NİŞANTAŞI ÜNİVERSİTESİ

ECV221 DİFFERANTİAL EQUATİONS

THURSDAY: 10:00-12:50
Asst. Prof. Dr. Merve Temizer Ersoy
e-mail:merve.temizerersoy@nisantası.edu.tr

Resource Book: umd_justin.pdf-https://www.math.umd.edu/~immortal/MATH246/

Contents

1 DIFFERENTIAL EQUATIONS and THEIR SOLUTIONS 2

1.1 CLASSIFICATION OF DIFFERENTIAL EQUATIONS; THEIR ORI-
GIN AND APPLICATION . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 SOLUTIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.3 INITIAL VALUE PROBLEMS, BOUNDARY VALUE PROBLEMS,
AND EXISTENCE OF SOLUTIONS . . . . . . . . . . . . . . . . . . . 6

2 FIRST ORDER EQUATIONS FOR WHICH EXACT SOLUTIONS ARE

OBTAINABLE 8
2.1 EXACT DIFFERENTIAL EQUATIONS AND INTEGRATING FAC-
TORS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2 SEPARABLE EQUATIONS AND EQUATION REDUCIBLE TO THIS
FORM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.3 LINEAR EQUATIONS AND BERNOULLI EQUATIONS . . . . . . . . 18
2.4 SPECIAL INTEGRATING FACTORS AND TRANSFORMATIONS . 25

3 APPLICATIONS OF FIRST ORDER EQUATIONS 32

3.1 ORTHOGONAL AND OBLIQUE TRAJECTORIES . . . . . . . . . . . 32
3.2 PROBLEMS IN MECHANICS . . . . . . . . . . . . . . . . . . . . . . . 35

4 EXPLICIT METHODS OF SOLVING HIGHER-ORDER LINEAR DIF-

FERENTIAL EQUATIONS 39
4.1 BASIC THEORY OF LINEAR DIFFERENTIAL EQUATIONS . . . . 39
4.2 THE HOMOGENEOUS LINEAR EQUATION WITH CONSTANTS
COEFFICIENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
4.3 THE METHOD OF UNDETERMINED COEFFICIENTS . . . . . . . 56
4.4 VARIATION OF PARAMETERS . . . . . . . . . . . . . . . . . . . . . 63
4.5 THE CAUCHY-EULER EQUATION . . . . . . . . . . . . . . . . . . . 70

5 APPLICATIONS OF SECOND ORDER LINEAR DIFFERENTIAL EQUA-

TIONS WITH CONSTANT COEFFICIENTS 73
5.1 THE DIFFERENTIAL EQUATION OF THE VIBRATIONS OF A
MASS ON A SPRING . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73

i
 1

5.2 FREE, UNDAMPED MOTION . . . . . . . . . . . . . . . . . . . . . . 73

6 THE LAPLACE TRANSFORM 75

6.1 DEFINITION, EXISTENCE, AND BASIC PROPERTIES OF THE
LAPLACE TRANSFORM . . . . . . . . . . . . . . . . . . . . . . . . . 75
6.2 THE INVERSE TRANSFORM AND THE CONVOLUTION . . . . . . 77
6.3 LAPLACE TRANSFORM SOLUTION OF LINEAR DIFFERENTIAL
EQUATIONS WITH CONSTANT COEFFICIENTS . . . . . . . . . . . 81

7 SERIES SOLUTIONS OF LINEAR DIFFERENTIAL EQUATIONS 86

7.1 POWER SERIES SOLUTIONS ABOUT AN ORDINARY POINT . . . 86
Chapter 1

DIFFERENTIAL EQUATIONS
and THEIR SOLUTIONS

1.1 CLASSIFICATION OF DIFFERENTIAL EQUATIONS; THEIR ORI-

GIN AND APPLICATION

A. Differential Equations and Their Classification

Definition 1. A differential equation involving ordinary derivatives of one or more depen-

dent variables with respect to a single independent variable is called an ordinary differential
equation.

Example 1.1.
2
d2 y

dy
+ xy = 0; y − dependent; x − independent (1.1)
dx2 dx

d4 x d2 x
+ 5 + 3x = sin t; x − dependent; t − independent (1.2)
dt2 dt2
Definition 2. A differential equation involving partial derivatives of one or more dependent
variables with respect to more than one independent variable is called a partial differential
equation.

Example 1.2.
∂v ∂v
+ = v; v − dependent; t, s − independent (1.3)
∂s ∂t

∂2u ∂2u ∂2u

+ 2 + 2 = 0; u − dependent; x, y, z − independent (1.4)
∂x2 ∂y ∂z
Definition 3. The order of the highest ordered derivative involved in a differential equation is
called the order of the differential equation.

Example 1.3. The ordinary differential equation (1.1) is of the second order, since the highest
derivative involved is a second derivative. Equation (1.2) is an ordinary differential equation
of the fourth order. The partial differential equations (1.3) and (1.4) are of the first and
second orders, respectively.

2
 3

Definition 4. A linear ordinary differential equation of order n, in the dependent variable y

and the in dependent variable x, is an equation that is in, or can be expressed in the form
dn y dn−1 y dy
a0 (x) + a1 (x) + · · · + an−1 (x) + an (x) y = b (x) ,
dxn dxn−1 dx
where a0 is not identically zero.
Example 1.4. The following ordinary differential equations are both linear.
d2 y dy
2
+5 + 6y = 0 (1.5)
dx dx

d4 y 3
2d y dy
4
+ x 3
+ x3 = xex (1.6)
dx dx dx
Definition 5. A nonlinear ordinary differential equation is an ordinary differential equation
that is not linear.
Example 1.5. The following ordinary differential equations are all nonlinear:
d2 y dy
2
+5 + 6y 2 = 0 (1.7)
dx dx
3
d2 y

dy
+5 + 6y = 0 (1.8)
dx2 dx
d2 y dy
+ 5y + 6y = 0 (1.9)
dx2 dx

B. Origin and Application of Differential Equations

Differential equations occur in connecting with numerous problems that are encountered in
the various branches of science and engineering. We indicate a few such problems in the
following list, which could easily be extended to fill many pages.
1. The problem of determining the motion of a projectile, rocket, satellite or planet.
2. The problem of determining the charge or current in an electric circuit.
3. The problem of the conduction of heat in a rod or in a slab.
4. The problem of determining the vibrations of a wire or a membrane.
5. The study of the rate of decomposition of a radioactive substance or the rate of growth
of a population.
6. The study of reactions of chemicals.
7. The problem of determination of curves that have certain geometrical properties.
The mathematical formulation of such problems give rise to differential equations.

1.2 SOLUTIONS

A. Nature of Solutions

Definition 6. Consider the n-th order ordinary differential equation

dn y
 
dy
F x, y, , · · ·, n = 0, (1.10)
dx dx
 4

dy dn y
where F is a real function of its (n + 2) arguments x, y, , · · ·, n .
dx dx
1. Let f be a real function defined for all x in a real interval I and having an n-th deriva-
tives. The function f is called an explicit solution of (1.10) on I if the following statements
hold:
dn f
 
df
a-) F x, f, , · · ·, n is defined for all x ∈ I.
dx dx
dn f
 
df
b-) F x, f, , · · ·, n = 0 for all x ∈ I.
dx dx
2. A relation g (x, y) = 0 is called an implicit solution of (1.10) if this relation defines at
least one real function f of the variable x such that this function is an explicit solution of
(1.10) on the interval I.
3. Both explicit solutions and implicit solutions will usually be called simply solutions.
Example 1.6. The function f defined for all real x by

f (x) = 2 sin x + 3 cos x (1.11)

is an explicit solution of the differential equation

d2 y
+y =0 (1.12)
dx2
for all real x. First note that f is defined and has a second derivative for all real x. Next
observe that
f 0 (x) = 2 cos x − 3 sin x,
f 00 (x) = −2 sin x − 3 cos x.
Putting last equation into the (1.12), we get the next identity

(−2 sin x − 3 cos x) + (2 sin x + 3 cos x) = 0,

which holds for all real x so that (1.11) is called an explicit solution of the differential equation
(1.12).
Example 1.7. The relation
x2 + y 2 − 25 = 0 (1.13)
is an implicit solution of the differential equation
dy
x+y =0 (1.14)
dx
on the interval I defined by −5 < x < 5. Solving (1.13) for y, we obtain
p p
f1 (x) = 25 − x2 and f2 (x) = − 25 − x2 ,

and both of these functions are explicit solutions of the differential equation (1.14). Since
−x
f10 (x) = √
25 − x2
for all x ∈ I, we observe that
 
p −x
x+ 25 − x2 √ = x − x = 0,
25 − x2
 5

which is identity. Thus the function f1 is an explicit solution of (1.14) on the interval I =
(−5, 5) .
Now consider the relation
x2 + y 2 + 25 = 0. (1.15)
d d
x2 + y 2 + 25 =


(0)
dx dx
dy dy x
2x + 2y =0⇒ =−
dx dx y
Substituting this into the differential equation (1.14), we obtain the formal identity
 
x
x+y − = 0.
y
Although (1.15) seems to be an implicit solution of (1.14), it is not. Because, solving (1.15)
for y, we find that p
y = ± −25 − x2 .
Since this expression yields non-real values of y for all real values of x, we conclude that the
relation (1.15) does not define any real function on any interval. Thus the relation (1.15) is
not truly an implicit solution but merely a formal solution of the differential equation (1.14).
Example 1.8. Consider the first order differential equation
dy
= 2x. (1.16)
dx
The function f0 defined for all real x by f0 (x) = x2 is a solution of (1.16). Consider f1 (x) =
x2 + 1, f2 (x) = x2 + 2, f3 (x) = x2 + 3 are also solutions of (1.16). Actually, for each real
number c, the function fc defined for all real x by
fc (x) = x2 + c
is a solution of (1.16), which is called one-parameter family of solutions.
We know consider the geometric significance of differential equations and their solutions.
We first recall that a real function F may be represented geometrically by a curve y = F (x)
in the xy plane and that the value of the derivative of F at x. F 0 (x) , may be interpreted as
the slope of the curve y = F (x) at x. Thus the general first order differential equation
dy
= f (x, y) , (1.17)
dx
where f is a real function, may be interpreted geometrically as defining a slope f (x, y) at every
point (x, y) at which the function f is defined. Now assume that the differential equation (1.17)
has a so-called one-parameter family of solutions that can be written in the form
y = F (x, c) ,
where c is the arbitrary constant or parameter of the family.

B. Methods of Solutions

When we sat that we will solve a differential equation we mean that we will find one or more
of its solutions. How is this done and what does it really mean? The greater part of this
text is concerned with various methods of solving differential equations. The method to be
employed depends upon the type of differential equation under consideration, and we will not
enter into the details of specific method here.
 6

1.3 INITIAL VALUE PROBLEMS, BOUNDARY VALUE PROBLEMS,

AND EXISTENCE OF SOLUTIONS

A. Initial Value Problems and Boundary Value Problems

Example 1.9. Find a solution f of the differential equation

dy
= 2x
dx
such that x = 1 this solution f has the value 4.

Solution: First, we must under the following statements:

1. f is continuously differentiable, that is, f 0 (x) = 2x for all x.
2. f satisfies the initial condition, namely, f (1) = 4.
dy
= 2x ⇒ dy = 2xdx ⇒ y = x2 + c
dx
The last equation is called general solution. Using the initial condition, we get
y (1) = 1 + c = 4 ⇒ c = 3.
Therefore y = x2 + 3 is the exact solution of the given differential equation.

If all of the associated supplementary conditions relate to one x value, the problem is called
an initial value problem (or one-point boundary value problem). If the conditions relate to
two different x values, the problem is called two-point boundary value problem (or simply
boundary value problem).
Example 1.10. The problem
d2 y

+ y = 0, y (1) = 3, y 0 (1) = −4
dx2
is an initial value problem (IVP).
Example 1.11. The problem
d2 y
 π 
+ y = 0, y (0) = 1, y =5
dx2 2
is a boundary value problem (BVP).
Definition 7. Consider the first order differential equation
dy
= f (x, y) , (1.18)
dx
where f is a continuous function of x and y in some domain D of the xy plane; and let (x0 , y0 )
be a point of D. The initial value problem associated with (1.18) is to find a solution φ of
the differential equation (1.18), defined on some real interval containing x0 , and satisfying the
initial condition
φ (x0 ) = y0 .
In the customary abbreviated notation this initial value problem may be written
 dy
dx = f (x, y) ,
y (x0 ) = y0 .
 7

B. Existence of Solutions

Theorem 1.1. (Basic Existence and Uniqueness Theorem) Consider the differential
equation
dy
= f (x, y) . (1.19)
dx
∂f
Let f is a continuous function of x and y in some domain D and is also continuous in
∂y
D; and let (x0 , y0 ) be a point in D. Then, there exists a unique solution φ of the differential
equation (1.19), defined on some interval |x − x0 | ≤ h, where h is a sufficiently small, that
satisfies the condition φ (x0 ) = y0 .

Example 1.12. Consider the initial value problem

 dy 2 2
dx = x + y ,
y (1) = 3.

∂f
f (x, y) = x2 + y 2 is continuous for all x and y. = 2y is also continuous for all x and y.
∂y
The initial condition y (1) = 3 means that x0 = 1 and y0 = 3, and the point (1, 3) certainly
lies in some such domain D. So, all conditions of Theorem 1.1 are satisfied. That is, there
is unique solution φ of the given differential equation, defined on some interval |x − 1| ≤ h,
which satisfies the initial condition, namely, φ (1) = 3.

Example 1.13. Consider the two problems:

dy y
1. = √ , y (1) = 2,
dx x
dy y
2. = √ , y (0) = 2.
dx x
Here
y ∂f (x, y) 1
f (x, y) = √ and =√ .
x ∂y x
For the first problem the Theorem 1.1 is applicable since both functions above satisfy the all
conditions of theorem. But, for the second problem we cannot say same statement. Because
both functions are not continuous at the point x = 0, so that we cannot apply the Theorem
1.1.
Chapter 2

FIRST ORDER EQUATIONS FOR

WHICH EXACT SOLUTIONS
ARE OBTAINABLE

2.1 EXACT DIFFERENTIAL EQUATIONS AND INTEGRATING FAC-

TORS

A. Standard Forms of First Order Differential Equations

The first order differential equations to be studied in this chapter may be expressed in either
the derivative form
dy
= f (x, y) , (2.1)
dx
or the differential form
M (x, y) dx + N (x, y) dy = 0. (2.2)

B. Exact Differential Equations

Definition 1. Let F be a function of two variables such that F has continuous first partial
derivatives in a domain D. The total differential dF of the function F is defined by the formula
∂F (x, y) ∂F (x, y)
dF (x, y) = dx + dy
∂x ∂y
for all (x, y) ∈ D.
Example 2.1. Let F be a function of two real variables defined by
F (x, y) = xy 2 + 2x3 y
for all real (x, y) . Then
∂F (x, y) ∂F (x, y)
= y 2 + 6x2 y, = 2xy + 2x3 ,
∂x ∂y
and the total differential dF is defined by
dF (x, y) = y 2 + 6x2 y dx + 2xy + 2x3 dy

for all real (x, y) .

