent of θ . This is possible only when n = 0 in
the general solution. Thus ,
(
u ( r , z ) j0 ( λ r ) Aeλ z + Be − λ z
= )
: u 0=
Using the BC= on z h , we obtain

=0 j0 ( λ r ) Aeλ h + Be − λ h ( )
λh −λh
Implying Ae + Be = 0, which yields
Ae λ h
B = − –λ h
e
Hence , the solution is
A
u( r ,z ) –λ h
j 0 ( λ r )  eλ ( z − h ) − e – λ ( z − h ) 
e
or
A
u ( r, z ) −λh
j 0 ( λ r ) sinh λ ( z − h )
e
 236 Differential Equations: Theory and Applications

Where 1
A = A / e λ h , Now, using the BC : u = 0 on the lateral surface ,
i.e., on r = a, we get
0 A1 J 0 ( λ a ) sinh λ ( z − h )

j0 ( λa ) = 0 .
Implying This has infinitely many positive roots; demoting
ξ n we shall have
them by
=ξ n λ=
a or λ ξ n / a

The solution now takes the form

ξ r  ξ 
u( r ,z ) A 1 j 0  n  sinh  n ( z − h )  ,
 a  a = n 1 , 2 , ,….
The principle of superposition gives
∞
ξ r ξ h 
u( r ,z ) ∑A n J 0  n  sin h  n ( z − h ) 
n =1  a   a 
 r2 
V 0  1− 2
u= 0
 on z =
The last BC :  a  yeilds
r2 ∞
 ξ h ξ r 
(1 − 2
V0 =
a
∑A
n =1
n Sinh  − n  J 0  n 
 a   a 
rj0 (ξ m r / a )
This is a Fourier-Bessel’s series . Multiplying both sides by
and integrating, we get
a
 r2   ξm r  ∞
 ξ h ξ r 
V 0 ∫  1− 2 rj
 0  dr ∑
= An sinh  − n  j0  n 
0
a   a  n =1  a   a 
Using the orthogonality property of Bessel functions
a  0, if i = j
 2
∫ xj ( α x ) jn ( a j x ) dx =  a 2
jn + 1 ( α i ) , if i ≠ j
n i
0 
 2
αi , α j jn ( x ) = 0 ,
Where are the zeros of we get
 Elliptic Differential Equation 237

 r2   ξn r  ξn h  a
a ∞ 2

V0 ∫  1 − 2 ∑  a  2 j1 ( ξ n )
2
 r j0   dr = A n sinh −
0
a   a  n =1  
which gives
2V0  r2   ξn r 
a
An
 ξn h  2 ∫0  1− a 2  r j0 
  a 
 dr
 j 1 ( ξn )
2
a sinh  −
 a 

Be letting ξ r a = x , this equation can be modified to

ξn
2V 0
An ∫ (ξ
2
− x 2 ) x j0 ( x ) dx
 ξ h
n
ξ n4 sinh  − n  j12 ( ξ n ) 0
 a 
Using the well-known recurrence relation
d
x α j α − 1 ( x )=  x α jα ( x )  for= α= 1 , 2 ,…
dx 
We get
∫ x j 0 ( x ) = xj 1 ( x ) , ∫ x 2 j1 ( x ) = x 2 j 2 ( x )

From these relations, we obtain

ξn
2V0
An = ∫ x 2 j1 ( x ) dx
 ξh  2
ξ n4 sinh  − n  J1 ( ξ n )
0
 a 
2V0 ξn
=  x 2 j2 ( x ) 
 ξ h  2 0
ξ n4 sinh  − n  J1 ( ξ n )
 a 
Thus,
4V 0 j2 ( ξ n )
An =
 ξ h
ξ n4 sinh  − n  J12 ( ξ n )
 a 
 238 Differential Equations: Theory and Applications

The recurrence relation

2n
J n −1 ( x ) + jn +1 ( x ) = jn ( x )
x
For n =1 gives
2
j0 ( ξ n ) + j2 ( ξ n ) = j1 ( ξ n )
ξn
Hence,
2
j2 ( ξ n ) = j1 ( ξ n )
ξn
j0 ( ξ n ) = 0 .
Since Therefore,
8V0 j1 ( ξ n )
An =
 h
ξ 3n sinh  − ξ n  j12 ( ξ n )
 a
Thus, the required potential inside the cylinder is
ξr   ξ 
8V0 j0  n  sinh  n ( z − h ) 
∞
u( r ,z )= ∑  a   a 
 ξ h 
n =1
ξ n3 j1 ( ξ n ) sinh  − n 
 a 

4.12. SOLUTION OF LAPLACE EQUATION IN

SPHERICAL COORDINATES
In example 3 the Laplace equation is expressed in spherical and has the
following form:
∂  2 ∂u  1 ∂  ∂u  1 ∂2 u
∇ 2u
=  r  +  sin θ  + = 0
∂r  ∂r  sin θ ∂θ  ∂θ  sin 2θ ∂φ 2 ( 88 )
Let us assume the separable solution in the form
u ( r ,θ ,φ ) = R ( r ) F ( θ ,φ ) ( 89 )

Subtracting of variables gives
 Elliptic Differential Equation 239

d  2 dR  − 1  ∂  sinθ ∂F  + 1 ∂ F 
2

r     
dr  dr  sin θ  ∂θ  ∂θ  sin θ ∂ φ 2 
= = −µ
R F
where µ is separation constant. Therefore,
1 d  2 dR 
r = −µ ( 90 )
R dr  dr 
1  ∂  ∂u  1 ∂2 F 
  sin θ  + =µ
F sinθ  ∂θ  ∂θ  sin θ ∂φ 2 

( 91 )

Equation
( 90 ) gives
2
d R dR
r2 2
+2r + µR =
0
dr dr

which is a Euler’s equation. Hence, using the transformation r = e , the

z

auxiliary equation can be written as

D ( D −1) + 2 D + µ = D 2 + µ = 0

d
D= .
where dz Its roots are given by
−1 ± 1 − 4 µ
D=
2
− α (α + 1 ) ;
µ=
Let then we get
 1  
2

 −1 ± 2  a +  
  2  1  1
D= =− ± a + 
2 2  2
– ( a + 1)
Hence, D = α and Therefore, the solution of Euler’s equation
is
R c1 e a + c2 r – ( a + 1 )
=

( 92 )
240 Differential Equations: Theory and Applications

µ a ( a +1 ) , Eq.
–=
Taking becomes
2
∂  ∂F  1 ∂ F
 sin θ + + a ( a +1 ) F sinθ =
0
∂θ  ∂θ  sin θ ∂ φ 2
Inserting

Into the above equation and separating the variables, we obtain

2
where v is another separation constant. Then

( 93 )
sin θ  d  dH   2
  sin θ  + a ( a +1 ) sin ( θ ) H  =
v
( 94 )
H  dθ  dθ  
The general solution of Eq. (93) is

0 , the solution is independent of φ which

Provided v ≠ 0 , if v =
corresponds to the ax symmetric case. Equation
( 94 ) becomes , for the ax
symmetric , case,
d  dH dx 
 sin θ  + a ( a +1 ) sin ( θ ) H =
0
dθ  dx dθ 
i.e.,
d  dH 

dx 
(
1 − cos 2 θ )
dx 
+ a ( a +1 ) H =
0

or
d  dH 

dx 
(
1− x 2
dx 
)+ a ( a +1 ) H =
0
 Elliptic Differential Equation 241

This is the well-known Legendre equation. Its general solution is given

by
=H c5 pa ( x ) + c6 Qa ( x ) , −1 ≤ x ≤1

p ,Q
where a a are Legendre function of the first and second kind respectively.
For connivance let α be a positive integer, say a = n . Then
=H c5 pn ( cos θ ) + c6 Qn ( cosθ ) ( 98 )

Continuity of implies the continuity of

H ( x ) at x = ±1 . Q ( x)
Since n has a singularity at x = 1, we choose
c6 = 0.
Therefore, in ax symmetric case the solution of Laplace equation in
spherical coordinates is given by
u ( r ,θ =
,φ ) {c r
1
a
+ c2 r − ( a + 1 ) } ( c )  c p ( cosθ ) 
3 5 n

After renaming the constant s and using the principle of superposition ,

we find the solution to be
∞
u ( r ,θ )
= ∑  A n r n + B n r – ( n + 1 )  p n ( cos θ )
n=0

Example. in a solid sphere of radius ' a′, the surface is maintained at the
temperature given by
 π
k cos θ , ≤ θ 2
f (θ )= 
 0 , π <θ < π
 2
Prove that the steady state temperature within the solid is
1 1 r  5 r
2
3 r
4

u ( r ,θ ) k  p0 ( cosθ ) +   p 1 ( cos θ ) +   p2 ( cosθ ) −   p4 ( cos θ ) +…
=
 4 2 a  16  a  32  a  

Solution It is known that the steady state temperature distribution is

governed by the Laplace equation. In spherical polar coordinates, the ax
symmetric solution of the Laplace equation in general with the assumption
that the temperature should be finite at the origin is given by Eq.(99)
In the form
 242 Differential Equations: Theory and Applications

∞
u ( r , θ ) = ∑ An r n pn ( cos θ )
n=0 (100)
u ( a, θ ) = f (θ ) ,
Using the given BC: we have
∞ ∞
u ( a=
, θ ) f=
(θ ) ∑ An a n pn ( cos
= θ) ∑b n pn ( cosθ )
=n 0=n 0

where This is a Fourier-Legendre series, where

1
2n + 1
f ( θ ) pn ( cos θ ) dθ
2 −∫1
bn =

In the present problem,

Let cos θ = x ,
1 1
1 1 k
=b0 ∫ . p0 ( x ) dx =
kx= ∫ kx .1 . dx
20 20 4
Hence, we get
k
A0 =
4
1
3 k
b=1
20∫k x . x . dx= = A1 a
2
Also,
Therefore,
k1
A1 =  
2a
1 1
5 5 3x 2 − 1 5
=b2 = ∫kxp2 ( x ) dx ∫=kx dx k
20 20 2 16
Thus,
5 1
A2 = k. 2
16 a
 Elliptic Differential Equation 243

Further,
1 1
7 7 5 x3 − 3x
( )
2 ∫0 2 ∫0
=b3 = kxp3 x dx =
kx dx 0
2
1
p4 ( =
x)
8
( )
35 x 4 − 30 x 2 + 3 ) ,
Similarly, noting that we get
3
b4 =
− k=A4 a 4
32
Hence,
3 1
− k . 4 ,…
A4 =
32 a
A , A1 , A2 ,…
Subtracting these values of 0 into Eq. (100), we obtain ,
finally, the required temperature as
1 1 r  
u ( r ,θ ) k  p0 ( cosθ ) +   p1 ( cosθ ) 
=
4 2 a  

5r
2
 3  r 4 
+   p2 ( cos θ ) +  −    p4 ( cosθ ) +…
16  a   32   a  
Example. find the potential at al point of space inside and outside of
a sphere of radius R = 1 which is maintained at a constant distribution of
u ( R=
, θ ) f=
(θ ) cos 2θ .
electric potential
Solution. It is known that the potential on the surface of a sphere is
governed by the Laplace equation. The Laplace equation in spherical polar
coordinates is
∂ 2 u 2 ∂u 1 ∂ 2 u cot θ ∂u 1 ∂ 2u
+ + + 2 + 0
=
∂r 2 r ∂r r 2 ∂θ r ∂θ r 2 sin 2 θ ∂φ 2
The possible general solution by variables separable method, after
using superposition principle , is given by Eq. (99). Thus have two possible
solution:
∞
u 1 ( r , θ ) = ∑ An r n pn ( cos θ )
n =0
(101)
 244 Differential Equations: Theory and Applications

∞
Bn
u 2 ( r ,θ ) = ∑ pn ( cos θ )
n=0 r n +1

(102 )
For point inside the sphere, we take the series (101). Why is this so?
Applying the BC : we obtain
∞
f ( θ ) = ∑ A n R n p n ( cos θ )
n=0

f (θ )
which is generalized Fourier series in terms of the Legendre
polynomials. Using the orthogonality property, we get
1
2n + 1
An R = ∫ f ( θ ) p ( x ) dx
n

2
n
−1

Let x = cos θ . Then we have

1
2n + 1
f ( θ ) p n ( cos θ ) sinθ dθ
2 −∫1
An =

For points outside the sphere, we takes the series (102). Why is this so ?
u ( R ,θ ) = f ( θ ) ,
Using the BC: we get
∞
Bn
f (θ ) = ∑ n +1
pn ( cosθ )
n=0 R

Again using the orthogonality property of Legendre polynomials, we

have
π
2n + 1 n + 1
Bn =
2
R ∫0 f ( θ ) p n ( cos θ ) sin θ dθ
In the present problem, it is assumed that at
= 1 , f ( θ =) cos 2θ= 2 cos 2 θ −=
R 1 2 x 2 −1 .
Hence
1
2n + 1
=An
2 −∫1
(
2 x 2 −1 Pn ( x ) dx )
However,
1
p2 ( x )
=
2
3 x 2 −1 ( )
 Elliptic Differential Equation 245

Therefore,
4 1
2 x 2 −1
= p2 ( x ) −
3 3
Thus,
2n + 1  4 
1 1 1
1
( ) ( ) ( )
2 −∫1  3 −∫1 3 −∫1
2
An  p 0 x dx − p0 x pn x dx 

Using the orthogonality property of Legendre polynomials, all integrals
=
vanish except those corresponding to n 0=and n 2 . We obtain therefore,
1
1 1 1
− . ∫ P 02 ( x ) dx =
A0 = −
2 3 −1 3
1
5 4 4
A2 . ∫ P 22 ( x ) dx
=
2 3 −1 3
Also ,
1
2n + 1
Bn =
2 −1 ∫ ( )
2 x 2 –1 pn ( x ) dx

2n + 1  4 
1 1
1
 ∫ 2 P ( x ) p ( x ) dx − ∫ P0 ( x ) pn ( x ) dx 
2  3 −1 3 −1
n

which, on using the orthogonality property , gives the non-vanishing
coefficients as
1 4
− , B2 =
B0 =
3 3
A0 A2 Eq . ( 101 ) ,
Subtracting these values of and into we obtain
1 4
u1 ( r , 0 ) =
− + r 2 p 2 ( cosθ )
3 3
Which gives the potential everywhere inside the sphere . Similarly.
B B Eq . ( 102 ) ,
Subtracting the values of 0 and 2 into we get
246 Differential Equations: Theory and Applications

1 4
u 2 ( r ,θ ) =
− + 2 P2 ( cosθ )
3r 3r
which gives the potential outside the sphere.
Example. Find a general spherically symmetric solution of the following

Helmholtz equation :
(
∇2 − k 2 u = 0 )
Solution In spherical polar coordinates , the Helmholtz equation can be
written as
∂ 2 u 2 ∂u 1 ∂ 2 u cot θ ∂u 1 ∂2u
+ + + + − k2u=
0
∂ r 2 r ∂θ 2 r 2 ∂θ 2 r 2 ∂θ r 2 sin 2 θ ∂ φ 2 (103)
In view of spherical symmetry , we look for u to be a function of r
alone. Hence . Eq. (103) becomes
∂ 2 u 2 ∂u
+ − k2u=
0
∂ r 2 r ∂r
Therefore , we have to solve
∂2 u ∂u
r2 2
+ 2r − k2 r2 u =
0
∂r ∂r
Let
1
u= F(r)
r
Differentiating twice with respect to r and rearranging, we obtain
∂u F ( r )
2r= + 2 r F '( r )
∂r 2 r
3
∂2u 3 −1/2
r2 =
− r F ( )
r − r −1/2
F '
( )
r + r 2
F ''
(r)
∂r2 4
Substituting the above relations, Eq. (104) becomes
 1
r 2 F '' ( r ) + r F ′ ( r ) −  k 2 r 2 +  F ( r ) = 0
 4
or
 Elliptic Differential Equation 247

 1 
2

r F ( r ) + rF ( r ) +  ( ik ) r −    F ( r ) =
2 '' ' 2 2
0
  2  

This is the Bessel equation whose solution is

F ( r ) A J 1 ( ikr ) + BY 1/2 ( ikr )
=
2

J 1 ,Y 1
where 2 2 are Bessel functions with imaginary arguments, and is

rewritten as
F ( r ) AI
= 1 ( kr ) + B K 1 ( kr )
2 2

Therefore,
 
=u ( r ) r −1/2  A I 1 ( kr ) + BK 1 ( kr ) 
 2 2 
But as r → ∞ , the solution should be finite, which is possible only if
A = 0 , . it is also known that for large z ,

π
K1 ( z )= e −z
2
2z
Thus the acceptable spherically symmetric solution of the Helmholtz
equation is given by
π – kz c − kz
=u ( r ) Br
= −1/2
e e
2 kr r
where
π
c=B
2k

4.13. MISCELLANEOUS EXAMPLE

Example. Show that the velocity potential for an irrational flow of an
incompressible fluids satisfies the Laplace solution.
Solution. Let us consider a closed surface ecnlosing a fixed volume in
the region occupied by a moving fluids .
248 Differential Equations: Theory and Applications

Figure 5

Let p be the density of the fluid. If n is a unit vector in the direction of

the normal to the element dS and q the velocity of the fluid at that point ,
then the inward normal velocity is
( − q . nˆ ) .. Hence the mass of the fluid
entering per unit time through the element dS is
( − q . nˆ ) dS . it follows
therefore that the mass of the fluid entering the surface S is unit time is
.

∬ p ( q . nˆ ) dS
S Also , the mass of the fluid within S is
.

∬∫ P dV
V

So the rate at which the mass goes on increasing is given by

. ..
∂ ∂P
∬
∂t V
∬
∫ P dV = ∫
∂ t
dV
V

By conservation of mass, the rate of generation of mass within a given

volume under the assumption that no internal sources are present is equal
to the net inflow of mass through the surface enclosing the given volume.
Thus,
. .
∂P
−∬ p ( q. nˆ ) dS
∬∫ ∂t dV =
V S
 Elliptic Differential Equation 249

−∬∫ div ( pq ) dV [ using the divergencetheorem ]

=
V

.
∂p
∬∫  ∂t + div ( pq )  dV =
0
Therefore, V  
Since the integrand is a continuous function and since this result is true
for any arbitrary volume element dV . it follows that the integrand is zero
. Therefore,
∂p
+ ∇ . pq =0
∂t
which is called the equation of continuity. For an incompressible fluids,
p = constant and therefore,
∇.q =
0

Further, if the flow is irrotational, i.e., there exist a velocity potential φ

such that
q = −∇ φ

Hence, ∇ . q =∇ . ∇ φ =∇ 2φ =0
Thus, an incompressible irrotational fluid satisfies the Laplace equation.
Example. A thin rectangular homogeneous thermally conducting plate
lies in the xy − plane defined by 0 ≥ x ≤ a , 0 ≤ y ≤ b . The edge y = 0 is held
Tx ( x − a ) ,
at the temperature where T is a constant , while the remaining
0
edges are held at 0 . The other faces are insulated and no internal sources
and sinks are present. Find the steady state temperature inside the plate.
Solution. Since no heat sources and sinks are present in the plate, the
2
steady state temperature u must satisfy ∇ u = 0 . hence the problem is to
solve
PDE : ∇ 2 u =
0
u ( 0=
, y ) 0 , u ( a=
, y ) 0 , u ( x=
, b ) 0 , u ( x=
, 0 ) Tx ( x − a )
BCs:
This is a typical Dirichlet problem . The general solution satisfying the
first three BCs is given by Eq. (47). Therefore,
 250 Differential Equations: Theory and Applications

∞
 nπ   nπ
u( x , y ) ∑A n sin  x  sinh  ( y − b ) 
n =1  a   s 
− nπ 2  nπ 
a
An sinh = ∫ f ( x ) sin  x  dx
a a0  a 
u ( x , 0 )= Tx ( x − a )= f ( x ) ,
Using the last BC: get
−nπ b 2  nπ 
a
An sinh
= ∫ Tx ( x − a ) sin  x  dx
a a0  a 

2T  nπ 
a
=
a ∫ x ( x − a ) d 
0
x  dx
a 

a 2T   nπ   
a

=
− .  ∫ x ( x − a ) d  cos  x 
nπ a  o   a 
a
2T  nπ nπ
a
  
=
−  ( 2 x – a ) sin  x   − ∫ ( 2 x − a ) d  sin  x  
nπ   a 0 o   a 
a
2 aT   nπ    nπ 
a

2 (
2 x – a ) sin  x   − ∫ 2 sin  x  dx
n π 
2
 a  0 o  a 

2aT  2a   nπ   
a

 a sin nπ + cos  x  
n2π 2  nπ   a   0 
=
2aT 2a 4a 2 T 
( −1 ) 3 3  (
−1 ) −1 
n
= cos nπ=
n 2 π 2 nπ n π 
Thus the required temperature distribution is given by
 nπ  4T a  nπ   nπ
∞ 2
u ( x , y) =
∑ cosech  − b  3 3 [ −1] sin  x  sinh  ( y − b )
n =1  a n π  a   a 
Example. Solve
∇ 2=
u 0 , 0≤ x≤ a , 0≤ y ≤b
Satisfying the BCs:
 Elliptic Differential Equation 251

u ( 0 , y ) 0=
= , u ( x ,0 ) 0=
, u ( x ,b ) 0

∂u πy
( a , y ) = T sin 3
∂x a
Solution . Using the variables separable method, one of the acceptable
general solutions is given by Eq. (38). Hence
u( x , y )=( )
c 1 e px + c2 e − px ( c3 cos py + c4 sin py )

u ( x ,0 ) = 0 ,
Using the BC: we get
= (
0 c3 c1 e px + c2 e − px )
Implying
c3 = 0 . Therefore,

u ( x , y ) c4 sin py ( c1 e px + c2 e – px
=
)
u ( x, b ) − 0,
Now, using the BC: we obtain
(
0 c4 sin pb c1 e px + c2 e px )
c4 = 0 , implying sin pb = 0 which gives
nπ
= p = , n 1, 2 , 3 ,… .
ob = nπ or b
 nπ 
= u ( x , y ) c 4 sin 
 b 
(
x  c1 e px + c2 e − px )
Thus
Renaming the constants, we have
 nπ    nπ   nπ  
u( x , y ) =
sin  y   A exp  x  + B exp  − x   ,n =
1 , 2 ,….
 b   b   b 
BC : u ( 0 , y ) = 0 ,
If we use the we get
 nπ 
0 sin  y ( A + B )
 b 

Giving A + B =0 ; therefore , A =− B . Thus,

252 Differential Equations: Theory and Applications

 nπ    nπ   nπ  
u ( x , y ) A sin  y   exp  x  − exp  – x 
 b   b   b 
 nπ   nπ 
2=
A sin  y  sinh  x  , n 1 ,2 , .
 b   b 
Differentiating with respect to x, we obtain
∂u nπ  nπ   nπ 
=2 A sin  y  cosh  x
∂x b  b   b 
The Last BC yields
πy nπ  nπ   nπ 
T sin 3 = 2A sin  y  cosh  x
a b  b   b 
The last BC yields
πy nπ  nπ   nπ 
T sin 3 = 2A sin  y  cosh  a
a b  b   b 
Form which we can determine 2 A. Hence, the required solution is
bT πy nπ  nπ 
u(x , y ) = sin 3 sech a sinh  x
nπ a b  b 
The principle of superposition gives the required solution as
∞
bT πy  nπ   nπ 
u(x , y )=∑ sin 3 sech  a  sinh  x
n =1 nπ a  b   b 
u( x , y ,z )
Example. Find the potential function in a rectangular box
defined by 0 ≤ x ≤ a , 0 ≤ y ≤ b , 0 ≤ z ≤ c , if the potential is zero on all sides
u= f ( x ,y)
and the bottom, while on the top of the box .
 Elliptic Differential Equation 253

Figure 6
Solution. The potential distribution in the rectangular box satisfies the
Laplace equation. Thus, the problem is to solve.
∇ 2 u= u xx + u yy + u zz= 0

Subject to the BCs:

( 0 , y , z ) u=
u= ( a , y ,z ) 0
u (=
y , 0 , z ) u=
( x ,b , z ) 0
u ( x , y ,0 ) = 0

u( x ,y ,z ) = f ( x ,y )

Following the variables separable method, let us assume the solution in

the form
u ( x , y , z ) = X ( x )Y ( y ) Z ( z )

Substituting into the Laplace equation, we get

X ''
( x ) Y ( y ) Z ( z ) + X ( x ) Y '' ( y ) Z ( z ) + X ( x ) Y ( y ) Z '' ( z ) =
0

which can also be written as

Y '' ( y ) Z '' ( z ) X ′′ ( x )
+ =
− λ12
=
Y( y) Z(z) X ( x)

where
λ1 is a separation constant. Thus, we have
X '' ( x ) + λ12 X ( x ) =
0 ( 105 )

 254 Differential Equations: Theory and Applications

After the second separation, we also have

Z (z) − λ 2=
''

−
Y ′′ ( y )
λ 22
=
Z(z) Y( y)
1

Y ′′ ( y ) + λ 22 Y ( y ) =
0 ( 106 )

Z '' ( z ) − λ 32 Z ( z ) =
0 ( 107 )

where
λ=2
3 λ 12 + λ 22 . The general solution of Eq. ( 105 ) − ( 107 ) are
X ( x ) c1 cos λ1 x + c2 sin λ1 x
=

Y ( y ) c3 cos λ2 x + c4 sin λ2 y
=

Z ( z ) c5 cosh λ3 z + c6 sinh λ3 z
=

From the BCs,

( 0 ) X=
X= (a) 0
( 0 ) Y=
Y= (b) 0
Z ( 0 ) =0

X ( 0 ) =0 c1 = 0
gives
X ( a )=0 λ1 a = mπ .
gives
Therefore,
mπ
λ1
= , m 1 , 2 , ….
=
a
Similarly,
Y ( 0 ) =0 c3 = 0
gives
Y (b) = 0 λ2 b = nπ
gives
Therefore,
nπ
λ2
= =, n 1 , 2 , ,…
b
 Elliptic Differential Equation 255

Z ( 0 )=0 c5 = 0
Also, gives Further, we note that
 m 2 n2 
λ32 = λ12 + λ22 = π 2  2
+ 2  = λ mn
2
( say )
 a b 
Then
m 2 n2
λ=
3 π + = λ mn
a 2 b2
The solution now take the form
mπ x
X (=
x ) c2 m sin , m 1 , 2 ,…
=
a
nπ y
Y (=
y ) c4π sin , n 1 , 2 ,…
=
b
Z ( z ) = c 6 mn sinh λ mn z

Let mn
c =c c c
2 m 4 n 6 mn ; then, after using the principle of superposition ,
the required solution is
∞ ∞
mπ x nπ y
u ( x , y , z ) X=
= ( x )Y ( y ) Z ( z ) ∑∑ c mn sin sin sinh λ mn z
= 1=
m n 1 a b (108)
f ( x , y ) = u ( x , y ,c ) ,
Using the final BC: we get
mπ x nπ y
f ( x , y ) = ∑ ∑ c mn sinh λ mn c sin sin
a b
which is a double Fourier sine series. Thus, we have
4 mπ x nπ y
a b
cmn sinh λ mn c = ∫ ∫ f ( x , y ) sin sin dx dy
ab 0 0 a b
(109)
, Eq . ( 108 ) and ( 109 )
Therefore constitute the required potential.
Example. Find the electrostatic potential u in the annular region
bounded by the concentric sphere r= a , r= b , 0 < a < b , if the inner outer
surfaces are kept at constant potential
u1 and u2 , u1 ≠ u2 .
256 Differential Equations: Theory and Applications

Solution. The electrostatic potential satisfies the Laplace equation

2
0
∇ u=
It is natural that we choose spherical polar coordinates. from the problem.
It is evident that we are looking for a solution with spherical symmetry
u =u ( r ) .
which is independent of θ and φ . Hence.

Figure 7
Thus, we have to solve
∂  2 ∂u 
PDE :  r =0
∂r  ∂r  (110)

Subject to
BCs ;=
u u1 at=
r a=
u u2 at=
r b

Eq. ( 110 )
Integrating with respect to r , we obtain
∂u
r2 =A
∂r (a constant of integration)
Again, integrating, we get
A
u=
− +B
r
Now, using the BCs : we have
A A
− + B , u2 =
u1 = − +B
a b
 Elliptic Differential Equation 257

Solving these equations, we get

u1  1 1  1 1 
A =u 2 −  u 1  −  + u2  −  
(1/ a ) − (1/ b )   r b   a r  
Example. A thermally conducting solid bounded by two concentric
spheres of radius a and b , a < b , is such that the internal boundary is
f (θ ) f (θ ) .
kept at 1 and the outer boundary at 2 Find the steady state
temperature in the solid .
Solution. It is known that the steady temperature T satisfies the Laplace
equationU?H. In the present problem,

Figure 8

T = T ( r ,θ )

Thus, we have to solve

2
0
PDE: ∇ T =
Subject to the boundary conditions
=T f 1 ( θ ) at r a
=

= 2 ( θ ) at r
T f= b

In spherical polar coordinates, for axially symmetric case, the solution

Eq . ( 99 )
of the Laplace equation is given by as follows :
 258 Differential Equations: Theory and Applications

∞
 Bn 
T ( r ,θ )
= ∑  A rn + pn ( cos θ )
r n +1 
n
n=0

Using the BCs:

∞
 Bn 
f1 ( θ )
= ∑  A a n
+ pn ( cos θ )
a n +1 
n
n =0 (111)
∞
 Bn 
f2 ( θ )
= ∑  An a n + pn ( cos θ )
n=0 b n +1  (112)

In order to find the coefficient n

A and B
n , we have to express

f 1 ( θ ) and f 2 ( θ )
in terms of Legendre polynomials and compare the
coefficient. In this process, the following orthogonality relation is useful:
π  0 , if m ≠ n

∫ p ( cosθ ) P (
m n cosθ ) sinθ dθ =  2
0  2n +1 , if m = n

Eq . ( 111 ) pm ( cosθ ) sinθ

Thus, multiplying both sides of by and
integrating, we obtain
π ∞ π
 B 
∫ f1 ( θ ) pm ( cosθ ) =
sin θ dθ ∑  An a n + nn+1  ∫ pn ( cos θ ) pm ( cosθ ) sin θ dθ
0 n=0 a 0

 B  2
=  An a n + nn+1 
 a  2m +1 (113)
Eq . ( 112 )
Similarly, gives
π ∞ π
 Bn 
∫ f 2 (θ ) pm ( cos θ =
) sin θ dθ ∑  An b + b m + 1  ∫ pn ( cos θ ) pm ( cosθ ) sin θ dθ
n

0 n=0 0

 m  2
=  A m b m + m +1 
 a  2m +1 (114)
Let
π
2m + 1
f1 (θ ) pm ( cosθ ) sin θ dθ = Cm
2 ∫0
 Elliptic Differential Equation 259

π
2m + 1
f 2 ( θ ) pm ( cos θ ) sin θ dθ = Dm
2 ∫0

Eq. ( 113 ) and ( 114 )

Then reduce to
B
Am a m + m +1
=
Cm
a
B
Am a n + m +1
=
Dm
b
Solving this pair of equations, we obtain
Cm a m + 1 − Dm b m + 1
Am =
a 2 m + 1 – b 2 m + 1 (115)

Bm =
(
a m +1 b m + 1 Cm b m − Dm a m )
2 m +1 2 m +1
b –a (116)
Hence, the require steady temperature is
∞
 Bm 
T ( r ,θ )
= ∑  A rm +  pm ( cos θ )
rm +1
m
m=0 

where
Am and Bm are given by Eq. ( 115 ) and ( 116 ) .

Example. A thin annulus occupies the region 0 < a ≤ r ≤ b , 0 < θ ≤ 2π .

0
The faces insulated. Along the inner edge the temperature is maintained at 0 ,
θ 
T = Kcos   , where K
while along the outer edge the temperature is held at 2
is a constant . Determine the temperature distribution in the annulus.
Solution. Mathematically, the problem is to solve

PDE: ∇ =
2
T 0 , a ≤ r ≤ b , 0 ≤θ ≤ 2π
T ( a ,θ ) = 0
BCs:
θ
T ( b ,θ ) = k cos
2
 260 Differential Equations: Theory and Applications

Eq. ( 57 )
The required general solution is given by in the form
T ( r ,θ ) = (
c1 r n + c2 r − n )( c 3 cos nθ + c4 sin nθ )

Using the first BC. We get

0= (
c1 a n + c2 a − n )( c 3 cos nθ + c4 sin nθ )

Implying thereby 1
c a n + c2 a − n = − c1 a 2 n . After adjusting the
0 , or c2 =
constants suitably, we have
 n a 2n 
T ( r ,θ ) =
 r − n  ( A cos nθ + B sin nθ )
 r 
The principal of superposition gives
∞
 n a 2n 
T ( r ,θ ) =
∑  r − n  ( A cos nθ + B sin nθ )
n = 1 r 
Now, using the second boundary condition, we obtain
θ ∞
( b ,θ ) K=
T= cos
2
∑( b
n =1
n
– b −n a 2n )( A n cos nθ + Bn sinnθ )

which is a full-range Fourier series . Hence,

2π
An (b n
–b −n
a 2n
) = π ∫ K cos θ2
1
cos nθ dθ
0

2π
k   1  1 
=
2π ∫  cos  n + 2 θ + cos  n − 2 θ  dθ
0

2π
k   1  1 
=
2π ∫  cos  n + 2 θ + cos  n – 2 θ  dθ
0

2π
 1  1 
  n + θ  n − θ 
k 2 2
 + 
2π  n + 1 1 
n−
 2 2  0

=0
 Elliptic Differential Equation 261

Implying
An = 0 . Also,
2π
k θ
( )
Bn b n − b − n a 2 n = ∫
π 0
cos sin nθ dθ
2
2π
k   1  1 
=
2π ∫  sin  n + 2 θ + sin  n − 2 θ  dθ
0

2π
  1  1 
 cos  n + θ cos  n − θ 
k  2  2
=
−  + 
2π  1 1 
n+ n−
 2 2 
0

 
k  1 1 1 1 
=
− − − − − 
2π  n + 1 n + 1 n − 1 n – 1 
 2 2 2 2 
 
k 1 1  k 2n
 + =1
π  n+ 1 n+ 1  π n2 −
 2 2  4
or
8kn
( )
Bn b n – b − n a 2 n =
(
π 4n 2 −1 )
Thus the temperature distribution in the annulus is given by
  r n  a  n 
 – 
8k ∞ n   a   r  
T ( r ,θ ) =
π ∑
sin nθ
n = 1 4n −1   b  a 
2 n −n

   −  
a  b  
Example. V is a function of r and θ satisfying the equation
∂ 2 V 1 ∂V 1 ∂ 2 V
+ + 0
=
∂r 2 r ∂r r 2 ∂ θ 2
 262 Differential Equations: Theory and Applications

=
within the region of the plane bounded by r a=, r b=,θ 0= ,θ π / 2 . its
π 
= r a is θ  −θ  ,
value along the boundary 2  along the other boundaries is
zero. Prove that

( r / b ) − (b / r )  sin ( 4n − 2 )θ 
4n − 2 4n − 2
2 ∞
V= ∑  
π n = 1 ( a / b ) 4 n − 2 − ( b / a )4 n − 2  ( 2n −1)
3


Solution. The task is to solve the PDE

2
∂ V 1 ∂V 1 ∂ 2V
+ + 0
=
∂ r 2 r ∂r r 2 ∂θ 2
Subject to the following boundary conditions:
( i ) V ( b ,θ ) = 0 , 0 < θ < π /2
( ii ) V ( r ,π / 2=) 0 , a<r ≤b

( iii ) V ( r , 0 ) = 0 , a<r ≤b
π 
( iv=
) V ( a ,θ ) θ ,  –θ  , 0 <θ < π / 2.
 2 
The three possible solution are given as follows:
(
c1r p + c2 r − p
V= )( c 3 cospθ + c4 sin pθ )

(
 c1 cos ( p In r ) + c2 sin ( p In r )  c3e pθ + c4 e − pθ
V= )
V =( c1 In r + c2 )( c3θ + c4 )

Since the problem is not defined for = r 0 , ∞ , the second and third
solutions are not acceptable .Hence , the generally acceptable solution is the
first one. The boundary condition (iii) gives
= (
0 c3 c1r p + c2 r − p )
Implying
c3 = 0 . The boundary condition (ii) implies
π
0 c4 sin p
w
(c r
1
p
+ c2 r − p )
 Elliptic Differential Equation 263

Therefore,
π
sin p= 0 or
= p 2n=
, n 1 , 2 ,…
2
Thus, the possible solution of the given equation has the form
= (
V ( r ,θ ) c4 sin ( 2nθ ) c1r 2 n + c2 r −2 n )
Now, applying the boundary condition
( i ) , we get
(
0 c4 sin ( 2nθ ) c1 b 2 n + c2 b −2 n )
which gives
c2 = − c1 b 4 n . Therefore,

=V ( r ,θ ) c1 c 4 sin ( 2nθ )  r 2 n − r − 2 n b 4 n 

Superposing all the solution, we obtain

∞

(
V ( r ,θ ) = ∑ cn sin ( 2nθ ) r 2 n – r −2 n b 4 n )
n =1

Satisfying boundary conditions (iv), we get

π   a 4n − b4n 
θ ∑ cn sin ( 2nθ ) 
−θ  = 2n 
2   a 
which is a Fourier sine series. Thus, we have
π

2 2
π   a 4n − b4n 
∫θ −θ  sin ( 2nθ ) =
cn  
π /2 0  2   an 
2

Integrating by parts , we obtain

π
 π 2 
 cos ( 2nθ )   π   sin ( 2nθ )   cos ( 2nθ )  2
 θ − θ   −  −  − 2θ   − 2  + ( −2 )  
 2  2n   2  4n   8n 3   0

π  a 4n − b4n 
= cn  
4 a 2n 
On simplification , we get
 264 Differential Equations: Theory and Applications

−
1
4n 3
{ ( )
−1
n
–1 =cn 
4 
}
π  a 4n – b 4n 
a 2n


Thus,
 1
π  a 4 n − b 4 n   3 , for n odd
cn   =  2n
4  a 2n   0 , for n even

Hence, the required solution is
4n − 2
∞
2 1 a  r 8 n − 4 − b8 n − 4 
V ( r ,θ ) ∑1 π ( 2n − 1) 3  r  sin ( 4n − 2 )θ  8 n − 4 
 a – b8 n − 4 
which can be recast in the form given in the problem .
Example. Determine the potential of grounded conducting sphere in a
uniform field defined by
2
PDE: ∇ =u 0 , 0 ≤ r < a , 0 <θ < π , 0 ≤ φ < 2π

BCs:
( i ) u ( a ,θ ) = 0 .
( iii ) u → E0 r cosθ as r →∞ ,

Solution. In spherical polar coordinates , with axial symmetry, the

solution of the Laplace equation is given by Eq. (99) in the form
∞
 Bn 
u ( r ,θ )
= ∑  A r n
+ Pn ( cos θ )
r n + 1 
n
n=0

Using the boundary conditions (ii), we have

∞
u ( r ,θ ) = ∑A r n
n
Pn ( cos θ ) = − E0 r cos θ
n=0

which is true only for

=n 1 , when p1 ( cos
= θ ) cos θ . Also=
, An 0 for n ≥ 2 .
Therefore ,
u ( r ,θ ) = A1 r cos θ = − E0 r cosθ

Implying
A1 = − E0 . Hence,
 Elliptic Differential Equation 265

∞
Bn
u ( r ,θ ) =
− E0 cosθ + ∑ pn ( cos θ )
n =1 r n +1

Now, applying the boundary condition

(i) , we get
∞
Bn
0 − E0 a cos θ + ∑ pn ( cos θ )
an +1 a n +1

Multiplying both sides of

pm (cos θ ) sin θ and integrating between the

limits 0 to π , we have
π ∞ π
Bn
E0 a ∫ cos (θ ) pm ( cosθ ) sinθ dθ = ∑ n +1 ∫ n (
p cosθ ) pm ( cosθ ) sinθ dθ
0 n =1 a 0

Using the orthogonality property

π  0 , for m ≠ n

∫ p ( cos θ ) p (
n m cos θ ) sin θ dθ =  2
0  2m +1 , for m = n

we obtain
π
Bm 2 E 0 a ∫ cos (θ ) pm ( cos θ ) sin θ dθ
m +1
a 2m + 1 = 0

or
π
2m + 1
Bm = E0 a m + 2 ∫cos (θ ) pm ( cosθ ) sin θ dθ
2 0

It can verified that the integral on the right-hand side of the Eq. ( 117 )
vanishes for all m except when m = 1 , I which case
B1 = E0 a 3
Therefore, the required potential is given by
u ( r ,θ ) =
− E 0 r cos θ + E0 a 3 r 2 cosθ

Example. The steady, two –dimensional, incompressible viscous fluid

flow past a circular cylinder, when the inertial terms are neglected ( stoke
flow ) , is governed by the inharmonic
 266 Differential Equations: Theory and Applications

PDE: where ψ is the stream function. Find its solution

subject to the BCs:
∂ψ
( i )ψ ( r ,=
θ) = 0
∂r on r =1
( ii )ψ ( r ,θ ) → r sinθ as r →∞

Solution. In view of the cylindrical geometry , we can write

∇4ψ =
∇2 ∇2 ψ =
0( )
where
 ∂2 1 ∂ 1 ∂2 
∇ 2 ψ=  2 + + 2 ψ
 ∂r r ∂r r ∂θ 2 
Using the variables separable method , let us look for a solution of the
form
ψ ( r ,θ ) = f ( r ) sin θ
Therefore,
∂ψ ∂ψ
f ' ( r ) sin θ ,
= f ( r ) cos θ
∂r ∂θ
∂2 ψ ∂ 2ψ
= f=''
( r ) sinθ , f ( r ) sinθ
∂r 2 ∂θ 2
Hence,
 1 1 
ψ  f '' ( r ) + f ' ( r ) − 2 f ( r )  sin θ
∇ 2=
 r r 
which can also be written in the form
F ( r ) sinθ
∇2 ψ =

where
1 1
F(r)=f '' ( r ) + f '
(r )− f (r)
r r2
 Elliptic Differential Equation 267

Therefore,
( )
∇ 4 ψ =∇ 2 ∇ 2 ψ =∇ 2  F ( r ) sin θ  =0

i.e.,
 '' 1 ' 1 
 F ( r ) + r F ( r ) − r 2 F ( r )  sin θ =
0

Implying
1 1
F '' ( r ) + F ' ( r ) 2 F ( r ) =
0
r r
d
= r e=
z
,D ,
Introducing the transformable dz the above equation
becomes

 D ( D –1 ) + D –1  F ( r ) =
0

or
(D 2
)
–1 F ( r ) = 0

Its complementary function is

B
F(r)=Ae z + Be – z =Ar +
r
or
1 1 B
f '' ( r ) + f ′( r ) − 2 f ( r ) =
Ar +
r r r
i.e.,
r 2 f '' ( r ) + rf ' ( r ) – f ( r ) =
Ar 3 + Br

which is a homogeneous ordinary differential equation. Again using the

d
= r − e=
z
,D
transformable dz . We get
 D ( D –1 ) + D –1  f =+
A 3 z Be z

or
 268 Differential Equations: Theory and Applications

(D 2
–1=)
f Ae 3 z + Be z

Its complementary function is

f (=
r ) C e z + De − z

while its particular integral is

1 Ae 3 z Bze z
D 2 −1
( )
Ae 3 z + Be z =
8
+
2
Therefore,
D A 3 B
f ( r ) = Cr + + r + r In r
r 8 2
Thus, we have
 A B D
ψ= 3
 r + r In r Cr +  sin θ
 8 2 r 

Now to satisfy the BC: ψ → r sin θ , as r →∞ and form physical

considerations, we choose A = 0 , Therefore,
B D
ψ  r in r + Cr +  sin θ
=
 2 r 
The boundary condition
ψ= 1 gives ( C + D ) sin θ =
0 on r = 0 , implying C =
−D .
Also , the
∂ψ
=0
boundary condition ∂r on r =1 gives
B B
+ C − D =0 = − 2 D
2 2

Implying B = 4 D.Hence , the general solution is

r 1
=ψ 2 D ( r in r − + sinθ
2 2r
Example. The problem of axisymmetric fluid flow in a semi-infinite
or in a finite circular pipe of radius a is described as follows in cylindrical
coordinates :
 Elliptic Differential Equation 269

PDE : ∇ 2=
u 0 , 0< r < a
∂u
(i) = 0=
at r 0
BCs: ∂r
∂u ∂u
( ii=
) 0=
, V ( z=
) at r a
∂z ∂r
Show that the speed of suction is given by
∞
V(z)=
− ∑ an ( An cosh an z + Bn sinh an z ) j1 ( an a )
n =1

Solution. In cylindrical coordinates

( r ,θ , z ) ,
2 ∂ 2 u 1 ∂u 1 ∂ 2 u ∂ 2u
∇ u= + + + = 0
∂ r 2 r ∂r r 2 ∂θ 2 ∂z 2
In axisymmetric case , the above equation becomes
∂ u 1 ∂u ∂ 2 u
2
+ + 0
=
∂r 2 r ∂r ∂z 2 (118)
u ( r , z ) = f ( r ) φ ( z ) which ,
Let when substituted into Eq. ( 118 ) ,
gives
1
f '' +   f '
r = φ ''
− = − a 2 ( say )
f φ
Then
φ − a2 φ =
''
0 (119)
1 ' 2
f '' + f +a φ =
0
r (120)
Eq . ( 119 )
The solution of is
φ = A cosh α z B sinh az
Equation (120) can be rewritten as
r 2 f '' + rf ' + α 2 r 2 f =
0
 270 Differential Equations: Theory and Applications

which is a Bessel’s equation of Zeroth order whose general solution is

=f j0 ( α r ) + DY0 ( α r )

j ( ar ) and Y0 ( ar )
Here, 0 are Zeroth order Bessel functions of the first
and second kind respectively , Therefore , the typical solution is
=u ( A cosh az + B sin α z )  j0 ( α ) 
Y (αr )
Now, 0 is infinite at r = 0 , and hence D = 0 , Therefore , the
possible solution is
=u ( A coshα z + B sinhα z ) j0 ( ar )
∂u
=0 j ' ( ar ) = 0
The condition ∂r at r = 0 is automatically satisfies , since 0
∂u
at r = 0 . Now the boundary condition ∂= z 0= at r a gives j0 ( aa ) = 0 ,
j .
implying that aa are the zeros of the Bessel function 0 Let these zeros be
n a ( n 0 ,1 , 2 ,… ) .
a=
Thus the appropriate solution is
∞
=u( r ,z ) ∑a n ( An cosh an z + B – n sinh an z ) j0 ( an r )
n =1

Using the fact that,

j0' = − j1 , the speed of suction is given by

 ∂u  ∞
V ( z) =
  − ∑ an ( An cosh an z + Bn sinh an z ) j1 ( an a )
=
 ∂r r = a n =1

Example. Solve the following Poisson equation:

∂ u ∂2 u
2
+ 2
=
∂x 2 ∂y 2
Subject to the boundary conditions
( 0 , y ) u=
u= ( 5 , y ) u=
( x ,0 ) u =
( x ,4 ) 0
Solution. We assume the solution of the form
u= v +ω (121)
 Elliptic Differential Equation 271

where v is a particular solution of the Poisson equation and ω is the solution

of the corresponding homogenous Laplace equation. That is,
∇2 v =
2 (122)
∇2 ω = 0 (123)
It is customary to assume that υ has the form
v ( x , y ) = a +bx +cy + dx 2 + exy + fy 2

Substituting this into Eq. (122), we get

2d + 2 f =
2

Let f = 0 . Then d =1 The remaining coefficients can be chosen

arbitrarily. Thus, we take
v(x , y )=− 5x + x2
(124)
So that v reduce to zero (satisfies the boundary conditions) on the sides
=x 0=
and x 5 .
Now, we shall find ω form
∇ 2 ω= 0 , 0 < x < 5 , 0 < y < 4 (125)
Satisfying
ω (0 , y ) =
−v( 0 , y )=
0

ω ( 5 , y )=
− v (5 , y ) =
0

ω ( x , 0 ) =− v ( x , 0 ) =− ( − 5 x + x 2 )

ω ( x , 4 ) =− v ( x , 4 ) =− ( − 5 x + x 2 )

Eq. ( 121 ) , ( 124 )

The above conditions are obtained by using and the
given boundary conditions. By using the superposition principle the general
Eq. ( 125 )
solution of is found to be
∞
 nπ x   nπ y   nπ  
=ω ( x , y) ∑ sin    an exp   + bn exp  − 
n =1 5   5   5   (126)
 272 Differential Equations: Theory and Applications

Now, applying the non-homogeneous BC: ω ( x , 0 ) =− − 5 x + x ,

2
( )
we get, after renaming the constants, the equation
 nπ x 
ω ( x , 0 ) =− ( − 5 x 2 + x 2 ) =∑ AN sin  
 5 
ω ( x , 0 ) =− ( − 5 x + x 2 ) ,
Also, applying the non-homogeneous BC :
we get , after renaming the constants , the equation
∞
4nπ 4nπ  nπ x
= (
− − 5x + x2 ) ∑  a n cosh
5
+ b n sinh
5 
 sin
5
n =1

which gives
5
2 nπ x
=an
50∫ (
5 x − x 2 sin
5
dx )
Now. Integrating by parts, the right-hand side yields
5
2 2  5 nπ   53 nπ x   53 nπ  
=
an
5
(
 5x − x  −
 nπ
)
cos x  − ( 5 − 2 x )  − 2 2 sin
5  n π 5
 + ( − 2 )  3 3 cos
n π
x 
5 0
  

2  2( 5 ) 
3
2  53
 – 3 3 cos nπ + 3  3 
5 n π n π  


=
( )
4 52  1
−
cosnπ  4( 5 )
2
 1 ( −1) n 
π n
3  3
n3 = π 3  3− 3 
 n n 

Hence,
 8 52

( )
an =  π 3 n 3 , when nis odd
 0 , when nis even
 (128)
Eq . ( 127 ) ,
Also, from we have
5
4nπ 4nπ 2  nπ 
an cosh
5
+ bn sinh
5
= ∫
5 0
5 x − x 2 sin  (
x  dx =
 5 
an )
 Elliptic Differential Equation 273

Therefore,
 4 
an 1 − cosh  nπ 
5 
bn = 
4 
sinh  nπ 
5  (129 )
an , b n Eq. ( 128 ) and ( 129 ) into Eq. ( 126 ) ,
Substituting from we get

∞
  nπ   4  4   nπ   nπ   nπ   4 
ω ( x , y ) ∑ a  cosh  y  sinh  nπ  − cosh  nπ  sinh  y  + sinh  y  sin  x  / sinh  nπ  
1   5  5  5   5   5   5   5 

or
∞
  nπ nπ  nπ 4
=ω( x ,y ) ∑  sinh  ( 4 – y ) + sinh  y   sin  x  / sinh  nπ 
n =1 5   5   5  5 

Eq. ( 124 ) , ( 128 ) and ( 130 ) ,

Combining the solution of the given
Poisson equation is
  π y 
8 × 52 ∞  sinh ( 2n –1 ) π ( 4 – y ) / 5 + sinh  ( 2n −1) 5  
 
( x , y ) x ( x –5 ) +
u= ∑ 
4
π3 n =1    
sinh  ( 2n –1 ) π 
  5  

 πx 
 sin [ (2n –1 ) 5 ]
× 
( 2n −1 )
3
 
 
p( x , y ,z)
Example. Let IR be a region bounded by ∂IR . Let be any
point in the interior of IR , as shown in Fig . Let φ be a harmonic function in
1
ψ= .
IR ; also , let r where r is the distance form p . Applying Green’s
second identity , show that
. .
1  1 ∂φ ∂  1 
=φ( p) ∫ ∫  r −φ   ds
4π ∂ IR
∂n ∂n  r  
274 Differential Equations: Theory and Applications

p( x , y ,z ) ,
Solution. Since ψ processes a point of discontinuity in IR at
Green’s second identity cannot be directly applied to φ and ψ . However ,
. .
1
∑.
∂IR ∪∂ ∑.
.
ψ= −∑ .
r is bounded in IR ò with the boundary , where ò is a ò

sphere of radius ε with centre at p . Now applying Green’s second identity

.
φ and ψ in IR − ∑ .,
( 19 ) to functions ò we get

Figure 9
.

 2 1  1  .  ∂  1   1  ∂φ  . ∂ 1  .
 1  ∂φ 
∫∫∫  φ∇   − = ∇ 2 φ  dV ∬  φ  −    ds + ∬. φ ∂n  r  dS – ∬    dS
.  r  r  ∂ IR  ∂n  r   r  ∂n    .  r  ∂n 
IR –
∑ ∂ .
∑ ∂ ∑.
ò ò ò
.

( 131 )
Eq . ( 131 ) ,
From the right-hand side of we observe that the last two
integrals depend only on ε . But in the direction of the exterior normal to
.
∂ ∑.,
ò we find that
∂  1 ∂  1 1
  . = −   =
∂n  r  ∂ .
∑
∂r  r  r =
ε
ε2
ò

Therefore,
 Elliptic Differential Equation 275

. .
∂ 1  1 4πε 2
∬φ =
  dA =2 ∬
φ dS = φ ( Q ) 4πφ * ( Q )
. ∂n  r  ε 2
. ε
∂ ∑. ∂ ∑.
ò ò

.
φ ( Q ) on ∂ ∑. .
φ* ( Q )
where is the average value of ò Further, the third
integral
. .
 1  ∂φ 1  ∂φ   ∂φ * 
− ∬  − ∬
dS = dS =
− 4πε  
.  r  ∂n ε .  ∂n   ∂n 
∂ ∑. ∂ ∑.
ò ò

.
∂∑ .
*
 ∂φ 
 
Where  ∂n  is an average value of the normal derivative on ò

1 .
∇ 2   = 0 in IR −∑ . ,
. Substituting these results and suing the fact that r ò
we obtain
.

 1 2 ,  ∂  1   1  ∂φ   ∂φ 
 dS + 4πφ ( p )+ 4πε  ∂n
*
∫∫∫  – ∇= φ dV ∬  φ   −  *
.  r  ∂IR  ∂n  r   r  ∂n   
IR − ∑
ò
.

(132)
Now, taking the limit as ε → 0 , and using the fact that φ is harmonic in
.
IR − ∑.
ò we arrive at the fundamental result
.
1  1 ∂φ ∂ 1 
=φ( p)
4π
∬  r  ∂n −φ ∂n  r   dS
∂IR    
Thus, the value of a harmonic function at any point of IR can be obtained
in terms of the values of φ and ∂φ / ∂n on the boundary ∂IR of the region
IR.
276 Differential Equations: Theory and Applications

4.14. SUMMARY AND DISCUSSIONS

In this chapter we discussed various properties and technique for solving
Laplace and Poisson equations which are elliptic in nature. Various physical
phenomena are governed by the well-known Laplace and poison equations.
A few of them, frequently encountered in applications are: study heat
condition, magnetic potential , torsion of prismatic shaft, etc.
2 ∂2 ∂2 ∂2
∇= div ∇= + +
∂x 2 ∂y 2 ∂z 2 is called the Laplace operator.
∇ 2V =
0 is called the Laplace equation
− 4π Gp is known as poisson’s equation .
∇2 V =

4.14.1. Harmonic Functions

Solutions of Laplace equation1 are called harmonic functions which possess
a number of interesting properties, and they are presented in the following
theorems.
Theorems. If a harmonic function vanishes everywhere on the boundary
, then it is identically zero everywhere.
If φ is a harmonic function in IR and ∂φ / ∂n =0 on ∂IR, then φ is a
constant in IR .

4.14.2. Separation Of Variables

The method of Separation of variables is applicable to a large number of
classified linear homogeneous equations. The choice of the coordinate
system in general depends on the shape of the boundary

4.14.3. Dirichlet Problem For A Rectangle

The Dirichlet problem for a rectangle is defined as follows:
2
PDE: ∇= u 0 , 0 ≤ x ≤ a ,0 ≤ y ≤ b

BCs:
( x, b ) u=
u= ( a , y ) 0, u=
( 0 , y ) 0 , u=
( x ,0 ) f ( x ) (41)
This is an interior Dirichlet problem. The general solution of the governing
PDE, using the method of variables separable
 Elliptic Differential Equation 277

4.14.4. The Neumann Problem For A Rectangle

The Neumann problem for a rectangle is defined as follows
PDE : ∇ 2 u= 0, 0 ≤ x ≤ a, 0 ≤ y ≤ b

x ( 0, y )
BCs : u= x ( a, y )
u= y ( x, y )
0, u= y ( x, b )
0 , u= f ( x)

4.14.5. Interior Dirichlet Problem For A Circle

The Dirichlet problem for the circle is defined as follows:

PDE: ∇ u= 0, 0 ≤ r ≤ a, 0 ≤ 0 ≤ 2
u ( a=
, θ ) f ( θ ) , 0 ≤ θ ≤ 2π
BC:

u ( r ,θ ) =
1
2π
(a 2
)
− r 2 f (φ )
2π ∫ a 2
− 2ar cos (φ − θ ) + r 2 
dφ
0 
This is known as Poisson’s integral formula for a circle, which gives a
unique solution for the Dirichlet problem. The solution can be interpreted
physically in many ways: it can be thought of as finding the potential
u ( r , θ ) as f (φ )
a weighted average of the boundary potential weighted by
the Poisson kernel p , given by
a2 − r 2
p= 2
 a − 2ar cos (φ − θ ) + r 2 

u ( r ,θ )
It can also be thought of as a steady temperature distribution
in a circular disc, when the temperature u on its boundary ∂IR is given by
u = f (φ )
which is independent of time.

4.14.6. Exterior Dirichlet Problem For A Circle

The exterior Dirichlet problem is described by
PDE : ∇ 2 u =
0
BC : u ( a, θ ) = f (θ )
, u must be bounded as r → ∞
 278 Differential Equations: Theory and Applications

4.14.7. Interior Neumann Problem For A Circle

The interior Neumann problem for a circle is described by
PDE : ∇ 2 u =
0, 0 ≤ r < a ; 0 ≤θ 2π
∂u ∂u ( a , r )
=
BC : = g= (θ ) ,r a
∂n ∂r

4.14.8. Laplace Equation In Cylindrical Coordinates

The Laplace equation in cylindrical coordinates assumes the following form:
1 1
∇ 2 u = urr + ur + 2 uθθ + u zz = 0
r r

4.14.9. Laplace Equation In Spherical Coordinates

In example 3 the Laplace equation is expressed in spherical and has the
following form:
∂  2 ∂u  1 ∂  ∂u  1 ∂2 u
∇2 u
=  r  +  sin θ  + = 0
∂r  ∂r  sin θ ∂θ  ∂θ  sin 2θ ∂φ 2
 CHAPTER

5
HYPERBOLIC DIFFERENTIAL
EQUATION

CONTENTS
5.1 Introduction...................................................................................... 280
5.2. Solution Of One-Dimensional Wave Equation by
Canonical Reduction .................................................................... 284
5.3. The Initial Value Problem; D’alembert’s Solution.............................. 288
5.4. Summary And Discussion................................................................ 297
280 Differential Equations: Theory and Applications

5.1 INTRODUCTION
One of the most important and typical homogenous hyperbolic differential
equations is the wave equation. It is of the form
∂2u
= c 2 ∇ 2u
2
∂r (1)
where c is the wave speed. This differential equation is used in many
branches of physics and longitudinal vibrations in a bar , propagation of
sound waves, electromagnetic waves, sea waves, elastic waves in solids, and
surface waves as in earthquakes. The solution of a waves equation is called
a wave function.
An example for inhomogeneous wave equation is
∂ 2u 2 2
−c ∇ u =
F
∂t 2 (2)
where F is a given function of spatial variables and time. In physical
problem F represents an external driving force such as gravity force.
Another related equation is

(3)
where γ is a real positive constant. This equation is called a wave equation
with damping term. The amplitude of which decreases exponentially as t
increases . , we shall derive the partial differential equation describing the
transverse vibration of a string .