8
 9

Definition 2. The expression

M (x, y) dx + N (x, y) dy (2.3)
is called an exact differential in a domain D if there exists a function F of two real variables
such that this expression equals the total differential F (x, y) for all (x, y) ∈ D. That is,
expression (2.3) is an exact differential in D if there exists a function F such that

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

for all (x, y) ∈ D.

If M (x, y) dx + N (x, y) dy is an exact differential, then the differential equation

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

is called exact differential equation.

Example 2.2. The differential equation

y 2 dx + 2xydy = 0

is an exact differential equation. Moreover

∂F (x, y) ∂F (x, y)
= M (x, y) = y 2 , = N (x, y) = 2x.
∂x ∂y
Theorem 2.1. Consider the differential equation

M (x, y) dx + N (x, y) dy = 0, (2.4)

where M and N have continuous first partial derivatives at all points (x, y) in a rectangular
domain D.
1. If the differential equation (2.4) is exact in D, then

∂M (x, y) ∂N (x, y)
= (2.5)
∂y ∂x

for all (x, y) ∈ D.

2. Conversely, if
∂M (x, y) ∂N (x, y)
=
∂y ∂x
for all (x, y) ∈ D, then he differential equation (2.4) is exact in D.

Proof. Part 1. By definition we have

∂F (x, y) ∂F (x, y)
= M (x, y) and = N (x, y) .
∂x ∂y
Then
∂ 2 F (x, y) ∂M (x, y) ∂ 2 F (x, y) ∂N (x, y)
= and = .
∂y∂x ∂y ∂x∂y ∂x
Since
∂ 2 F (x, y) ∂ 2 F (x, y)
= ,
∂y∂x ∂x∂y
 10

we have that
∂M (x, y) ∂N (x, y)
= . N
∂y ∂x

Part 2. Now, we will do converse of what we did in the part 1. The hypothesis is

∂M (x, y) ∂N (x, y)
= . (2.6)
∂y ∂x

We need to show that M (x, y) dx + N (x, y) dy = 0 is an exact differential equation. This

means we must prove that there exists a function F such that

∂F (x, y)
= M (x, y) (2.7)
∂x
and
∂F (x, y)
= N (x, y) (2.8)
∂y
in some domain D. Assume that F (x, y) satisfies (2.7). Then
Z
F (x, y) = M (x, y) ∂x + φ (y) .

Taking the partial derivative of the last equation with respect y, we obtain
Z
∂F (x, y) ∂ dφ (y)
= M (x, y) ∂x + .
∂y ∂y dy

Now, if (2.8) is to be satisfied, we must have

Z
∂ dφ (y)
N (x, y) = M (x, y) ∂x +
∂y dy
or Z
dφ (y) ∂
= N (x, y) − M (x, y) ∂x.
dy ∂y
dφ (y)
Since φ is a function of y only, the derivative must also be independent of x. That is,
dy
Z
∂
N (x, y) − M (x, y) ∂x
∂y
must be independent of x. So, it is enough to prove that
 Z 
∂ ∂
N (x, y) − M (x, y) ∂x = 0.
∂x ∂y

We at once have
∂2
 Z  Z
∂ ∂ ∂N (x, y)
N (x, y) − M (x, y) ∂x = − M (x, y) ∂x.
∂x ∂y ∂x ∂x∂y

If (2.7) and (2.8) are to be satisfied, then using the hypothesis (2.6), we must have

∂2 ∂ 2 F (x, y) ∂ 2 F (x, y) ∂2
Z Z
M (x, y) ∂x = = = M (x, y) ∂x.
∂x∂y ∂x∂y ∂y∂x ∂y∂x
 11

Thus we obtain
∂2
 Z  Z
∂ ∂ ∂N (x, y)
N (x, y) − M (x, y) ∂x = − M (x, y) ∂x
∂x ∂y ∂x ∂y∂x

∂N (x, y) ∂M (x, y)
= − .
∂x ∂y
Using the hypothesis (2.6), we get
 Z 
∂ ∂
N (x, y) − M (x, y) ∂x = 0.
∂x ∂y

Thus we may write  Z 

∂
dφ (y) = N (x, y) − M (x, y) ∂x dy
∂y
or Z  Z 
∂
φ (y) = N (x, y) − M (x, y) ∂x dy.
∂y
Therefore Z Z  Z 
∂
F (x, y) = M (x, y) ∂x + N (x, y) − M (x, y) ∂x dy
∂y
This F (x, y) thus satisfies both (2.7) and (2.8) for all (x, y) ∈ D, and so M (x, y) dx +
N (x, y) dy = 0 is an exact differential equation.

C. The Solution of Exact Differential Equations

Theorem 2.2. Suppose that M (x, y) dx+N (x, y) dy is an exact differential equation in some
rectangular domain D. Then there exists a solution of this differential equation F (x, y) = c
such that
∂F (x, y) ∂F (x, y)
= M (x, y) and = N (x, y) ,
∂x ∂y
where c is an arbitrary constant.

Example 2.3. Solve the equation

3x2 + 4xy dx + 2x2 + 2y dy = 0.

(2.9)

Solution: Standard Method

We have that
M (x, y) = 3x2 + 4xy and N (x, y) = 2x2 + 2y.
Since
∂M (x, y) ∂N (x, y)
= 4x = ,
∂y ∂x
(2.9) is an exact differential equation.

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

∂F (x, y)
= 3x2 + 4xy ⇒ ∂F (x, y) = 3x2 + 4xy ∂x


∂x
 12

Z
3x2 + 4xy ∂x = x3 + 2x2 y + φ (y) .


F (x, y) =

∂F (x, y) ∂ dφ (y)
x3 + 2xy + φ (y) = 2x2 +


=
∂y ∂y dy
∂F (x, y) dφ (y)
= N (x, y) ⇒ 2x2 + = 2x2 + 2y
∂y dy
dφ (y)
⇒ = 2y ⇒ φ (y) = y 2 + c1
dy
Therefore
F (x, y) = x3 + 2x2 y + y 2 + c1
or
x3 + 2x2 y + y 2 = c
is the general solution of the given differential equation.
Method of Grouping:

3x2 + 4xy dx + 2x2 + 2y dy = 0

3x2 dx + 4xydx + 2x2 dy + 2ydy = 0

d x3 + d 2x2 y + d y 2 = d (c)

Taking the integration of both sides, we obtain

x3 + 2x2 y + y 2 = c.

Example 2.4. Solve the initial value problem

2x cos y + 3x2 y dx + x3 − x2 sin y − y dy = 0, y (0) = 2.

(2.10)

Solution: Standard Method

We have that

M (x, y) = 2x cos y + 3x2 y and N (x, y) = x3 − x2 sin y − y.

Since
∂M (x, y) ∂N (x, y)
= −2x sin y + 3x2 = ,
∂y ∂x
(2.10) is an exact differential equation.

∂F (x, y) ∂F (x, y)
= M (x, y) , = N (x, y)
∂x ∂y
∂F (x, y)
= 2x cos y + 3x2 y ⇒ ∂F (x, y) = 2x cos y + 3x2 y ∂x


∂x
Z
2x cos y + 3x2 y ∂x = x2 cos y + x3 y + φ (y) .


F (x, y) =

∂F (x, y) ∂ dφ (y)
x2 cos y + x3 y + φ (y) = −x2 sin y + x3 +


=
∂y ∂y dy
∂F (x, y) dφ (y)
= N (x, y) ⇒ −x2 sin y + x3 + = x3 − x2 sin y − y
∂y dy
 13

dφ (y) y2
⇒ = −y ⇒ φ (y) = − + c1
dy 2
Therefore
y2
F (x, y) = x2 cos y + x3 y − + c1 .
2
Using the initial condition y (0) = 2, we get

22
0+0− + c1 = 0 ⇒ c1 = −2.
2
So
y2
x2 cos y + x3 y − = −2
2
is the exact solution of the given differential equation.
Method of Grouping:

2x cos y + 3x2 y dx + x3 − x2 sin y − y dy = 0

2x cos ydx + 3x2 ydx + x3 dy − x2 sin ydy − ydy = 0

2x cos ydx − x2 sin ydy + 3x2 ydx + x3 dy − ydy = 0
 2
2

3

y
d x cos ydx + d x y dx + d − dy = d (c1 )
2
Taking the integration of both sides, we obtain

y2
x2 cos y + x3 y − = c1 .
2
Using the initial condition y (0) = 2, we get the same solution.

D. Integration Factors

∂M (x, y) ∂N (x, y)
Consider M (x, y) dx + N (x, y) dy = 0. What happens if 6= ?
∂y ∂x
Let us try to solve
ydx + 2xdy = 0. (2.11)
∂M (x, y) ∂N (x, y)
= 1 6= 2 =
∂y ∂x
So (2.11) is not exact differential equation. Multiplying both sides of (2.11) by y, we obtain

y 2 dx + 2xydy = 0. (2.12)

Now,
∂M (x, y) ∂N (x, y)
= 2y = ,
∂y ∂x
(2.12) is an exact differential equation. We call y an integrating factor of (2.11).
 14

2.2 SEPARABLE EQUATIONS AND EQUATION REDUCIBLE TO THIS

FORM

A. Separable Equations

Definition 3. An equation of the form

F (x) G (y) dx + f (x) g (y) dy = 0 (2.13)

is called separable equation.

In general separable equation is not exact. Why?

M (x, y) = F (x) G (y) ; N (x, y) = f (x) g (y)

∂M (x, y) ∂N (x, y)
= F (x) G0 (y) 6= f 0 (x) g (y) =
∂y ∂x
Dividing both sides by f (x) G (y) , we obtain

F (x) g (y)
dx + dy = 0. (2.14)
f (x) G (y)

Now, (2.14) is an exact, since

   
∂ F (x) ∂ g (y)
=0= .
∂y f (x) ∂x G (y)

Let
F (x) g (y)
M (x) = and N (y) = .
f (x) G (y)
Then Z Z
M (x) dx + N (y) dy = c

is the general solution of the differential equation (2.14). Here,

1
µ (x, y) =
f (x) G (y)

is the integrating factor of (2.13), where f (x) 6= 0 and G (y) 6= 0.

Example 2.5. Solve the equation

(x − 4) y 4 dx − x3 y 2 − 3 dy = 0.


(2.15)

Solution: Divide both sides by x3 y 4

x3 y 2 − 3


(x − 4) y 4
dx − dy = 0,
x3 y 4 x3 y 4

where x3 6= 0 and y 4 6= 0.
x−4 y2 − 3
dx − dy = d (c)
x3 y4
Z Z Z
−2 −3 −2 −3



x − 4x dx − y − 3y dy = d (c)
 15

1 2 1 1
∴− + 2 + − 3 =c
x x y y
is the general solution of the given differential equation. Now, consider y = 0 ⇒ y 4 = 0. It is
not a member of the one-parameter family of solutions which we obtained. However, writing
the original differential equation of the problem in the derivative form
dy (x − 4) y 4
= 3 2 .
dx x (y − 3)
It is obvious that y = 0 is a solution of the original equation. We conclude that it is a solution
which was lost in the separation process.
Example 2.6. Solve the initial value problem
π
x sin ydx + x2 + 1 cos ydy = 0, y (1) = .


2

Solution: Divide both sides by x2 + 1 sin y

x2 + 1 cos y


x sin y
dx + 2 dy = 0,
(x2 + 1) sin y (x + 1) sin y
where x2 + 1 6= 0 and sin y 6= 0.

Z Z Z
xdx cos ydy
+ = d (c)
x2 + 1 sin y
Thus
1  2 
ln x + 1 + ln |sin y| = c
2
is the general solution of the given differential equation. Let us do more computations. Since
x2 + 1 > 0, we have that

ln x2 + 1 + 2 ln |sin y| = 2c ⇒ ln x2 + 1 + ln |sin y|2 = 2c

or
ln x2 + 1 + ln sin2 y = 2c ⇒ ln x2 + 1 sin2 y = 2c.

Then
eln(x ) sin2 y = e2c ⇒ x2 + 1
sin2 y = c ; 2c = c .
2 +1
1 1
π
Using the initial condition y (1) = , we get the exact solution of the given differential
2
equation
x2 + 1 sin2 y = 2.

Homogeneous Equation

Definition 4. The first order differential equation

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

is said to be homogeneous if, when written in the derivative form

dy
= f (x, y) ,
dx
y
there exists a function ϕ such that f (x, y) can be expressed in the form ϕ .
x
 16

Example 2.7. Consider the differential equation

x2 − 3y 2 dx + 2xydy = 0.


(2.16)

(2.16) is neither exact nor separable equation.

3y 2 − x2 3y 2 x2
 
dy 3y x 3 y 1 1
= = − = − = −
dx 2xy 2xy 2xy 2x 2y 2 x 2 y/x

Therefore (2.16) is a homogeneous equation.

A function F is called homogeneous of degree n if

F (tx, ty) = tn F (x, y) .

For example, F (x, y) = x2 + y 2 is homogeneous of degree 2, since

F (tx, ty) = (tx)2 + (ty)2 = t2 x2 + y 2 = F (x, y) .

Theorem 2.3. If
M (x, y) dx + N (x, y) dy = 0 (2.17)
is a homogeneous equation, then the change of variables y = vx transforms (2.17) into a
separable equation in the variables v and x.

Proof: Since (2.17) is homogeneous, it mat be written in the form

dy y
=ϕ .
dx x
Let y = vx. Then
dy dv
=v+x .
dx dx
dv dv
⇒v+x = g (v) ⇒ v − g (v) + x = 0 ⇒ [v − g (v)] dx + xdv = 0
dx dx
[v − g (v)]
Z Z Z
x dx dv
⇒ dx + dv = 0 ⇒ + = d (c)
x [v − g (v)] x [v − g (v)] x [v − g (v)]
R dv
Denoting F (v) = , we obtain
[v − g (v)]
y
F (v) + ln |x| = c ⇒ F + ln |x| = c. N
x
Example 2.8. Solve the equation

x2 − 3y 2 dx + 2xydy = 0.

Solution: We know that

 
dy 3 y 1 1
= − .
dx 2 x 2 y/x
y
Let y = vx ⇒ v = . Then
x
dy dv
=v+x .
dx dx
 17

dv 3v 1 dv v 1 dv v2 − 1
v+x = − ⇒x = − ⇒x =
dx 2 2v dx 2 2v dx 2v
Z Z
2vdv dx  2   2 
⇒ 2
= ⇒ ln v − 1 = ln |x| + ln |c| ⇒ v − 1 = |cx|
v −1 x
 2 
y
⇒  2 − 1 = |cx| ⇒ y 2 − x2  = |cx| x2
  

x
If y ≥ x ≥ 0, then this may be expressed somewhat more simply as

y 2 − x2 = cx3 .