5.1.1. Derivation of one-Dimensional wave equation

Suppose a flexible string is stretched under tension τ between two points at
a distance L apart as shown in . We assumed the following
• The mention takes place in one planes only and in the plane each
particle moves in a direction perpendicular to the equilibrium
position of the string.
• The tension τ in the string is constant.
• The gravitational force is neglected as compared with tension τ
of the string.
• The slope of the deflection curve is a small .
 Hyperbolic Differential Equation 281

A ( L ,0)
Let two fixed ends of the string be at the origin O and which
lies along the x − axis in its equilibrium position. Consider an infinitesimal
segment PQ of the string. Let ρ be the mass per unit length of the string. If
the string is set vibrating in the xy – plane , the subsequent displacement,
y from the equilibrium position of a point p of the string will be a function
of and time t , while an element of lenfght dx is stretched into an element
of length ds given by

 ∂y 
2
 1  ∂y  2 
1+ 
ds = 1+ 
 dx =   dx
 ∂x   2  ∂x  
The elementary elongation is given by
2
1  ∂y 
= – dx
dL ds=   dx
2  ∂x 
While the work done by this element against the tension τ is
2
1  ∂y 
τ   dx
2  ∂x 

Therefore, the total work done, W , for the whole string is

2
1
L
 ∂y 
W = ∫τ   dx
2 0  ∂x 

If U is the potential energy of the string, then

2
1
L
 ∂y 
U= W= ∫ τ 
2 0  ∂x 
 dx

Also, the total kinetic energy of the string is given by

2
1
L
 ∂y 
T= ∫ ρ
2 0  ∂t 
 dx

Using Hamilton’s principle, we have

282 Differential Equations: Theory and Applications

t1

δ ∫ ( T –U ) dt = 0
t0

i.e.,
t1

∫ ( T –U ) dt
t0

Is stationary. In other words,

1
t1 L
  ∂y  2  ∂y  
2

2 ∫t ∫0  ρ  ∂t  −τ  ∂x   dx dt
0  
∫ ∫ F ( x , t , y , y x , yt ) dx dt

Noting that x and t are independent variables, form the Euler-

Ostrogradsky equation, we have
∂F ∂  ∂F  ∂  ∂F 
−  −  =0
∂y ∂ t  ∂yt  ∂x  ∂y x 
which gives
∂  ∂y  ∂  ∂y 
ρ − τ =0
∂t  ∂t  ∂x  ∂x 
If the string is homogenous, then ρ and τ are constants, in which case
the governing equation representing the transverse vibration of a string is
given by
∂2 y 2
2 ∂ y
=c
∂t2 ∂x 2 (4)
where
τ
c2 =
ρ (5)
Example. Consider Maxwell’s equations of electromagnetic theory
given by
4πρ
∇.E =
 Hyperbolic Differential Equation 283

∇. H =
0
1 ∂H
∇× E = −
c ∂t
4π i 1 ∂E
∇×=
H +
c c ∂t
where E is an electric field, ρ is electric density, H is the magnetic field, i
is the current density, and c is the velocity of light . Show that in the absence
of charges, i.e., when ρ = i = 0, E and H satisfy the wave equations.
Solution. Given
1 ∂H
curl E = ∇× E = −
c ∂t
Taking its curl again, we get
 1 ∂H  ∂1 
∇× ( ∇ × E ) = ∇×  −  = −  ∇× H 
 c ∂t  ∂t  c 
1 ∂  1 ∂H  ∂ 1 
=− −  =−  ∇ × H 
c ∂t  c ∂t  ∂t  c 
1 ∂  1 ∂E  1 ∂2 E
=
−   =
−
c ∂t  c ∂t  c 2 ∂t 2
Moreover, using the identity
∇× ( ∇ × E ) = ∇ ( ∇ . E ) −∇ 2 E = −∇ 2 E

It follows directly that

2 1 ∂2 E
∇ E=
c 2 ∂t 2
Similarly, we can observe that the magnetic field H also satisfies
1 ∂2 H
∇2 H =
c 2 ∂t 2
which is also a wave equation.
284 Differential Equations: Theory and Applications

5.2. SOLUTION OF ONE-DIMENSIONAL WAVE

EQUATION BY CANONICAL REDUCTION
The one-dimensional wave equation is
utt − c 2 u xx =
0
(6)
Choosing the characteristic lines
ξ = x – ct , η= x + ct (7)
The chain rule of partial differential gives
uξ ξ x + uη η x =
ux = uξ + uη

ut = uξ ξt + uη ηt = c ( uη − uξ )
In the operator notation we have
∂ ∂ ∂ ∂  ∂ ∂ 
= + , =
c − 
∂x ∂ξ ∂η ∂t  ∂η ∂ξ 
Thus, we get
2
∂ 2u  ∂ ∂ 
= +  u=uξξ + 2uξη + uηη
 ∂ξ ∂η 
2
∂x (8)
∂ 2u
= c 2 ( uξξ − 2uξη − uηη )
∂t 2 (9)
Substituting Eq. (8) and (9 ) into Eq. ( 6 ) , we obtain
4u ξη = 0
(10)
Integrating, we get

where φ and ψ are arbitrary functions. Replacing ξ and η as defined in

Eq., we have the general solution of the wave equation in the form
u ( x ,=
t ) φ ( x – ct ) +ψ ( x + ct )
(11)
The two terms a wave traveling to the right with speed c whose shapes
does not change as it travels, the initial shape being given by a known
 Hyperbolic Differential Equation 285

φ ( x).
function In fact, by setting t = 0 in the form argument of φ , it can
be observed that the initial wave profile is given by
u1 ( x , 0 ) = φ ( x )

1
t= ,
At time c
 1
u1  x, = φ ( x −1)
 c
'
Let x = x –1 . Then
( )
φ ( x –1 ) = φ x ' .
That is, the same shape is
retained even if the origin is shifted by one unit along the x − axis . in other
 1
words, the graph of 
u  x , 
c  is the same as the graph of the original wave
1

2  2
t= , u1  x , 
profile translated one unit to the right . At c the graph of  c
is the graph of the wave profile translated two units to the right. Thus, in
= t 1, we haveu1 ( x = ,1) φ ( x − c ) .
particular, at Hence in one unit of time,
the profile has moved c units to the right. Therefore, c is the waves speed
or speed of propagation Using similar argument, we can conclude that the
u ( x=
, t ) ψ ( x + ct )
equation 2 is also a wave profile travelling to the left
c
with speed along the x − axis . Hence the general solution ( 11 ) of one-
dimensional wave equation represents the superposition of two arbitrary
wave profiles, both of which are travelling with a common speed but in the
opposite directions along the x – axis , while their forms remain unaltered
as they travel. This situation us described in
Let k be an arbitrary real parameter. Consider then
u (=
x, t ) φ  k ( x – ct )  +ψ  k ( x + ct ) 
(12)
This is also a solution of the one-dimensional wave equation. Further,
ω
let = kc. Then
u ( x , t ) φ ( kx –ωt )  +ψ  k + ( kx + ωt ) 
=
(13)
 286 Differential Equations: Theory and Applications

A function of the type given in Eq. (13) is a solution of one-dimensional

wave equation if ω = kc. Therefore, waves travelling with speeds which are
not the same as c cannot be described by the solution of the wave equation
(6). Here,
( kx + ωt ) is called the phase for the left travelling wave. We have
already noted that x ± ct are the characteristics of the one-dimensional
wave equation.
Example. Obtain the periodic solution of the wave equation in the form
u ( x , t ) = Ae i ( kx ± ωt )

ω
i= −1 , k =
± , A
where c is constant; and hence define various term
involved in wave propagation.
u ( x , t ) = f ( x ) e ± iωt
Solution. Let be a solution of the wave equation
utt = c 2 u x x
then
u xx f '' ( x ) e ± iωt , utt f ( x ) ω 2 e ± iωt
=

Substituting into the wave equation , we get

ω2
f ′′ ( x ) + f ( x)=
0
c2
Its general solution is found to be
 ω    ω 
=f ( x ) c1 exp i   x  + c2 exp  – i   x 
 c     c  
Therefore, the required solution of the wave equation is
  ω     ω  
=u ( x, t ) c1 exp i   x  + c2 exp  – i   x  e ± iωt
   c     c  
ω
k=± ,
Since c the time- dependent wave function are of the form
u ( x, t ) = Aei( kx ±ωt )
 Hyperbolic Differential Equation 287

u ( x, t ) = Ae ( )
i kx ±ωt
Hence, is a solution of the wave equation, and is called
a wave function. It is also called a plane harmonic wave or monochromatic
wave. Here, A is called the amplitude, ω the angular or circular frequency,
and k is the wave number, defined as the number of waves per unit distance.
By Taking the real and imaginary parts of the solution. We find the linear
combination of term of the form
A cos ( kx ± ωt ) , A sin ( kx ± ωt )

Representing periodic plane waves. For instance, consider the function

u ( x, t ) = A sin ( kx –ωt ) .
This is a sinusoidal wave profile moving towards
the right along the x − axis with speed c. Defining the wave length λ as
the length over which one full cycle is completed . we have λ = 2π / k .
Thereby implying that k = 2π / λ .

Suppose an observe is stationed at a fixed point

x0 ; then ,

 λ   λ 
+  A sin  kx0 − ω t −ω 
u  x0 , t=
 c   c 
= A sin ( kx0 − ωt −=
2π ) A sin ( kx0 –ωt )

Thus, we have
 λ 
u  x0 , t +  =u ( x0 , t )
 c 
λ
,
T=
Hence, exactly one computer wave passes the observer in time c
which is called the period of the wave. The reciprocal of the period is called
frequency and is denoted by
1
f=
T
π 
u A cos ( kx −=
ωt ) A sin  + kx − ωt  ,
The function,  2  also represents
π
a wave train except that it differ in phase by 2 form the sinusoidal wave.
 288 Differential Equations: Theory and Applications

Now consider the superposition of the sinusoidal waves having the same
amplitude, speed , frequency , but moving in opposite directions. Thus, we
have
u ( x=
, t ) A sin  k ( x – ct )  + A sin  k ( x + ct ) 

2 A sin kx cos ( kct ) = 2 A cos ( kct ) sin kx

 2 A cos ( kct ) 
Its amplitude factor  varies sinusoidal with
frequency ω . This situation is described as standing wave. The point
nπ
xn = , n = 0 , ± 1 , ± 2 ,…
k are called nodes . No displacement takes place
at a node . Therefore,
u ( xn , t ) = 0 for all t

The nth standing wave profile will have

( n −1 ) equally spaced nodes
in a given interval as shown in

5.3. THE INITIAL VALUE PROBLEM;

D’ALEMBERT’S SOLUTION
Consider the initial value problem of Cauchy type described as
2
PDE : ut t − c= u xx 0 , − ∞ < x < ∞ , t ≥ 0 (14)

= u ( x , 0 ) η=
( x ) , ut ( x , 0 ) v ( x ) (15)
ICs:
η ( x ) and v ( x )
where the curve on which the initial data are prescribed
x axis. ( x ) v( x )
is the and are assumed to be twice continuously dif-
u ( x ,t )
ferentiable . Here, the string considered is of an infinite extent. Let
denote the displacement for any x and t . At t = 0 , let the displacement
and velocity of the string be prescribed. We have already noted that the gen-
eral solution of the wave equation is given by
(16)
where φ and ψ are arbitrary functions. Substituting the ICs (15) into Eq.
(16), we obtain
 Hyperbolic Differential Equation 289

φ ( x ) +ψ ( x ) =
η ( x)
(17)
Integrating the second equation of (17) , we have
1
x
φ ( x ) –ψ ( x ) = v ( ξ ) dξ
c ∫0
Addition and substituting of this equation with the first relation of Eq.
(17) yield
η( x) 1
x

( x) v ( ξ ) dξ
2c ∫0
φ= +
2

η( x) 1
x

( x) v ( ξ ) dξ
2c ∫0
ψ= −
2

φ( x)
Respectively, substituting these relations for and x
ψ ( x ) into Eq. ( 16 ) ,
we at once have
x + ct
1 1
u( x =
,t ) η ( x + ct ) +η ( x − ct )  + ∫ v ( ξ ) dξ
2 2c x − ct (18)
Thus is known as the D’ALEMBERT’s Solution of the one-dimensional
wave equation. If v = 0 , i. e. , if the string is released form rest, the required
solution is
1
u( x =
,t ) η ( x + ct ) +η ( x − ct ) 
2 (19)
The D’ALEMBERT’s solution has an interesting interpretation as given
in figure
p ( x0 , t0 )
Consider the xt − plane and a point . Draw two characteristics
p
through backwards, until they intersect the initial line, i.e., the x − axis at
A and B. The equation of these two characteristics are
x ±ct = x 0 ± ct0
290 Differential Equations: Theory and Applications

u ( x , t ) at P ( x0 , t0 )
Equation (18) reveals that the solution can be
η
obtained by averaging the value of at A and B and integrating along v
the x − axis between Aand B. Thus, to find the solution of the eave
equation at a given point p in the xt − plane we should know the initial
data on the segment AB of the initial line which is obtained by drawing
the characteristics backward from p to the initial line. Here the segment
AB of the initial line, on which the value of u ( x, t ) at p depend is called
the domain of dependence of p, and the triangle PAB is called the
characteristics triangle , which is also called the domain of determinacy of
the interval.
Example. A stretched string of finite length L is held fixed at its ends
πx 
u ( x, 0 ) = u0 sin  .
and is subjected to an initial displacement  L  The string is
released from this position with zero initial velocity. Find the resultant time
dependent motion of the string.
Solution. One of the practical application of the theory of wave motion
is the vibration of a stretched string, say, that of a musical instrument. In
the present problem , let us consider a stretched string of finite L, which
is subjected to an initial disturbance. The governing equation of motion is

PDE:
utt − c 2u xx= 0 , 0 ≤ x ≤ L , t > 0 (1)

BCs:
( 0 , t ) u=
u= ( L ,t ) 0 (2)
 πx 
u ( x , 0 ) = u0 sin  ,
ICs:  l  (3)
∂u
( x ,0 ) = 0
∂t (4)
We have shown the solution of the one-dimensional wave equation by
canonical reduction as
u( x=
, t ) φ ( x – ct ) +ψ ( x + ct )
(5)
One of the known method for solving this problem is based on trial
function approach. Let us choose a trial function of the form
 Hyperbolic Differential Equation 291

π π
( x , t ) A  sin ( x + ct ) + sin ( x − ct )
u=
 L L  (6)
where A is an arbitrary constant. Now, we rewrite Eq. (6) as
 πx   cπ t 
u ( x , t ) = 2 A sin   cos  
 L   L  (7)
u0
A= ,
Obviously, (7) satisfies the initial condition (3) with 2 while the second
initial condition (4) is satisfied identically. In fact Eq. (7) also satisfies the
boundary condition (2). Therefore, the final solution is found to be
 πx   cxπ 
u ( x , t ) = u0 sin   cos  
 L   L  (8)
It may be noted that the trial function approach is easily adoptable if
the initial condition is specified as a sin function . However, it is difficult if
the initial conditions are specified as a general function such as f ( x ) . In
such case, it is better to follow variable separable method as explained in
previous sections .
Example. Solve the following initial value problem of the wave equation
(Cauchy problem), described by the inhomogeneous wave equation
f ( x ,t )
utt − c 2 u xx =
PDE:
subject to the initial conditions
u ( x , 0 ) η=
= ( x ) , ut ( x , 0 ) v ( x )
u= u + u u
Solution. To make the task easy, we shall set 1 2 , so that 1 is

a solution of the homogenous wave equation subject to the general initial

conditions given above. Then
u2 will be a solution of

∂ 2u 2 2
2 ∂ u2
− c f ( x, t )
=
∂t 2 ∂x 2 (20)
subject to the homogenous ICs
∂u 2
u2 ( x , 0 ) 0=
= , ( x ,0 ) 0
∂t (21)
292 Differential Equations: Theory and Applications

p ( x0 , t0 ) ,
To obtain the value of u at we integrate the partial differential
equation ( 20 ) over the region IR as shown in , we obtain
. .
 ∂ 2 u2 2
2 ∂ u2 
∫  ∂t 2
∫
IR 
− c
∂x 2


dx dt ∫IR ∫ f ( x , t ) dx dt
=

Using Green’s theorem in a plane to the left-hand side of the above

equation to replace the surface integral over IR by a line integral around the
boundary ∂IR of IR , the above equation reduces to
. .
 ∂  2 ∂ u2  ∂  ∂u2  
−∫ ∫ 
IR 
c
∂x 
−
∂x  ∂t
   dx dt =
 ∂t  
∫ ∫ f ( x , t ) dx dt
IR

And finally to
. .
 ∂u 2 ∂u 
∫ ∫ f ( x , t ) dx dt
∫∂IR  ∂t2 dx + c ∂x2  dt =
IR (22)
Now, the boundary ∂IR comprises three segments BP, PAand AB .
dx dx
= −c; PB. = c.
Along BP , dt along dt Using these results , Eq. (22)
. .
 ∂u ∂u   ∂u ∂u 
∫BP c  ∂t2 dt + ∂x2 dx  − PA∫ c  ∂t2 dt + ∂ x2 dx 
. .
 ∂u ∂u 
− ∫  2 dx + c 2 2 dt  =
∫ ∫ f ( x , t ) dx dt
AB  ∂t ∂x  IR

The integrands of the first two integrals are simply the total differentials,
while in the third integral, the first term vanishes on AB in view of the
second IC in Eq. (21), and the second term also vanishes because AB is
dt
= 0.
directed along the x − axis on which dx Then we arrive at the result
. .

∫ c du2 – ∫ ∫ f ( x , t ) dx dt
BP IR

which can be rewritten as

 Hyperbolic Differential Equation 293

.
cu2 ( p ) − cu2 ( B ) + cu2 ( p ) − cu2 ( A ) =
∫ ∫ f ( x, t ) dxdtIR (23)
2 ( A)
u= 2 ( B)
u= 0,
Using the first IC of Eq. ( 21), we get and hence
Eq. (23) becomes
.
1
( p)
u2= ∫ ∫ f ( x , t ) dx dt
2c IR

with the help of , we deduce

t0 x 0 + ct0 − c t
1
u2 ( p ) = ∫ ∫ f ( x , t ) dx dt
2c 0 x 0 − c t 0 + ct
(24)

Now, using the fact that 1 2 u= u + u ,

as also using Eq. (24) and
D’Alembert’s solution (18) the required solution of the inhomogeneous
wave equation subject to the given ICs is given by
x + ct t x + ct − ct
1 1 10 0 0
u ( x,=
t)
2
{η ( x + ct ) + η ( x – ct ) +
c ∫ v (ξ ) d ξ +
2 ∫0 x0 − ct∫0 + ct
f ( x, t ) dxdt
x − ct
(25)
The solution is known as the Riemann-Volterra solution.
Following Tychonov and Samarski, it is known that transverse vibration
of a string is normally generated in musical instruments. We distinguish the
string instrument depending on whether the string is plucked as in the case
of guitar or struck as in the case of harmonic or piano. In the case of string
which are struck we give a fixed initial velocity but does not undergo any
initial displacement. In the case of plucked instruments, the string vibrate
form a fixed initial displacement without any initial velocity .
Let a thin homogeneous string which is perfectly flexible under uniform
tension lie in it equilibrium position along the x − axis . The ends of the
string are fixed at x = 0 and x = L . The string is pulled aside a short distance
and released. If no external forces are present which correspond to the case
of free vibrations, the subsequent motion of the string is described by the
u ( x, t )
solution of the following problem:
utt − c 2 u= 0, 0 ≤ x ≤ L, t > 0
PDE: tt (26)
u ( 0,=
t ) 0, t > 0
BCs:
 294 Differential Equations: Theory and Applications

u ( L,=
t ) 0, t > 0
(27)
= u ( x , 0 ) f ( x ) , ut ( x, 0 ) − g ( x )
ICs: (28)
To obtain the variables separable solution, we assume
u ( x, t ) = X ( x ) T ( t )
(29)
Substituting into Eq. (26), we obtain
d 2T 2 d2X
X = c T
dt 2 dx 2
i.e.,
d 2 X / dx 2 d 2 / dt 2
= = k
X c 2T (a separation constant)
Case I. When K > 0 , we have k = λ . Then
2

d2X
2
−λ2X =
0
dx
d 2T
2
− c 2 λ 2T =
0
dt
Their solution can be put in the form
X c1eλ x +c2 e − λ x (30)
=

=T c3ecλt + c4 e − cλt (31)

Therefore,
u ( x, t ) = ( )(
c1eλ x + c2 e − λ x c3ecλt + c4 e − λt ) (32)
Now, use the BCs:
u ( 0, t ) = (
( c1 + c2 ) c3eλt + c4e−λt
0= ) (33)
0. Also, u ( L, t ) = 0 gives
which imply that c1 + c2 =
c1e − λ L + c2 e − λ L =
0
(34)
Equation (33) and (34) possess a non- trivial solution if
 Hyperbolic Differential Equation 295

1 1
λL
= e − λ L − eλ L = 0
e e−λ L
or
1 − e 2 λ=
L
0 implying e 2 λ=
L
1 or λ =
L 0

This implies that λ = 0, since L cannot be zero, which is against the

assumption as in the Case I, Hence, the solution is not acceptable.
Case II. Let K = 0 . Then we have
d2X d 2T
= 0,
= 0
dx 2 dt 2
Their solutions are found to be
Ax B, T =
X =+ ct + D
Therefore, the required solution of PDE (26) is
u ( x ,t ) =( Ax + B )( ct + D )
Using the BCs, we have
u ( 0, t ) =B ( ct + D ) , implying B =
0= 0

u ( L ,t) =AL ( ct + D ) , implying A =

0= 0

Hence, only a trivial solution is possible. Since we are looking for a non-
trivial solution, consider the following case.

Case III. When k = 0, say K = −λ , the differential equations are

2

d2X d 2T
2
+ λ 2
X= 0, 2
+ c 2 λ 2 T= 0
dx dt
Their general solution gives
u ( x, t ) =
( c1 cos λ x + c2 sin λ x )( c3 cos cλt + c4 sin cλt ) (35)
u ( 0, t ) = 0 c1 = 0. Also, using the BC:
Using the BC: we obtain
u ( L , t ) = 0, nπ
we get sin λ L = 0 implying that= λ , n 1 , 2 ,….
=
which are
L
the eigenvalue. Hence the possible solution is
296 Differential Equations: Theory and Applications

nπ x  nπ ct nπ ct 
u n ( x, t ) =
sin  A cos + B sin , n =
1 , 2 ,…
L  L L  (36)
Using the superposition principle, we have

(37)
The initial condition give
∞
nπ x
u ( x=
, 0 ) f=
( x) ∑ A sin n
n =1 L
which is half – range Fourier sine series, where
2 nπ x
An = f ( x ) sin dx
L L (38)
Also,
∞
nπ x  nπ 
ut ( x=
, 0 ) g=
( x) ∑B n sin  c
n =1 L  L 
which is also a half range sine series , where
2 nπ x
L
g ( x ) sin
nπ c ∫
Bn = dx
0
L
Hence the required physically meaningful solution is obtained from Eq.
A B u ( x, t ) givenby Eq.
(37), where n and n are given by Eq. (38) and (39). n
nπ c
= ωn= , n 1 , 2 ,…
(36) are called normal modes vibrations and L are
called frequencies .
Example. Obtain the solution of the wave equation
utt = c 2u xx
under the following conditions:

(i)
( 0, t ) u=
u= ( 2, t ) 0
πx
u ( x, 0 ) = sin 3
(ii) 2
 Hyperbolic Differential Equation 297

ut ( x, 0 ) = 0.
(iii)
Solution. We have noted in Example 4 that the physically acceptable
solution of the wave equation is given by Eq. (35), and is of the form
u ( x ,t) ( c1 cos λ x c2 sin λ x ) c3 cos ( cλt ) + c4 sin ( cλt )
u (0 , t ) , c = 0. Also, condition (iii)
Using the condition we obtain 1
c = 0. The condition u ( 2, t ) = 0 gives
implies 4
sin 2λ = 0,
Implying that
nπ
λ
= ,=
n 1, 2 ,…
2
Finally, using condition (ii), we obtain
∞
nπ x nπ ct
∑ A sin
n =1
n
2
cos
2 (40)
Finally, using the condition (ii), we obtain
∞
nπ x 3 πx 3 πx 1 3π x
∑A sin=
n =1 2
n sin
=
2 4
sin − sin
2 4 2

3
A1 =
which gives 4.

5.4. SUMMARY AND DISCUSSION

One of the most important and typical homogenous hyperbolic differential
equations is the wave equation. It is of the form
∂2u
= c 2 ∇ 2u
∂r 2
where c is the wave speed. This differential equation is used in many
branches of physics and longitudinal vibrations in a bar , propagation of
sound waves, electromagnetic waves, sea waves, elastic waves in solids, and
surface waves as in earthquakes. The solution of a waves equation is called
a wave function.
 298 Differential Equations: Theory and Applications

∂ 2u 2 2
−c ∇ u = F
An example for inhomogeneous wave equation is ∂t 2
,Where F is a given function of spatial variables and time. In physical
problem F represents an external driving force such as gravity force.

Another related equation is ,Where γ is a

real positive constant . This equation is called a wave equation with damping
term. The amplitude of which decreases exponentially as t increases . ,
we shall derive the partial differential equation describing the transverse
vibration of a string.
( x , t ) φ  k ( x – ct )  +ψ  k ( x + ct ) 
u=

This is also a solution of the one-dimensional wave equation. Further,

let ω = kc. Then
u ( x , t ) φ  ( kx –ω t )  +ψ  k + ( kx + ωt ) 
=

A function of the type given in above equation is a solution of one-

dimensional wave equation if ω = kc. Therefore, waves travelling with
speeds which are not the same as c cannot be described by the solution of
the wave equation. Here,
( kx + ωt ) is called the phase for the left travelling
wave. We have already noted that x ± ct are the characteristics of the one-
dimensional wave equation.
Solution of the wave equation is
  ω    ω  
=u ( x , t )  c1 exp  i   x  + c2 exp  – i   x   e ± iωt
   c     c  
ω
k=± ,
Since c the time- dependent wave function are of the form
u ( x, t ) = Aei( kx ±ωt )

u ( x, t ) = Aei( kx ±ωt )
Hence, is a solution of the wave equation, and is called
a wave function . it is also called a plane harmonic wave or monochromatic
wave. Here, A is called the amplitude , ω the angular or circular frequency,
and k is the wave number, defined as the number of waves per unit distance
 Hyperbolic Differential Equation 299

. By Taking the real and imaginary parts of the solution. We find the linear
combination of term of the form
A cos ( kx ± ωt ) , A sin ( kx ± ωt )

Representing periodic plane waves. For instance , consider the function

u ( x, t ) = A sin ( kx –ωt ) .
This is a sinusoidal wave profile moving towards
the right along the x − axis with speed c. Defining the wave length λ as the
2π
λ=
length over which one full cycle is completed . we have k . Thereby
2π
k= .
implying that λ
x + ct
1 1
u( x =
,t ) η ( x + ct ) +η ( x − ct )  + ∫ v ( ξ ) dξ
2 2c x − ct

Thus is known as the D’ALEMBERT’s Solution of the one-dimensional

wave equation. If v = 0 , i. e. , if the string is released form rest, the required
solution is
1
u( x =
,t ) η ( x + ct ) +η ( x − ct ) 
2
The D’ALEMBERT’s solution has an interesting interpretation.
x + ct t x + ct − ct
1 1 1 0 0 0
u ( x,=
t)
2
{η ( x + ct ) + η ( x – ct ) +
c ∫ v (ξ ) d ξ +
2 ∫0 x0 − ct∫0 + ct
f ( x, t ) dx dt
x − ct

The solution is known as the Riemann-Volterra solution

 CHAPTER

6
PARABOLIC DIFFERENTIAL
EQUATIONS

CONTENTS
6.1. Introduction..................................................................................... 302
6.2. Boundary Conditions....................................................................... 304
6.3. Elementary Solutions Of The Diffusion Equation ............................. 305
6.4. Dirac Delta Function....................................................................... 310
6.5. Separation Of Variables Method....................................................... 316
6.6. Maximum-Minimum Principle and Consequences........................... 340
6.7. Miscellaneous Example. ................................................................. 343
6.8. Boundary Conditions....................................................................... 352
302 Differential Equations: Theory and Applications

6.1. INTRODUCTION
The diffusion phenomena such as conduction of heat in solids and diffusion
of vorticity in the case of viscous flow past a body are governed by a partial
differential equation of parabolic type. For example , the flow of heat in a
conducting medium is governed by the parabolic equation
∂T
ρ  = div ( K ∇T ) + H ( r.T , t )
∂t (1)
where ρ is the density,  is the specific heat of the solid, T is the temperature
at a point with position vector r. K is the thermal conductivity , t is the time
H ( r,T , t )
,and is the amount of heat generated per unit time in the element
dV situated at a point ( x, y, z ) whose position vector is r. This equation is
known as diffusion equation or heat equation. We shall now derive the heat
equation form the basic concepts.
Let V be an arbitrary domain bounded by a closed surface
S and letV= V ∪ S . Let T ( x, y, z , t ) be the temperature at a point ( x, y, z )
at time t. if the temperature is not constant , heat flows from a region of high
temperature to a region of low temperature and flows the Fourier law which
q ( r ,t )
states that heat flux across the surface element dS with normal n̂
is proportional
q ( r , t ) =− k ∇T ( r , t )
(2)
where k is the thermal conductivity of the body. The negative sign indicates
that the heat flux vector points in the direction of decreasing temperature .
Let n̂ be the outward unit normal vector and q be the heat flux at the surface
element dS . Then the rate of heat flowing through the element surface dS
in unit time as shown in
dQ = ( q . nˆ ) dS
(3)
Heat can be generated due to nuclear reactions or movement of
mechanical parts as in inertial measurement unit ( IMU ) , or due to chemical
sources which may be a function of position, temperature and time and
H ( r ,T , t ) .
may be denoted by We also may be function of position ,
temperature and time may be substance as the amount of heat needed to
 Parabolic Differential Equations 303

raise the temperature of a unit mass by a unit dm = ρ dV the value T is

given by therefore,
Q = ∫∫∫ C ρT dV
V

dQ ∂T
= ∫∫∫  ρ dV
dt V
∂t

These energy balance equation for a small control volume V is: The
rate of energy storage in V equate to the sum of rate of heat entering V
theorugh its bounding surface and the rate of heat generation in V . Thus,
∂T ( r , t ) .

∫∫∫  ρ dV = − −∬q . nˆ dS + ∫∫∫ H ( r ,T , t ) dV

∂t
V S V (4)
Using the divergence theorem, we get
 ∂T ( r , t ) 
∫∫∫   ρ dV = − div q ( r , t ) + H ( r ,T , t )  dv =
0
∂t 
V (5)
Since the volume is arbitrary , we have
∂T ( r , t )
ρ − div q ( r , t ) + H ( r ,T , t )
=
∂t (6)
Substituting Eq. (2) into Eq. (6), we obtain
∂T ( r , t )
ρ  = div q ( r , t ) + H ( r ,T , t )
∂t (7)
If we define thermal diffusivity of the medium as
k
α=
ρ
Then the differential equation of heat conduction with heat source is
1 ∂T ( r , ) H ( r,T , t )
∇ 2T ( r , t ) +
=
α ∂t K (8)
In the absence of heat sources, Eq. (8) reduces to
304 Differential Equations: Theory and Applications

∂T ( r , t )
= α∇ 2T ( r , t )
∂t (9)
This is called Fourier heat conduction equation or diffusion equation.
The fundamental problem of heat conduction is to obtain the solution of Eq.
(8) subject to the initial and boundary conditions which are called initial
boundary value problems, hereafter referred to as IBVPs.

6.2. BOUNDARY CONDITIONS

The heat conduction equation may have numerous solution unless a set of
initial and boundary conditions are specified .The boundary conditions are
mainly of three types which we now briefly explain.
Boundary Condition 1: The temperature is prescribed all over the
T ( r ,t )
boundary surface. That is, the temperature is a function of both
T = G ( r ,t )
position and time. In other words, which is some prescribed
function on the boundary. This type of boundary condition is called the
Dirichlet condition. Specification of boundary conditions depends on the
problem under investigation. Sometimes the temperature on the boundary
surface is a function of position only or is a function of time only or a constant.
T ( r, t ) = 0
A special case includes on the surface of the boundary, which
is called a homogeneous boundary condition Specification of boundary
conditions depends on the problem under investigation. Sometimes the
temperature on the boundary surface is a function of position only or is a
T ( r ,t ) = 0
function of time only or a constant . A special case includes
on the surface of the boundary, which is called a homogeneous boundary
condition.
Boundary Condition II: The flux of heat, i.e., the normal derivative of
∂T
,
the temperature ∂n is prescribed on the surface of the boundary. It may
be a function of both position and time, i.e.,
∂T
= f ( r ,t )
∂n
This is called the Neumann condition. Sometimes, the normal derivatives
of temperature may be a function of position only or a function of time
 Parabolic Differential Equations 305

only. A special case includes boundary. This homogenous

boundary condition is also called insulated boundary condition which states
the heat flow is zero.
Boundary condition III. A linear combination of the temperature and
its normal derivatives is prescribed on the boundary , i.e.,
∂T
K G ( r ,t )
+ hT =
∂n
where K and h are constants. This type of boundary condition is called
Robin’s condition. It means that the boundary surface dissipates heat by
convection. Following Newton’s law of the cooling, which states that the rate
at which heat is transferred from the body to the surrounding is Proportional
to the difference in temperature between the body and the surrounding , we
have
∂T
−K =h ( T − Ta )
∂n
As a special case, we may also
∂T
K 0
+ hT =
∂n
which is homogeneous boundary. This mean that heat is converted by
dissipation from the boundary surface into a surrounding maintained at zero
temperature.
The other boundary conditions such as the heat transfer due to radiation
obeying the fourth power temperature law and those associated with change
of phase .like melting, ablation, etc, give rise to non-linear boundary
conditions.

6.3. ELEMENTARY SOLUTIONS OF THE DIFFU-

SION EQUATION
Consider the one-dimensional diffusion equation
∂ 2 T 1 ∂T
= , − ∞ < x ∞ ,t 0
∂ x 2 α ∂t (10)
 306 Differential Equations: Theory and Applications

The function
1
T ( x ,t ) exp  − ( x –ξ ) / ( 4at ) 
2
=
4 x at  
(11)
where ξ is an arbitrary real constant , is a solution of Eq. (10) . It can be
verified easily as follows:

∂T 1  ( x –ξ ) 2 
  exp[ − ( x –ξ ) 2 / ( 4at )]
∂t 4 xat  4α t 

∂T 1  −2 ( x –ξ ) 
 exp [ − ( x –ξ ) / ( 4at )]
2

∂t 4 xat  4α t 
Therefore
 ( x –ξ ) 
2
∂2T 1 1 1 ∂T
 exp [ − ( x –ξ ) / ( 4at ) ] =
2
= – +
 2α t 4α 2 t 2
2
∂x 4 xat  a ∂t

which show that the function (11) is a solution of Eq. (10). The function (11),
known as Kernel, is the elementary solution or the fundamental solution of
t > 0 , the Kernel T ( x , t )
the heat equation for the infinite interval. For is
T ( x ,t )
an analytic function of x and t and it can also be noted that is
positive for every x . Therefore, the region of influence for the diffusion
x →∞ ,
equation includes the entire x-axis. It can be observed that as the
amount of heat transported decrease exponentially .
In order to have an idea about the nature of the solution to the heat
equation, consider a one-dimensional infinite region which is initially at
f ( x)
temperature . Thus, the problem is described by
∂T ∂2T
=
PDE : α 2 , − ∞ < x ∞ ,t 0
∂t ∂x

IC:
( x , 0 ) f ( x ) , − ∞ < x=
T= < ∞ ,t 0

Following the method of variables separable, we write

 Parabolic Differential Equations 307

(14)
Eq . ( 12 ) .
Substituting into We arrive at
X '' 1 β ′
= =λ
X α β (15)
where λ is a separation constant. The separated solution for β gives
β = C e aλt (16)

If λ > 0 , we have β and , therefore, T growing exponentially with

time. From realistic physical considerations, it is reasonable to assume that
f ( x ) → 0 as x →∞ , T ( x , t ) < M as x →∞ . T ( x ,t )
while But for
to remain bounded , λ should be negative and thus we take λ = − µ .
2

Eq . ( 15 ) .
Now from we have
X '' + µ 2 X =
0
Its solution is found to be
=X c1 cos µ x +c 2 sin µ x
Hence
T ( x , t , µ ) = A cos µ x B sin µ x ) e − aµ
2
t
(17)
Eq . ( 12 ) ,
is a solution of where Aand B are arbitrary constants . Since
f ( x)
is in general not periodic , it Is natural to use Fourier integral instead
of Fourier series in the present case . Also, since A and B are arbitrary, we
may consider them as functions of µ and take
= A A= ( µ ) ,B B( µ ) .
In this particular problem, since we do not have any boundary conditions
which limit our choice of µ , we should consider all possible values. From
the principles of superposition, this summation of all the product solution
will give in the relation
∞ ∞
T ( x, t )
= T ( x, t , µ ) d µ ∫  A ( µ ) cos µ x + B ( µ ) sin µ x  e
∫=
−αµ 2 t
dµ
0 0 (18)
 308 Differential Equations: Theory and Applications

Eq . ( 12 )
which is the solution of From the initial condition (13) , we have
∞
T (=
x, 0 ) f=
( x) ∫  A ( µ ) cos µ x + B ( µ ) sin µ x  d µ
0 (19)
In addition, if we recall the Fourier integral theorem, we have
1
∞
 ∞ 
f ù( t )  ùf ( x ) cos ( t − x ) dx  d
π ∫0  −∫∞  (20)
Thus, we may write
1  ∞ ∞

f ( x)= ∫  ∫ f ( y ) cos µ ( x – y ) dy  d µ
π 0  −∞ 

1  ∞
∞

( )( )
π ∫0  −∫∞
 f y cos µ x cos µ y + sin µ x sin µ y dy  dµ


1  
∞ ∞ ∞

∫  cos µ x ∫ f ( y ) cos µ dy + sin µ x ∫ f ( y ) sin µ ydy  d µ

π 0 
−∞ −∞
(21)
Let
∞
1
A( µ ) = ∫ f ( y ) cos µ dy
π −∞

∞
1
B( µ ) = ∫ f ( y ) sin µ y dy
π −∞

Then Eq. (21 ) can be written in the form

∞
=f ( x) ∫  A ( µ ) cos µ x + B ( µ ) sin µ x  d µ
0 (22)
Eq . ( 19 ) and ( 22 ) ,
Comparing we shall write relation (19) as
 ∞ 1
∞

T (=
x, 0 ) f=
( x) ( ) ( )
π ∫0  −∫∞
 f y cos µ x – y dy  dµ
 (23)
 Parabolic Differential Equations 309

Assuming that the conditions for the formal interchange of orders of

integration are satisfied, we get
∞1
∞

=T ( x, t )
π ∫0
f ( y )  ∫ exp −αµ 2
t cos µ ( x – )(d µ  dy )
−∞  (25)

Using s = µ α t , and choosing

x− y
b=
2 αt
Equation (26) becomes
∞
π
cos µ ( x − y ) = exp  − ( x – y ) / ( 4at ) 
2
∫e
−αµ 2t
dµ
4α t  
0 (27)
Eq . ( 27 ) Eq . ( 25 ) ,
Substituting into we obtain
∞
1
T ( x, t ) ∫ f ( y ) exp − ( x – y ) / ( 4at ) dy
2

4απ t −∞ (28)
f ( y)
Hence, if is bounded for all real values of y , Eq. ( 28 ) is the
solution of the problem described by Eq.( 12 ) and ( 13 ) .
Example. In a one-dimensional infinite solid, − ∞ < x < ∞ , the surface
a < x < b is initially maintained at temperature T0 and at zero temperature
everywhere outside the surface. Show that
T0   b − x   a−x 
T ( x ,t ) =  erf  – erf  
2   4at   4at  
where erf is an error function.
Solution. The problem is described as follows:
=Tt α T xx , − ∞ < x < ∞
PDE:
=T T0 , a< x<b
IC: .outside the above region
The general solution of PDE is found to be
∞
1
T ( x, t ) ∫ f (ξ ) exp − ( x − ξ ) / ( 4at ) dξ
2
=
4at −∞
 310 Differential Equations: Theory and Applications

Introducing the new independent variable η defined by

x −ξ
η= −
4α t
And hence
d ξ = 4α t dη
The above equation becomes
( b − x ) / ( 4 at )
T0
T ( x ,t ) = ∫
2
e −η dη
π ( a − x ) / ( 4 at )

 2 ( b – x )/( 4a t ) ( a − x ) / ( 4 at ) 
T0 2
∫ ∫
−η 2 2
 e dη − e −η dη 
2  π 0 π 0 
Now we introduce the error function defined by defined by
2
z
Erf ( z ) ∫ exp ( −η ) dη
2
=
π 0

Therefore the required solution is

T0   b –x   a−x 
T ( x ,t )
=  erf   − erf  
2   4at   4α t  

6.4. DIRAC DELTA FUNCTION

According the notion mechanic, we come across a very large force (ideally
infinite) acting for a short duration (ideally zero time ) knows as impulsive
force. Thus we have a function which is non-zero in a very short interval. The
Dirac delta function may be thought of as a generalization of this concept.
This Dirac delta function and its derivative play a useful role in the solution
of initial boundary value problem (IBVP).
Consider the function having the following property:
 1 2 ε , t < ε
δε (t ) = 
 0, t < ε (29)
Thus,
 Parabolic Differential Equations 311

∞ ε
1
ε ( t ) dt
∫ δ= ∫= dt 1
2ε
−∞ −ε (30)
f (t ) ( –ε , ε ) .
Let be any function which is integrable in the interval
Then using the Mean –value theorem of integral calculus , we have
∞ ε
1
∫ f ∞ ( t ) δ ε ( t )=
dt ∫ε f ( t )=
dt f (ξ ) , − ε < ξ < ε
2ε
−∞ − (31)
δ (t ) δε ( t )
Thus, we may regard as limiting function approached by
as ε → 0 , i.e.,
δ ( t ) = LTε → 0 δ ε ( t )
(32)
Eq . ( 29 ) and ( 30 ) ,
As ε → 0 , we have , from the relations
( inthe same of being very large )

LTε → 0δ ε ( t ) =  ∞, if t = 0
δ (t ) =  0, if t ≠ 0
 (33)
∞

∫ δ ( t ) dt 1
−∞ (34)
δ (t ) Eq. ( 33 ) ( 34 ) is known
This limiting function defined by and
as Dirac delta function or the unit impulse function. Its profile is depicted in
Dirac originally called it an improper function as there is no proper function
with these properties . In fact, we can observed that
∞ .
=1 ∫ δ ( t ) dt LTε → 0
= ε ( t ) dt
∫ ε δ= =
LT ε →0 0 0
−∞ t >

δ (t )
Obviously, this contradiction implies that cannot be a function in
the ordinary sense. Some important properties of Dirac delta function are
presented now:
312 Differential Equations: Theory and Applications

Property 1:
∞

∫ δ ( t ) dt =1
−∞

f (t ) ,
Property II: For any continuous functions
∞

∫ f ( t ) δ ( t ) dt = f ( 0 )
−∞

Proof (of property 2) Consider the equation

∞
LTε → 0 ∫ f ( t=
−∞
) δ ε ( t ) dt LT ε → 0 f ( ξ ) − ε < ξ < ε

As ε → 0 , we have ξ → 0 . Therefore,
∞

∫ f ( t ) δ ( t ) dt = f ( 0 )
−∞

f (t )
Property III: Let be any continuous function. Then
∞

∫ δ ( t – a ) f ( t ) dt = f ( a )
−∞

Proof
Consider the function
1
 . a <1 < a + ε
δe ( t –a ) =  ε
0 , elsewhere

Using the mean-value theorem of integral calculus, we have

∞ a+ε
1
∫ δε ( t – a ) f ( t ) dt=
−∞
ε ∫ f ( t ) dt=
a
f ( a + θε ) , 0 <θ <1

Now, taking the limit as ε → 0, we obtain

∞

∫ δ ( t − a ) f ( t ) dt =
−∞
f (a)
 Parabolic Differential Equations 313

f ( t ) by δ ( t − a )
Thus, the operation of multiplying and integrating
over all t is equivalent to substituting a for t in the original function .
Property IV:
δ ( −t ) =
δ (t )
Property V:
1
δ ( at )
= δ ( t ) , a >0
a
δ (t )
Property VI: If is a continuously differentiable. Dirac delta
function vanishing for large , then
t
∞ ∞
∞
f ( t ) δ ' ( t ) dt  f ( t ) δ ( t )  − ∞ −
∫= ∫ f ( t ) δ ( t ) dt
'

−∞ −∞

Since δ t → 0 as t →∞ , − ∞ , we have
∞

∫ f ( t ) δ ( t ) dt = − f ( 0 )
' '

−∞

Property VII:
∞

∫ δ ( t − a ) f ( t ) dt =
− f (a)
' '

−∞

Having discussed the one-dimensional Dirac delta function , we

can extend the definition to two dimension, Thus, for every f which is
continuous over the region S containing the point
( ξ ,η ) , we define
δ ( x –ξ , y –η )
in such a way that
.

∬δ ( x –ξ y –η ) f ( x , y ) dσ = f ( ξ ,η )
S

δ ( x − ξ , y −η )
Note that is a form a limit of a sequence of ordinary
functions, i.e.,
314 Differential Equations: Theory and Applications

δ ( x − ξ , y −η ) =
LTε → 0 δ ε ( r )
(36)
r 2 = ( x − ξ ) + ( y −η ) .
2 2

where Also observe that

∬δ ( x − ξ ) δ ( y − η ) f ( x , y ) dx dy =
f (ξ ,η )
(37)
Eq . ( 35 ) and ( 37 ) ,
Now, comparing we see that
δ ( x − ξ , y –η ) =
δ ( x − ξ ) δ ( y −η )
(38)
Thus, a two –dimensional Dirac delta function can be expressed as the
product of two one –dimensional delta functions. Similarly, the definition
can be higher dimensions.
Example. A one-dimensional infinite region − ∞ < x < ∞ is initially kept
at zero temperature. A heat source of strength s units, situated at x = ξ
g
releases its heat instantaneously at time t = τ . Determine the temperature in
the region for t >τ .
Solution. Initially, the region − ∞ < x < ∞ is at zero temperature. Since
the heat source is situated at x = ξ and releases heat instantaneously at
t = τ , the released temperature at x = ξ and t = τ is a δ − function type.
Thus, the given problem is a boundary value problem described by
∂ 2 T g ( x , t ) 1 ∂T
2
+ = , −∞ x ∞ , t 0
PDE: ∂ x k a ∂t
T ( x ,=
t) F(= x ) 0 , − ∞ < x < ∞=
,t 0
IC:
The general solution to this problem as given , after using the initial
F ( x) = 0 ,
condition is
∞
α t
d t'
∫g ( x′, t ′) exp  – ( x – x ) / {4α ( t − t ′)} dx′
2
T ( x, t ) ∫
'

k t ′ =0 (
4πα t − t '
) x′
(39)
Since the heat source term is of the Dirac delta function type, substituting
g (=
x , t ) g s δ ( x −ξ ) δ ( t –τ )
into Eq. (39), and integrating we get, with the help of properties of delta
function, the relation
 Parabolic Differential Equations 315

T ( x ,t ) =
α gs
t
 { (
exp  – ( x –ξ ) 2 / 4α t − t '  )}
 δ t –τ dt '
∫ ( )
k 4πα 0 4πα ( t –τ )

Therefore, the required temperature is

α g s exp  – ( x –ξ ) / { 4α ( t –τ )}
2

T ( x ,t ) =
k 4πα ( t −τ )
for t > τ
Example. An infinite one-dimensional solid defined by −∞ < x < ∞ is
maintained at zero temperature initially. There is a heat source of strength
gs ( t )
units, situated at x = ξ , which releases costant heat continuously
for t > 0 . Find an expression for the temperature distribution in the solid
for t > 0 .
g ( x , t ) = g s ( t ) δ ( x –ξ )
Solution. This problem is similar to, except that
is a Dirac delta function type. The solution to this IBVP is
α t
g s (t′)
T ( x, t ) ∫ ( )
exp  − x − ξ 2 / {4 α ( t − t ′ )} dt ′
k t ′=0 4πα ( t − t ′ )
(40)
= g ( t ) constant
= g x ( say ) .
It is given as s Let us introduce a new
η
variable defined by
1 ( x –ξ )
2
x −ξ
=η = or t − t ′
4α ( t − t ′ ) η 2 4α

Therefore,

dt =' 1 ( x –ξ ) 2 dη
η3 2α
, Eq . ( 40 )
Thus becomes

T ( x ,t ) = gs
x −ξ
∞
(
exp −η 2 ) dη
2K π ∫ η 2
( x –ξ )/ 4 at

However,
316 Differential Equations: Theory and Applications

d  e− n  e− n
2 2
2
 – =  + 2e − η
dη  η  η 2

Hence,

x −ξ   e− η 2 
∞
∞ 
T ( x, t )  –
∫ e −η dη 
2
gs  −2
2K π   η x − ξ / ( x –ξ )/ 
 4 at 4 at

Recalling the definitions of error function and its complement
2
x
Erf ( x ) =
π ∫
−η 2
e dη ,
0 erf
( ∞ ) =1
∞ x

erfc ( x ) =
1 − erf ( x ) = (
2 / π  ∫ exp −η 2 dη − ∫ exp −η 2 dη  ) ( )
0 0 

2
x
exp ( − n ) dη
π ∫
2

The temperature distribution can be expressed as

α gs  t  ( x –ξ ) 2  x –ξ  x −ξ 
T (=
x ,t )  exp  − −  1 – erf 
K  2π  4at  2α  4at  

Alternatively, the required temperature is

α gs  t  ( x –ξ ) 2  x –ξ x –ξ 
( x ,t )
T=  exp  − − erfc 
K  2π  4at  2α 4at 
 

6.5. SEPARATION OF VARIABLES METHOD

Consider the equation
∂T ∂2T
=α
∂t ∂ x 2 (41)
Among the many methods that are available for the solution of the above
parabolic partial differential equation. The method of separation of variables
 Parabolic Differential Equations 317

T ( x ,t )
of as follows. Let
T ( x ,t ) = X ( x ) β ( t )
(42)
Be a solution of the differential Eq.(41). Substituting Eq.(42) into (41),
X '' 1 β′
= = K ,a
we obtain X a β separation constant. Then we have
d 2X
– KX = 0
d x2 (43)
dβ
− ak β =0
dt (44)
In solving Eq. (43) and (44), three distinct case arise:

Case I when K is positive , say λ , the solution of Eq. (43) and (44)
2

will have the form

2
c1 e λ x + c2 e – λ x . β =
X= c3 e a λ t
(45)

Case II when K is negative , say – λ , then the solution of Eq. (43)

2

and (44) will have the form

(46)
Eq. ( 43 ) ( 44 ) can have
Case III when K is zero, the solution of and
the form
c1 c + c2 . β =
X= c 3 (47)
Thus, various possible solutions of the heat conduction equation (41)
could be the following
T (=
x ,t ) (c )
2
'
1 e λ x + c 2' e − λ x e a λ t

T ( x , t ) ( c1' cos λ x + c 2' sin λ x ) e − aλ

2
= t
(48)
T ( x,=
t ) c x +c 1
' '
2

where
=c 1' c=
1c 3 , c '2 c 2 c 3
 318 Differential Equations: Theory and Applications

Solve the one-dimensional diffusion equation in the region

0 ≤ x ≤ π , t ≥ 0 , subject to the conditions

( i ) T remains finite as t →∞
( ii=
) T , if x 0 and π for all t
0=

 π
 x , 0 ≤ x ≤ 2
=t 0= ,T 
π – x , π ≤ x ≤ π .
( iii ) At  2
Solution. Since T should satisfy the diffusion equation, the three
possible solutions are:
T (=
x ,t ) (c e )
2
λx
1 + c2 e − λ x e a λ t

T ( x ,=
t ) c1 x + c2 )

The first condition demand that T should remain finite as t →∞ , we

therefore reject the first solution. In view of BC
( ii ) , the third solution
gives
0 =c1 . 0 +c 2 , 0 =c1 .π +c 2

c and c 2
Implying thereby the both 1 are zero and hence T = 0 for all t.
This is a trivial solutions. Since we are looking for a non-trivial solutions,
we reject the third solution too. Thus, the only possible solution satisfying
the first condition is
T ( x ,t ) ( c1 cos λ x + c2 sin λ x ) e − αλ
2
= t

Using the BC
( ii ) , we have
=0 c1 cos λ x + c 2 sin λ x x=0

c1= 0 .
Implying Therefore, the possible solution is
 Parabolic Differential Equations 319

T ( x , t ) = c2 e − αλ t sin λ x
2

BC: T 0=
Applying the= when x , we get
sin λπ =⇒
0 λπ = nπ
where nis an integer. Therefore,
λ =n
Hence the solution is found to be of the form
T ( x , t ) = ce an t sin nx
2

Nothing that the heat conduction equation is linear, its most general
solution is obtained by applying the principle of superposition. Thus,
∞
T ( x , t ) = ∑c n e − an
2
t
sin nx
n =1

Using the third condition, we get

∞
T ( x , 0 ) = ∑ c n sin nx
n =1

which is a half-range Fourier-sine series and, therefore,

π
2 cos nx sin nx  
π π /2
2 
( ) + − (π – x )
π ∫0 ∫
=c T x , 0 sin =
nxdx  + 2  
π  0  n n π /2 

or
 nπ 
4sin  
cn =  2 
n 2π
Thus, the required solution is
2  nπ 
e − αn t
sin  
4 ∞
 2 
T ( x ,t ) = ∑
π n =1 n2 π

Example. A uniform rod of length L whose surface is thermally insulated

is initially at temperature
θ =θ 0 . At time t = 0 , one end is suddenly cooled
 320 Differential Equations: Theory and Applications

to θ = 0 and subsequently maintained at this temperature ; the other end

θ ( x ,t ) .
remains thermally insulated. Find the temperature distribution
Solution. The initial boundary value problems IBVPs of heat conduction
is given by
λθ ∂ 2θ
= a 0 ≤ x ≤ L ,t > 0
PDE: ∂t ∂ x2
θ ( 0= ,t ) 0 , t ≥ 0
BCs:
∂θ
(L= ,t ) 0 , t > 0
∂x
θ ( x , 0= ) θ0 , 0 ≤ x ≤ L
IC:
θ ( x , t ) = e − λ t ( A cos λ x B sin λ x )
2

Using the first boundary condition, we obtain A = 0. Thus the acceptable

solution is
2
θ = Br a λ t sin λ x
∂θ 2
= λ Be − aλ t cosλ x
∂x
Using the second boundary condition, we have
2
0 = λ Be aλ t
cos λ L
Implying cos λ L = 0. Therefore, the eigen value and the corresponding
eigen functions are

=λn
( 2n +=
1)π
n 0 ,1 , 2 ,…
2I .
Thus the acceptable solution is of the form
  2n + 1  2   2n + 1 
2

=θ B exp  − α   π t  sin  π x
  2L    2 L 
Using the principle of superposition, we obtain
∞   2n + 1  2 2   2n + 1 
θ ( x ,t )
= ∑ n  − a  2l  π t  sin  2 L π x 
B exp
n=0  
 Parabolic Differential Equations 321

Finally, using the initial condition , we have

∞
 2n + 1 
θ 0 = ∑ B n sin  πx
n=0  2L 
which is a half range Fourier-sine series and, this
2  2n + 1 
L
Bn =
L ∫θ
0
0 sin 
 2L
π x  dx


2  2L   2n + 1   
L

=  –θ 0 cos π x  
L  ( 2n +1 ) π   2 L   0 

4θ 0  ( 2n + 1 ) π  4θ 0
=
− [ cos   − cos 0 =
( 2n + 1 ) π  2  ( 2n + 1 ) π
Thus, the required temperature distribution is
∞
4θ 0   ( 2 n +1 ) 
2
 2 n +1
θ ( x ,t ) = ∑ exp  –α   π 2
t  sin  
πx
n = 0 ( 2n + 1 ) π   2L    2L 

Example. A conducting bat of uniform cross-section lies among the

x − axis with ends= at x 0= and x L . it kept initially at temperature 00
and its lateral surface is insulated. There are no heat sources in the bar. The
0
end x = 0 is kept at 0 , and heat is suddenly applied at the end x = L , so
q0
that there is a constant flux at x = L . find the temperature distribution
in the bar for t > 0 .
Solution. The given initial boundary value problem can be described as
follows:
∂T ∂2T
=α
PDE: ∂t ∂ x2
T ( 0=
,t ) 0 , t > 0
BCs:
∂T
(=
L , t ) q0 , t > 0
∂x
T(x=
,0 ) 0 , 0≤ x≤ L
IC:
 322 Differential Equations: Theory and Applications

Prior to applying heat suddenly to the end x = L , when t = 0 , the heat

flow in the bar is independent of time (steady state condition ) . Let
T (=
x , t ) T s ( x ) + T1 ( x , t )

T
where s is a steady part and
T1 is the transient part of the solution .
Therefore,
∂ 2 Ts
=0
∂ x2
whose general solution is
T=
s Ax + B

when
= x 0=
,T s 0 , implying B = 0 . Therefore,
T s = Ax
∂T s
= q0 . A = q0 . hence , the steady state
Using the other BC: ∂x w get
solution is
T s = q0 x
For the transient part, the BCs and IC are redefined as
( i ) T1 ( 0 , t ) = T ( 0 , t ) – T s ( 0 ) = 0 − 0 = 0
( ii ) ∂T1 ( L , t ) / ∂ x ∂T / ∂x – ∂T s ( L t ) / ∂x = q 0 − q 0 = 0
=
( iii ) T 1 ( x , 0 ) =T ( x , 0 ) − T s ( x ) =− q 0 x , 0 < x < L .