Example 2.9. Solve the initial value problem

 p 
y + x2 + y 2 dx − xdy = 0, y (1) = 0.

Solution: We have seen that the differential equation is homogeneous. As in the last
example, we write in the form p
dy y + x2 + y 2
= .
dx x
√
Since the initial x value is 1, we consider x > 0 and take x = x2 and obtain
r
dy y  y 2
= + 1+ .
dx x x
Let y = vx. Then
dv p
v+x = v + 1 + v2
dx
or
dv p
x = 1 + v2.
dx
Separating variables, we find
dv dx
√ = .
1+v 2 x
Using tables, we perform the required integrations to obtain
 p 
ln v + v 2 + 1 = ln |x| + ln |c| ,
 

or p
v+ v 2 + 1 = cx.
y
Now replacing v by , we obtain the general solution of the differential equation in the form
x
r
y y2
+ + 1 = cx
x x2
or p
y+ x2 + y 2 = cx2 .
The initial conditions requires that y = 0 when x = 1. This gives c = 1 and hence
p
y + x2 + y 2 = x2 ,

from which it follows that

1 2

y= x −1 .
2
 18

Example 2.10. Solve the differential equation

 y 
x tan + y dx − xdy = 0.
x
Example 2.11. Solve the differential equation
 p   p 
x3 + y 2 x2 + y 2 dx − xy x2 + y 2 dy = 0.

Example 2.12. Solve the initial value problem

x2 + 3x2 dx − 2xydy = 0, y (2) = 6.

Example 2.13. Solve the initial value problem

3x2 + 9xy + 5y 2 dx − 6x2 + 4xy dy = 0, y (2) = −6.

2.3 LINEAR EQUATIONS AND BERNOULLI EQUATIONS

A. Linear Equations

Definition 5. A first order ordinary differential equation is called linear in the dependent
variable y and the independent variable x, if it can be written in the form
dy
+ P (x) y = Q (x) . (2.18)
dx

dy dy
+ P (x) y = Q (x) ⇒ + P (x) y − Q (x) = 0
dx dx

⇒ [P (x) y − Q (x)] dx + dy = 0 (2.19)

Here M (x, y) = P (x) y − Q (x) and N (x, y) = 1.

∂M (x, y) ∂N (x, y)
= P (x) ; =0
∂y ∂x

If P (x) 6= 0 (2.19) is not exact. Let µ (x) be an integrating factor of (2.19). Then

[µ (x) P (x) y − µ (x) Q (x)] dx + µ (x) dy = 0. (2.20)

Suppose (2.20) is exact differential equation. Then

∂ ∂
[µ (x) P (x) y − µ (x) Q (x)] = [µ (x)]
∂y ∂x
or
d
µ (x) P (x) = [µ (x)] .
dx
Solving it, we get Z Z
dµ dµ
µP (x) = ⇒ = P (x) dx
dx µ
Z R
⇒ ln |µ| = P (x) dx ⇒ µ = e P (x)dx ,
 19

R
where µ > 0. So, the integration factor is µ = e P (x)dx . Multiplying both sides by µ, we get

P (x)dx dy
R R R
P (x)dx P (x)dx
e +e P (x) y = e Q (x)
dx
or
d h R P (x)dx i R
e y = e P (x)dx Q (x) .
dx
A one-parameter family of solutions of this equation is
R Z R
P (x)dx
ye = e P (x)dx Q (x) dx + c;

that is, Z 
R R
− P (x)dx P (x)dx
y=e e Q (x) dx + c .

Theorem 2.4. The lienal differential equation

dy
+ P (x) y = Q (x)
dx
has an integrating factor of the form
R
P (x)dx
µ (x) = e .

A one-parameter family of solutions of this equation is

R
Z R 
− P (x)dx P (x)dx
y=e e Q (x) dx + c .

Furthermore, it can be shown that this one-parameter family of solutions of the linear equation
(2.18) includes all solutions of (2.18).

Example 2.14. Solve the differential equation

dy (2x + 1)
+ y = e−2x . (2.21)
dx x

(2x + 1)
Solution: In this problem P (x) = and Q (x) = e−2x . It is easy to see that (2.21)
x
is not exact differential equation. Therefore, we need to find an integrating factor to solve
given problem. We have that
2x + 1 R
R dx
x P (x)dx
µ (x) = e =e
Z Z  
(2x + 1) dx 1
= 2+ dx = 2x + ln |x| .
x x
Thus
µ (x) = e2x+ln|x| = e2x · eln|x| = xe2x .
Multiplying (2.21) by xe2x , we get
 
2x dy (2x + 1) −2x
xe + y=e ,
dx x
 20

dy (2x + 1)
xe2x + xe2x y = xe2x · e−2x ,
dx x
or
dy d 
xe2x + e2x (2x + 1) y = x ⇒ yxe2x = x ⇒ d yxe2x = xdx.
  
dx dx
Taking the integration of both sides, we obtain

x2
Z Z
2x
= xdx ⇒ yxe2x =
 
d yxe +c
2
or
1 c
y = xe−2x + e−2x .
2 x
Example 2.15. Solve the initial value problem

dy
x2 + 1 + 4xy = x, y (2) = 1.
dx

Solution: First, we need to divide both sides by x2 + 1. Hence

dy 4x x
+ 2 y= 2 . (2.22)
dx x + 1 x +1
is linear equation. Here
4x x
P (x) = and Q (x) = 2 .
x2 + 1 x +1
4xdx
R R
µ (x) = e P (x)dx = e x2 +1
u = x2 + 1
Z   Z
4xdx du x + 1 = ln x2 + 1 2
 2 

= = 2 = 2 ln |u| = 2 ln
x2 + 1 du = 2xdx u
2 2
2
∴ µ (x) = eln(x +1) = x2 + 1

2
Multiplying (2.22) by x2 + 1 , we get
 
2

2 dy 4x x
x +1 + y= 2 ,
dx x2 + 1 x +1
or

2 dy
x2 + 1 + 4x x2 + 1 y = x x2 + 1 .



dx
Z h Z
d h 2
i
2 i
y x2 + 1 = x3 + x ⇒ d y x2 + 1 x3 + x dx



=
dx

2 x4 x2
y x2 + 1 = + +c
4 2
Using the initial condition y (2) = 1, we get

25 = 4 + 2 + c ⇒ c = 19.

Therefore
x4 x2
 
1
y= + + 19
(x2 + 1)2 4 2
is the exact solution of the given differential equation.
 21

Example 2.16. Solve the differential equation

cos2 x − y cos x dx − (1 + sin x) dy = 0.

(2.23)

Solution: Let us write (2.23) in the derivative form, namely

dy cos x cos2 x
+ = . (2.24)
dx 1 + sin x 1 + sin x
Here
cos x cos2 x
P (x) = and Q (x) = .
1 + sin x 1 + sin x
So (2.24) is linear differential equation.
cos xdx
R R
µ (x) = e P (x)dx = µ (x) = e 1+sin x
Z   Z
cos xdx u = 1 + sin x du
= = = ln |u| = ln |1 + sin x| = ln (1 + sin x)
1 + sin x du = cos dx u
∴ µ (x) = eln(1+sin x) = 1 + sin x
Multiplying (2.24) by 1 + sin x, we get

dy
(1 + sin x) + cos x = cos2 x.
dx
Z Z
d
[y (1 + sin x)] = cos x ⇒ d [y (1 + sin x)] = cos2 xdx
2
dx
Z
y (1 + sin x) = cos2 xdx
Z Z   Z Z
2 1 + cos 2x 1 1
cos xdx = dx = dx + cos 2xdx
2 2 2
x 1
= + sin 2x + c.
2 4
Thus
x 1
y (1 + sin x) = + sin 2x + c
2 4
is the general solution of the given equation.

Example 2.17. Solve the differential equation

y 2 dx + (3xy − 1) dy = 0.

Solution. Solving for dy/dx, this becomes

dy y2
= ,
dx 1 − 3xy
which is clearly not linear in y. Writing this equation in the derivative form, we obtain
dx 1 − 3xy
=
dy y2
 22

or
dx 3 1
+ x = 2. (2.25)
dy y y
Now observe that the equation (2.25) is of the form

dx
+ P (y) x = Q (y)
dy
and so is linear in x. Thus the integrating factor is
3
R R 3
P (y)dy dy
µ (y) = e =e y = eln|y| = y 3 .

Multiplying (2.25) by y 3 , we obtain

dx
y3 + 3y 2 x = y
dy
or
d  3
xy = y.
dy
Integrating, we find the solutions in the form
1
xy 3 = y 2 + c
2
or
1 c
x= + ,
2y y 3
where c is arbitrary constant.

B. Bernoulli Equations

Definition 6. An equation of the form

dy
+ P (x) y = Q (x) y n
dx
is called a Bernoulli differential equation.

If n = 0 ⇒ Linear Differential Equation.

If n = 1 ⇒ Separable Differential Equation.
If n 6= 0 or 1 ⇒ Bernoulli Differential Equation.

Theorem 2.5. Suppose n 6= 0 or 1. Then the transformation v = y 1−n reduces the Bernoulli
equation
dy
+ P (x) y = Q (x) y n (2.26)
dx
to a linear equation in v.
 23

Proof. We first multiply equation (2.26) by y −n , thereby expressing it in the equivalent

form
dy
y −n + P (x) y 1−n = Q (x) . (2.27)
dx
If we let v = y 1−n , then
dv dy
= (1 − n) y −n
dx dx
and equation (2.27) transforms into

1 dv
+ P (x) v = Q (x)
1 − n dx
or, equivalently,
dv
+ (1 − n) P (x) v = (1 − n) Q (x) .
dx
Letting P1 (x) = (1 − n) P (x) and Q1 (x) = (1 − n) Q (x) , this may be written

dv
+ P1 (x) v = Q1 (x)
dx
which is linear in v.

Example 2.18. Solve the differential equation

dy
+ y = xy 3 . (2.28)
dx

Solution: Since n = 3 (2.28) is a Bernoulli equation. First, we must divide both sides by
y3.
1 dy 1
3
+ 2 =x
y dx y
Let v = y 1−n = y −2 . Then
dv dy −2 dy 1 dy 1 dv
= −2y −3 = 3 ⇒ 3 =− .
dx dx y dx y dx 2 dx
Therefore  
1 dv dv
2· − +v =x ⇒ − 2v = −2x (2.29)
2 dx dx
is a linear differential equation with P (x) = −2 and Q (x) = −2x. The differential equation
(2.29) has an integration factor such that
R R
−2dx
µ (x) = e P (x)dx
=e = e−2x .

So
dv d  −2x 
e−2x − 2e−2x v = −2xe−2x ⇒ e v = −2xe−2x .
dx dx
Now, we will take the integration of both sides.
 
u = −2 ⇒ du = −2dx
Z
−2x
−2xe dx = −2x
dv = e−2x ⇒ v = − e 2
  Z  
1 −2x 1 −2x
= (−2x) − e − − e (−2) dx
2 2
 24

e−2x
Z  
−2x −2x −2x −2x 1
= xe − e dx = xe + +c=e x+ +c
2 2
Thus  
−2x −2x 1 1
e v=e x+ + c ⇒ v = x + ce2x .
2 2
1
But v = . So
y2
1 1
2
= x + + ce2x
y 2
is the general solution of the given differential equation.
Example 2.19. Solve the initial value problem
dy
x2 + xy = xy 3 , y (1) = 1.
dx
Solution:
dy 1 1
+ y = y 3 ; n = 3 ⇒ Bernoulli Differential Equation
dx x x
Let v = y 1−n −2
= y . Then
dv dy −2 dy 1 dy 1 dv
= −2y −3 = 3 ⇒ 3 =− .
dx dx y dx y dx 2 dx
Therefore
1 dy 1 1 1 1 dv 1 1
3
+ 2
= ⇒− + v=
y dx x y x 2 dx x x
or
dv 2 2
− v=− (2.30)
dx x x
2
is linear differential equation with P (x) = − = Q (x) . So
x
R R dx −2 ln 1 1 1
µ (x) = e P (x)dx
= e−2 x = e−2 ln|x| = eln|x| =e |x|2 = eln x2 =
x2
is the integrating factor of the linear differential equation (2.30).
  Z   Z  
1 dv 2 2 d v 2 v 2
− v=− 3 ⇒ = − 3 ⇒ d 2 = − 3 dx
x2 dx x3 x dx x2 x x x
v 1 1 1
2
= 2 +c⇒ 2 2 = 2 +c
x x x y x
Using the initial condition y (1) = 2, we get
1 1 1
= +c⇒c= .
4 2 4
Thus
1 1 1
= 2+
x2 y 2 x 4
is the exact solution of the given differential equation.
Example 2.20. Solve the initial value problem
dy y3
x2
+ xy = , y (1) = 1.
dx x
Example 2.21. Solve the differential equation
dy
x + y = −2x6 y 4 .
dx
 25

2.4 SPECIAL INTEGRATING FACTORS AND TRANSFORMATIONS

A. Finding Integrating Factors

Theorem 2.6. Consider the differential equation

M (x, y) dx + N (x, y) dy = 0. (2.31)

If  
1 ∂M (x, y) ∂N (x, y)
−
N (x, y) ∂y ∂x
depends upon x only, then
Z   
1 ∂M (x, y) ∂N (x, y)
µ (x) = exp − dx
N (x, y) ∂y ∂x

is an integrating factor of equation (2.31). If

 
1 ∂N (x, y) ∂M (x, y)
−
M (x, y) ∂x ∂y

depends upon y only, then

Z   
1 ∂N (x, y) ∂M (x, y)
µ (y) = exp − dy
M (x, y) ∂x ∂y

is an integrating factor of equation (2.31).

Example 2.22. Solve the differential equation

2x2 + y dx + x2 y − x dy = 0.



(2.32)

Solution: First, we should check that whether (2.32) is exact differential equation or not.
We have that
M (x, y) = 2x2 + y and N (x, y) = x2 y − x.
Since
∂M (x, y) ∂N (x, y)
= 1 6= 2xy − 1 =
∂y ∂x
(2.32) is not exact. So we need to find an integrating factor.
 
1 ∂M (x, y) ∂N (x, y) 1
− = 2 [1 − 2xy + 1]
N (x, y) ∂y ∂x x y−x

2 − 2xy −2 (xy − 1) 2
= = =−
x (xy − 1) x (xy − 1) x
Hence R R dx 1
µ (x) = e P (x)dx
= e−2 x = .
x2
1
Multiplying both sides of (2.32) by , we obtain
x2
1 1
2x2 + y dx + 2 x2 y − x dy = 0



x2 x
 26

or
y 1
2dx + 2
dx + ydy − dy = 0.
x x
From that it follows
y2
Z Z   Z  y Z
2 d (x) + d + d − = d (c) .
2 x
Thus
y2 y
2x + − =c
2 x
is the general solution of (2.32).
Example 2.23. Solve the differential equation

y 2 [x + 1] + y dx + (2xy + 1) dy = 0.