Thus, for the transient part, we have to solve the given PDE subject to
Eq. ( 48 ) , i. e.
these conditions. The acceptable solution is given by
T1 ( x , t ) e − aλ ( A cosλ x + B sin λ x )
2t
=

Applying the BC
( i ) , we get A = 0 . Therefore ,

T 1 ( x , t ) = Be – a λ
2
t
sin λ x )

Applying the BC
( i ) , we get A = 0 . Therefore ,
 Parabolic Differential Equations 323

T1 ( x , t ) = Be – a λ t sin λ x
2

And using the BC (ii), we obtain

∂T 1 2
= x=L λ – a λ t cos λ L 0
B=
∂x
π
λL
= ( 2n –1 ) =,n 1 , 2 , ,…
Implying 2 Using the superposition
principle , we have
∞   2n −1  2 
2
 2n −1 
T1 ( x , t ) = ∑B n exp  –α   π t  sin  πx
n =1   2L    2L 

Now, applying the IC

( iii ) , we obtain
∞
 2 n −1 
T1 ( x , 0 ) = ∑
− q0 x = B n sin  πx
n =1  2L 
 2m −1 
sin  π x
Multiplying both sides by  2 L  and integrating between 0
L
to and noting that
 0 , n≠ m
 2n −1   2m −1 
L

∫0 B n sin  2L  π x sin  2L π x  dx =  Bm L , n = m
 2

We get at once , after integrating by parts, the equation

4 L2   2m −1   L
− q0  sin    =
B
( 2m −1 ) π 2   2 L  
2
2
m

which gives
( −1 ) 8Lq 0
m

Bm =
( 2m −1 ) π 2
2

Hence, the required temperature distribution is

324 Differential Equations: Theory and Applications

8 Lq0 ∞  ( −1 )   ( 2m –1) 2   2m − 1  
m

T ( x, t ) =
q0 x + 2 ∑ exp  − a   π 2
t  sin  nπ  
π n =1  ( 2m − 1)2   L    2 L 
 
Example. The end A and B of a rod, 10 cm on length , are kept at
0 0
temperature 0 C and 100 C until the steady state condition prevails.
0
Suddenly the temperature at the end A is increased to 20 C , and the end
0
B is described to 60 C . find the temperature distribution in the rod at time
t.
Solution. The problem is described by
∂T ∂2 T
= α , 0 < x <10
PDE: ∂t ∂x 2
= T ( 0 , t ) 0=
, T ( 10 , t ) 100
BCs:
Prior to change in temperature at the ends of the rod, the heat flow in
the rod is independent of time as steady condition prevails. For study state,
d 2T
=0
d x2
whose solution is
T=
s Ax +B

when=x 0=
,T 0 implying B = 0 . Therefore,
T s = Ax

when= x 10= ,T 100 , implying A = 10 . Thus, the initial steady

temperature distribution in the rod is
T s ( x ) =10 x

Similarly, when the temperature at the ends Aand B are changed to

20 and 60, the final steady temperature in the rod is
T s ( x ) =10 x
 Parabolic Differential Equations 325

which will be attained after a long time, To get the temperature distribution
T ( x, t )
in the intermediate period. Counting time the moment the end
temperature were changed, we assume that
T ( x , t ) = ( 4 x + 20 ) + e − α λ ( B cos λ x + c sin λ x )
2
t

=
Using the BC: T 20
= when x 0 , we obtain
2
= 20 + Be − αλ
20 t

Implying B = 0 . Using the BC: T = 60 when x =10 , we get

nπ
sin10
= λ 0 , imlying
= λ =, n 1 , 2 ,…
10
The principal of superposition yields
∞   nπ  2   nπ 
T ( x , t ) = ( 4 x + 20 ) + ∑c n exp  − a   t  sin  x
n =1   10    10 

the IC: T 10=

Now using= x , whent 0 , we obtain

 nπ 
10 x= 4 x + 20 + ∑ cn sin  x
 10 
or
 nπ 
6 x – 20 = ∑ cn sin  x
 10 
where
10
2  nπ  1 n 800 200 
c n =∫ ( 6 x – 20 ) sin  −  ( −1 )
x  dx = −
10 0  10  5 nπ nπ 
Thus required solution is
1 ∞  n 800 200    nπ  2   nπ 
T ( x, t ) =
4 x + 20 – ∑ ( ) nπ nπ  −α  10  t  sin  10
−1 – exp x
5 n =1    

Example. Assuming the surface of the earth to be flat , which is initially

at zero temperature and for times t > 0 , the boundary surface is being
subject is being subject to a periodic heat flux
g 0 cos ω t . investigate the
326 Differential Equations: Theory and Applications

penetration of these temperature variations into the earth’s surface and show
that at a depth x, the temperature fluctuates and the amplitude of the steady
temperature is given by
g0 2α   ω  
exp  −  x 
2 ω   2a  

Solution. The given IBVP is described by

∂T ∂2 T
=α
PDE: ∂t ∂x (49)
∂T
− =g 0 cos ωt at x =0 , t > 0
BC: ∂x (50)
T ( x ,0 ) = 0
IC: (51)
Eq . ( 49 ) – ( 51 )
We shall introduce an auxiliary function T satisfying
and then define the complex function Z such that
Z= T + i T
We can easily verify that Z satisfies
∂Z ∂ 2Z
=α
PDE: ∂t ∂x 2 (52)
∂Z
= g 0 e ωt at=
x 0 ,t > 0
BC:- ∂x
IC: Z = 0 in the region. t = 0
Eq. ( 52 )
Let us assume the solution of in the form
Z = f ( x ) e iωt

f ( x)
where satisfies
d2 f (x ) ω
−i f ( x)=
0
dx 2
α (53)
df ( x )
− = g 0 at x = 0
dx (54)
 Parabolic Differential Equations 327

Also,
f ( x)
is finite for large x. ( 55 )
Eq . ( 53) , BC ( 55 ) ,
The solution of satisfying the is
  iω  
f ( x ) A exp  − 
= x 
  α  

BC ( 54 ) .
The constant A can be determined by using the Therefore ,
1 α   iω  
=f ( x) g0 exp  −   x
i ω   α  

Thus,
a 1   iω  
Z = g0 exp iωt –  x 
ω i   α  
(56)
It can be shown for convenience that
1+ i 1 1− i
=i = ,
2 i 2
q. ( 56 )
Thus, can be written in the form
g 0 2a  ω   ω 
Z= exp  − x  (1 – i ) exp i (ωt − x
2 ω  2a   2a 

 2a    ω   ω 
g 2a  − x  (1 − i ) cos  ωt – x  + i sin  ωt − x 
= 0 ω  2a  2a  
2 ω exp    
Its real part gives the fluctuation in temperature is
g0 2a  ω  ω  ω 
( x, t )
T= exp  − x  cos(ωt – x + sin  ω − x 
2 ω  2a   2a  2a  

g0 2a  ω   ω π 
 − x  cos  ωt − x − 
2 ω exp  2a   2a 4
328 Differential Equations: Theory and Applications

Hence, the amplitude of the steady temperature is given by the factor

g0 2a  ω 
exp  − x
2 ω  2a 

Example. Find the solution of the non-dimensional diffusion equation

satisfying the following BCs.
(i) T is bounded as t →∞
∂T
( ii ) =0 ,
∂x x=0 ,
for all t
∂T
( iii )
∂x x=0 = 0 , for all t
Solution. This is an example with insulated boundary conditions. It
can be seen that a physically acceptable general solution of the diffusion
equation is
T ( x , t ) exp α λ 2 t ( ) ( A cos λ x + B sin λ x )
Thus,
∂T
∂x
( )
=exp −αλ 2t (− Aλ sin λ x + Bλ cos λ x)
(57)
Using the BC (ii ) , Eq. ( 57 ) , gives B = 0 , Since we are looking for a
non-trivial solution, the use of BC ( iii ) into Eq. (57) at once gives
λα n=
sin λα = 0 implying= π , n 0 ,1 , 2 ,…
Using the principle of superposition, we get
∞   nπ  2   nπ 
T ( x, t ) = ( ∑
∑ An exp −αλ 2t cos λ x = An )
exp  −α   t  cos   x.
n=0   a    a 

The boundary condition (iv ) gives

  nπ  2 
∞
 nπ 
T ( x, 0 ) =
x (a – x) =
A0 + ∑ A n exp  − a   t  cos  x
n =1   a    a 

where
 Parabolic Differential Equations 329

2 a2
a
A0 =
a ∫(
0
)
ax − x 2 dx =
6

2 a2
a
An =
a ∫(
0
)
ax − x 2 dx =
6

2a2 2a 2 
( ) 1 + ( −1 ) 
n
= 1 + cos
= nπ 2 
n π
2 2
n π
2 
Therefore,
 4a 2
− , for n even
A n =  n 2π 2
 0 , for n odd

Hence, the required solution is
a 2 4a 2 ∞
1  nπ    nπ  2 
T ( x, t ) = − 2 ∑ cos   x exp  −α   t
6 π =n 2 , 4 ,… even n 2
 a    a  

Example. The boundaries of the rectangle 0 ≤ x ≤ a, 0 ≤ y ≤ b are

maintained at zero temperature. if at t = 0 the temperature T has the
F(x ,y) ,
prescribed value show that for t > 0 , the temperature at a point
within the rectangle is given
4 ∞ ∞
mπ x nπ y
=T ( x, y , t )
ab
∑∑ f ( m, n ) exp ( −aλ t ) sin
= 1=
m n 1
2
mn
a
sin
b
where
mπ x nπ y
a b
f ( m , n ) = ∫ ∫ f ( x , y ) sin sin dx dy
0 0
a b
and
 m2 n2 
λ
= π  2+ 2
2
mn
2

 a b 
Solution. The problem is to solve the diffusion equation described by
330 Differential Equations: Theory and Applications

∂T  ∂ 2T ∂ 2 T 
:
PDE = α 2 +  , 0 < x < a , 0 < y b ,t 0
∂t  ∂x ∂ y2 
T ( 0 , y=
,t ) T ( a , y =
, t ) 0 , 0 < y b ,t 0
BCs :
( x , 0 , t ) T=
T= ( x ,b ,t ) 0 , 0 < x a ,t 0

IC:
( x , y ,0 ) f ( x , y ) ,
T= 0 < x < a ,0 < y < b

Let the separable solution be

T = X ( x )Y ( y ) β ( t )

Substituting into PDE , we get

X Y '' 1 β '
''
+ = = −λ 2
X Y a β
Then
β +aλ 2 β =
'
0
X '' Y ''
− (λ 2 +
= − p 2 ( say )
=
X Y
Hence,
X + p2 X =
''
0
Y ''
− q 2 ( say )
−λ 2+ p 2 =
=
Y
Therefore ,
Y +q2Y =
''
0
Thus, the general solution of the given PDE is
T ( x, y , t ) =
( A cos px + B sin px )( c cos qy + D sin qy ) e ^ −a 2
t

where
λ=2 p 2 + q 2

Using the BC:

( 0 , y , t ) 0=
T= , we get A 0 .
Then, the solution is of
the form
 Parabolic Differential Equations 331

T ( x , y , t ) B sin px ( c cos qy + D sin qy ) e − aλ

2
=

Applying the BC:

( x , 0 , t ) 0=
T= , we get c 0
Thus , the solution is
given by
T ( x , y , t ) = BDsin px sin qye − aλ
2
t

T ( a , y ,t ) =0
Applying the BC : gives
sin pa = 0 , implying pa = nπ
or
nπ
=p , n 1 , 2 ,…
=
a
Using the principle of superposition the solution can be written in the
form
∞
 nπ 
( x , y ,t ) = ∑ A n
2
sin  x  sin qye − aλ t
n =1  a 
Using the solution is found to be
∞ ∞
 nπ   mπ  − aλ 2 t
T ( x , y , t ) = ∑ ∑ A mn sin  x  sin  e
m 1=
= n 1  a   b 
 m2 n2 
λ 2 = p2 +q2 = π 2  2
+ 2
 b a 
Finally, using the IC , we get
 nπ   mπ 
T ( x , y , 0 ) = f ( x , y ) = ∑ Amn sin  x  sin  y
 a   b 
which is a double Fourier series , where
2 2  mπ   nπ 
a b
Amn = . ∫ ∫ f ( x , y ) x  sin  y  dx dy
a a00  a   b 
Hence, the required general solution is
∞ ∞
 mπ   nπ 
T ( x, y , t ) = ∑ ∑ f ( m, n ) e − aλ 2 t
sin  x  sin  y  dxdy
= 1=
m n 1  a   b 
 332 Differential Equations: Theory and Applications

and
 m2 n2 
=λ2 π 2 2 + 2 
 b a 

6.5.1. Solution of Diffusion Equation in Cylindrical Coordi-

nates
Consider a three dimensional equation
∂T
= a∇ 2T
∂t

In cylindrical coordinates
( r ,θ , z ),
it becomes
1 ∂T ∂ 2 T 1 ∂T 1 ∂ 2 T ∂ 2 T
= + + +
a ∂t ∂ r 2 r ∂r r 2 ∂θ 2 ∂z 2 (58)
T = T ( r ,θ , z , t )
where .
Let us assume separation of variables in the form
T ( r ,θ , z , t ) = R ( r ) H ( θ ) Z ( z ) β ( t )

Substituting into Eq. (58) , it becomes

1 ' 1 β'
R′′ HZ β + R HZ β + 2 H '' RZ β + Z '' RH β = RHZ
r r α
or
R '' 1 R ' 1 H ′′ Z ′′ 1 β '
+ + 2 + = = −λ 2
R r R r H Z a β

where – λ is separation constant . Then

2

β ' + aλ 2 β =
0 (59)
R '' 1 R′ 1 H ′′ Z ′′
+ + 2 + λ2=
− − µ 2 ( say )
=
R R R r H Z

Thus, the equation determining Z , R and H become

Z '' − µ 2 Z =
0 (60)
 Parabolic Differential Equations 333

R′′ 1 R′ 1 H ''
+ + +λ 2 +µ 2=
0
R r R r2 H
or
R '' R' H '' 2
r2
R
+r
R
(
+ λ 2 + µ 2 r2 =
−
H
= )
v ( say )

Therefore ,
H +v2 H =
''
0 (61)
1 '  2 v2 
R′′ +
r
(
R +  λ + µ2 − 2  R =
r 
0 )
 (62)

Equation
( 59 ) − ( 61 ) have particular solution of the form
2
β = e − aλ t

=H c cos cθ + D sin vθ
=Z Ae µ z + Be – µ z
The differential equation ( 62 ) is called Bessel’s equation of order v and
its general solution is known as
R (=
r ) c1 jv ( )
λ 2 + µ 2 r + c 2 Yv ( λ2 +µ2 r )
jv ( r ) and Y v ( r )
where are Bessel functions of order v of the first and
Eq. ( 62 )
second kind, respectively. Of course, is singular when r = 0 . The
physically meaningful solutions must be twice continuously differentiable
Eq. ( 62 )
in 0 ≤ r ≤ a . Hence, has only one bounded solution, i.e.,
R( r ) j v
= ( λ2 +µ2 r )
Eq. ( 58 )
Finally, the general solution of is given by
e − aλ t  Ae µ z + Be − µ z  [C cos vθ + D sin vθ ] jv
T ( r ,θ , z , t ) =
2

( λ2 + µ2 r )
334 Differential Equations: Theory and Applications

T ( r ,t )
Example. Determine the temperature in the infinite cylinder
0 ≤ r ≤ a when the initial temperature is T ( r , 0 ) = f ( r ) , and the surface
r = a is maintained at 00 temperature.
Solution. The governing PDE from the data of the problem is
∂T
= a∇ 2T
∂t
where T is a function of r and t only. Therefore,
∂ 2 T 1 ∂T 1 ∂T
+ =
∂r 2 r ∂r a ∂t (63)
The corresponding boundary and initial coordination are given
T ( a ,t ) = 0
BC:
T ( r,0 ) = f ( r )
IC: (64)
Eq. ( 63 )
The general solution of is
T (= ( )
r , t ) A exp − a λ 2 t j 0 ( λ r )

Using the BC ( 64 ) , we obtain

j0 ( λa ) = 0

ξ a (=
n 1 , 2 ,…. , ∞ ) .
which has an infinite number of roots, n Thus , we get
from the superposition principle the equation
∞

( )
T ( r , t ) = ∑ An exp – a ξ n2 t j 0 ( ξ n r )
n =1

T ( r , 0 ) = f ( r ) , we get
Now using the IC:
∞
f ( r ) = ∑ A n j0 ( ξ n r )
n =1

A ,
To compute n we multiply both side of the above equation by
rj0 ( ξ m r )
and integrate with respect to r to get
 Parabolic Differential Equations 335

a ∞ a

∫r f ( r ) j 0 ( ξm r ) dr = ∑ An
0 n =1
∫ r j ( ξ r ) J ( ξ r ) dr
0=
0 m 0 n

 0 for n ≠ m

=  a2  2
 Am  2  J 1 (ξ m a ) for n = m
  
which gives
2
a
Am =
a 2 j12 ( ξ m a ) ∫uf ( u ) j ( ξ u ) du
0
0 m

Hence, the final solution of the problem is given by

2 ∞
j0 ( ξ m r ) a 
=T ( r ,t )
a2
∑ j 1 ( ξm a )
2
exp − αξ 2
m (
t )
 ∫uf ( u ) j0 ( ξ u ) du 
m =1 0 

6.5.2. Solution of Diffusion Equation In spherical Coordinates

In this section, we shall examine the solution of diffusion or heat conduction
equation in the spherical coordinates system. Let us consider the three –
T = T ( r ,θ ,φ , t ) .
dimensional diffusion Eq. ( 9 ) and let In the spherical
coordinates system Eq. ( 9 ) can be written as
∂ 2 T 2 ∂T 1 ∂  ∂T  1 ∂ 2T 1 ∂T
+ +  sin θ + =
∂ r 2 r ∂r r 2sin θ ∂θ  ∂θ  r 2 sin 2 θ ∂φ 2 α ∂t (65)
This equation is separated by assuming the temperature function T in
the form
T = R ( r ) H ( θ )φ ( φ ) β ( t )
(66)
Eq. ( 66 ) into Eq. ( 65 )
Substituting , we get
R′′ 2 R′ 1 1 d  dH  1 d 2φ 1 β ′
+ + 2  sin θ  + = = −λ 2 ( say )
R r R r sin θ H dθ  dθ  φ r 2 sin 2 θ dφ 2 a β

where λ is a separation constant. Thus,

2

d β + λ 2 αβ =
0
336 Differential Equations: Theory and Applications

whose solution is
2
β = c1 e aλ t
(67)
Also,

.
ich gives

whose solution is

(68)
Now, the other separated equation is
1  d 2 R 2 dR  1 d  dH  m2
 sin θ  +λ =
2
 + +
R  dr 2 r dr  Hr 2sin θ dθ  dθ  r 2 sin 2 θ
or
r2  2  m2 1 d  dH 
′′ ′
 R + R  + λ r =2 –
2 2
 sin θ 
R r  sin θ H sin θ dθ  dθ 
On re-arrangement, this equation can be written as
2  n ( n +1 ) 
R′′ + R′ +  λ 2 – 2
0
R=
r  r  (69)
and
1  d2H dH  m2
−  sin θ + cos θ  + n ( n +1 )
=
H sin θ  dθ 2 dθ  sin 2 θ
or
d 2H dH  m2 
+ cot θ +  n ( n +1 ) − 2 0
H =
dθ dθ  sin θ  (70)
Let then Eq. (69 ) becomes
 Parabolic Differential Equations 337

   1 
2

   n+   
ψ ' ' ( r ) + 1 ψ  2  2 
1
( λr )
−
( r ) + λ – =
ψ
'
2 0
 r r 2

   
   

Since
(λr ) ≠ 0 , we have
  1 
  n+ Z2 
1  2 
ψ ′′ ( r ) + ψ ′ ( r ) +  λ 2 –  ψ ( r ) =
0
r  r2 
 

 1
 n+  ,
which is Bessel’s differential equation of order  2  whose solution is
ψ ( r ) Aj 1 ( λ r ) + BY n+ 1 ( λ r )
n+
2 2

Therefore
1  
R( r ) ( λr )  Ajn + 1 ( λ r ) + BY n + 1 ( λ r ) 
−
2

 2 2  (71)

where
jn and Yn are Bessel functions of first and second kind, respectively .
Eq. ( 70 )
Now, can be put in a more convenient from by introducing a new
independent variable , µ = cos θ .
So that
cot θ = µ / 1 − µ
2

dH dH
= − 1− µ 2
dθ dµ

d 2H 2
2 d H dH
dθ 2
1
= − µ
dµ 2(−µ
dµ
)
Eq. ( 70 )
Thus, becomes
 338 Differential Equations: Theory and Applications

d 2H dH  
m2
(1− µ 2 ) dµ 2
− 2 µ +
dµ 
 (
n n + 1) − H=
1− µ 2
0
 (72)
which is an associated Legendre differential equation whose solution is
H (θ ) A′ pnm ( µ ) + B ' Qnm ( µ )
=
(73)
where p mn ( µ ) and Q mn ( µ )
are associated Legendre functions of degree n
and of order m , of the and second kind, respectively. Hence the physically
meaningful general solution of the diffusion equation in spherical geometry
is of the form
. 1
T ( r ,θ ,φ , t ) = ∑ Aλ mn ( λ r ) ( λ r ) pnm (cos θ )e± imφ − aλ t
– 2
2 j 1
n+
λ ,m ,n 2 (74)
1
(λr ) ( λr )
–
2 Y
Q mn ( µ ) n+
1
In this general solution, the functions and 2

are exclude because these functions have poles at µ = ±1 and r = 0

respectively.
Example. Find the temperature in a sphere of radius a. when its surface
f ( r ,θ ) .
is kept at zero temperature and its initial temperature is
Solution. Here, the temperature is governed by three-dimensional heat
equation in spherical polar coordinates independent of . Therefore, the task
is to find the solution of PDE
1 ∂T ∂ 2T 2 ∂T 1 ∂  ∂T 
= + + sin θ
α ∂t ∂r 2 r ∂r r 2 sin θ ∂θ  
∂θ  (75)

Subject to
T ( a ,θ , t ) = 0
BC:
T ( r ,θ , 0 ) = f ( r ,θ )
IC: (76)
Eq. ( 75 ) Eq. ( 74 ) ,
The general solution of with the help of can be
written as
. 1
T ( r ,θ , t ) = ∑ Aλ n ( λ r ) ( λ r ) pn ( cos θ ) e− a λ
– 2
t
2 j 1
n+
λ ,n 2 (78)
 Parabolic Differential Equations 339

BC ( 76 ) ,
Applying the we get
j
n+
1 ( λa ) = 0
2

This equation has infinitely many positive roots. Denoting them by

ξi ,
we have
∞ ∞ 1
T ( r ,θ , t ) ∑∑Ani (ξi r ) (ξi r ) Pn ( cos θ ) exp ( −αξt 2t )
−
2 J 1
n+
n 0=
= n 1 2 (79)
Now applying the IC and denoting cos θ by µ , we get
∞ ∞

( )
f r , cos −1 ( µ ) = ∑∑ Ani (ξ1r )
−1/2
J
n+
1 (ξi r ) pn ( µ ) .
n 0=i 1
= 2

pn ( µ ) d µ
Multiplying both side by and integrating between the limits,
−1to1, we obtain
1 ∞ ∞ 1

( −1
)
∫ f r , cos ( µ ) pm ( µ ) d µ = ∑∑Ani (ξi r )
−1/2
jn +1/2 (ξi ) ∫ pm ( µ ) pn ( µ ) d µ
−1 n 0=i 1
= −1

∞ ∞ 1
2 
= ∑∑Ani (ξi r ) ( ξi r ) 
−
2 j 1 
n 0=i 1
=
n+
2  2n + 1 
or
1
 2n + 1  ∞ 1


 2  −1
 ∫ pn ( µ ) f r , cos ( =
−2
(
µ) dµ ) ∑A (ξ r )
ni i
2 j
n+
1 (ξi r=
) for n 0 ,1, 2,3, …
i =1 2

Ani ,
Now, to evaluate the constants we multiply both sides of the above
3
r j 2
n+
1 (ξ r )
j
equation by and integrating with respect to r between limits
2

0 to a and use the orthogonality property of Bessel functions to get

1 a 3
 2n + 1  2 1 
ξi   ∫r jn + 1 (ξ j r ) dr  ∫ pn ( µ ) f r , cos ( µ ) d µ  ( )
2 −1

 2 0 2 − 1 
 340 Differential Equations: Theory and Applications

∞ a
= ∑Ani ∫rj
n+
1 ( ξ i r ) j n + 1 ( ξ j r ) dr
i =1 0 2 2

2
1 ∞  
= ∑ Ani  j ' 1 ( ξi=
r )  , n 0 ,1 , 2 , 3 ,…
2 i =1  n+ 2  (80)
Thus, Eq. (79 ) and ( 80 ) together constitute the solution for the given
problem.

6.6. MAXIMUM-MINIMUM PRINCIPLE AND

CONSEQUENCES
Theorem. (Maximum- minimum principle ) .
u ( x ,t )
Let be a continuous function and a solution of
ut = α u xx (81)

for 0 ≤ x ≤ l , 0 ≤ t ≤ T , whereT > 0 is a fixed time. Then the maximum and

minimum values u are attained either at time t = 0 or at the end points
x = 0 and x = l at some time in the interval 0 ≤ t ≤ T .
Proof. To start with, let us assume that the assertion is false . Let the
u ( x, t ) t 0 ( 0 ≤ x < l ) or for x= 0 or x= l ( 0 ≤ t ≤ T )
maximum value of for =
u ( x ,t )
be denoted by M . we shall assume that the function attains
its maximum at some interior point
( x0 , t0 ) in the rectangle defined by
0 ≤ x 0 ≤ l , 0 ≤ t0 ≤ T , and them arrive at a contradiction. This mean that

u ( x0 , t 0 =
) M +ε (82)

Now, we shall compare the signs in Eq. (81) at point

( x0 , t0 ) ,
it is well
u ( x, t )
known from calculus that the necessary condition for the function
(x , t )
to posses maximum at 0 0 is.
∂u ∂2u
( 0 0)
x , t = 0 , ( x0 , t0 ) ≤ 0
∂x ∂ x2 (83)
 Parabolic Differential Equations 341

u ( x0 , t0 ) t = t0 ,
In addition, attains maximum for implying
∂u
( x0 , t 0 ) ≥ 0
∂t (84)
Thus, with the help of Eq. (83) and (84) we observe that the signs on the
left - and right - hand sides of Eq. (81) are different. However, we cannot
claim that we have reached contradiction, since the left- and right hand sides
can simultaneously by zero. To complete the proof, let us consider another
∂2 u ∂u
( xl , tl ) 2
≤ 0 and > 0.
point at which ∂x ∂t
( x , t ) u ( x , t ) + λ ( t0 – t )
v=
(85)
v ( x 0 ,t 0 ) =
M + ε and λ ( t 0 – t ) ≤ λT .
Where λ is a constant. Obviously ,
ε
λ< ;
Suppose we choose λ > 0 , such that 2T then the maximum of
ε
v ( x ,t ) . M+
for t = 0 or for=x 0= , x l cannot exceed the value 2 But
v ( x ,t ) ( x ,t )
is a continuous function and , therefore ,a point 1 1 exists at
which it assumes its maximum. It implies
ε
M+ ≤ v ( x1 , t1 ) ≥ v ( x 0 , t 0 ) = M + ε
2
This pair of inequalities is inconsistent and therefore contradicts the
assumption that v takes on its maximum at
( x 0 , t 0 ) . Therefore, the
assertion that u attains its maximum either at t = 0 or at end points is true .
u ( x , t ) . if u
We can establish a similar results for minimum values of
satisfies Eq. ( 81 ) , −u also satisfies Eq. ( 81 ) . Hence , both maximum
and minimum values are attained either initially or at the end points. Thus
the proof is complete. We Shall give some of the consequences of the
maximum-minimum principle in the following theorems We Shall give some
of the consequences of the maximum-minimum principle in the following
theorems.
 342 Differential Equations: Theory and Applications

Theorem. (Uniqueness theorem). Given a rectangular region defined

by 0 ≤ x ≤ l , 0 ≤ t ≤ T , and a continuous function
( u , x , t ) defined on the
boundary of the rectangle satisfying the heat equation
u t = α u xx
This equation possess one and only one solution satisfying the initial
and boundary conditions
u ( x ,0 ) = f ( x )

u ( 0 , t ) g=
= 1 ( t ) ,u ( l ,t ) g2 ( t )

f ( x ) , g1 ( t ) , g 2 ( t )
where are continuous on their domains of definition.
u ( x ,t ) ,u 2 ( x , t )
Proof. Suppose there are two solution 1 satisfying
the heat equation as well as the same initial and boundary conditions. Now
let us consider the difference
v ( x , t ) u2 ( x , t ) − u 1 ( x , t )
=

It is also a solution of the heat conduction equationfor 0 ≤ x ≤ l , 0 ≤ t ≤ T and is

v ( x , t=) 0 , 0 ≤ x ≤ l and v ( l , t =
) 0 ,0 ≤ t ≤T .
continuous in x and t . Also ,
v ( x ,t )
Hence . satisfies the condition required for the application of
v ( x ,t ) = 0
maximum –minimum principle .Thus , in the rectangular region
u ( x ,t ) = u 2 ( x , t ) .
defined by 0 ≤ x ≤ l , 0 ≤ t ≤ T . it follows therefore that 1
Another important consequence of the maximum-minimum principle is
the stability property which stated in the following theorem without proof .
u ( x ,t )
Theorem. (stability) theorem . The Solution of the Dirichlet
problem
=u t α u xx , 0 ≤ x ≤ l , 0 ≤ t ≤ T

u (=
x ,0 ) f ( x ) , 0 ≤ x ≤ l

u (=
0 , t ) g ( t ) , u (=
l ,t ) h ( t ) , 0 ≤ t ≤T

depend continuously on the initial and boundary conditions.