Solution: Here M (x, y) = y 2 [x + 1] + y and N (x, y) = 2xy + 1.

 
1 ∂M (x, y) ∂N (x, y) 1
− = [2y (x + 1) + 1 − 2y]
N (x, y) ∂y ∂x 2xy + 1
1
= [2xy + 2y + 1 − 2y] = 1
2xy + 1
So R
dx
µ (x) = e = ex .
Multiplying both sides of the given differential equation, we obtain

ex y 2 [x + 1] + y dx + ex (2xy + 1) dy = 0

or
ex y 2 [x + 1] + ex ydx + ex 2xy + ex dy = 0.
The last differential equation is exact. Using method of grouping, we get

d ex xy 2 + d (yex ) = d (c) .

Taking the integration of both sides, we obtain

ex xy 2 + yex = c ⇒ xy 2 + y = ce−x .

Example 2.24. Solve the differential equation

(2x + tan y) dx + x − x2 tan y dy = 0.

B. A Special Transformation

Theorem 2.7. Consider the differential equation

(a1 x + b1 y + c1 ) dx + (a2 x + b2 y + c2 ) dy = 0. (2.33)

where a1 , b1 , c1 , a2 , b2 and c3 are constants.

Case 1. If
a2 b2
6= ,
a1 b1
 27

then the transformations

x = X + h,
y = Y + k,

where (h, k) is the solution of the system


a1 h + b1 k + c1 = 0,
a2 h + b2 k + c2 = 0,

reduces equation (2.33) to the homogeneous equation

(a1 X + b1 Y ) dX + (a2 X + b2 Y ) dY = 0

in the variables.
Case 2. If
a2 b2
= = k,
a1 b1
then the transformation
z = a1 x + b1 y
reduces the equation to a separable equation in the variables x and z.

Example 2.25. Solve the differential equation

(x − 2y + 1) dx + (4x − 3y − 6) dy = 0.

Solution: Here a1 = 1, b1 = −2, c1 = 1 and a2 = 4, b2 = −3, c2 = −6.

a2 4 b2 −3 3 3
= = 4; = = ⇒ 4 6=
a1 1 b1 −2 2 2
Therefore, we make transformation

x = X + h,
y = Y + k,

where (h, k) is the solution of the system

  
h − 2k = −1, −4h + 8k = 4, k=2
⇒ ⇒
4h − 3k = 6. 4h − 3k = 6. h=3

So

x = X + 3,
y = Y + 2.

We have the following homogeneous equation

(X − 2Y ) dX + (4X − 3Y ) dY = 0

or
Y


dY 1−2 X
= Y

.
dX 3 X −4
 28

Let Y = vX. Then

dv 1 − 2v
v+X = .
dX 3v − 4
This reduces to
(3v − 4) dv dX
2
=− .
3v − 2v − 1 X
Integrating both sides, we obtain
1 15
− ln |v − 1| + ln |3v + 1| = − ln |X| + ln |c1 | .
4 4
− ln |v − 1| + 15 ln |3v + 1| = −4 ln |X| + 4 ln |c1 |
ln |v − 1|−1 + ln |3v + 1|15 = ln |X|−4 + ln c41 
 
 
1 1  (3v + 1)15   4
c 
15 4
 = ln  14 
 
ln + ln |3v + 1| = ln + ln c1 ⇒ ln 
  
|v − 1| 4  v−1  X
|X|

X 4 |3v + 1|15 = c |v − 1| ; c = c41

 15    15
4  3Y
  Y  4  3Y + X 
  |Y − X|
X  + 1 = c  − 1 ⇒ X 
   =c
X X X  X
⇒ |3Y + X|15 = cX 10 |Y − X|
Putting X = x − 3 and Y = y − 2, we obtain

|3y − 6 + x − 3|15 = c (x − 3)10 |y − 2 − x + 3|

or
|x + 3y − 9|15 = c (x − 3)10 |y − x + 1| .

Example 2.26. Solve the initial value problem

 
1
(6x + 4y + 1) dx + (4x + 2y + 2) dy = 0, y = 3.
2

Solution: Here a1 = 6, b1 = 4, c1 = 1 and a2 = 4, b2 = 2, c2 = 2.

a2 4 2 b2 2 1 2 1
= = ; = = ⇒ 6=
a1 6 3 b1 4 2 3 2
So we make a transformation x = X + h and y = Y + k such that (h, k) is the solution of the
system
h = − 32
  
6h + 4k = −1, 6h + 4k = −1
⇒ ⇒
4h + 2k = −2. −8h − 4k = 4 k=2
3
Hence x = X − and y = Y + 2.
2
       
3 3
6 X− + 4 (Y + 2) + 1 dX + 4 X − + 2 (Y + 2) + 2 dY = 0
2 2

[6X − 9 + 4Y + 8 + 1] dX + [4X − 6 + 2Y + 4 + 2] dY = 0

Y


dY 6+4 X
(6X + 4Y ) dX + (4X + 2Y ) dY = 0 ⇒ =− Y

(2.34)
dX 4+2 X
 29

So, (2.34) is homogeneous differential equation. Let Y = vX. Then

dv 6 + 4v dv 6 + 4v
v+X =− ⇒X =− −v
dX 4 + 2v dX 4 + 2v
dv −6 − 4v − 4v − 2v 2 dv 2v 2 + 8v + 6
⇒X = ⇒X =− .
dX 4 + 2v dX 4 + 2v
Z Z
(2 + v) dv dX
⇒ 2
= −
v + 4v + 3 X
2+v A B A (v + 3) + B (v + 1)
= + =
(v + 1) (v + 3) (v + 1) (v + 3) (v + 1) (v + 3)
2 + v = Av + 3A + Bv + B = v (A + B) + 3A + B

A+B =1 1 1
⇒ ⇒A= , B=
3A + B = 2 2 2
Z Z Z
(2 + v) dv 1 dv 1 dv
2
= +
v + 4v + 3 2 v+1 2 v+3
Thus
 2
1 1 c 
ln |v + 1| + ln |v + 3| = − ln |X| + ln |c1 | ⇒ ln |v + 3| |v + 1| = ln  12  .
2 2 X
  
Y Y
X 2 |v + 3| |v + 1| = c ; c = c21 ⇒ X 2  + 3  + 1 = c

X X
  
 3X + Y   X + Y 
X 2    = c ⇒ |3X + Y | |X + Y | = c
X  X 
So   
 5   1 
3x + y + x+y− =c
 2  2
 
1
is the general solution. Using the initial condition y = 3, we get
2
  
3
 + 5 + 5   1 + 5 − 1  = c ⇒ c = 45.
 
2 2 2
  2
Hence   
 5   1 
3x + y + x+y− = 45
 2  2
is the exact solution of the given problem.
Example 2.27. Solve the differential equation

(x + 2y + 3) dx + (2x + 4y − 1) dy = 0.

Solution: Here a1 = 1, b1 = 2, c1 = 3 and a2 = 2, b2 = 4, c2 = −1.

a2 2 4 b2
= =2= =
a1 1 2 b1
So we make a transformation z = a1 x + b1 y = x + 2y.

(z + 3) dx + (2z − 1) dy = 0
 30

1 1
z = x + 2y ⇒ y = (z − x) ⇒ dy = (dz − dx)
2 2
 
dz − dx
(z + 3) dx + (2z − 1) =0
2
2 (z + 3) dx + (2z − 1) (dz − dx) = 0
2zdx + 6dx + 2zdz − 2zdx − dz + dx = 0 ⇒ 7dx + (2z − 1) dz = d (c)
Thus last equation is separable.
Z Z Z
7dx + (2z − 1) dz = d (c) ⇒ 7x + z 2 − z = c

Therefore
x2 − 4y 2 + 4xy + 6x − 2y = c
is the general solution of the given equation.
Example 2.28. Solve the initial value problem
(2x + 3y + 1) dx + (4x + 6y + 1) dy = 0, y (3) = −2.

Solution: Here a1 = 2, b1 = 3, c1 = 1 and a2 = 4, b2 = 6, c2 = 1.

a2 4 6 b2
= =2= =
a1 2 3 b1
So we make a transformation z = 2x + 3y, namely
(z + 1) dx + (2z + 1) dy = 0.
Here we need to write dy in terms of dz to take the integration of both sides.
z − 2x (dz − 2dx)
z = 2x + 3y ⇒ y = ⇒ dy =
3 3
(dz − 2dx)
(z + 1) dx + (2z + 1) =0
3
(3z + 3) dx + (2z + 1) (dz − 2dx) = 0
3zdx + 3dx + 2zdz + dz − 4zdx − 2dx = 0 ⇒ (1 − z) dx + (2z + 1) dz = 0
Z Z Z
2z + 1
⇒ dx + dz = d (c)
1−z
Z Z Z Z
(2z + 1) dz 2zdz dz 2zdz
= + = + ln |1 − z|
1−z 1−z 1−z 1−z
Z Z  
2zdz 2
= −2 + dz = −2z + 2 ln |1 − z| = −2 (z − ln |1 − z|)
1−z 1−z
Hence
x − 2z + 3 ln |1 − z| = c.
x − 4x − 6y + 3 ln |1 − 2x − 3y| = c ⇒ −3x − 6y + 3 ln |1 − 2x − 3y| = c
Using the initial condition y (3) = −2, we get
−9 + 8 + 3 ln |1 − 6 + 6| = c ⇒ c = −1.
Therefore
−3x − 6y + 3 ln |1 − 2x − 3y| = −1
is the exact solution of the given problem.
 31

C. Other Special Types and Methods; An Important Reference

Many other special types of first-order equations exist for which corresponding special methods
of solution are known. We will not go into such highly specialized types in this book. Instead
we refer the reader to the book Differentialgleichungen: Losungsmethoden und Losungen, by
E. Kamke (Chelsea, New York, 1948). This remarkable volume contains discussions of a large
number of special types of equations and their solutions.
Chapter 3

APPLICATIONS OF FIRST
ORDER EQUATIONS

3.1 ORTHOGONAL AND OBLIQUE TRAJECTORIES

A. Orthogonal Trajectories

Definition 1. Let
F (x, y) = c (3.1)
be a given one-parameter family of curves in the xy plane. A curve that intersects the curves
of the family (3.1) at right angles is called an orthogonal trajectory of the given family.
Example 3.1. Consider the family of circles
x2 + y 2 = c2 (3.2)
with center at the origin and radius c. Each straight line through the origin
y = kx, (3.3)
is an orthogonal trajectory of the family of circles (3.2). Conversely, each circle of the family
(3.2) is an orthogonal trajectory of the family of straight lines.
The problem of finding orthogonal trajectories of a given family of curves arises in many
physical situation. For example, in a two-dimensional electric field the lines of force (flux
lines) and the equipotential curves are orthogonal trajectories of each other.
We now proceed to find the orthogonal trajectories of a family of curves (3.1). We obtain
the differential equation of the family (3.1) by first differentiating equation (3.1) implicitly with
respect to x and then eliminating the parameter c between the derived equation so obtained
and the given equation (3.1) itself. We assume that the resulting differential equation of the
family (3.1) can be expressed in the form
dy
= f (x, y) . (3.4)
dx
Thus the curve C of the given family (3.1) which passes through the point (x, y) has the slope
f (x, y) there. Since an orthogonal trajectory of the given family intersects each curve of the
family at right angles, the slope of the orthogonal trajectory to C at (x, y) is
1
− .
f (x, y)

32
 33

Thus the differential equation of the family of orthogonal trajectories is

dy 1
=− . (3.5)
dx f (x, y)

A one-parameter family
G (x, y, c) = 0
or
y = F (x, c)
of solutions of the differential equation (3.5) represents the family of orthogonal trajectories
of the original family (3.1), except possibly for certain trajectories that are vertical lines.
We summarize this procedure as follows:

Procedure for Finding the Orthogonal Trajectories of a Given Family of Curves

Step 1. From the equation

F (x, y, c) = 0
of the given family of curves, find the differential equation
dy
= f (x, y)
dx
of this family.
Step 2. In the differential equation dy/dx = f (x, y) replace f (x, y) by its negative
reciprocal −1/f (x, y) . This gives the differential equation

dy 1
=−
dx f (x, y)

of the orthogonal trajectories.

Step 3. Obtain a one-parameter family G (x, y, c) = 0 or y = F (x, c) of solutions of the
differential equation (3.5).

Example 3.2. Find the orthogonal trajectories of the family of circles x2 + y 2 = c2 .

Solution: Step 1.
d d 2
dy dy x
x2 + y 2 =


c ⇒ 2x + 2y =0⇒ = − = f (x, y)
dx dx dx dx y
Step 2.
1 −1 y
− = x = .
f (x, y) −y x
Step 3.
Z Z
dy y dy dx
= ⇒ = ⇒ ln |y| = ln |x| + ln |k| ⇒ ln y = ln xk ⇒ y = kx
dx x y x

Example 3.3. Find the orthogonal trajectories of the family of parabolas y = cx2 .
 34

Solution: Step 1.
dy 2y
= 2cx = = f (x, y)
dx x
Step 2.
1 1 x
− = − 2y = − .
f (x, y) x
2y
Step 3. Z Z Z
dy x
d k 2 ⇒ x2 + 2y 2 = k 2


=− ⇒ xdx + 2ydy =
dx 2y

B. Oblique Trajectories

Definition 2. Let
F (x, y, c) = 0 (3.6)
be a one-parameter family of curves. A curve that intersects the curves of the family (3.6) at
a constant angle α 6= 90◦ is called an oblique trajectory of the given family.

Suppose the differential equation of a family is

dy
= f (x, y) . (3.7)
dx
Then the curve of the family (3.7) through the point (x, y) has the slope f (x, y) at (x, y) and
hence its tangent line has angle of inclination tan−1 [f (x, y)] there. The tangent line of an
oblique trajectory that intersects this curve at the angle α will thus have angle of inclination

tan−1 [f (x, y)] + α

at the point (x, y) . Hence the slope of this oblique trajectory is given by

f (x, y) + tan α
tan tan−1 [f (x, y)] + α =

.
1 − f (x, y) tan α

Thus the differential equation of such a family of oblique trajectories is given by

dy f (x, y) + tan α
= .
dx 1 − f (x, y) tan α

Thus to obtain a family of oblique trajectories intersecting a given family of curves at the
constants angle α 6= 90◦ , we may follow the three steps in the above procedure for finding the
orthogonal trajectories, except that we replace Step 2 by the following step:
Step 20 . In the differential equation dy/dx = f (x, y) of the given family, replace f (x, y)
by the expression
f (x, y) + tan α
.
1 − f (x, y) tan α
Example 3.4. Find a family of oblique trajectories that intersect the family of straight lines
y = cx at angle 45◦ .

Solution: Step 1. Find a differential equation.

dy dy y
=c⇒ = = f (x, y)
dx dx x
 35

Step 2. Replace f (x, y) with following formula.

y y
f (x, y) + tan α + tan 45◦ 1+
= x y ◦
= x
y
1 − f (x, y) tan α 1 − x tan 45 1− x

Step 3. Solve the differential equation.

y
dy 1+ x
= y
dx 1− x

We have homogeneous differential equation. Let y = vx. Then

dy dv
=v+x .
dx dx
Thus
dv 1+v dv 1+v
v+x = ⇒x = − v.
dx 1−v dx 1−v
1 + v − v + v2 1 + v2 (1 − v) dv
Z Z
dv dv dx
x = ⇒x = ⇒ 2
=
dx 1−v dx 1−v 1+v x
(1 − v) dv
Z Z Z
dv vdv 1 
= arctan v − ln 1 + v 2 

2
= 2
− 2
1+v 1+v 1+v 2
So
1
ln 1 + v 2 = ln x + ln c.


arctan v −
2
Putting v = y/x, we obtain

y2
 
y 1
arctan − ln 1 + 2 = ln x + ln c
x 2 x
or
y
ln c2 x2 + y 2 − 2 arctan = 0.


x

3.2 PROBLEMS IN MECHANICS

A. Newton’s Law of Cooling

In the Newton’s law of a cooling the temperature of a body changes at a rate that is propor-
tional to the difference in temperature between the outside medium and itself. We assume
here that the constant of proportionality is the same whether the temperature is increasing
or decreasing.
Suppose that, for example, a thermometer which has been of reading, 70◦ F in a house,
is placed outside where the air temperature is 25◦ F. 3 minutes later it is found that the
temperature is 10◦ F. We wish to predict. Let u◦ (F ) represent the temperature of the
thermometer at time t (minute), the time measure of is placed outside. We are given that
when t = 0, u = 70◦ and t = 3, u = 25◦ F. According to Newton’s law, the time rate of
du
change of temperature, , is proportioned to the temperature difference (u − 10) . Since the
dt
thermometer temperature is decreasing, it is convenient to choose (−k) as the constant of
proportionality. Thus the u is be determined from the differential equation.
du
= −k (u − 10)
dt
 36

and the conditions that u (0) = 70 and u (3) = 25.

du
+ ku = 10k
dt
R
 Z R 
− kdt kdt
u (t) = e c+ e 10kdt
 Z   
−kt
=e c+ e 10kdt = e−kt c + 10ekt = e−kt c + 10
kt

Using initial condition u (0) = 70, we get

u (0) = c + 10 = 70 ⇒ c = 60.

So
u (t) = 60e−kt + 10.
Using boundary condition u (3) = 25, we get
1
u (3) = 60e−3k + 10 = 25 ⇒ 60e−3k = 15 ⇒ e−3k = .
4

For these types of problems we will be assuming that the question involves the temperature
T of a certain body placed in a medium of constant temperature M and as time t varies, so
does T, (so T has a rate of change with respect to t). In this case Newton’s law of cooling
tells us the following
dT
= k (T − M )
dt
for some constant k.

Example 3.5. A boiling 100◦ C solution is set on a table where room temperature is assumed
to be constant at 25◦ C. The solution cooled to 60◦ C after 5 minutes.
a-) Find a formula for the temperature T of the solution, t minutes after it is placed on
the table.
b-) Determine how long it will take for the solution to cool to 22◦ C.

Solution. a-) We are asked to find an explicit formula for T in terms of t. We know this
is a heating and cooling question Newton’s law of cooling question so Newton’s law of cooling
question tells us
dT
= −k (T − M )
dt
for some constant k. So letting M = 20, we have
dT
= −k (T − M )
dt
for some constant k. Recognizing this as a separable differential equation
Z Z
1 1
dT = −kdt ⇒ dT = −kdt
T − 20 T − 20

⇒ ln (T − 20) = −kt + ln C
⇒ eln(T −20) = e−kt+C
 37

⇒ T − 20 = e−kt eln C
⇒ T = Ce−kt + 20.
Since the initial temperature of the solution was 100◦ C, we know that T = 100 when t = 0,
so the last above gives:
100 = Ce(0)(−k) + 20 ⇒ C = 80
So we now have
T = 80e−kt + 20. (3.8)
Now using the fact that 5 minutes (i.e. when t = 5), T = 60 we have
   
−5k −5k 1 1 1 1
60 = 80e + 20 ⇒ e = ⇒ −5k = ln ⇒ k = − ln .
2 2 5 2
So
k ≈ 0.0113863.
Substituting this into (3.8), we then have

T = 80e−0.0113863t + 20

which is require formula for T.

b-) We wish to find out what t is when T is 22. We use the formula just we found in part
(a). Therefore
22 = 80e−0.0113863t + 20 ⇒ 2 = 80e−0.0113863t
 
2 −0.0113863t 1
⇒ =e ⇒ ln = −0.0113863t ⇒ t = 26.6.
80 40

B. Simple Chemical Conversion

It is known from the results of chemical experimentation that, in certain reaction in which a
substance A is being converted into another substance, the time rate of change of amount x
of amount of unconverted substance be known at same specified time; that is, let x = x0 at
t = 0. Then the amount x any time t, is determined by the differential equation
dx
= −k (3.9)
dt
and the condition x = x0 when t = 0. Hence the amount x is decreasing as the time increases,
the constant of proportionality in equation (3.9) is take to be (−k) . From (3.9) it follows that
dx
= −kdt ⇒ x (t) = ce−kt .
x
Put x = x0 when x = 0. Hence

x (0) = c, x (t) = x0 e−kt . (3.10)

Let us now add another condition, which will enable us to determine k. Suppose it is known
that at the end of half a minute, at t = 30 (sec), two-thirds of the original amount x0 has
already been converted. Let us determined how much unconverted substance remaining at
t = 60 (sec).
1
t = 30; x (30) = x0
3
 38

1 1
x0 = e−3k x0 ; k = ln 3
3 30
 
1
x = x0 − + ln 3
30
1 x0
x (60) = x0 e− 30 60 ln 3 = x0 e−2 ln 3 = x0 e− ln 9 =
9

C. Price of Commodities

We consider an economic model of a certain commodity market. We assume that the price
P, the supply S, and the demand D of that the rate change of the price is proportional to the
difference between the demand and the supply. That is
dP
= k (D − S) .
dt
We further assume that the constant k is positive, so that the price will increase if the demand
exceeds the supply. Different models of the commodity market will results upon the nature
of the demand and supply function that are indicated. If, for example, we assume that

D = c − dP,
(3.11)
S = a + bP,

where a, b, c, d > 0 constants. We obtain the differential equation

dP
= k [(c − a) − (d + b) P ] .
dt
dP c−a
+ k (d + b) P = k (c − a) ⇒ P (t) = c1 e−k(d+b)t +
dt d+b
c−a c−a
t = 0, P (0) -is given. ⇒ P (0) = c1 + ⇒ c1 = P (0) −
d+b d+b
Therefore  
c−a c−a
P (t) = P (0) − e−k(d+b)t + .
d+b d+b
Passing to limit when t → ∞, we obtain
c−a
P (t) = .
d+b
Chapter 4

EXPLICIT METHODS OF
SOLVING HIGHER-ORDER
LINEAR DIFFERENTIAL
EQUATIONS

4.1 BASIC THEORY OF LINEAR DIFFERENTIAL EQUATIONS

A. Definition and Basic Existence Theorem

Definition 1. A linear ordinary differential equation of order n in the dependent variable y

and the independent variable x is an equation that is in, or can be expressed in, the form

dn y dn−1 y dy
a0 (x) + a1 (x) + · · · + an−1 (x) + an (x) y = F (x) , (4.1)
dxn dxn−1 dx
where a0 (x) 6≡ 0. We shall assume that a0 (x) 6= 0 for any x on a ≤ x ≤ b. The right-hand
member F (x) is called the nonhomogeneous term. If F is identically zero, equation (4.1)
reduces to
dn y dn−1 y dy
a0 (x) n + a1 (x) n−1 + · · · + an−1 (x) + an (x) y = 0 (4.2)
dx dx dx
and is then called homogeneous.

For n = 2, equation (4.1) reduces to second-order nonhomogeneous linear differential

equation
d2 y dy
a0 (x) 2 + a1 (x) + a2 (x) y = F (x) (4.3)
dx dx
and (4.2) reduces to the corresponding second-order homogeneous equation

d2 y dy
a0 (x) 2
+ a1 (x) + a2 (x) y = 0. (4.4)
dx dx
Example 4.1. The equation
d2 y dy
+ 3x + x3 y = ex
dx2 dx
is a linear ordinary differential equation of the second order.

39
 40

Example 4.2. The equation

d3 y d2 y dy
3
+ x 2
+ 3x2 − 5y = sin x
dx dx dx
is a linear ordinary differential equation of the third order.

Theorem 4.1. Consider the n-th order linear differential equation (4.1), where a0 , a1 , · · ·, an
and F are continuous real functions on a real interval [a, b] and a0 (x) 6= 0 for any x on [a, b] .
Let x0 be any point of the interval [a, b] , and let c0 , c1 , · · ·, cn−1 be arbitrary real constants.
Then there exists a unique solution of (4.1) such that

f (x0 ) = c0 , f 0 (x0 ) = c1 , · · ·, f (n−1) (x0 ) = cn−1 ,

and this solution is defined over the entire interval [a, b] .

Proof. Proof of this theorem follows from condition of general theorem about existence
and uniqueness for nonlinear differential equation. It will be considered in further.

Example 4.3. Consider the initial value problem

d2 y dy
2
+ 3x + x3 y = ex , y (1) = 2, : y 0 (1) = −5.
dx dx
Here a0 (x) = 1, a1 (x) = 3x, a2 (x) = x3 are continuous on (−∞, ∞) . F (x) = ex is also
continuous functions. 1, 2, −5 are constants. By last theorem there exists and unique solution
y (x) of the given initial value problem.

Example 4.4. Consider the initial value problem

d3 y d2 y dy 7
2 + x + 3x2 − 5y = sin x, y (4) = 3, y 0 (4) = 5, y 00 (4) = − .
dx3 dx2 dx 2
Here a0 (x) = 2, a1 (x) = x, a2 (x) = 3x2 , a3 (x) = −5 are continuous on (−∞, ∞) . F (x) =
7
sin x is also continuous functions. 3, 4, 5 and − are constant. So the given initial value
2
problem has also a unique solution.

B. The Homogeneous Equation

We now consider the fundamental results concerning the homogeneous equation

dn y dn−1 y dy
a0 (x) + a1 (x) + · · · + an−1 (x) + an (x) y = 0. (4.5)
dxn dxn−1 dx
We first state the following basic theorem:

Theorem 4.2. (Basic Theorem on Linear Homogeneous Differential Equations)

Let f1 , f2 , · · ·, fm be any m solution of the homogeneous linear differential equation (4.5).
Then
c1 f1 + c2 f2 + · · · + cm fm
is also a solution of (4.5), where c1 , c2 , · · ·, cm are arbitrary constants.
 41

Definition 4.2. If f1 , f2 , · · ·, fm are given m functions and c1 , c2 , · · ·, cm are m constants,

then the expression
c1 f1 + c2 f2 + · · · + cm fm
is called a linear combination of f1 , f2 , · · ·, fm .
In terms of this concept, Theorem 4.2 may be stated as follows:
Theorem 4.3. (RESTATED) Any linear combination of solutions of the homogeneous
linear differential equation (4.5) is also a solution of (4.5).

Proof. We must prove that y (x) = c1 f1 (x) + c2 f2 (x) + · · · + cm fm (x) is satisfied the
differential equation (4.5). It is easy to wire that

dy df1 (x) df2 (x) dfm (x)

= c1 + c2 + · · · + cm .
dx dx dx dx
Hence  n
dn f2 (x) dn fm (x)

d f1 (x)
a0 (x) c1 + c2 + · · · + cm
dxn dxn dxn
 n−1
dn−1 f2 (x) dn−1 fm (x)

d f1 (x)
+a1 (x) c1 + c2 + · · · + cm
dxn−1 dxn−1 dxn−1
 
df1 (x) df2 (x) dfm (x)
+ · · · +an−1 (x) c1 + c2 + · · · + cm
dx dx dx
+an (x) [c1 f1 (x) + c2 f2 (x) + · · · + cm fm (x)]
n dn−1 f1 (x)
 
d f1 (x) df1 (x)
= c1 a0 (x) + a1 (x) + · · · + an−1 (x) cm + an (x) f1 (x)
dxn dxn−1 dx
dn f2 (x) dn−1 f2 (x)
 
df2 (x)
+c2 a0 (x) + a1 (x) + · · · + an−1 (x) cm + an (x) f2 (x)
dxn dxn−1 dx
+···+
dn fm (x) dn−1 fm (x)
 
dfm (x)
+cm a0 (x) + a1 (x) + · · · + an−1 (x) cm + an (x) fm (x) .
dxn dxn−1 dx
Since f1 , f2 , · · ·, fm be any m solution of (4.5) the last equation is identically zero. This
completes the proof of Theorem 4.2.
In particular, any linear combination

c1 f1 (x) + c2 f2 (x) + · · · + cm fm (x)

of m solutions f1 , f2 , · · ·, fm of the second order homogeneous linear differential equation

d2 y dy
a0 (x) 2
+ a1 (x) + a2 (x) y = 0 (4.6)
dx dx
is also a solution of (4.6).
Example 4.5. Consider the differential equation

d2 y
+ y = 0.
dx2
It is easy to see that sin x and cos x are solutions of last equation. By Theorem 4.2 the linear
combination c1 sin x + c2 cos x is also a solution of the given equation.
 42

Example 4.6. Consider the differential equation

d3 y d2 y dy
− 2 − + 2y = 0.
dx3 dx2 dx
The functions ex , e−x and e2x are solutions of the given equation. By Theorem 4.2 the linear
combination c1 ex + c2 e−x + c3 e2x is also a solution of the given equation.

Definition 4.3. The n functions f1 , f2 , · · ·, fn are called linearly dependent on [a, b] if

there exist constants c1 , c2 , · · ·, cn not all zero, such that

c1 f1 (x) + c2 f2 (x) + · · · + cn fn (x) = 0 ∀ x ∈ [a, b] .

Example 4.7. Consider the functions sin2 x, cos2 x, 1. They are linearly dependent. Why?
Because
c1 sin2 x + c2 cos2 x + c3 · 1 = 0
⇒ c1 = 1 6= 0, c2 = 1 6= 0, c3 = −1 6= 0.
Example 4.8. The functions x and 2x are linearly dependent. Why? Because

c1 x + c2 · 2x = 0 ⇒ c1 = 2 6= 0, c2 = −1 6= 0.