 Parabolic Differential Equations 343

6.7. MISCELLANEOUS EXAMPLE.

Example. A homogenous solid sphere of radius R has the initial temperature
f ( r ) ,0 ≤ r ≤ R .
distribution where r is the distance measured form
0
the centre. The surface temperature is maintained at 0 . Show that the
T ( r ,t )
temperature in the sphere is the solution of
 2 
=T1 c 2  Trr + Tr 
 r 
2
where c is a constant. Show also that the temperature in the sphere for
t > 0 is given by
1 ∞
 nπ  cnπ
T ( r ,t )
= ∑B n sin  − λ 2n t , ) λn
r  exp ( =
r n =1  R  R
Solution. The temperature distribution in a solid sphere is governed by
the parabolic heat equation
T=
t c 2 ∇ 2T

From the data, T is a function of r and t alone. In view of the symmetry

of the sphere the above equation with the help of Eq. reduce to
 2 
=Tt c 2  Trr + T r 
 r  (86 )
Setting v = rT , the given BC gives
( R , t ) rt=
v= ( R ,t ) 0
which the IC gives
v ( r , 0 ) rT
= = ( r , 0 ) rf ( r )
Since T must be bounded at r = 0 , we require
v ( 0 ,t ) = 0

Now,
344 Differential Equations: Theory and Applications

Similarly, finding
Trr and substituting into Eq. (86), we obtain

vt = c 2 v rr

v ( r , t ) = R ( r )τ ( t )
Using the variables separable method, we may write
and get
R ( r ) A cos kr + B sin kr
=

( t ) exp ( − c 2 k 2 t )
τ=

Thus, using the principle of superposition , we get

∞
v ( r , t ) =∑( An cos kr + Bn sin kr ) r =0 =0
n =1

v ( 0 ,t ) = 0 ,
Also, using we have
( An cos kr + Bn sin kr ) r =0 0
=

Implying
An = 0. Also, v ( R , t ) = 0 gives Bn sin kr = 0 , implying
sin kr = 0 , as Bn ≠ 0 . Therefore,
nπ
kR nπ =
= , k =, n 1 , 2 ,…
R
Thus, the possible solution is
∞
 nπ   c 2 n 2π 2 t 
v ( r , t ) = ∑ Bn sin  r  exp  – 
n =1  R   R2 

v ( r , 0 ) = rf ( r ) ,
Finally, applying the IC: we get
∞
 nπ 
rf ( r ) = ∑ Bn sin  r
n =1  R 
which is a half-range Fourier series . Therefore,
2  nπ 
R
Bn =
R ∫ rf ( r ) sin 
0
r  dr
R 
 Parabolic Differential Equations 345

v ( r , t ) = rT ( r , t ) .
But Hence, the temperature in the sphere is given
by
1 ∞
 nπ   c2 n2 π 2 t 
T ( r ,t ) ∑ Bn sin  r  exp −  2 
r
n =1  R   R 
Example. A circular cylinder of radius a has its surface kept at a
T.
constant temperature 0 If the initial temperature is zero throughout the
cylinder , prove that for t > 0 .
 2 ∞
j 0 (ξ n a ) 
T ( r , t ) T0 1 – ∑ξ j1 ( ξ n a )
( )
exp − ξ n2 kt 
 a n =1 n 

where
± ξ1 , ± ξ 2 ,…., ± ξ n are the roots of j0 ( ξ a ) = 0 , and k is the thermal
conductivity which is a constant.
Solution. It is evident that T is a function of r and t alone and ,
therefore , the PDE to be solved is
∂ 2 T 1 ∂T 1 ∂T
+ =
∂ r 2 r ∂r k ∂t (87)
Subject to
IC : T ( r , 0=
) 0 , 0≤ r < a
BC : T ( a=
, t ) T0 , t ≥ 0

Let
T ( r ,t =
) T0 + T1 ( r , t )
So that
T1 ( r , 0 ) = − T0
(88)
T1 ( a , t ) = 0
(89)

where
T1 is the solution of Eq. (87). By the variables separable method we
have ,
T1 ( r , t ) Aj0 ( λ r ) exp − λ 2 kt
= ( )
 346 Differential Equations: Theory and Applications

T1 ( a , t ) = 0 ,
Using the BC: we get
Aj0 ( λ a ) exp − λ 2 kt =
0 ( )
j0 ( λ a ) =
0 as A ≠ 0 .let ξ1 ,ξ 2 ,…., ξ n ,
which gives be the roots of
j0 ( λ a ) = 0 .
Then the possible solution using the superposition principle is
∞
=T1 ( r , t ) ∑A j (ξ r ) exp ( − ξ
n 0 n
2
n kt )
n =1 (90)
IC : T1 ( r , 0 ) = − T0
Using the into Eq. ( 90 ) , we obtain
∞

∑A j ( ξ r ) =
n =1
n 0 n − T0

rj0 ( ξ m r )
Multiply both sides by and integrating, we get
a ∞ a
− T0 ∫ rj 0 ( ξ m r ) dr =
∑An ∫ r j0 ( ξm r ) j0 ( ξn r ) dr
0 n =1 0

m ∫ rj0 ( ξ m r ) if m n ; otherwise 0
2
A=
0

a2 2
= Am j 1 ( ξm a )
2
But,
a ξm a
x dx
− T0 ∫rj0 ( ξ m r ) dr =
− T0 ∫ j 0( x ) ( x=
ξm r )
0 0
ξm ξm
ξm a
T0 d
= − ∫  xj1 ( x )  dx
ξ 2
m 0
dx 
T aT
− 02  xj1 ( x )  ξ0m a =
= − 0 j1 ( ξ m a )
ξ m ξm
Therefore,
 Parabolic Differential Equations 347

a2 2 aT 0
Am J 1 ( ξm a ) = − j (ξ a )
2 ξm 1 m
or
2T0 1
An = −
nξ n j1 (ξ n a )

Hence, Eq. (90) becomes

2 ∞ j0 ( ξ n r ) exp −ξ n kt
T1 ( r , t ) = − T0 ∑
2
( )
a n =1 j1 (ξ n a ) ξn
Finally, the complete solution is found to be

T (r=
 2
, t ) T0 1 −
∞ 2
(
j0 ( ξ n r ) exp –ξ n kt  )
 a
∑
n =1 j1 ( ξ n a ) ξn



Example. Determine the temperature in a sphere of radius a, when

its surface is maintained at zero temperature while its initial temperature if
f ( r ,θ )
.
Solution. Here the temperature is governed by the three-dimensional
heat equation in polar coordinates independent of φ , which is given by
1 ∂u ∂ 2 u 2 ∂u 1 ∂  ∂u 
= 2+ + 2  sinθ 
k ∂t ∂r r ∂r r sin θ ∂θ  ∂θ  (91)
Let
u ( r ,θ , t ) = R ( r ) H ( θ ) T ( t )

By the variables separable method, the general solution of Eq. (91) is

found to be
. . 1
v ( r ,θ , t ) ∑∑Aλn ( λ r ) ( λ r ) pn ( cos θ ) exp ( −k λ 2t )
−
2 j 1
n+
λ n 2 (92)
In the present problem, the boundary and initial conditions are
BC : u ( a ,θ , t ) = 0
(93)
348 Differential Equations: Theory and Applications

IC : u ( r ,θ , 0 ) = f ( r ,θ )
(94)
Substituting the BC (93) into Eq. (92), we get
j
n+
1 ( λa ) = 0
2 (95)

Let 1 2 ξ a , ξ a,… . ,ξ a ,…
1 be the roots of Eq. (95). Then the general
solution can be put in the form
∞ ∞ 1
u ( r ,θ , t ) ∑∑A (ξ r ) (ξl r ) pn ( cos θ ) exp ( − kξ t2t )
–
ni l
2 j 1
n+
n 1=
= i 1 2 (96)
Now using the IC, we obtain
∞ ∞ 1
−
f ( r ,θ ) = ∑∑ Ani ( ξ i r ) 2
J 1 ( ξi r ) pn ( cosθ )
n+
n 1=
= i 1 2

pn ( cos θ ) d ( cos θ )
Multiplying both sides by and integrating, we have
1 ∞ ∞ 1 1

∫ pn ( cosθ ) f ( r ,θ ) d ( cos θ ) = ∑ ∑ Ani ( ξi r ) ( ξi r ) ∫ p n2 ( cosθ ) d ( cos θ )

−
2 j 1
n+
−1 n 1=
= i 1 2 −1

Using the orthogonality property of Legendre polynomials, we get

1 ∞ ∞ 1
2 
pn ( cosθ ) f ( r ,θ ) d ( cosθ ) = ∑∑ A ni (ξi r ) ( ξi r ) 
−
∫
−1 n= 1=
i 1
2 j
n+
1
2

 2n + 1 
Rearranging and multiplying both sides of the above equation by
r J 1 (ξi r )
3/2
n+
2 and integrating between the limits 0 to a with respect to
r , we get

1
2n + 1 32 1
a a
−
1 ( i ) n( ) ( )
2 ∫0 ∫
ξ θ θ θ ni ∫ ξ
2
r j r dr p cos f r , d (cos ) = A rj 1 ( i
2
dr
n+ n+
2 −1 0 2

2
a2  
= Ani  j ' 1 (ξi a ) 
2  n+ 2 
Therefore,
 Parabolic Differential Equations 349

( 2n + 1) ξi1/2 a 1
Ani =
  0
2 ∫ J n+1/2 (ξi r ) dr ∫ pn ( cos θ ) f ( r ,θ ) d ( cos θ )
−1
a  j ' 1 (ξi a ) 
2

 n+ 2  (97)
Hence, we obtain the solution to the given problem for Eq. (96), where
Ani is given by Eq. (97).
Example. The heat conduction in a thin round insulated rod with heat
sources present is described by the PDE

ut − au xx =
F
( x, t ) , 0
ρc 0 < x l,t 0
(98)
Subject to ‘
( 0, t ) u=
BCs : u= (l, t ) 0
IC : u ( =
x, 0 ) f ( x ) , 0 ≤ x ≤ l
(99)
where ρ and c are constant and F is a continuous function of x and t. Find
u ( x, t )
.
Solution. It can be noted that the boundary conditions are of homogeneous
type. Let us consider the homogeneous equation
ut − au xx = 0 (100)
u ( x, t ) = X ( x ) T ( t ) , we get
Setting
T ′ X ′′
= = −λ 2 ( say )
αT X (101)
which gives X ′′ + λ X =
2
0. The corresponding BCs are
( 0 ) X=
X= (l ) 0
The solution of Eq. (101) gives the desired eigen functions and eigen
values, which are
2
 nπ 
=X n ( x ) sin
= λn x, λn2   , n ≥ 1
 l  (102)
For the non-homogenous problem (98), let us propose a solution of the
form
 350 Differential Equations: Theory and Applications

∞
u ( x, t ) = ∑ Tn ( t ) X n ( x )
n =l (103)
It is clear that Eq. (103) satisfies the BCs (99).From the orthogonality of
eigen functions, it follows that
2
l
Tm ( t ) = u ( x, t ) X m ( x ) dx
l ∫0
However,
1
2  mπ 
Tm ( 0 ) = ∫ f ( x ) sin  x  dx
l 0  l  (104)
Tm ( t ) .
which is an IC for introducing Eq. (103) into the governing equation
(98) . we get
∞ ∞
F ( x, t )
=
∑ T
n 1=
n
'
X n − a ∑
n 1
Tn X n'' =
ρ c (105)

Now, we shall expand F x, t ) / ρ c, so that it is represented by a

0 < x l,t 0
convergent series on in the form
∞
F
= ∑ qn ( t ) X n ( x )
ρ c n =1 (106)
where
2 F ( x, t )  nπ
1

qn ( t ) =
l ∫0 ρ c
sin  x  dx
 l  (107)
qn ( t )
Thus, is known. Now, Eq. (105), with the help of Eq.(101)
becomes
∞

∑X ( T
n =1
n n
'
+ λn2 α Tn − qn =
0 )
Therefore, it follows that
Tn' ( t ) + λn2 α Tn ( t ) =
qn ( t )
(108)
Its solution with the help of IC (104) is
 Parabolic Differential Equations 351

( )
( t ) Tn ( 0 ) exp −λn2α t + ∫ exp λn2 α (τ − t ) qn (τ ) dτ
Tn =
0 (109)
From Eq. (103) and (109) , the complete solution is found to be
∞  1

u ( x, t )
= ∑ T
 n ( 0 ) exp − λn
2
α (
t + ∫ )
exp[ λn2α (τ − t ) qn (τ ) dτ  X n ( x )
n =1  0 
In the expanded form, it becomes
∞   2 l 
u ( x, t ) ∑   ∫ f (ξ ) X n (ξ ) d ξ  exp(−λn at )
2

n =1  l
  0 

} ∫ (ρ c ) X (ξ ) dξ dτ  X ( x )
2
l l
F ξ ,τ 
l 0
{
+ ∫ exp −λn2 a (τ − t ) n n
0  (110)
It can be verified that the series in Eq. (110) converges uniformly for
t > 0. By changing the order of integration and summation in Eq.(110) , we
get
l
u ( x, t ) = ∫  ∑
( )
 ∞ exp −λn2 at X n ( x ) X n (ξ ) 
 f (ξ ) d ξ
0 
1/ 2 
 n =1

11
+ ∫∫  ∑
{ }
 ∞ exp − λn2 a ( l − t ) X n ( x ) X n (ξ )  F (ξ , t )
 d ξ dt
00 
l/2  ρc
 n =1

which can also be written in the form
1 11
F ( ξ ,τ )
u ( x, t )
= ∫G ( x,ξ ; t ) f (ξ ) dξ + ∫∫G ( x, ξ ; t − τ ) d ξ dτ
ρc
0 00 (111)
where

G ( x ,ξ ; t ) = ∑
∞
( )
exp −λn2 at X n ( x ) X n (ξ )
n =1 l/2
is called Green’s function. More details on Green’s function are given .
352 Differential Equations: Theory and Applications

6.7.1. Summary and Discussion

The diffusion phenomena such as conduction of heat in solids and diffusion
of vorticity in the case of viscous flow past a body are governed by a partial
differential equation of parabolic type. For example, the flow of heat in a
conducting medium is governed by the parabolic equation
∂T
ρ  = div ( K ∇T ) + H ( r.T , t )
∂t (1)
where ρ is the density,  is the specific heat of the solid, T is the temperature
at a point with position vector r. K is the thermal conductivity , t is the time
H ( r,T , t )
,and is the amount of heat generated per unit time in the element
dV situated at a point ( x, y, z ) whose position vector is r. This equation is
known as diffusion equation or heat equation. We shall now derive the heat
equation form the basic concepts.
Then the differential equation of heat conduction with heat source is
1 ∂T ( r , ) H ( r,T , t )
∇ 2T ( r , t ) +
=
α ∂t K (2)
In the absence of heat sources, Eq. (8) reduces to
∂T ( r , t )
= α∇ 2T ( r , t )
∂t (3)
This is called Fourier heat conduction equation or diffusion equation.
The fundamental problem of heat conduction is to obtain the solution of Eq.
(2) subject to the initial and boundary conditions which are called initial
boundary value problems, hereafter referred to as IBVPs.

6.8. BOUNDARY CONDITIONS

The heat conduction equation may have numerous solution unless a set of
initial and boundary conditions are specified. The boundary conditions are
mainly of three types which we now briefly explain.
Boundary Condition 1: The temperature is prescribed all over the
T ( r ,t )
boundary surface. That is, the temperature is a function of both
T = G ( r ,t )
position and time. In other words, which is some prescribed
function on the boundary .This type of boundary condition is called the
 Parabolic Differential Equations 353

Dirichlet condition. Specification of boundary conditions depends on the

problem under investigation.
T ( r ,t ) = 0
A special case includes on the surface of the boundary,
which is called a homogeneous boundary condition.
Boundary Condition II : The flux of heat , i.e., the normal derivative of
∂T
,
the temperature ∂n is prescribed on the surface of the boundary . it may
be a function of both position and time, i.e.,
∂T
= f ( r ,t )
∂n
This is called the Neumann condition .Sometimes, the normal derivatives
of temperature may be a function of position only or a function of time only.
A special case includes
∂T
= 0 onthe
∂n boundary
This homogenous boundary condition is also called insulated boundary
condition which sates the heat flow is zero.
Boundary condition III. A linear combination of the temperature and its
normal derivatives is prescribed on the boundary , i.e.,
∂T
K G ( r ,t )
+ hT =
∂n
Where K and h are constants. This type of boundary condition is called
Robin’s condition. it means that the boundary surface dissipates heat by
convection .Following Newton’s law of the cooling, which states that the rate
at which heat is transferred from the body to the surrounding is Proportional
to the difference in temperature between the body and the surrounding , we
have
∂T
−K =h ( T − Ta )
∂n
As a special case, we may also
∂T
K 0
+ hT =
∂n
 354 Differential Equations: Theory and Applications

which is homogeneous boundary . This mean that hear is converted by

dissipation from the boundary surface into a surrounding maintained at zero
temperature.
The other boundary conditions such as the heat transfer due to radiation
obeying the fourth power temperature law and those associated with change
of phase .like melting, ablation, etc., give rise to non-linear boundary
conditions.
δ (t ) Eq. ( 33 ) ( 34 ) is known as
This limiting function defined by and
Dirac delta function or the unit impulse function. Its profile is depicted in .
Dirac originally called it an improper function as there is no proper function
with these properties . In fact, we can observed that
∞ .
=1 ∫ δ ( t ) dt LTε → 0
= ∫ ε ( t ) dt
δ= =
LT ε →0 0 0
−∞ t >ε

δ (t )
Obviously , this contradiction implies that cannot be a function
in the ordinary sense. Some important properties of Dirac delta function are
presented now:
∞

∫ δ ( t ) dt =1
Property 1: −∞

∞
f (t ) , ∫ f ( t ) δ ( t ) dt = f ( 0 )
Property II: For any continuous functions −∞

.
f (t )
Property III: let be any continuous function . Then
∞

∫ δ ( t – a ) f ( t ) dt = f ( a )
−∞ .
δ ( −t ) =
δ (t )
Property IV:
1
= δ ( at ) δ ( t ) , a >0
Property V: a
 Parabolic Differential Equations 355

δ (t )
Property VI: if is a continuously differentiable
. Dirac delta function Vanishing for large t, then
∞ ∞
∞
f ( t ) δ ' ( t ) dt  f ( t ) δ ( t )  − ∞ −
∫= ∫ f ( t ) δ ( t ) dt
'

−∞ −∞

∫ f ( t ) δ ( t ) dt = − f ( 0 )
' '

Since δ t → 0 as t →∞ , − ∞ , we have −∞

∫ δ ( t − a ) f ( t ) dt =
− f (a)
' '

operty VII: −∞

Having discussed the one-dimensional Dirac delta function , we

can extend the definition to two dimension, Thus, for every f which is
continuous over the region S containing the point
( ξ ,η ) , we define
δ ( x –ξ , y –η )
in such a way that
.

∬δ ( x –ξ y –η ) f ( x , y ) dσ = f ( ξ ,η )
S

δ ( x − ξ , y −η )
Note that . is a forma limit of a sequence of ordinary
functions, i.e.,
δ ( x − ξ , y −η ) =
LTε → 0 δ ε ( r ) r 2 = ( x − ξ ) + ( y −η ) .
2 2

,Where
 CHAPTER

7
LAPLACE TRANSFORM
METHODS

CONTENTS
7.1. Introduction..................................................................................... 358
7.2. Transform of Some Elementary Functions......................................... 362
7.3. Properties of Laplace Transform....................................................... 364
7.4. Transform of A Periodic Function..................................................... 370
7.5. Transform of Error Function.............................................................. 372
7.6. Transform of Bessel’s Function......................................................... 374
7.7. Transform of Dirac Delta Function................................................... 376
7.8. Convolution Theorem (Faltung Theorem).......................................... 382
7.9. Convolution Theorem ( Faltung Theorem)......................................... 387
358 Differential Equations: Theory and Applications

7.1. INTRODUCTION
Laplace transform is essentially a mathematical tool which can be used to
solve several problems in science and engineering. This transform was first
introduced was first introduced by Laplace, a French mathematician, in the
year 1790 in his work on probability theorem. This technique became popular
when Heaviside applied to the solution of an ordinary differential equation
referred hereafter as ODE. representing a problem in electrical engineering.
To the basic question as to why one should learn Laplace transform technique
when other technique available. The answer is very simple. Transform are
used to accomplish the solution of certain problems with less effort and
in a simple routine way. To illustrate, consider the problem of finding the
value of x from the equation simple routine way. To illustrate, consider the
problem of finding the value of x from the equation
x 1 . 85 = 3 (1)
It is an extremely tedious task to solve this problem algebraically.
However, taking logarithms on both side, we have the transformed equation
as
1.85 In x = In 3 (2)
In this transformed equation, the algebraic operation and exponentiation
have been changed to multiplication which immediately gives
In 3
x=
1.85
In
To get the required result, it is enough if we take the antilogarithm on
both side of the above equation, which yields
 In 3 
x = In −1  
 1.85 
with the help of any ordinary calculator , we can now compute x. Following
this simple example, the Laplace transform method reduce the solution of
an ODE to the solution of an algebraic equation. In fact, this method has a
particular advantage in finding the solution of an ODE with appropriate. ICs,
without first finding the general solution and then using ICs for evaluating
the arbitrary constants. Also, when the Laplace transform technique is
applied to a PDE, it reduce the number of independent variables by one.
 Laplace Transform Methods 359

f (t )
Definition. Suppose is a piecewise continuous function and if it
has an additional property that there exists a real numbers
ã 0 and a finite
positive number M such that
Ltt →∞ f ( t ) e −γ t ≤ M for γ > γ 0

And the limit does not exist when

γ < γ 0 , then such a function is said to

be of exponential order
γ 0 . Also written as

(
f ( t ) = 0 e γ 0t )
Variables such as velocity and current are always finite. Which means that
f (t ) f ( t ) , | f ( ( t ) e −γ t → 0
is bounded . Thus for any bounded function
for all γ > 0 . The order of such a function in zero. However , variables such
as electrical charge and mechanical displacement may increase without limit
but of course proportional to t. Such functions are also of exponential order.
For illustration, let us consider the following examples:
(i ) Ltt →∞ te − γ t = 0
n
The fact that t is of exponential order zero can be seen as follows:
n −γt  tn   nt n − 1 
=
Lt t →∞ t e =
Lt t →∞  γt  Lt t →∞  γt 
 e   γ e  ( using L’ Hospital’s rule )
Applying the L’ Hospital rule repeatedly, we get
e – γ t 0 if γ > a
LTt → ∞ t n=

Thus the function e is of exponential order a.

at

(iii) exp
( t ) ( n >1 ) is not of exponential order, since
n

LT exp ( t ) e = LT exp  t – γ )  = ∞
t →∞
n −γt
t →∞
n− 1

for any finite value of γ .

f (t )
Definition. Let be a continuous and single-valued function of the
real variable t definition for all t , 0 < t < ∞, and is of exponential order.
360 Differential Equations: Theory and Applications

f (t ) F (s)
Then the Laplace transform of is defined as a function denoted
by the integral
∞
L  f ( t )= (s)
; s  F= ∫e
− st
f ( t ) dt
0 (3)
For any finite value of γ .
Over the range of values of s for which the integral exist. Here, s is
L  f ( t ) : s 
a parameter, real or complex . Obviously,  is a function of s
Thus,
f ( t ) into F ( s )
Where L is the operator which transform called laplace
−1
transform operator, and L is the inverse Laplace transform operator.
The Laplace transform belongs to the family of “ integral transform”.
F (s) f (t )
An integral transform pf the function is defined by an integral
of the form
b

∫k ( s, t ) f ( t ) dt = F ( s )
c (4)
k ( s, t ) ,
Where a function of two variables s and t , is called the Kernel
of the integral transform The kernel and limits of integration of various
integral transform are given in Table 1 ( which is not exhaustive ).
 Laplace Transform Methods 361

Table 1. Kernel and Limits for Various Integral Transforms

k ( s, t ) a b
Name of the transform

Laplace transform e − st 0 ∝

Fourier transform e ist / 2π −∞ ∞

2
sin st
Fourier sine transform π 0 ∞
2
cos st
Fourier cosine transform π 0 ∞
tJ n ( st ) 0 ∞
Hankel transform

Mellin transform t s −1 0 ∞
The integral transform defined above are applicable , either for semi-
infinite or infinite domains. Similarly , finite integral transform can be
defined domains. Now, we are in a position to verify the following important
result.
f (t )
Theorem. If is piecewise continuous in the range t ≥ 0 and is of
F ( s ) of f ( t )
exponential order γ . then the Laplace transform exists for
all s > γ .
Proof. From the definition of Laplace transform,
∞ T
L  f ( t ) ; s = ∫ e f ( t ) dt=
− st
∫e
− st
f ( t ) dt= I 1 + I 2
0 0

f (t ) 0 < t < T , I1
Since is piecewise continuous on every finite interval
exists, whereas
∞
I 2 ≤ ∫ | e − st f ( t ) | dt
T

But
(t) is a function of exponential order’ therefore,
 362 Differential Equations: Theory and Applications

f ( t ) < Meλt for λ real

Hence,
e − st
f ( t ) < Me −( s – λ )t

Thus,
Me – ( s – γ ) T
∞
=I2 e − ( s − γ )t M dt
∫= , s >γ
T
s −γ

In other words,
l2 can be made as small as we like provided T is large

l2 L  f ( t ) ; s  s >γ .
enough and , therefore exists. Hence exist for

7.2. TRANSFORM OF SOME ELEMENTARY FUNC-

TIONS
Following the definition of Laplace transform by the integral (3), we shall
compute the Laplace transform of some elementary functions.
Example. Find the Laplace transform of
( i )1 . ( ii ) 0 , ( iii ) t , ( iv ) eat , ( v ) e – at
Solution. Using the definition of Laplace transform, we have
∞ ∞
 e − st  1
( i ) L [=
l ;s ] ∫ e −st
(=
1 ) dt  =  if s > 0
0  − s 0 s
∞
L [ 0 ; s )]
= e ( 0 ) dt
∫=
− st
0
(ii) 0

∞ ∞ ∞ ∞
 e − st   e − st  e − st 1
L[ t ; s ]= ∫e
− st
. dt = ∫ td  =
  t  − ∫ dt = 2
(iii) 0 0  − s   −s  0 0 − s s
∞
∞ ∞
 e − ( s − a )t  1
L [=
t ;s ] ∫e
– st
=at
e dt ∫e
−( s − a ) t
=dt  =  s>a
(iv) 0 0  − ( s − a )  0 s – a
 Laplace Transform Methods 363

∞ ∞
1
∫e = e –( s +a ) d t
∫0=
− st
e − at d t
L  e − at ; s  0 s+a
(v) =
Example. Find the Laplace transform of
(i ) cos at , ( ii ) sin at.
Solution Following the definition of Laplace transform, we have
∞ ∞

[cos at ; s ] ∫ e=
( i ) L= cos at dt
− st
=
Re ∫e e dt Re L e ; s 
− st iat iat

0 0

1 s + ia s
=
Re =
Re 2 2
s – ia s +a s +a2
2

s + ia a
L [ sin
= at ; s ] Im L =
 e ; s  Im
iat
= 2 2
(ii) s +a s + a2
2

Example. Find the Laplace transform of

(i ) cosh at , ( ii ) at .
Solution. Using the result established in , we have
 e at + e − at  1
() [
i L=cosh a t ; s ] 
L=
 2

 2
{
L  e at ; s  + L  e − at ; s  }
1  1 1  s
=  + = 2
2  s − a s + a  s − a2 c

 e at − e − at  1
L [sinh at ; s ] L  =
=
 2
;s 
 2
{
L  e at ; s  − L  e − at ; s  }
(ii)
1 1 1  a
=  − = 2 2
2  s –a s + a  s −a

Example. Find the Laplace transform of t , where n is a positive integer.

n

Solution . Using the definition of Laplace transform , we have

364 Differential Equations: Theory and Applications

∞ ∞ ∞ ∞ ∞
 e − st   n e − s t  n n n − 1 − st
∫0
L  t ; s  = ∫0  − s  = ∫0 t e dt = s ∫0
− st n −1 − st
n
e t dt = tn
d 
n
t  + t e dt
 − s 0 s

Hence ,
n
L  t n ; s  = L  t n − 1 ; s 
s
Similarly, we can prove the following
n −1
L t n –1 ; s  = L t n − 2 ; s 
s
n –2
L t n − 2 ; s  = L t n − 2 ; s 
s

2
L t 2 ; s  = L [ t ; s ]
s
1
L[ t ; s ]=
s2
n n −1 n − 2 2 1 n!
L  =
t n ; s 
. . …= . 2 n +1
Therefore, s s 2 s s s which can be
expressed in Gamma functions as
n +1
L  t n ; s  = n + 1
s

7.3. PROPERTIES OF LAPLACE TRANSFORM

We present a few important properties of the Laplace transform in the
following theorems which will be enable us to find the Laplace transform of
a combination of functions whose transform are known.
c and c
Theorem. (Linearity property ) . if 1 2 are any two constants

F ( s ) and F2 ( s )
and if 1 are the Laplace transforms, respectively of
f1 ( t ) and f 2 ( t ) ,
then
 Laplace Transform Methods 365

L  { c1 f 1 ( t ) + c2 f 2 ( t )} ; s  =
c1 L  f1 ( t ) ; s  + c2 L  f 2 ( t ) ; s  =
c1 F1 ( s ) + c2 F2 ( s )

Proof. Following the definition of Laplace transform , we have

∞
L  { c1 f1 ( t ) + c2 f 2 ( t )} ; s  = ∫e − st { c1 f 1 ( t ) + c2 f 2 ( t )} dt
0

∞ ∞
c1 ∫ e − st
f1 ( t ) dt + c2 ∫e − st f 2 ( t ) dt
0 0

c1 L  f 1 ( t ) ; s  + c2 L  f 2 ( t ) ; s 

= c1 F1 ( s ) + c2 F2 ( s )

Theorem. (shifting property ) . if a function is multiplied by e , the

at

transform of the resultant is obtained by replacing s by

( s − a ) in the
transform of the original function. That is , if
L  f ( t ) ; s  = F ( s )

Then
L  e at f ( t ) =
; s  F ( s − a )

Proof . From the definition of Laplace transform,

∞ ∞
L  e at
f (t )=
; s  ∫e
− st
e at
f (t =
) dt ∫e
–(s −a) t
f ( t=
) dt F ( s − a )
0 0

Similarly,
L  e – at f ( t )=
; s F ( s + a ) 

Theorem . (Multiplication by power of ) . if
L  f ( t ) ; s  = F ( s )

then
n
n d
L  t f ( t ); s  =
 n
 ( −1 ) n F ( s ) =
( −1) n F ( n ) ( s )
ds
=
where n 1 , 2 , 3 ,…
 366 Differential Equations: Theory and Applications

Proof. Form the definition of Laplace transform

∞
F ( s ) L=
=  f ( t ) ; s  ∫e
− st
f ( t ) dt
0

Hence,
d  − st 
∞
d
 F ( s )  =  ∫e f ( t ) dt 
dt ds  0 
Interchanging the operations of differentiation and integration for which
we assume that the necessary conditions are satisfied and since there are two
variables s and t , we use the notation of partial differentiation and obtain
∞ ∞
d ∂
dt
{ F ( s )} =
∫ ∂ s
{
e – st f ( t ) dt = }
− ∫e − st tf ( t )}dt =
− L tf ( t ) ; s 
0 0

Therefore ,

L  tf ( t ) ; s  = − F(s)
ds
By repeated application of the above result, it can be shown that
n
n d
L  t n f ( t ) ; s  =
( −1 ) n F ( s ) =
( −1 ) F n ( s )
n

ds
Theorem. ( Differentiate property) , if
L  f ( t ) ; s  = F ( s )

Then
f n t ; s  s n F ( s ) − s n − 1 f ( 0 ) − s n − 2 f
L  = '
( 0 ) −…− s f n − 2 ( 0 ) − f n − 1 ( 0 )
Proof Form the definition of Laplace transform , we have
∞ ∞
L  f ( t ) ; s 
= '
e f ( t ) dt
∫=
– st
 e
− st
f ( t )  + s ∫ e − st f ( t ) dt
∞
0
0 0

Similarly , it can be shown that

L  f n ( t ) ; s = sL  f ' ( t ) ; s  – f '
( 0 )= s { sF ( s ) − f ( 0 )} − f ' ( 0 )= s 2 F ( s ) − sf ( 0 ) − f ' ( 0 )
 Laplace Transform Methods 367

L  f m
( t ) ; s  = s 3 F ( s ) − s 2 f ( 0 ) − sf ' ( 0 ) – f ''
(0)
Thus, in general,
f n t ; s  s n F ( s ) − s n − 1 ( 0 ) − s n − 2 f ( 0 ) −…− sf n – 2 ( 0 ) − f n − 1 ( 0 )
L  =

This property is very useful for solving differential equations.