Definition 4.4. The n functions f1 , f2 , ···, fn are called linearly dependent on [a, b] if they
are not linearly dependent there. That is, The functions f1 , f2 , · · ·, fn are linearly independent
on [a, b] if the relation
c1 f1 (x) + c2 f2 (x) + · · · + cn fn (x) = 0
∀ x ∈ [a, b] implies that
c1 = c2 = · · · = cn = 0.
Example 4.9. The functions 1, x and x2 are linearly independent on [0, 1] . Why? Because
?
c1 + c2 x + c3 x2 = 0 ⇒ c1 = c2 = c3 = 0.

If x = 0, then c1 = 0.
If x = 1, then c2 + c3 = 0.
c2 c3
If x = 12 , then + = 0. Hence c1 = c2 = c3 = 0.
2 4
Theorem 4.4. The n-th order homogeneous linear differential equation (4.5) always possesses
n solutions that are linearly independent. Further, if f1 , f2 , · · ·, fn are n linearly independent
solutions of (4.5), then every solution f of (4.5) can be expressed as a linear combination

c1 f1 + c2 f2 + · · · + cn fn

of these n linearly independent solutions by proper choice of the constants c1 , c2 , · · ·, cn .

Example 4.10. We have observed that sin x and cos x are solutions of
d2 y
+y =0
dx2
∀ x ∈ (−∞, ∞) . Using Theorem 4.3 we can write for formula general solution of given
equation
y (x) = c1 sin x + c2 cos x,
where c1 and c2 are arbitrary constants.
 43

Definition 4.5. If f1 , f2 , · · ·, fn are n linearly independent solutions of the n-th order ho-
mogeneous differential equation (4.5) on [a, b] , then the set f1 , f2 , ···, fn is called a fundamental
set of solutions of (4.5) and the function f defined by
f (x) = c1 f1 (x) + c2 f2 (x) + · · · + cn fn (x) , a ≤ x ≤ b,
where c1 , c2 , · · ·, cn are arbitrary constants, is called a general solution of (4.5) on [a, b] .
Example 4.11. Consider the differential equation
d3 y d2 y dy
3
− 2 2
− + 2y = 0.
dx dx dx
y = c1 ex + c2 e−x + c3 e2x is the general solution of the given differential equation, where c1 , c2
and c3 are arbitrary constants.

Definition 4.6. Let f1 , f2 , · · ·, fn be n real functions each of which has an (n − 1)st

derivative on a real interval [a, b] . The determinant
· · · fn
 
 f1 f2 
 0 0 0 
 f1 f 2 · · · f n

W (f1 , f2 , · · ·, fn ) = 
 ,
· · ·
 (n−1) · · · · · · · · · 

(n−1) (n−1) 
 f
1 f2 · · · fn
is called the Wronskian of these n functions. We observe that W (f1 , f2 , · · ·, fn ) is itself
a real function defined on [a, b] . Its value at x is denoted by W (f1 , f2 , · · ·, fn ) (x) or by
W [f1 (x) , f2 (x) , · · ·, fn (x)] .
Theorem 4.5. The n solutions f1 , f2 , ···, fn of the n-the order homogeneous linear differential
equation (4.5) are linearly independent on [a, b] if and only if the Wronskian of f1 , f2 , · · ·, fn
is different from zero for some x on [a, b] .
Theorem 4.6. The Wronskian of n solutions f1 , f2 , · · ·, fn of (4.5) is either identically zero
on [a, b] or else is never zero on [a, b] .
Example 4.12. Show that the solutions sin x and cos x of
d2 y
+y =0
dx2
are linearly independent.

Solution:
 
 sin x cos x  = − sin2 x − cos2 x = −1 6= 0

W (sin x, cos x) = 
cos x − sin x 

So sin x and cos x are linearly independent.

Example 4.13. Show that the solutions ex , e−x and e2x of
d3 y d2 y dy
3
− 2 2
− + 2y = 0
dx dx dx
are linearly independent.

Solution:
e−x e2x
 x   
 e   1 1 1 
W ex , e−x , e2x =  ex −e−x 2e2x  = e2x  1 −1 2  = −6e2x 6= 0

   
  
 ex e−x 4e2x   1 1 4 

So, they are linearly independent.

 44

C. Reduction of Order

We will study for obtaining explicit solutions of higher order linear differential equation. First,
we find that the following theorem on reduction of order is often quite useful.
Theorem 4.7. Let f be a nontrivial solution of the n-the order homogeneous linear differential
equation (4.5). Then the transformation y = f (x) v reduces equation (4.5) to an (n − 1)st
order homogeneous linear differential equation in the dependent variable w = dv/dx.

Let n = 2. Suppose f is a known nontrivial solution of the second order homogeneous

linear differential equation
d2 y dy
a0 (x) 2
+ a1 (x) + a2 (x) y = 0. (4.7)
dx dx
Let us make the transformation
y = f (x) v, (4.8)
where f is the known solution of (4.7) and v is a function of x that will determined. Then,
differentiating, we obtain
dy dv
= f (x) + f 0 (x) v, (4.9)
dx dx

d2 y d2 v dv
2
= f (x) 2
+ 2f 0 (x) + f 00 (x) v. (4.10)
dx dx dx
Substituting (4.8), (4.9), (4.10) into (4.7), we obtain

d2 v
 
0 dv 00
a0 (x) f (x) 2 + 2f (x) + f (x) v
dx dx
 
dv 0
+a1 (x) f (x) + f (x) v + a2 (x) f (x) v = 0
dx
or
d2 v   dv
a0 (x) f (x) 2 + 2a0 (x) f 0 (x) + a1 (x) f (x)
dx dx
00 0
 
+ a0 (x) f (x) + a1 (x) f (x) + a2 (x) f (x) v = 0.
Since f is a solution of (4.7), the coefficient of v is zero, and so the last equation reduces to

d2 v   dv
a0 (x) f (x) 2
+ 2a0 (x) f 0 (x) + a1 (x) f (x) = 0.
dx dx
Letting w = dv/dx, this becomes
dw 
+ 2a0 (x) f 0 (x) + a1 (x) f (x) w = 0.

a0 (x) f (x) (4.11)
dx
This is a first order homogeneous linear differential equation in the dependent variable w. The
equation is separable; thus assuming f (x) 6= 0 and a0 (x) 6= 0, we may write
 0 
dw f (x) a1 (x)
=− 2 + dx.
w f (x) a0 (x)
Thus interacting, we obtain
Z
2 a1 (x)
ln |w| = − ln [f (x)] − dx + ln |c|
a0 (x)
 45

or h R i
a1 (x)
c exp − a0 (x) dx
w= 2 .
[f (x)]
This is the general solution of (4.11); choosing the particular solution for c = 1, recalling that
dv/dx = w, and integrating again, we now obtain
h i
Z exp − R a1 (x) dx
a0 (x)
v= dx. (4.12)
[f (x)]2
Finally, from (4.8), we obtain
h i
Z exp − R a1 (x)
a0 (x) dx
y = f (x) 2 dx. (4.13)
[f (x)]
The function is defined in the right member of (4.13), which we shall henceforth denote by
g, is actually a solution of the original second order equation (4.7). Furthermore, this new
solution g and the original known solution f are linearly independent, since
   
 f (x) g (x)   f (x) f (x) v 
W (f, g) (x) =  0 0
=
0 0 0

f (x) g (x)   f (x) f (x) v + f (x) v 
 Z 
2 0 a1 (x)
= [f (x)] v = exp − dx 6= 0.
a0 (x)
Thus the linear combination
c1 f + c2 g
is the general solution of equation (4.7). We now summarize this discussion in the following
theorem.
Theorem 4.8. Let f be a nontrivial solution of the second-order homogeneous linear differ-
ential equation (4.6). Then;
1. The transformation y = f (x) v reduces equation (4.6) to the first-order homogeneous
linear differential equation
dw 
+ 2a0 (x) f 0 (x) + a1 (x) f (x) w = 0

a0 (x) f (x) (4.14)
dx
in the dependent variable w, where w = dv/dx.,
2. The particular solution
R a1 (x)
− dx
e a0 (x)
w=
[f (x)]2
of equation (4.14) gives rise to the function v where
R a1 (x)
Z − dx
e a0 (x)
v= dx.
[f (x)]2
The function g defined by g (x) = f (x) v (x) is then a solution of the second-order equation
(4.6).
3. The original known solution f and the ”new” solution g are linearly independent
solutions of (4.6), and hence the general solution of (4.6) may be expressed as the linear
combination
y (x) = c1 f (x) + c2 g (x) .
 46

Example 4.14. Given that y = x is a solution of

d2 y dy
x2 + 1 2
− 2x + 2y = 0, (4.15)
dx dx
find a linearly independent solution by reducing the order.

Solution: First observe that y = x does satisfy the given equation. Then let
y = xv.
Then
dy dv d2 y d2 v dv
=v+x and 2
= x 2
+2 .
dx dx dx dx dx
dy 2
d y
Substituting the expressions for y, and into (4.15), we obtain
dx dx2
 2   
2

d v dv dv
x +1 x 2 +2 − 2x v + x + 2xv = 0
dx dx dx
or

d2 v dv
x x2 + 1 2
+2 = 0.
dx dx
Letting
dv dw d2 v
w= ⇒ = 2
dx dx dx
we obtain the first-order homogeneous linear equation

dw
x x2 + 1 + 2w = 0.
dx
Treating this as a separable equation, we obtain
dw 2dx
=−
w x (x2 + 1)
or  
dw 2 2x
= − + 2 dx.
w x x +1
Integrating, we obtain the general solution
c x2 + 1


w= .
x2
Choosing c = 1, we recall that w = dv/dx and integrate obtain the function v given by
1
v (x) = x − .
x
Now forming g = f v, where f (x) denotes the known solution x, we obtain the function g
defined by  
1
g (x) = x x − = x2 − 1.
x
By Theorem 4.8. we know that this is the desired linearly independent solution. The
general
solution of (4.15) may thus be expressed as the linear combination c1 x + c2 x2 − 1 of the
linearly independent solutions f and g. We thus write the general solution of the given
equation as
y (x) = c1 x + c2 x2 − 1 .


 47

Example 4.15. Given that y = x + 1 is a solution of

d2 y dy
(x + 1)2 2
− 3 (x + 1) + 3y = 0,
dx dx
find a linearly independent solution by reducing the order. Write the general solution.

D. The Nonhomogeneous Equation

We now return briefly to the nonhomogeneous equation

dn y dn−1 y dy
a0 (x) n
+ a1 (x) n−1
+ · · · + an−1 (x) + an (x) y = F (x) .
dx dx dx
The basic theorem dealing with this equation is the following.

Theorem 4.9. Let v be any solution of the given (nonhomogeneous) n-th order linear differ-
ential equation (4.1). Let u be any solution of the corresponding homogeneous equation (4.5).
Then u + v is also a solution of the given (nonhomogeneous) equation (4.1).

Example 4.16. Observe that y = x is a solution of the nonhomogeneous equation

d2 y
+ y = x.
dx2

Solution: y = sin x is a solution of the corresponding homogeneous equation

d2 y
+ y = 0.
dx2
Then by Theorem 4.8. the sum
sin x + x
is also a solution of the given nonhomogeneous equation

d2 y
+ y = x.
dx2

Theorem 4.10. Let yp be a given solution of the n-th order nonhomogeneous linear equation
(4.1) involving no arbitrary constants. Let

yc = c1 y1 + c2 y2 + · · · + cn yn

be the general solution of the corresponding homogeneous equation (4.5). Then every solution
φ of the n-th order nonhomogeneous equation (4.1) can be expressed in the form

yc + yp ,

that is,
c1 y1 + c2 y2 + · · · + cn yn + yp
for suitable choice of the n arbitrary constants c1 , c2 , · · ·, cn .
 48

Definition 4.7. Consider the n-th order nonhomogeneous linear differential equation

dn y dn−1 y dy
a0 (x) n
+ a1 (x) n−1
+ · · · + an−1 (x) + an (x) y = F (x)
dx dx dx
and the corresponding homogeneous equation

dn y dn−1 y dy
a0 (x) n
+ a1 (x) n−1 + · · · + an−1 (x) + an (x) y = 0.
dx dx dx

1. The general solution of (4.5) is called complementary function of (4.1). We will denote
this by yc .
2. Any particular solution of (4.1) involving no arbitrary constants is called a particular
integral of (4.1). We will denote this by yp .
3. The solution yc + yp of (4.1), where yc is the complementary function and yp is a
particular integral of (4.1), is called the general solution of (4.1).

Example 4.17. Consider the differential equation

d2 y
+ y = x.
dx2
The complementary function is the general solution

yc = c1 sin x + c2 cos x

of the corresponding homogeneous equation

d2 y
+ y = 0.
dx2
A particular integral is given by
yp = x.
Thus the general solution of the given equation may be written

y = yc + yp = c1 sin x + c2 cos x + x.

Theorem 4.11. Let f1 be a particular integral of

dn y dn−1 y dy
a0 (x) n
+ a1 (x) n−1
+ · · · + an−1 (x) + an (x) y = F1 (x) . (4.16)
dx dx dx
Let f2 be a particular integral of

dn y dn−1 y dy
a0 (x) n
+ a1 (x) n−1
+ · · · + an−1 (x) + an (x) y = F2 (x) . (4.17)
dx dx dx
Then k1 f1 + k2 f2 is a particular integral of

dn y dn−1 y dy
a0 (x) + a1 (x) + · · · + an−1 (x) + an (x) y = k1 F1 (x) + k2 F2 (x) , (4.18)
dxn dxn−1 dx
where k1 and k2 are constants.
 49

Example 4.18. Suppose we seek a particular integral of

d2 y
+ y = 3x + 5x2 . (4.19)
dx2
We may then consider the two equations

d2 y
+y =x (4.20)
dx2
and
d2 y
+ y = x2 . (4.21)
dx2
We have already noted in last example that a particular integral of (4.20) is given by

y = x.

Using the Theorem 4.11 we can find the particular integral of (4.19)

y = 3x + 5x2 − 10.

4.2 THE HOMOGENEOUS LINEAR EQUATION WITH CONSTANTS

COEFFICIENTS

A. Introduction

We consider the following differential equation

dn y dn−1 y dy
a0 + a1 + · · · + an−1 + an y = 0, (4.22)
dxn dxn−1 dx
where a0 , a1 , · · ·, an are real constants. We show that the general solution of this equation can
be found explicitly.
Assume that y = emx is a solution for certain m. Then

y 0 = memx , y 00 = m2 emx , · · ·, y (n) = mn emx .

Substituting in (4.22), we obtain

a0 mn emx + a1 mn−1 emx + · · · + an−1 memx + an emx = 0

or
emx a0 m + a1 mn−1 + · · · + an−1 m + an = 0.

Since emx 6= 0, we obtain the polynomial equation the unknown m,

a0 m + a1 mn−1 + · · · + an−1 m + an = 0. (4.23)

This equation is called the auxiliary equation or the characteristic equation of (4.22).
If y = emx is a solution of (4.22). Then we see that the constant m satisfies (4.23). Hence
to solve (4.23) we write the auxiliary equation (4.23) and solve it for m. Observe that (4.23)
is formally obtained from (4.22) by merely replacing the k-th order derivative in (4.22) by m
(k = 0, 1, · · ·, m). There cases arise, according as the roots of (4.23) are real and distinct and
repeated, or complex.
 50

B. Case 1. Distinct Real Roots

Suppose that the roots of (4.23) are then distinct real numbers m1 , m2 , · · ·, mn . Then
em1 x , em2 x , · · ·, emn x
are n distinct solutions of (4.22). Further, using the Wronskian determinant one may show
that these n solution are linearly independent. Thus we have the following results.
Theorem 4.12. Consider the n-th order homogeneous linear differential equation (4.22) with
constant coefficients. If the auxiliary equation (4.23) has the n distinct real roots m1 , m2 , · ·
·, mn , then the general solution of (4.22) is
y = c1 em1 x + c2 em2 x + · · · + cn emn x ,
where c1 , c2 , · · ·, cn are arbitrary constants.
Example 4.19. Solve the differential equation
d2 y dy
−3 + 2y = 0.
dx2 dx
Solution: The auxiliary equation is
m2 − 3m + 2 = 0.
Hence
(m − 1) (m − 2) = 0 ⇒ m1 = 1, m2 = 2.
Since the roots are distinct, the general solution of the given differential equation is
y = c1 ex + c2 e2x .
It is easy to see that  x
e2x

x 2x
 e
 = e3x 6= 0.


W e ,e =  x
e 2e2x 

Therefore, ex and e2x are linearly independent.

Example 4.20. Solve the differential equation
d3 y d2 y dy
3
− 4 2
+ + 6y = 0.
dx dx dx
Solution: The auxiliary equation is
m3 − 4m2 + m + 6 = 0.
Hence
(m + 1) (m − 2) (m − 3) = 0 ⇒ m1 = −1, m2 = 2, m3 = 3.
Since the roots are distinct, the general solution of the given differential equation is
y = c1 e−x + c2 e2x + c3 e3x .
It is easy to see that
 e−x e2x e3x
 

W e−x , e2x , e 3x
=  −e−x 2e2x 3e3x  = 12e4x 6= 0.

 

 e−x 4e2x 9e3x 

Therefore, e−x , e2x and e3x are linearly independent.

 51

C. Case 2. Repeated Real Roots

Example 1. Solve the differential equation

d2 y dy
2
−6 + 9y = 0. (4.24)
dx dx

Solution: The auxiliary equation is

m2 − 6m + 9 = 0

or
(m − 3)2 = 0.
The roots of this equation are m1 = 3 and m2 = 3. These are real but not distinct. We must
find a linearly independent solution. Since we already know the one solution e3x , we may
apply Theorem 4.8 and reduce the order. We let y = e3x v, where v is to be determined. Then
dy dv
= e3x + 3e3x v,
dx dx
d2 y 2
3x d v dv
= e + 6e3x + 9e3x v.
dx2 dx2 dx
Substituting into equation (4.24) we have
2
   
3x d v 3x dv 3x 3x dv
e + 6e + 9e v − 6 e + 3e v + 9e3x v = 0
3x
dx2 dx dx
or
d2 v
e3x = 0.
dx2
Letting w = dv/dx, we have the first-order equation

dw dw
e3x = 0 or simply = 0.
dx dx
The solutions of first-order equation are simply w = c, where c is an arbitrary constant.
Choosing the particular solution w = 1 and recalling that dv/dx = w, we find

v (x) = x + c0 ,

where c is an arbitrary constant. By Theorem 4.8 we know that for any choice of the constant
c0 ,
v (x) = (x + c0 ) e3x
is a solution of the given second-order equation (4.24). Further, by Theorem 4.8, we know
that this solution and the previously known solution e3x are linearly independent. Choosing
c0 = 0 we obtain the solution
y = xe3x ,
and thus corresponding to the double root 3 we find the linearly independent solutions e3x
and xe3x of equation (4.24).
Thus the general solution of (4.24) may be written

y = c1 e3x + c2 xe3x = (c1 + c2 x) e3x .

 52

Theorem 4.13. 1. Consider the n-the order homogeneous linear differential equation (4.22)
with constant coefficients. If the auxiliary equation (4.23) has the real root m occurring k
times, then the part of the general solution of (4.22) corresponding to this k-fold repeated root
is  
c1 + c2 x + c3 x2 + · · · + ck xk−1 emx .

2. If, further, the remaining roots of the auxiliary equation (4.23) are the distinct real
numbers mk+1 , · · ·, mn , then the general solution of (4.22) is
 
y = c1 + c2 x + c3 x2 + · · · + ck xk−1 emx + ck+1 emk+1 x + · · · + cn emn x .

3. If, however, any of the remaining roots are also repeated, then the parts of the general
solution of (4.22) corresponding to each of these other repeated roots are expressions similar
to that corresponding to m in part 1.

Example 4.21. Find the general solution of

d3 y d2 y dy
3
− 4 2
−3 + 18y = 0.
dx dx dx

Solution: The auxiliary equation is

m3 − 4m2 − 3m + 18 = 0

has the roots 3, 3, −2. The general solution is

y = c1 e3x + c2 xe3x + c3 e−2x = (c1 + c2 x) e3x + c3 e−2x .

Example 4.22. Find the general solution of

d4 y d3 y d2 y dy
4
− 5 3
+ 6 2
+4 − 8y = 0.
dx dx dx dx

Solution: The auxiliary equation is

m4 − 5m3 + 6m2 + 4m − 8 = 0.

with roots 2, 2, 2, −1. The part of the general solution corresponding to th three-fold root 2 is

y1 = c1 + c2 x + c3 x2 e2x

and that corresponding to the simple root −1 is simply

y2 = c4 e−x .

Thus the general solution is y = y1 + y2 , that is,

y = c1 + c2 x + c3 x2 e2x + c4 e−x .


 53

D. Case 3. Conjugate Complex Roots

Now suppose that auxiliary equation has the complex number a + bi (a, b real, i2 = −1, b 6= 0)
as a nonrepeated root. Then, since the coefficients are real, the conjugate complex number
a − bi is also a nonrepeated root. The corresponding part of the general solution is

k1 e(a+bi)x + k2 e(a−bi)x ,

where k1 and k2 are arbitrary constants. The solutions defined by e(a+bi)x and e(a−bi)x are
complex functions of the real variable x. It is desirable to replace these by two real linearly
independent solutions. This can be accomplished by using Euler’s formula,

eiθ cos θ + i sin θ,

which holds for all real θ. Using this we have:

k1 e(a+bi)x + k2 e(a−bi)x = k1 eax ebix + k2 eax e−bix

h i
= eax k1 eibx + k2 e−ibx
= eax [k1 (cos bx + i sin bx) + k2 (cos bx − i sin bx)]
= eax [(k1 + k2 ) cos bx + i (k1 − k2 ) sin bx]
= eax [c1 sin bx + c2 cos bx] ,

where c1 = i (k1 − k2 ) and c2 = k1 + k2 are two new arbitrary constants. Thus the part of the
general solution corresponding to the nonrepeated conjugate complex roots a ± bi is

eax [c1 sin bx + c2 cos bx] .

Theorem 4.14. 1. Consider the n-th order homogeneous linear differential equation (4.22)
with constant coefficients. If the auxiliary equation (4.23) has the conjugate complex roots
a + bi and a − bi, neither repeated, then the corresponding part of the general solution of
(4.22) may be written
y = eax (c1 sin bx + c2 cos bx) .

2. If, however, a + bi and a − bi are each k-fold roots of the auxiliary equation (4.23), then
the corresponding part of the general solution of (4.22) may be written
h 
y = eax c1 + c2 x + c3 x3 + · · · + ck xk−1 sin bx
  i
+ ck+1 + ck+2 x + ck+3 x2 + · · · + c2k xk−1 cos bx .

Example 4.23. Find the general solution of

d2 y
+ y = 0.
dx2

Solution: The auxiliary equation is

m2 + 1 = 0 ⇒ m2 = −1 ⇒ m = ±i.

Hence
y = c1 sin x + c2 cos x
is the general solution of the given equation.
 54

Example 4.24. Find the general solution of

d2 y dy
−6 + 25y = 0.
dx2 dx

Solution: The auxiliary equation is

m2 − 6m + 25 = 0 ⇒ m = 3 ± 4i ⇒ a = 3, b = 4.

Hence
y = e3x (c1 sin 4x + c2 cos 4x)
is the general solution of the given equation.
Example 4.25. Find the general solution of
d4 y d3 y d2 y dy
4
− 4 3 + 14 2 − 20 + 25y = 0.
dx dx dx dx

Solution: The auxiliary equation is

m4 − 4m3 + 14m2 − 20m + 25 = 0 ⇒ m = 1 ± 2i ⇒ a = 1, b = 2.

Hence
y = ex [(c1 + c2 x) sin 2x + (c3 + c4 x) cos 2x]
is the general solution of the given equation.

E. An Initial-Value Problem

We now apply the results concerning the general solution of a homogeneous linear equation
with constant coefficients to an initial-value problem involving such an equation.
Example 4.26. Solve the initial-value problem
d2 y dy
−6 + 25y = 0, y (0) = −3, y 0 (0) = −1.
dx2 dx

Solution: The auxiliary equation is

m2 − 6m + 25 = 0 ⇒ m = 3 ± 4i ⇒ a = 3, b = 4.

Hence
y = e3x (c1 sin 4x + c2 cos 4x)
is the general solution of the given equation. From this we find
dy
= e3x [(3c1 − 4c2 ) sin 4x + (4c1 + 3c2 ) cos 4x] .
dx
Now, we will apply the initial conditions.

−3 = e0 (c1 sin 0 + c2 cos 0) ⇒ c2 = −3

−1 = e0 [(3c1 − 4c2 ) sin 0 + (4c1 + 3c2 ) cos 0] ⇒ c1 = 2

Therefore
y = e3x (2 sin 4x − 3 cos 4x)
is the exact solution of the given IVP.
 55

Exercises

Solve the following initial-value problems.

d2 y dy
1. 2
− − 12y = 0, y (0) = 3, y 0 (0) = 5.
dx dx
Solution: The auxiliary equation is

m2 − m − 12 = 0 ⇒ (m + 3) (m − 4) = 0 ⇒ m1 = −3, m2 = 4.

So, the general solution of the given IVP is

y = c1 e−3x + c2 e4x .

From that it follows

y 0 = −3c1 e−3x + 4c2 e4x .
Using the given initial conditions, we get
 
y (0) = c1 + c2 = 3, c1 = 1,
0 ⇒
y (0) = −3c1 + 4c2 = 5, c2 = 2.

Therefore
y = e−3x + 2e4x
is the exact solution of the given IVP.

d2 y dy
2. 2
+6 + 9y = 0, y (0) = 2, y 0 (0) = −3.
dx dx
Solution: The auxiliary equation is

m2 + 6m + 9 = 0 ⇒ (m + 3)2 = 0 ⇒ m1,2 = −3.

So, the general solution of the given IVP is

y = (c1 + c2 x) e−3x .

From that it follows

y 0 = e−3x (−3c1 + c2 − 3c2 x) .
Using the given initial conditions, we get
 
y (0) = c1 = 2, c1 = 2,
0 ⇒
y (0) = −3c1 + c2 = −3, c2 = 3.

Therefore
y = (2 + 3x) e−3x
is the exact solution of the given IVP.

d2 y dy
3. 2
−4 + 29y = 0, y (0) = 0, y 0 (0) = 5.
dx dx
Solution: The auxiliary equation is

m2 − 4m + 29 = 0.
 56

Using the quadratic formula

∆ = B 2 − 4AC = 16 − 116 = −100 < 0.

Thus √ √
−B ± ∆ 4 ± −100 4 ± 10i
m1,2 = = = = 2 ± 5i.
2A 2 2
That means a = 2 and b = 5. So, the general solution of the given IVP is

y = e2x (c1 sin 5x + c2 cos 5x) .

From that it follows

y 0 = 2e2x (c1 sin 5x + c2 cos 5x) + e2x (5c1 sin 5x − 5c2 cos 5x) .

Using the given initial conditions, we get

 
y (0) = (c1 · 0 + c2 · 1) = 0, c2 = 0,
0 ⇒
y (0) = 2c1 · 0 + 5c1 · 1 = 5, c1 = 1.

Therefore
y = e2x sin 5x
is the exact solution of the given IVP.

4.3 THE METHOD OF UNDETERMINED COEFFICIENTS

A. Introduction; An Illustrative Example

We now consider the (nonhomogeneous) differential equation

dn y dn−1 y dy
a0 n
+ a1 n−1
+ · · · + an−1 + an y = F (x) , (4.25)
dx dx dx
where the coefficients a0 , a1 , · · ·, an are constants but where the nonhomogeneous term F is
(in general) a nonconstant function of x. Recall that the general solution of (4.25) may be
written
y = yc + yp ,
where yc and yp are complementary and particular solutions, respectively.
Up to now we learned how to find the complementary function. Now we consider methods
of determining a particular solution (integral).

Example 4.27. Introductory Example

d2 y dy
2
−2 − 3y = 2e4x (4.26)
dx dx
We proceed to seek a particular solution yp , but what type of function might by possible candi-
date for such a particular solution? The differential equation (4.26) requires a solution which
is such that its second derivative, minus twice its first derivative, minus three times the solu-
tion itself, add up to twice the exponential function e4x . Thus we assume a particular solution
of the form
yp = Ae4x . (4.27)
 57

From that it follows

yp0 = 4Ae4x and yp00 = 16Ae4x .
Then, substituting into (4.26), we obtain

16Ae4x − 2 4Ae4x − 3Ae4x = 2e4x

or
2
5Ae4x = 2e4x ⇒ A = .
5
So the particular solution is
2
yp = e4x .
5

B. The Method

We begin by introducing certain preliminary definitions.

Definition 4.8. We will call a UC function if it is either (1) a function defined by one
of the following:
(i) xn , where n is a positive integer or zero,
(ii) eax , where a is a constant 6= 0,
(iii) sin (bx + c) , where b and c are constants, b 6= 0,
(iv) cos (bx + c) , where b and c are constants, b 6= 0,

or (2) a function defined as a finite product of two or more functions of these four types.

Example 4.28. Examples of UC functions of the four basic types (i), (ii), (iii), (iv) of the
previous definition are those defined respectively,
 
3x  π
x3 , e−2x , sin , cos 2x + .
2 4

Examples of UC functions defined as finite products of two or more of these four basic types
are those defined respectively by

x2 e3x , x cos 2x, e5x sin 2x, sin 2x cos 5x, x3 e4x sin 5x.