Example. Find the Laplace transform of
( i ) e at cos bt , ( ii ) e at sin bt , ( iii ) e at cosh bt , ( iv ) e at t n , and ( v ) cos at cosh bt.

Solution. Using the shifting property

s s−a
( i ) L=
 e at sin bt ; s  =
s + b 2 s → ( s –a )
2
( s − a ) 2 +b 2
b s –a
( ii ) L=
 e at sin bt ; s  =
( s − a)
2
s −b 2 s → ( s –a )
2
+ b2

s s−a
L e=
at
cosh bt ; s  =
( iii ) (s − a)
2
s − b2 s → ( s – a )
2
−b 2

n! n!
( iv ) L=
 e at t n ; s  =
( s − a)
n +1
s n + 1 s →( s − a )

  ebt − e − bt  
( v ) L [ cos at cosh bt ; s ] = L  cos at  ;s 
  2  
1
2
{
L  ebt cos at ; s  − L  e – bt cos at ; s  }
1  s−a s +b 
 − 
2  ( s – b )2 + a 2 ( s + b ) 2 + a 2 
 
Example. Find the Laplace transform of the following :
(i ) t 2 e at , ( ii ) t sin at , ( iii ) t 2 cos at , ( iv ) t n e − at
368 Differential Equations: Theory and Applications

Solution. Using the result established in the Theorem , we have

2 d
2
d 2  1  d  −1  2
( i ) L  t 2 ( −1 ) 2 L  e at ; s  =
e at ; s  = 2  = 

=

d s  s − a  ds ( s − a )
 ( s –a )
2 3
ds

Alternatively,
2! 2
L =
e at t 2 ; s  =
( s−a )
3
s 3 s → ( s –a )
(using the shifting property)
1 d d  a  2as
( ii ) L [ t sin at ; s ] =
( −1 ) L [ sin at ; s ] =
−  2 2 =
ds ds  s + a  s 2 + a 2 2 ( )
2 2
2 d d  s 
( −1 ) 2 L [ cos at ; s ] =
L t 2 cos at ; s  = 2  2 
(iii) ds ds  s + a 2 

 
d  a 2 − s 2  2 s 3 − 6 sa 2
=
(
ds  s 2 + a 2 2  ) ( )
3

  s2 + a2

n! n!
( iv ) L =
 e − at t n ; s  =
( s + a)
n +1
s n +1 s → ( s + a )
( using the shifting
property )
Example. Verify the initial value theorem for the function
f (t ) =
1 + e − t ( sin t + cos t )

f ( t ) = 1 + e − t (sin t + cos t ) , we have

Solution . Given
F (s) = (
L [ 1 ; s ] + L  e − t sin t + e − t cost ; s 
L  f ( t ) ; s  = )
1  1 s 
=+  2 + 1 + 2 
s s s +1  s → ( s + 1 )

1  2+s 
= + 
s  s 2 + 2 s +2 
 Laplace Transform Methods 369

Hence,
2s + s 2
sF ( s ) = 1 + 2
s + 2s + 2
Therefore,
2
Lts → ∞ sF ( s ) = Lt s → ∞ 1 + s +1 = 1+1= 2
2 2
1+ + 2
s s
f ( 0 ) =1 +1 = 2 .
But Thus,
Lt s → ∞ sF ( s ) = f ( 0 )

Hence the result

Theorem . ( Division by t ) . if
L  f ( t ) ; s = F ( s ) 

Then
 f (t )  ∞
L ; s  = ∫ F ( s ) ds
 t  s
Proof. From the definition of Laplace transform
∞
L  f ( t )= (s)
; s  F= ∫e
− st
f ( t ) dt
0

Integrating the above with respect to s between the limits s to ∞, we get

∞
 ∞ − st  ∞
 ∞ − st 
∫s
= F ( s ) ds =∫e f ( t ) dt  ds ( )
∫0  ∫0e ds  dt
f t
s 
(by changing the order of integration )
f (t )  f (t )
∞ ∞ ∞
 e − st 
∫ ( t )   dt
f= ∫= e dt – st
L  ;s
0  −t  s t 0  t 
Hence the result .
 370 Differential Equations: Theory and Applications

f (t )
Note: In applying this rule , one should be careful . Since t may
f (t )
have an infinite discontinuity at t = 0. it may not be integrable. If t is
not integrable, then its Laplace transform does not exist. For example , at
t = 0 , the function sin t t doses not have an infinite discontinuity , while
t
cos has an
the function t infinite discontinuity .

7.4. TRANSFORM OF A PERIODIC FUNCTION

f ( t ) is T , if ( t + T ) =
f (t )
A function called periodic with period
for all values of t and T > 0. For example, the trigonometric functions
sin t and cos t are periodic functions of periods 2π . Periodic functions
occur very often in a variety of engineering problems.
if f ( t )
Theorem. is a periodic function with periods T , then
T
L  f ( t ) ; s 
= ∫e
− st
(
f ( t ) dt / 1 − e – st )
0

Proof. From the definition of Laplace transform , we have

∞ T ∞
L  f ( t ) ; s 
= e f ( t ) dt ∫ e
∫=
− st − st
f ( t ) d t + ∫ e − st f ( t ) dt
0 0 T

If we substitute t= u + T in the second integral on the right-hand side

and write dt = du. we obtain
T ∞
L  f ( t ) ; s  =∫ e – st
f ( t ) d t + ∫ e − s( u + T ) f ( u + T ) du
0 0

T ∞
= ∫e
– st
f ( t ) dt + e − st
∫e
– st
f ( u ) du
0 0

T
= ∫e
– st
f ( t ) dt + e – sT L  f ( t ) ; s 
0
 Laplace Transform Methods 371

Rearranging , we get
T

( 1− e ) L  f ( t ) ; s  =
sT
∫ e f ( t ) dt
− st

Thus,
T
L  f ( t ) ; s  = ∫ e – st f ( t ) dt
0 /
(1 – e )
−st

Hence the result.

Example. Obtain the Laplace transform of the periodic saw-tooth wave
function given by
t
f (t ) =
T of period T , 0 < t < T
Solution . The graph of the periodic saw-tooth function is described in
f (t )
. since is periodic with period T , we have
1
T
t 1 1
T T
 e − st 
L f (t ) ; s ]
= = − st ∫
e − st dt = − sT ∫
e − st tdt ∫0  −s 
td
1− e 0 T T 1− e 0 ( )
T 1 − e − st ( )
1   te − st T 1 T 
 + ∫ e dt 
− st

(
T 1 − e − st )   − s 0 s 0 

1  Te − st 1 − st 
=  − e − 1  ( )
( )
T 1 − e − st  − s s 2


Therefore,
1 e − st
L  f ( t ) ;=
s  −
(
s 2T s 1 − e − st )
Example. Find the Laplace transform of
1 , 0 ≤ t < 2
f (t ) = 
−1, 2 ≤ t < 4
f ( t + 4) =
f (t )
 372 Differential Equations: Theory and Applications

f (t )
Solution. In this problem, is a periodic function of period 4; we,
therefore, have \
4
1
L  f ( t ) ; s  = − 4s ∫
e − st f ( t ) dt
1− e 0

1  − st 
2 4

( ) ( )
1 − e −4 s  ∫0 ∫2
=  e 1 dt + e − st
−1 dt 


1  −2 e − 2 s e −4 s 1 
=  + + 
1− e − 4 s  s s s

7.5. TRANSFORM OF ERROR FUNCTION

The error function denoted by erf
(t ) is defined as
2
t
erf ( t ) = ∫e
− u2
du
π 0

This function occurs in many branches of science and engineering ; for

example , in probability theory, the theory of heat condition, and so on . In
terms of the power series , we have

erf ( t ) =
2
t
( −1 ) n t 2 n + 1
∞

π ∫0 n∑= 0 n !( 2n +1)
We can easily verify that these series converge everywhere and, therefore
erf ( t ) is
, an entire function .From the definition , it can be verified at once
that
erf ( 0 ) = 0
∞
2 1/ 2
erf (=
∞) ∫e
−u2
=
du = 1
π 0 π
 Laplace Transform Methods 373

The graph of error function is shown in

In solving heat conduction equation, it has been found useful to introduce
the complementary error Function defined as
2  
∞ ∞
2
t
erfc ( t )
= = ∫ e du
−u 2
 ∫ e du − ∫ e
−u2 –u 2
du 
π 0 π 0 0 
Therefore,
erfc ( t ) = 1 – erf ( t )

Now we shall the Laplace transform of erf (t): From the definition Of
Laplace Transform
∞
2
t
L  erf ( t ) ; s  = ∫ e − st
∫e
−u2
du dt
0 π 0

Changing the order on integration , we obtain

∞ ∞
2
L  erf ( t ) ; s  = ∫e ∫e
2
−u – st
dt du
π 0 0

∞
2 (
− u 2 + su )
=
s π ∫e
0
du

2
s2 8 −u + s 
2  
∫e
2 
= e 4 
du
s π 0

s
x= u + ,
Setting 2 we get
s2 ∞
2
L  erf ( t ) ; s  = e 4
∫e
− x2
dx
s π s
2

1 2 s
= e s /4 erfc  
s 2

Example. Find the Laplace transform of erf

(t ).
1/2

Solution. From the definition of Laplace transform , we have

 374 Differential Equations: Theory and Applications

r1/2
 1
 ∞ − st 2
L erf (t ; s  = ∫ e
2
∫e
−u2
du dt
  0 π 0

Changing the order of integration, we get

∞ r1/2
2 2
L erf t ( ) ; s  =
1/2

π ∫
e − st

π ∫e
− n2
du dt
0 0

∞ ∞
2
∫e du ∫ e dt
2
−u − st
=
s π 0 u2

∞
2
π ∫
− u 2 − su 2
e du
s 0

∞
2 −(1+ s )u 2

π ∫
e du
s 0

Setting
( 1 + s )=
u2 x 2 or 1 +=
su x,
we have
du = dx / 1+ s
Then
  1  2
∞
1  2
∞

∫ s 1 + s  π ∫0
− x2 − x2
L erf  t  ; s  =
= 2
e dx  e dx 
    s π 1 + S 0 
or
1
( )
L erf t1/2 ; s  =
s 1+ s

7.6. TRANSFORM OF BESSEL’S FUNCTION

Bessel function arise in several problem involving circular or cylindrical
geometry, it is therefore useful to find the Laplace transform of Bessel
functions if the first kind .
Example. Find the Laplace transform of
( i ) j0 ( t ) , ( ii ) tj0 ( t ) , ( iii ) e−at j0 ( t ) .
 Laplace Transform Methods 375

Solution.
( i ) From the definition of the Bessel function, we have
( −1)
r n+2r
∞
t
jn ( t ) = ∑  
r =0 r ! n + r + 1  2 

For n = 0, we have
( ( −1)  t 
−1)  t 
2r r 2r r
∞ ∞
=j0 ( t ) ∑   =∑ 2  
= r! r +1  2 
r 0= r 0 ( r !)  2 

t2 t4 t6
= 1− + − +…
2 2 2 2 ×4 2 2 2 ×4 2 ×6 2
Thus,
1 1
) ; s  L [1; s ] −
L  j0 ( t= 2
L t 2 ; s  + 2 2 L t 4 ; s  −…
2 2 ×4
1 1 2! 1 4! 1 6!
= − 2 3 + 2 2 5 − 2 2 2 7 +…
s 2 s 2 ×4 s 2 ×4 ×6 s

1  1  1  1× 3  1  1×3 × 5  1  
2 3

 1 −  2 +  2 −  2  +…
s  2  s  2 × 4  s  2 ×4 × 6  s  
−1/2
1 1 1
=1 + 2  =
s s  1 + s2
Hence,
1
L  j0 ( t ) ; s  =
1 + s2
(ii) From the properties of Laplace transform, we have
dn
L t n f ( t ) ; s  = ( −1) F (s)
n

ds n
Therefore,
d d  1 
L tj0 ( t ) ; s  =
( −1)  L  j0 ( t ) ; s   =
−  
ds ds  1 + s 2 
376 Differential Equations: Theory and Applications

Thus,
s
L tj0 ( t ) ; s  = 3

(
1 + s2 ) 2

(iii) From the shifting property of Laplace transform , we have

L  e − at f ( t ) =
; s  F ( s + a )

Therefore,
1 1
e j0 ( t ) ; s 
− at
L= =
1 + s 2 s →( s + a ) 1+ (s + a)
2

7.7. TRANSFORM OF DIRAC DELTA FUNCTION

The concept of impulse function or Dirac delta function has been introduced
in itself. In certain applications involving a sudden excitation of a system or
a large voltage over a short interval of time, the Laplace transform of Dirac
Delta function is useful. Form the property of Dirac Delta function, we have
∞

∫δ ( t − a ) f ( t ) dt =
0
f (a)

f ( t ) = e − st ,
In particular , if then
∞
L δ ( t − a ) ; s  = ∫e
− st
δ ( t − a ) dt = e − as , a > 0
0

7.7.1. Inverse Transform

So far we have discussed various properties of the Laplace transform and
studies the Laplace transform of some simple functions. However, if the
Laplace transform technique is to be useful in applications, we have to
consider the reverse problem too, i.e., we have to find the original function
f (t ) F ( s ).
x when we know its Laplace transform Thus, if
L  f ( t ) ; s  = F ( s )

Then
 Laplace Transform Methods 377

f ( t ) = L−1  F ( s ) ; t 
In other words , the inverse Laplace transform of a given function
F (s) f (t ) F ( s ).
is that function whose Laplace transform is It can be
f (t ) −1
established that is unique . Here, L is known as inverse Laplace
transform operator. Form the elementary definition (24) and form the results
obtained thus far in finding the Laplace transform of some elementary
functions , we can immediately generate the following table of transforms:

Table of Laplace Transform

f (t ) L  f ( t ) ; s  (s)
F= F (s) L−1  F ( s ) ; t  f ( t )

0 0 0 0
1 1
1 s s 1
1 1
e at (s − a) (s − a) e at
1 1
−at
e (s + a) (s + a) e −at
t 1/ s 2 1/ s 2 t
n!
tn s n +1 1/ s n +1 t n / n!
a a
sin at s2 + a2 s2 + a2 sin at
s s
cos at s2 + a2 s2 + a2 sinh at
a a
sinh at s2 − a2 s − a2
2
sinh at
s s
cosh at
s − a2
2
s − a2
2
cosh at
2as 2as
(s ) (s )
2 2
t sin at t sin at
2
+ a2 2
+ a2
2 2
s a s2 2

(s ) (s )
2 2
2 2
t cos at +a 2
+ a2 t cos at
378 Differential Equations: Theory and Applications

L  f ( t ) ; s 
In most of the problems we have considered earlier,  is a
simple rational function. The linearity property holds true even in the case
F (s) F (s)
of inverse transform. That is, if 1 and 2
f1 ( t ) and f 2 ( t ) c1 and c2 are any
Are the Laplace transform of , and if
two constants, then
L−1 { c1 F1 ( s ) ± c2 F2 ( s )} ; t =
 c1 L−1  F1 ( s ) ; t  ± c2 L−1  F2 ( s ) : t 

L  f ( t ) ; s 
By expressing  as partial fractions, we should be able to
recognize them as the transform of some known function , with the helps of
which we can write down the inverse transform . Similarly shifting property
is also useful in constructing the inverse transform of some functions, which
is stated in the following theorem:
Theorem.
L  f ( t ) ; s  = F ( s ) ,
If then
L−1  F ( s + a ) ; t  =
e − at L−1  F ( s ) ; t 

Proof.
L  f ( t ) ; s  = F ( s ) ,
Since we have
L−1  F ( s ) ; t  = f ( t )

Recalling the shifting property of Laplace transform, we find that

L e − at ( t ) =
; s  F ( s + a )

L−1  F ( s + a ) ; t  =
e − at f ( t )

Thus,
L−1  F ( s + a ) ; t  =
e at L−1  F ( s ) ; t 

Example. Obtain the inverse Laplace transform of

4 s 2 − 3s + 5
(
( s + 1) s 2 − 2s + 2 )
 Laplace Transform Methods 379

Solution Using partial fraction expansion , we can write

4 s 2 − 3s + 5 A Bs + C
= + 2
( )
( s + 1) s − 2s + 2 s + 1 s − 2s + 2
2

Therefore,
4 s 2 − 3s + 5 A Bs + C
= +
( )
( s + 1) s 2 − 2s + 2 s + 1 s − 2s + 2
2

Therefore,
( )
5 A s 2 − 2 s +2 + ( Bs + C )( s + 1)
4 s 2 − 3s +=

12
A= .
Let s = −1 ; then 5 Equating the coefficient of s on both sides,
we have
9
B+C =
5
Equating the coefficient of constant on both sides, we get 2 A + C =
5
1 8
C= , B= .
which gives 5 and hence 5 The given expression can now be
written as
4 s 2 − 3s + 5 12 1 8 s 1 1
= + +
( 2
)
( s + 1) s − 2s +2 5 s + 1 5 ( s − 1) +1 5 ( s − 1)2 + 12
2 2

Thus, we find that

 4 s 2 − 3s + 5  12  1  8 −1  ( s − 1) + 1  1 −1  1 
L−1  ;t = L−1  ;t + L  ; t  + L  ; t 
( 2
)
 ( s + 1) s − 2 s + 2  5  s + 1  5  ( s − 1) +12  5  ( s − 1) + 12 
2 2

1  1 
+ et L−1  2 ; t 
5  s + 1  ( by using the shifting property)
12 − t 8 t 1
= e + e ( cos t sin t ) + et sin t
5 5 5
12 8 9
= e − t + et cos t + et sin t
5 5 5
 380 Differential Equations: Theory and Applications

Theorem. (change of scale property). If

L−1  F ( s ) , t  = f ( t )

then
1 t 
L−1  F (α s ) ; t  = f 
α α 
Proof. Form the definition of Laplace transform, we have
∞
F ( s ) = ∫e − st f ( t ) dt
0

Therefore,
∞
f (α s ) = ∫ e −α st f ( t ) dt
0

dx
dt = .Then
Let α t = x, so that a we get
∞
1 − sx  x  1  x  1   t  
F (α s )
= ∫
= e f   dx =L f  ; s L f  ;
a0 a a   a   a   a  
Thus,
1 t
L−1  F ( as ) ; t  = f 
a a
Example. Find the inverse Laplace transform of
s2 + 1
( ii ) cot −1 
s
, 
( i ) In s ( s + 2) k
Solution. (i) From Theorem , we have
L t n f ( t ) ; s  = ( −1) F n (s)
n

In particular n = 1 gives
L tf ( t ) ; s  = − F ′ ( s )
 Laplace Transform Methods 381

i.e.,
d
F ( s ) = − L  tf ( t ) ; s 
ds
Let
s 2 +1
L  f ( t ) ; s  = In
s ( s +1)
Then,
d d 
ds
=F(s)
ds 
( )
In s 2 +1 − I n s − In ( s +1 ) 

2s 1 1
= 2
− − = L tf ( t ) ; s 
s +1 s s −1
We get
1 1 2s
L  ft ( t ) ; s  =+ − 2
s s +1 s +1
Hence,
1   1  1   s 
tf ( t ) =
L−1  ; t  + L−1  ; t  ; t  − 2 L−1  2 ;t 
s   s +1  s +1   s +1 
1 e − t − 2 cos t
=+
Therefore,
 s 2 +1  1 + e − t − 2 cost
=f ( t ) L=
−1
 In ;t
 s ( s +1)  t

−1  s 
f ( t ) ; s  cot
L = =   F (s)
( ii ) Let k
Then ,
d d  −1 s  k
F(s)=  cot = − 2 2
ds ds  k k +s
Now we obtain
d k
F(s) =
= L  tf ( t ) ; s 
ds k + s2
2
 382 Differential Equations: Theory and Applications

Therefore,
−1  k 
=tf ( t ) L=  k + s 2 ; t  sin kt
2
 
Hence,
−1  −1  s   sin kt
f (t ) L= cos  k  ; t 
    t

7.8. CONVOLUTION THEOREM (FALTUNG

THEOREM)
We often across function which are not the transform of some known
function, but then, they can possibly be expressed as a product of two
functions, each of which is the transform of a known function. Thus we may
F ( s )G ( s ).
be able to write the given function as where of a known to be
f (t ) g (t ) ,
transform of the functions and respectively.
F ( s ) and G ( s )
Theorem. If are the Laplace transform of
f ( t ) and g ( t ) F (s)G (s)
respectively, then is the Laplace transform of
t

∫ f ( t – u ) g ( u ) du
0

t
L  F ( s ) G ( s )  = ∫ f ( t – u ) g ( u ) du
−1

i.e., 0

This integral is called the convolution of f and g and is denoted by the

*
symbol f g .
Proof. Form the definition of Laplace transform ,we have
∞  ∞ 
F ( s ) G ( s ) =  ∫ e − sv f ( v ) dv   ∫ e − su g ( u ) du 
0  0 
∞∞
= ∫∫ e − s( v + u ) f ( v ) g ( u ) dv du
0 0
 Laplace Transform Methods 383

∞
 ∞ ( v + u ) 
= ∫ g ( u )  ∫e f ( v ) dv  du
0  0 
Let u + v =t in the inner integral. Then,
∞
 ∞ – st 
F ( s )G ( s ) ∫0 g ( u )  ∫ e f ( t – u ) dt  du
 0 
Change the order of integration
Then, we get
 t − st
∞

F ( s ) G ( s ) ∫  ∫ e f ( t − u ) g ( u ) du  dt
=
0 0 

 
∞ t
= ∫ e − st  ∫ f ( t – u ) g ( u ) du  dt
0  0 
t 
= L  ∫ f ( t – u ) g ( u ) du ; t 
0 
Hence the result.
f ( t ) and g ( t )
Definition. We define the Laplace convolution of by the
integral
t

∫ f ( t − u ) g ( u ) du
*
=
f g
0

It can be verified that f and g can be interchanged in the convolution, i.e.,

f and g are commutative Let t − u = v so that – du = dv . Then ,
Therefore,
f *g = g* f
Example. Prove that
t

∫ j ( t ) j ( t − u ) du =
0
0 0 sin t

Solution. Form table ,we not that

 384 Differential Equations: Theory and Applications

 1 
L−1  2 ; t  = sin t
 s +1 
We shall also write
1 1 1
2
=
s +1 s 2 +1 s 2 +1
Now taking
1 1
=F(s) = ,G ( s )
s 2 +1 s +12

Their inverse transform give

 1 
L− 1  ; t  = j0 ( t )
2
 s +1  = f= (t ) g(t )
Hence, using the convolution theorem, we have
 1
t

∫ j ( t ) j ( t − u=
) du
−1
L  2 = ;t  0 0 sin t
 s +1  0

7.8.1. Summary and Discussion

Laplace transform is essentially a mathematical tool which can be used to
solve several problems in science and engineering . This transform was
first introduced was first introduced by Laplace, a French mathematician,
in the year 1790 in his work on probability theorem. This technique became
popular when Heaviside applied to the solution of an ordinary differential
equation referred hereafter as ODE. representing a problem in electrical
engineering . To the basic question as to why one should learn Laplace
transform technique when other technique available. The answer is very
simple. Transform are used to accomplish the solution of certain problems
with less effort and in a simple routine way.
f (t )
Definition. Let be a continuous and single-valued function of the
real variable t definition for all t , 0 < t < ∞, and is of exponential order.
f (t ) F (s)
Then the Laplace transform of is defined as a function denoted
by the integral
 Laplace Transform Methods 385

∞
L  f ( t )= (s)
; s  F= ∫e
− st
f ( t ) dt
0 (3)
For any finite value of γ .
Over the range of values of s for which the integral exist. Here, s is
L  f ( t ) : s 
a parameter, real or complex . Obviously,  is a function of s
Thus,
f ( t ) into F ( s )
Where L is the operator which transform called laplace
−1
transform operator, and L is the inverse Laplace transform operator.
The Laplace transform belongs to the family of “ integral transform”.
F (s) f (t )
An integral transform pf the function is defined by an integral
of the form
b

∫k ( s, t ) f ( t ) dt = F ( s )
c (4)
k ( s, t ) ,
where a function of two variables s and t , is called the Kernel of
the integral transform The kernel and limits of integration of various integral
transform are given in Table 1 ( which is not exhaustive ) .
f (t )
Theorem. If is piecewise continuous in the range t ≥ 0 and is of
F ( s ) of f ( t )
exponential order γ . then the Laplace transform exists for
all s > γ .
∞ ∞
 e − st  1
( i ) L [=
l ;s ] ∫ e −st
(=
1 ) dt  =  if s > 0
0  − s 0 s
∞
L [ 0 ; s )]
= e ( 0 ) dt
∫=
− st
0
(ii) 0

∞ ∞ ∞ ∞
 e − st   e − st  e − st 1
L[ t ; s ]= ∫e
− st
. dt = ∫ td  =
  t  − ∫ dt = 2
(iii) 0 0  − s   −s  0 0 − s s
386 Differential Equations: Theory and Applications

∞
∞ ∞
 e − ( s − a )t  1
L [=
t ;s ] ∫e
– st
=
e dt at
∫e
−( s − a ) t
=dt  =  s>a
0 0  − ( s − a )  0 s – a
(iv)
∞ ∞
–( s +a ) 1
∫e =
e d t ∫=
− st − at
e dt
L  e − at
; s  s+a
(v) = 0 0

∞ ∞

[cos at ; s ] ∫ e=
( i ) L= cos at dt − st
=
Re ∫e e dt Re L e ; s 
− st iat iat

0 0

1 s + ia s
=
Re =
Re 2 2
s – ia s +a s +a2
2

s + ia a
L [ sin
= at ; s ] Im L =
 e ; s  Im
iat
= 2 2
(ii) s +a s + a2
2

 e at + e − at  1
() [
i L=cosh a t ; s ] 
L=
 2

 2
L  e at ; s  + L  e − at ; s  { }
1  1 1  s
=  + = 2
2  s − a s + a  s − a2 c

 e at − e − at  1
L [sinh at ; s ] L  =
=
 2
;s 
 2
L  e at ; s  − L  e − at ; s  { }
(ii)
1 1 1  a
=  − = 2 2
2  s –a s + a  s −a
 Laplace Transform Methods 387

Table Of Laplace Transform

f (t ) L  f ( t ) ; s  (s)
F= F (s) L−1  F ( s ) ; t  f ( t )

0 0 0 0
1 1
1 s s 1
1 1
e at (s − a) (s − a) e at
1 1
e −at (s + a) (s + a) e −at
t 1/ s 2 1/ s 2 t
n!
tn s n +1 1/ s n +1 t n / n!
a a
sin at s + a2
2
s + a2
2
sin at
s s
cos at s + a2
2
s + a2
2
sinh at
a a
sinh at s − a2
2
s − a2
2
sinh at
s s
cosh at
s − a2
2
s − a2
2
cosh at
2as 2as
(s ) (s )
2 2
2
t sin at + a2 2
+ a2 t sin at
2 2 2 2
s a s
(s ) (s )
2 2
2 2 2
t cos at +a + a2 t cos at

7
 CHAPTER

8
GREEN’S FUNCTION

CONTENTS
8.1. Introduction..................................................................................... 390
8.2. The Eigenfunction Method............................................................... 403
8.3. Summary and Discussion................................................................. 406
References.............................................................................................. 408
390 Differential Equations: Theory and Applications

8.1. INTRODUCTION
Consider the differential equation

Lu ( x ) = f ( x ) L−1 (1)

Where L is an ordinary linear differential operator, f ( x ) is a known

function, while is a unknown function. To solve the above differential
equation, one method is to find the operator in the form of an integral
operator with a kernel G ( x,ξ ) such that

u ( x ) = L−1 f ( x ) = ∫ G ( x, ξ ) f (ξ ) d ξ (2)
The kernel of this integral operator is called Green’s function for the
differential operator. Thus the solution of the non-homogenous differential
equation can be written down, once the Green’s function for the problem is
known. Applying the differential operator to both sides of , we get

f ( x ) = LL−1 f ( x ) = ∫ LG ( x ,ξ ) f ( ξ ) d ξ (3)

This equation is satisfied if we choose G ( x, ξ ) such that

LG ( x =
, ξ ) δ ( x − ξ ) (4)

Where δ ( x − ξ ) is a Dirac δ - function. The solution of Eq. is called a

singularity solution of Eq. In previous section , we have already introduced
the concept of Dirac δ − function and studied its various properties. Now,
they become handy to understand more about Green’s function

Definition . Let us consider an auxiliary function φ ( x ) of a real

variable x which possess derivative of all orders and vanishes outside a
finite interval. Such function are called test functions.
Now we shall introduced the concept of the derivative of a δ − function
in terms of the derivative of a test function . we say that δ ′ ( x ) is the
derivative of δ ( x ) if
 Green’s Function 391

∫ δ ( x ) φ ( x ) dx = −φ ′ ( 0 )
'

−∞ (5)

For every test function φ ( x ) . similarly , we define φ ′′ ( x ) by

∫ δ ( x )φ ( x ) dx = −φ ′′ ( 0 ) (6)
''

−∞

With this definition of a derivative, we can show that the δ − function

is the derivative of a Heavside unit step function H ( x ) defined by
 1 for x ≥ 0
H ( x) = 
 0 for x < 0 (7)
To see this result, we consider

∞ ∞ ∞

∫ H ′ ( x )φ ( x ) dx =
−∞
− ∫ H ( x )φ ′ ( x ) dx =
−∞
− ∫ φ ′ ( x ) dx =
0
φ(0)

By comparing the above result with property III of i.e., with

∞

∫ δ ( x ) φ ( x ) dx = φ ( 0 )
−∞

we obtain
H ′( x) = δ ( x)
(8)
Similarly, the notion of δ − function and its derivative enables us to
give a meaning to the derivative of a function that has a jump discontinuity
at x = ξ of magnitude unity. Let
 1, x ≥ ξ
H (x −ξ ) =

0, x < ξ
φ ( x).
Then, for any test function we have
∞ ∞

∫ H ′ ( x − ξ )φ ( x ) dx ∫ δ ( x –ξ ) φ ( x ) dx =φ ( ξ )
−∞ = −∞ (9)
 392 Differential Equations: Theory and Applications

It can also be noted that

d
 ( x –ξ ) H ( x − ξ )  = ( x − ξ ) H ' ( x −ξ ) + H ( x − ξ )
dx
Integrating both sides with respect to x between the limits − ∞ to ∞ ,we
get
∞ ∞

( x − ξ ) H=
( x –ξ ) ∫ ( x –ξ ) H ' ( x –ξ ) dx + ∫ H ( x –ξ ) dx
−∞ −∞

∞
= ∫ H ( x –ξ ) dx
−∞ (10)
We now consider an example to illustrate the inversion of a differential
operator by considering the BVP:
d2 u
= f ( x ) , u=
( 0 ) u=
(1) 0
dx 2
In this Eq. (4) becomes
d 2G
=
LG = δ ( x –ξ )
d x2 (11)
Noting that the δ − function is the derivative of the Heaviside unit step
function and integrating
d
G=( x , ξ ) H ( x –ξ ) + C1 ( ξ )
dx 
C (ξ )
where 1 is an arbitrary function. Integrating the above result once with
respect to x. we get
G ( x ,ξ ) =∫ H ( x –ξ ) dx + C1 ( ξ ) x + C 2 (ξ )
= ( x − ξ ) H ( x − ξ ) + C1 ( ξ ) x + C 2 ( ξ )
C ( ξ ) and C2 ( ξ )
where 1 can be determined form the boundary conditions.
Thus , from Eq. (2) we have
x ∞ ∞
u ( x)
= ∫ ( x –ξ ) H ( x –ξ ) f ( ξ ) d ξ + x ∫ C ( ξ ) f ( ξ ) d ξ + ∫ C ( ξ ) f ( ξ ) d ξ
1 2
0 −∞ −∞

: u ( 0 ) = 0,
Now. Using the boundary condition we get
 Green’s Function 393

∞
0 = 0 + 0 + ∫ C2 ( ξ ) f ( ξ ) d ξ
−∞

C2 (ξ ) = 0.
Implying that Using the second boundary condition :
u (1) = 0.
we have
1 ∞
0 ∫ ( 1 –ξ ) f ( ξ ) dξ + ∫ C1 ( ξ ) f ( ξ ) dξ
0 −∞

C1 (ξ ) =− (1 − ξ ) , 0 ≤ ξ ≤ 1,
Implying that and zero for all other values
of ξ . Hence
x 1
u( x) ∫ ( x –ξ ) H ( x –ξ ) f ( ξ ) d ξ – x ∫ ( 1 − ξ ) f (ξ ) d ξ
0 0 (12)
Comparing this result with Eq. (2) , we have the Kernel of the integral
operator, which is known as Green’s function or source function given by
G ( x, ξ ) = ( x − ξ ) H ( x − ξ ) − x (1 − ξ ) , 0 ≤ ξ ≤ 1
(13)

Satisfying the boundary condition:

( 0, ξ ) G=
G= (1, ξ ) 0
This concept can be extended to partial differential equation also. To
make the ideas clearer, let us consider
L u ( X )  = f ( X )
(14)
where L is some linear v partial differential operator in three independent
variables x , y , z and X is a vector in three-dimensional space. Then the
G( X; X )
Green’ which satisfies the equation.
L G ( X ; X=
′ )  δ ( X − X ′ )
(15)
On expansion , this equation becomes
L G ( x, y, z; x′ , y′, z ′ )  =
δ ( x − x′ ) δ ( y − y ′ ) δ ( z − z ′ )
(16)
δ ( X − X ')
Here the expansion is the generiliazation of the concept
G( X, X )
of delta function in three-dimesnional space IR and represents
394 Differential Equations: Theory and Applications

the effect at that point X due to a source function or delta function input
applied at ′ .Equation has the following interpretation in heat conduction or
G( X; X )
electrostatic: can be viewed as the temperature ( the electrostatic
potential) at any point X in IR due to a unit source ( due to a unit charge )
f (X )
located at X , we get Eq. (15) on both sides by and integrating over
the volume V with respect to X ′. we get
.  .
L  ∫ G ( X ; X ) f ( X ) dV=
X  ∫ f ( X ) δ ( X − X ′) dV
x
'
= f (X )
Vx  VX

Comparing with Eq. (14) ,we arrive at

.
u ( X ) = ∫ G ( X ; X ) f ( X ) dVx
VX
(17)
Which is the solution of Eq. (14) . This leads to the simple definition that
u ( x, y, z ; x′, y′, z ′ )
a function is a fundamental solution of the equation , for
2
0 if is a solution of the non-homogenous equation
example, ∇ u =
∇ 2 u = δ ( x, y , z; x′ , y′, z ′ )
This idea can be easily extended to higher dimension . Thus the Green’s
function technique can be applied, In principle , to find the solution of
any linear non-homogenous partial differential equation . Although a neat
formula (17) has been for solution of non-homogenous PDE, in practice , it
is not any easy task to construct the Green’s function . We shall now present
a few singularity solutions. Also called fundamental solutions to the well-
known operators. These will guide us to construct Green’s function for the
solutions of partial differential equations which occurs most frequently in
mathematical physics.
To start with, the fundamental solution for a three-dimensional potential
problem satisfies
δ (X )
∇ 2u =
u (18)
or
div grad u = δ ( X )
 Green’s Function 395

where u can be interpreted, for example, as the electrostatic potential. We

r=X ;
seek a solution which depends only on the source distance thus, for
r > 0, u ( r )
satisfies
1 ∂  2 ∂u 
=∇2 u 2
= r  0
r ∂r  ∂r 
On integration, we get
A
u= +B
r
Using the fact that the potential vanishes at infinity ,we have
A
u=
r
Integrating Eq. (18) the potential vanishes at infinity ,we have
.

∫ [ div grad u ] dV = 1
Rε

Using the divergence theorem, we obtain

.
∂u
∫
σ ∂r r =ε
dS = 1
ε

Where
σ ò is the surface of Rε .Hence ,

Therefore,
1
A= −
4π
2
0 is
Thus, the singularity solution or the fundamental solution of ∇ u =
1
u= −
4π r (19)
The two-dimensional case of Eq. (18) is
 396 Differential Equations: Theory and Applications

1 ∂  ∂u 
(r  =  0, r > 0
r ∂r  ∂r 
On integration, we get
=u A In r + B

Integrating Eq. (18) over a disc

Rò of small radius ò, whose surface is
σ ò , we get
.

∫ div grad u = 1
Rε

Hence,
.
∂u A A
∫
σ ∂r r =ò
dS =
∫
rò
dS =
ò
×2π ò =
1
ε

1
A= .
Therefore, 2π The constant B remains arbitrary and can be set
equal to zero for convenience Thus, the fundamental solution is
1
u (r ) = In r
2π (20)
if r ( x, y , z ) and r ′ ( x′, y′, z ′ )
If are two distinct points in three-
dimensional space IR, then the singularity solution of Laplace equation is
1
u=
4π r − r ′
(21)
Similarly, the singularity solution for the diffusion equation
ut − k ∇ 2 u =0
In three-space variables assumes the form

1  − r − r′ 2 
exp  
 4k ( t − τ ) 
3/2
8 π k ( t − τ ) 
 
(22)
For Helmholtz equation in three space variables, viz.
 Green’s Function 397

∇2 u + k 2 u =
0
The singularity solution is
ik r − r
e
r − r′
(23)
Before we attempt to solve Eq.(17) we shall examine the form of three-
dimensional δ − function in genral curvilinear coordinates as a preparation
to study the solution of partial differential equations in polar coordinates,
sphereical polar coordinates etc. suppose we are looking for a transformation
form Cartesian coordinates , x, y , to curvilinear coordinates ξ ,η through
the relations
f ( ξ ,η ) , y = g (ξ ,η )
x= (24)
where f and g are single-valued, continuously differentiable functions of
their arguments. Suppose that under this transformation,
=ξ β=
1 and η β2

correspond to
x = α1 and y = α 2 respectively Also,

 dx   fξ fη   d ξ  ∂ ( f , g )  dξ 
=
  g g =   = j 
 dy   ξ η   dη  ∂ (ξ ,η )  dη 
If we transform the coordinates following Eq. (24) , the relation
∫ ∫ φ ( x, y ) δ ( x − α1 ) δ ( y − α 2 ) dx dy =φ (α1 ,α 2 )

Becomes
∫ ∫ φ ( f , g ) δ  ( ξ ,η ) − a1  δ  g (ξ ,η ) – a 2  J= dη
= φ ( α1 , α 2 )
(25)
where j is the Jacobian of the transformation Eq. (25) states that the Dirac
δ -function δ  f ( ξ ,η ) − a1  ∂  g (ξ ,η ) − a2  j assign to any test function
φ( f , g) = f a=1 ,g a2 , i.e.,
the value of that test function at the point where
at the points
=ξ β=
1 ,η β . Thus we may write
δ  f ( ξ ,η ) − a1  δ  g (ξ ,η ) − a2  j =δ ( x − a1 ) δ ( y − a2 ) j =δ ( ξ − β1 ) δ (η − β 2 )

Hence
398 Differential Equations: Theory and Applications

δ (ξ − β1 ) δ (η − β 2 )
δ ( x − a1 ) δ ( y − a2 ) =
j
(26)
In the next few sections, we shall discuss the Green’s function method
for solving partial differential equations with particular emphasis on elliptic
equations. The discussion on wave equation and heat equation is also
included in respectively .

8.1.1. Green’s Function for Laplace Equation

To find analytic solution of the boundary value problems , the Green’
function method of the convenient technique. In this section, we shall give
the definition of Green’s function for the Laplace equation and study its
basic properties . To begin with, we shall define Green’s function for the
2
Dirichlet problem, i.e., we shall find u such that ∇ u = 0 is valid inside
some finitely bounded region IR, enclosed by ∂IR, as sufficiently smooth
surface , when u = f is prescribed on the boundary ∂IR .
Suppose that u is known at every poibt of the boundary ∂IR and that it
satisfies the relation
∇2 u =
0 in IR
u ( p ) when P ∈ IR
The task is to find . Let = r , and let c be a sphere
with centre at p and radius . ò. Also , let ∑ be denoted by ∂ ∑ , and
1
u′ =
r − r′
(27)
where r ′ is another point Q either in ∑ or on the boundary ∂ ∑ . if u
∂
and ∂n are twice continuously differentiable functions in IR and have first
order derivatives on ∂IR, then by Green ‘s theorem the region IR we have ,
form Eq. (19) , the relation
. . .
 ∂u ' ′ ∂u 
( )
∫ ∫ ∫ u ∇ 2u ′ − u ′ ∇ 2 u dV =
∫∫ u
∂n
−u
∂n
 dS
IR ∂ IR   (28)
 Green’s Function 399

Here, n I the unit vector normal to dSIR drawn outwards from and
∂
∇ 2u ′ = 0 within
∂n denotes differentiation in that direction. Since ∇ 2 u =
∂ ∑ , we have , in the region ∑ , the relation
. . .
 ∂u ′ ∂u   ∂u ′ ′ ∂u 
∫C  u ∂n − u′ ∂n  dS ′ + ∫∫
∂ IR
u
 ∂n
−u
∂n 
0
 dS =

Inserting u ' form relation (27), the above equation reduces to

.  ∂  1 1 ∂u  
∫ ∫  u ( r ′ )  −  ( r ′ )  dS ′
C   ∂n  r − r ' r − r ' ∂n   .
. . 
∂  1  1 ∂u 
+ ∫∫  u ( r )   − ( r ) dS =
0
∂ IR 
 ∂n  r − r  r − r ∂n  . (29)
When Q is on C , we have

Also, dS ′, the surface element on C , is ò sin θ dθ dφ . d on C ,

2

u (=
r ′ ) u ( r ) + du

or
Therefore,

Now,
400 Differential Equations: Theory and Applications

Employing these results and taking the limit as Eq. (29) becomes
. .  
∂  1  1 ∂
4π u ( r ) + ∫∫ u ( r )   − u ( r ′ )  dS =
0
∂ IR 
 ∂n  r − r ′  r − r ' ∂n 

Implying thereby
1
. .  1 ∂u 
∂  1
u (r ) ∫∫  ( r ) − u ( r )   dS

4π  r − r ' ∂n
∂ IR 
∂n  r − r '
  (30)
Therefore , the value of at an point of the region IR is determined , if
∂u
u and ∂n are known on the boundary , ∂IR . This lead to the conclusion
∂u
that both the values of u and ∂n are required solution to obtain the solution
of Dirichlet’ s problem. But this is not so, as can be seen from the concept
H ( r, r )
of the Green’s function defined as follows : Let be a function h u
armonic in IR. Then the Green’s function for the Dirichlet problem involving
the Laplace operator is defined by are the G, the two point function for
the Dirichlet problem involving the Laplace operator is defined by point
function of position, as
1
G (=
r, r′) + H ( r , r ')
r −r'
(31)
H ( r, r′)
where satisfies the following
 Green’s Function 401

 ∂2 ∂2 ∂2 
 2 + + 2 
H ( r, r′) =
0
(i )  ∂x ∂y 2
∂z  .
1
G= + H ( r, r ) =
0 on ∂IR
( ii ) r −r'
. (33)
Thus the Green’s function for the Dirichlet problem involving the
Laplace operator is a function
G ( r, r )
which satisfies the following properties:
(i ) ∇ G ( r, r ) =
2
δ ( r − r ′ ) in IR
(34)
( ii ) G ( r=
, r ) 0 on ∂IR
(35)
( iii ) G is symmetric, i.e.,
(36)
( iv ) G is continuous , but ∂G / ∂n . has . dcontinuity at the point r. which
is given by the equation
.
∂G
Ltò → 0 ∫ ∫ dS = 1
∂n
C (37)
Following the procedure adopted in the derivation of Eq. (30), replacing
u ′ by G ( r , r ) ,
we can show that
. .
1  ∂u ∂G
u (r ) =
− ∫∫  G (r , r ) ( r ) −u ( r ) ( r , r ) dS
4π ∂ IR  ∂n ∂n 
. (38)
From Eq. (31) and (33) we can see that= G 0 on ∂IR. . Thus the solution
u . at an interior point is given by the relation
. .
1 ∂G
u (r ) = − ∫∫ u ( r ) ∂n ( r , r ′) dS
4π ∂ IR (39)
And therefore, the solution of the interior Dirichlet ‘ s problem is reduced
G ( r, r′) .
to the determination of Green’s function
402 Differential Equations: Theory and Applications

Green’s function can be interpreted physically as follows: Let ∂IR be

a grounded electrical conductor ( boundary potential zero ) and if a unit
charge is located athe source point r, then G . is the sum of the potential at
the point r ' . due to the charge at the source point r in the free space and the
potential due to the charges induced on ∂IR. Thus
1
G ( r, r ) = + H ( r, r′)
r, r '
(40)

Hence, property
( i ) , viz, . Eq. (34) , essentially mns that ∇ 2G =
0
everywhere except at the source point (r) ..
Example. Consider a sphere with centre at the origin and radius ‘a’.
Apply the divergence theorem to the sphere and show that
1
−4πδ ( r )
∇2   =
r ..
δ (r )
is a Dirac delta function.
Solution . Applying the divergence theorem to
1 1
grad   = ∇  
r r .
t
. .
1 1
∫ ∫ ∫ ∇.∇   dV = ∫ ∫ ∇  r  .nˆ dS
V r S

u = u ( r ,θ , φ ) ,
where nˆ is an outward drawn normal . if then
∂u 1 ∂u 1 ∂u
eˆr
grad u = + eˆθ + eθ sin θ
∂r r ∂θ r ∂φ
. .
1 1 1
∫S ∫ grad  r  . eˆr dS =
∫S ∫ r 2 dS =
− 2 ×4π a 2 =
a
−4π

1
∇2  
Thus, we observe that  r  has the following properties :
 Green’s Function 403

( i ) it is undefined at the origin .

( ii ) it vanishes if r ≠ 0.
( iii )
its integral over any sphere with centre at the origin is – 4π . Hence,
we conclude that
.
1
∫ ∫ ∫ ∇ 2   dV = −4πδ ( r )
V r
Now, we shall prove the symmetric property through the following
theorem.

8.2. THE EIGENFUNCTION METHOD

Let us consider the Dirichlet boundary value problem described by
∇2 u =f (41)
Valid in certain region IR, subject to the boundary condition
u = g (42)

On ∂ IR, the boundary of IR.

From the definition of Green’s function which has been introduced in
The Green’s function must satisfy the relations
∂ ( x − ξ , y − η ) in IR
∇2 G =
(43)
=G 0 , on ∂IR (44)
2
Now, consider the eigenvalue problem associated with the operator ∇
in the domain IR,
∇ 2 φ + λφ =i
0 nIR (45)
φ 0 on ∂IR (46)
=
λ φ
Let mn be the eigenvalue and mn be the corresponding eigen functions.
Suppose we give Fourier series expansion to G and δ in terms of the eigen
functions
φmn in the following form:
404 Differential Equations: Theory and Applications

. .
G ( x, y ; ξ ,η ) = ∑∑ amn ( ξ ,η ) φmn ( x, y )
m n (47)
. .
δ ( x − ξ , y −η ) =
∑∑ bmn ( ξ ,η ) φmn ( x, y )
m n (48)
where
1
.
φmn ( ξ ,η )
b=
mn
φmn
2 ∫ ∫ δ ( x − ξ , y − η )φmn ( x, y ) dx =
IR
dy
φmn 2
(49)
.
φmn 2= ∫ ∫φ
2
mn dx dy
IR

Now, substituting Eq. (47) and ( (48) into Eq. (43) and (44) and noting
that Eq. (45) has the form

(50)
We obtain
. . . .
∇ 2 ∑∑amn (ξ ,η ) φmn ( x, y ) =
∑∑ bmn ( ξ ,η )φmn ( x, y )
m n m n

Using Eq. ( 49) and (50) , the above equation reduce to

. . . .
φ ( ξ ,η )φmn ( x, y )
− ∑∑ λmn amn (ξ ,η ) φmn ( x, y ) =
∑∑ mn 2
m n m n φmn
.
From which
φmn ( ξ ,η )
amn ( ξ ,η ) = − 2
λmn φmn
(51)
Hence, Eq.(47) and (51) give the required Green’s function for the
Dirichlet problem, in the form
. .

G ( x, y; ξ ,η ) = −
∑ ∑ φ (ξ ,η ) φ ( x, y ) m n mn mn
2
λmn φmn
(52)

Example. find the Green’s function f

0 ≤ y ≤ b, ∇ 2 + λ u =0 in IR
or
( )
 Green’s Function 405

theirichlet problem on the rectangle IR : 0 ≤ x ≤ a, described by the PDE

(53)
u 0 on ∂ IR.
And the BC=
Solution. The eigen functios of the given PDE can be obtained easily
by using the variables separable method. Let us assume the solution o the
u ( x, y ) = X ( x ) Y ( y ) .
given PDE in the form Substituting into the given
PDE, we obtain
X ′′  Y ′′ 
= −  + λ  = −v
X Y  ( a separation constant ) (54)
Since u is zero on the boundary ∂ IR , X satisfies
X ′′ +
= ( 0 ) X=
vX 0, X= (a) 0 (55)
Its solution, in general , is
X ( x ) A cos v x + B sin vx
=

nπ
X (0) = 0 = A 0.= X (a) 0 = sin va 0,= implying v .
implies gives a
The corresponding real valued eigen functions are
 nπ x 
X n sin 
= = , n 1 , 2 ,… while
 a  the eigen values are
2 π
2
=
vn n , n 1 , 2 ,…
=
a Now, the factor Y satisfies
Y ′′ + ( λ − vn ) Y =
0, Y ( 0 ) = Y ( b ) = 0

Following the above procedure, we can show at once that the eigen functions
are given by
mπ y
Y=
m sin , m
= 1, 2 …,
b
And the corresponding eigen values are
m 2π 2
λ − vn = , m = 1, 2 ,…
b2
406 Differential Equations: Theory and Applications

Thus, we obtain the eigen functions to the given problem in the form
mπ x nπ y
φmn ( x, y ) = sin sin , m = 1, 2,…; n = 1 , 2,..
a b (56)
while the eigen values are given by
m 2π 2 n 2π 2 2m n2 
2
λmn − 2 + 2 + π  2 + 2 
a b a b 
(57)
φmn
Computation of gives
mπ x nπ x
ab
2 ab
φmn ∫∫=
2 2
sin sin dx dy
a b 4
00 (58)
Hence , the Green’s function for the given Dirichlet problem can be
obtained with the help of Eq. (56) and (57) as
 mπ x  nπ x mπξ nπη
sin   sin sin sin
4ab ∞
 a 
∞
b a b
G ( x , y ,ξ ,η ) = − 2 ∑∑
π m= 1=n 1 2 2
m b +n a 2 2
(59)

8.3. SUMMARY AND DISCUSSION

Consider the differential equation
Lu ( x ) = f ( x )
(1)
f ( x)
Where L is an ordinary linear differential operator, is a known
u ( x)
function, while is a unknown function. To solve the above differential
−1
equation, one method is to find the operator L in the form of an integral
G ( x, ξ )
operator with a kernel such that
u ( x ) = L−1 f ( x ) = ∫ G ( x, ξ ) f (ξ ) d ξ
(2)
The kernel of this integral operator is called Green’s function for the
differential operator .
φ ( x ) of
Definition . Let us consider an auxiliary function a real
variable x which possess derivative of all orders and vanishes outside a
finite interval. Such function are called test functions.
 Green’s Function 407

The idea of Green’s function can be easily extended to higher dimension

. Thus the Green’s function technique can be applied, In principle , to find
the solution of any linear non-homogenous partial differential equation
. Green’s function can be applied for the solutions of partial differential
equations which occurs most frequently in mathematical physics.
Green’s function for the Dirichlet problem involving the Laplace
operator is a function
G ( r, r )
which satisfies the following properties:
(i ) ∇ G ( r, r ) =
2
δ ( r − r ′ ) in IR

( ii ) G ( r=
, r ) 0 on ∂IR
.
( iii ) G is symmetric, i.e., G ( r , r ′ ) = G ( r ′, r )
( iv )G is continuous , but ∂G / ∂n has a discontinuity at the point r.
.
∂G
Ltò → 0 ∫ ∫ 1
dS =
∂n
which is given by the equation C .
408 Differential Equations: Theory and Applications

REFERENCES
1. C.W. Gear Numerical Initial Value Problems in ordinary Differential
Equation, Prentice-Hall, 1971.
2. E Hairer , S. P. Norsett and G Wanner , Solving Ordinary Differential
Equations I, Springer-Verlag, 1987.
3. Differential equation with boundary value Problem, D . G Zill and M.
R. Cullen, Brooks/Cole, USA
4. Paul Blanchard, Robert L. Devaney, Glenn R. Hall, Differential
Equations (Preliminary Edition),PWS Publishing, Boston, 1996.
5. William H. Boyce and Richard C. Diprima, Elementary Differential
Equations and Boundary Value Problems (6th Edition), Wiley, New
York, 1996.
6. C.H. Edwards, Jr., David E. Penney, Elementary Dfifferential Equations
with Applications (Third Edition), Prentice-Hall, Englewood Cliffs,
NJ, 1996.
7. R. Kent Nagle and Edward B. Saff, Fundamentals of Differential
Equations (Third Edition), Addison Wesley, Reading, MA, 1993.
 INDEX

A Constant temperature 345

Across function 382, 387 D
Arbitrary function 20
Dependent function 112, 141
B Differential equation 2, 3, 4, 5, 6,
7, 10, 13, 16, 19, 23, 24, 29,
Boundary condition 196, 197, 205,
31, 41, 42, 44, 47, 50, 51, 52,
213, 215, 260, 262, 263, 265,
53, 57, 60, 65, 77, 91, 92, 94,
268, 270
95, 97, 98, 99, 100, 101, 102,
Boundary curve 131, 137
103, 104, 108
Boundary surface 304, 305, 325,
Diffusion Equation 332, 335
352, 353, 354
Diffusion phenomena 302, 352
Boundary value problems (BVPs)
Distance Dependent Acceleration
195
153, 187
C
E
Canonical equation 117, 119, 120,
Electrical engineering 358, 384
121, 122, 123, 124, 125, 127,
Electric field 283
128
Elliptic parabolic 113, 141
Constant Acceleration 144
Exact equation 19, 20, 25, 27, 28
Constant horizontal velocity 161,
187
410 Differential Equations: Theory and Applications

F Laplace transform 358, 360, 361,

362, 363, 364, 365, 366, 367,
Flexible string 280
369, 370, 371, 373, 374, 375,
G 376, 377, 378, 380, 382, 384,
385, 388
Gravity 145, 161, 186, 187
Laplace transform technique 358,
H 376, 384
Legendre polynomial 244, 245, 258
Harmonic function 199, 204, 273, Linear combination 305, 353
275, 276 Linear homogeneous equation 210,
Heat conduction equation 304, 317, 276
319, 335, 342, 352
Homogeneous 13, 14, 15, 16, 17, M
24, 26, 51, 56, 61, 85, 87, 91,
Magnetic potential 192
92, 100, 101, 104, 106
Mathematical formulation 5, 99
Homogenous hyperbolic 280, 297
Mathematical tool 197, 358, 384
Horizontal range 166, 182, 184,
Monochromatic wave 287, 298
188
Horizontal velocity 166, 177, 188 N
Hyperbolic equation 117, 132, 138
Normal direction 196
Hyperbolic system 132
Notion mechanic 310
I
O
Inclined plane 167, 168, 182
One-dimensional Dirac delta func-
Independent variable 112, 113, 115,
tion 313, 355
122, 141, 142
One-dimensional wave 284, 285,
Initial boundary value 304, 310,
286, 289, 290, 298, 299
320, 321, 352
Ordinary calculator 358
Initial boundary value problem
Ordinary differential equation
(IBVP) 310
(ODE) 27, 102
Initial temperature 334, 338, 343,
Original equation 114
345, 347
Integrate function 122 P
L Parabolic equation 118
Parabolic of hyperbolic 113, 142
Laplace equation 194, 196, 199,
Partial derivative 112, 133, 141
205, 206, 207, 208, 212, 215,
Partial differential equation 112,
225, 229, 231, 235, 241, 243,
130, 132, 137, 141
249, 253, 256, 257, 264, 271,
Physical phenomena 192, 276
276, 278
 Index 411

Physical situation 192 T

Plane harmonic wave 287, 298
Temperature distribution 225, 231,
Potential 192, 193, 194, 196, 222,
234, 241, 250, 259, 261
234, 235, 238, 243, 245, 246,
Time Dependent Acceleration 148,
247, 249, 252, 253, 255, 256,
187
264, 265, 276, 277
Transformation 113, 114, 115, 119,
Prismatic beam 192
120, 121, 122, 123, 126, 128,
Probability theorem 358, 384
142
Problem under investigation 304,
Transformed equation 113, 114,
353
119
R Transformed equation assume 113
Tthermal conductivity 302, 345,
Riemann-Green function 136
352
S
U
Several problem 358, 384
Uniform cross-section 321
Spatial variable 280, 298
Specific heat 302, 352 V
Steady temperature distribution
Velocity Dependent Acceleration
222, 277
150, 187
Subsequent analysis 113
Supplementary boundary 7, 100
Surrounding medium 196, 197