The method of undetermined coefficients applies when the nonhomogeneous function F in the
differential equation is a finite linear combination of UC functions.

Definition 4.9. Consider a UC function f . The set of consisting of f itself and all
linearly independent UC functions of which the successive derivatives of f are either constant
multiples or linear combinations will be called the UC set of f.
 58

UC function UC
 n setn−1 n−2
xn

1 x ,x ,x , · · ·, x, 1
2 eax {eax }
3 sin (bx + c) or cos (bx + c) {sin
 n (bx + c) , cos (bx + c)}
4 xn eax x e , xn−1 eax , · · ·, xeax , eax
ax

{xn sin (bx + c) , xn cos (bx + c) ,

xn−1 sin (bx + c) , xn−1 cos (bx + c) ,
5 xn sin (bx + c) or xn cos (bx + c) · · ·,
x sin (bx + c) , x cos (bx + c) ,
sin (bx + c) , cos (bx + c)}
6 eax sin (bx + c) or eax cos (bx + c) {eax sin (bx + c) , eax cos (bx + c)}
{xn eax sin (bx + c) , xn eax cos (bx + c) ,
x n−1 eax sin (bx + c) , xn−1 eax cos (bx + c) ,
7 xn eax sin (bx + c) or xn eax cos (bx + c) · · ·,
xeax sin (bx + c) , xeax cos (bx + c) ,
eax sin (bx + c) , eax cos (bx + c)}

Example 4.29. The function f defined for all real x by f (x) = x3 is a UC function. Com-
puting derivatives of f, we find
f 0 (x) = 3x2 , f 00 (x) = 6x, f 000 (x) = 6, f (n) (x) = 0 for n > 3.
The linearly independent UC functions of which the successive derivatives of f are either
constants multiples or linear combinations are those given by
x2 , x, 1.
Thus the UC set of x3 is the set S = x3 , x2 , x, 1 .


Example 4.30. The function f defined for all real x by f (x) = sin 2x is a UC function.
Computing derivatives of f, we find
f 0 (x) = 2 cos 2x, f 00 (x) = −4 sin 2x, · · ·.
The only linearly independent UC function of which the successive derivatives of f are either
constants multiples or linear combinations is that given by cos 2x.Thus the UC set of sin 2x is
the set S = {sin 2x, cos 2x} .
Example 4.31. The function f defined for all real x by f (x) = x2 sin x is the product of
the two UC functions defined by x2 and sin x. Hence f is itself a UC function. Computing
derivatives of f , we find
f 0 (x) = 2x sin x + x2 cos x,
f 00 (x) = 2 sin x + 4x cos x − x2 sin x,
f 000 (x) = 6 cos x − 6x sin x − x2 cos x, · · · .
No ”new” types of functions will occur from further differentiation. Each derivative of f is a
linear combination of certain of the six UC functions given by
x2 sin x, x2 cos x, x sin x, x cos x, sin x and cos x.
Thus the set
S = x2 sin x, x2 cos x, x sin x, x cos x, sin x, cos x


is the UC set of x2 sin x.

 59

C. Examples

A few illustrative examples, with reference to the above outline, should make the procedure
clear.

Example 4.32. Solve the differential equation

d2 y dy
−2 − 3y = 2ex − 10 sin x.
dx2 dx

Solution: The corresponding homogeneous equation is

d2 y dy
2
−2 − 3y = 0.
dx dx
The auxiliary equation is

m2 − 3m − 3 = 0 ⇒ m1 = −1, m2 = 3.

Hence the complementary solution is

yc = c1 e−x + c2 e3x ,

where c1 and c2 are arbitrary constants. Now we will find a particular solution. It is easy to
see that
yp = Aex + B sin x + C cos x.
Then
yp0 = Aex + B cos x − C sin x,
yp00 = Aex − B sin x − C cos x.
Actually substituting, we find

(Aex − B sin x − C cos x) − 2 yp0 = Aex + B cos x − C sin x

−3 (Aex + B sin x + C cos x) = 2ex − 10 sin x.

From last equation, we find that
1
A = − , B = 2, C = −1.
2
Hence we obtain the particular solution
1
yp = − ex + 2 sin x − cos x.
2
Therefore the general solution of the differential equation under consideration is
1
y = yc + yp = c1 e−x + c2 e3x − ex + 2 sin x − cos x.
2
Example 4.33. Solve the differential equation

d2 y dy
2
−3 + 2y = 2x2 + ex + 2xex + 4e3x .
dx dx
 60

Solution: The corresponding homogeneous equation is

d2 y dy
2
−3 + 2y = 0.
dx dx
The auxiliary equation is

m2 − 3m + 2 = 0 ⇒ m1 = 1, m2 = 2.

Hence the complementary solution is

yc = c1 ex + c2 e2x ,

where c1 and c2 are arbitrary constants. Now we will find a particular solution.
1. from the UC set for each of these functions. We have

S1 = x2 , x, 1 ,


S2 = {ex } ,
S3 = {xex , ex } ,
S4 = e3x .


2. We note that S2 is completely included in S3 , so S2 is omitted from further considera-

tion, leaving the three sets

S1 = x2 , x, 1 , S3 = {xex , ex } , S4 = e3x .
 

3. We now observe that S3 = {xex , ex } includes ex , which is included in the comple-

mentary solution and so is a solution of the corresponding homogeneous differential equation.
Thus we multiply each member of S3 by x to obtain the revised family

S30 = x2 ex , xex ,


which contains no members that are solutions of the corresponding homogeneous equation.
4. Thus there remain the original UC sets

S1 = x2 , x, 1 and S4 = e3x
 

and the revised set

S30 = x2 ex , xex .


These contain the six elements

x2 , x, 1, e3x , x2 ex , xex .

We form the linear combination

Ax2 + Bx + C + De3x + Ex2 ex + F xex

of these six elements.

5. Thus we take as our particular solution

yp = Ax2 + Bx + C + De3x + Ex2 ex + F xex .

 61

From this, we have

yp0 = 2Ax + B + 3De3x + Ex2 ex + 2Exex + F xex + F ex ,

yp00 = 2A + 9De3x + Ex2 ex + 4Exex + 2Eex + F xex + 2F ex .

We substitute yp , yp0 , yp00 into the differential equation for y, dy/dx, d2 y/dx2 , respectively, to
obtain:
2A + 3De3x + Ex2 ex + (4E + F ) xex + (2E + 2F ) ex
−3 2Ax + B + 3De3x + Ex2 ex + (2E + F ) xex + F ex
 

+2 Ax2 + Bx + C + De3x + Ex2 ex + F xex

= 2x2 + ex + 2xex + 4e3x ,

or
(2A − 3B + 2C) + (2B − 6A) x + 2Ax2 + 2De3x + (−2E) xex + (2E − F ) ex
= 2x2 + ex + 2xex + 4e3x .
Equating coefficients of like terms, we have:


 2A − 3B + 2C = 0,



 2B − 6A = 0,
2A = 2,


 2D = 4,
−2E = 2,




2E − F = 1.


From this 

 A = 1,



 B = 3,
C = 7/2,


 D = 2,
E = −1,




F = −3,


and so the particular solution is

7
yp = x2 + 3x + + 2e3x − x2 ex − 3xex .
2
Therefore the general solution is
7
y = yc + yp = c1 ex + c2 e2x + x2 + 3x + + 2e3x − x2 ex − 3xex .
2
Example 4.34. Find the general solution of

d4 y d2 y
+ = 3x2 + 4 sin x − 2 cos x.
dx4 dx2

Solution: The corresponding homogeneous equation is

d4 y d2 y
+ = 0.
dx4 dx2
 62

The auxiliary equation is

m4 + m2 = 0 ⇒ m2 m2 + 1 = 0 ⇒ m1,2 = 0, m3,4 = ±i.

Hence the complementary solution is

yc = c1 + c2 x + c3 sin x + c4 cos x,

where c1 , c2 , c3 and c4 are arbitrary constants. The nonhomogeneous term is the linear com-
bination
3x2 + 4 sin x − 2 cos x
of three UC functions given by
x2 , sin x, and cos x.

1. From the UC set for each of these three functions. These sets are, respectively,

S1 = x2 , x, 1 ,


S2 = {sin x, cos x} ,
S3 = {cos x, sin x} .

2. Observe that S2 and S3 are identical and so we retain only one of them, leaving the
two sets
S1 = x2 , x, 1 , S2 = {sin x, cos x} .


3. Now observe that S1 = x2 , x, 1 includes 1 and x, which, as the complementary



solution shows, are both solutions of the corresponding homogeneous differential equation.
Thus we multiply each member of the set S1 by x2 to obtain revised set

S10 = x4 , x2 , x2 ,


none of whose members are solutions of the homogeneous differential equation. We observe
that multiplication by x instead of x2 would not be sufficient, since the resulting set would
be x3 , x2 , x , which includes the homogeneous solution x. Turning to the set S2 , observe

that both of its members, sin x and cos x, are also solutions of the homogeneous differential
equation. Hence we replace S2 by the revised set

S20 = {x sin x, x cos x} .

4. None of the UC sets remain here. They have been replaced by the revised sets S10 and
S20 containing the five elements

x4 , x3 , x2 , x sin x, x cos x.

We form a linear combination of these

Ax4 + Bx3 + Cx2 + Dx sin x + Ex cos x,

with undetermined coefficients A, B, C, D, E.

5. We now take this as our particular solution

yp = Ax4 + Bx3 + Cx2 + Dx sin x + Ex cos x.

 63

Then
yp0 = 4Ax3 + 3Bx2 + 2Cx + Dx cos x + D sin x − Ex sin x + E cos x,
yp00 = 12Ax2 + 6Bx + 2C − Dx sin x + 2D cos x − Ex cos x − 2E sin x,
yp000 = 24Ax + 6B − Dx cos x − 3D sin x + Ex sin x − 3E cos x,
yp(4) = 24A + Dx sin x − 4D cos x + Ex cos x + 4E sin x.
Substituting into the differential equation, we obtain

24A + Dx sin x − 4D cos x + Ex cos x + 4E sin x

12Ax2 + 6Bx + 2C − Dx sin x + 2D cos x − Ex cos x − 2E sin x

= 3x2 + 4 sin x − 2 cos x.
equating coefficients, we find 

 24A + 2C = 0,
6B = 0,



12A = 3,
−2D = −2,




2E = 4.


Hence 

 A = 1/4,
 B = 0,


C = −3,
D = 1,




E = 2,


and the particular solution is

1
yp = x4 − 3x2 + x sin x + 2x cos x.
4
Therefore the general solution is
y = yc + y p
1
= c1 + c2 x + c3 sin x + c4 cos x + x4 − 3x2 + x sin x + 2x cos x,
4
where c1 , c2 , c3 and c4 are arbitrary constants.

Example 4.35. Solve the initial-value problem

d2 y dy
2
−8 + 15y = 9xe2x , y (0) = 5, y 0 (0) = 10.
dx dx

4.4 VARIATION OF PARAMETERS

A. The Method

While the process of carrying out of the method of UC is actually quite straightforward
(involving only techniques of college algebra and differentiation), the method applies in general
to a rather small class of problems. For example, it would not apply to the apparently simply
equation
d2 y
+ y = tan x.
dx2
 64

We thus seek a method of finding a particular integral that applies in all cases (including
variable coefficients) in which the complementary function is known. Such a method is the
method of variation of parameters, which we now consider.
We will develop this method in connection with the general second order linear differential
equation with variable coefficients

d2 y dy
a0 (x) 2
+ a1 (x) + a2 (x) y = F (x) . (4.28)
dx dx
Suppose that y1 and y2 are linearly independent solutions of the corresponding homogeneous
equation
d2 y dy
a0 (x) 2 + a1 (x) + a2 (x) y = 0. (4.29)
dx dx
Then the complementary solution of (4.28) is

c1 y1 (x) + c2 y2 (x) ,

where c1 and c2 are arbitrary constants. The procedure in the method of variation of pa-
rameters is to replace the arbitrary constants c1 and c2 in the complementary solution by
respective functions v1 and v2 which will be determined so that the resulting function, which
is defined by
v1 (x) y1 (x) + v2 (x) y2 (x) , (4.30)
will be particular solution of (4.28) (hence the name, variation of parameters).
We have at our disposal the two functions v1 and v2 with which to satisfy the one condition
that (4.30) be a solution of (4.28). Since we have two functions but only one condition on
them, we are thus free to impose a second condition, provided this second condition does not
violate the first one. We will see when and how to impose additional condition as we proceed.
We thus assume a solution of the form (4.30) and write

yp = v1 (x) y1 (x) + v2 (x) y2 (x) . (4.31)

Differentiating (4.31), we have

yp0 = v1 (x) y10 (x) + v2 (x) y20 (x) + v10 (x) y1 (x) + v20 (x) y2 (x) (4.32)

where we use primes to denote differentiations. At this point we impose the aforementioned
second condition; we simplify yp0 by demanding that

v10 (x) y1 (x) + v20 (x) y2 (x) = 0. (4.33)

With this condition imposed, (4.32) reduces to

yp0 = v1 (x) y10 (x) + v2 (x) y20 (x) . (4.34)

Now, differentiation (4.34), we obtain

yp00 = v1 (x) y100 (x) + v2 (x) y200 (x) + v10 (x) y10 (x) + v20 (x) y20 (x) . (4.35)

We substitute (4.31), (4.34) and (4.35) for y, y 0 and y 00 , respectively, into (4.28) and obtain
the identity

a0 (x) v1 (x) y100 (x) + v2 (x) y200 (x) + v10 (x) y10 (x) + v20 (x) y20 (x)
 
 65

+a1 (x) v1 (x) y10 (x) + v2 (x) y20 (x)

 

+a2 (x) [v1 (x) y1 (x) + v2 (x) y2 (x)] = F (x) .

This can be written as

v1 (x) a0 (x) y100 (x) + a1 (x) y10 (x) + a2 (x) y1 (x)

 

+v2 (x) a0 (x) y200 (x) + a1 (x) y20 (x) + a2 (x) y2 (x)
 

+a0 (x) v10 (x) y10 (x) + v20 (x) y20 (x) = F (x) .
 
(4.36)
Since y1 and y2 are solutions of the corresponding homogeneous differential equation (4.29),
the expressions in the first two brackets in (4.36) are identically zero. This leaves merely

F (x)
v10 (x) y10 (x) + v20 (x) y20 (x) = . (4.37)
a0 (x)

So, we have  0
v (x) y1 (x) + v20 (x) y2 (x) = 0,
 1



F (x)
 v10 (x) y10 (x) + v20 (x) y20 (x) = .


a0 (x)
Using Crammer’s rule, we get
 

 0 y2 (x) 

 F (x)
0

 a (x) y2 (x)
 
0
 F (x) y2 (x)
v10 (x) =   =− ,
 y1 (x) y2 (x)  a0 (x) W [y1 (x) , y2 (x)]
 y 0 (x) y 0 (x)
 

1 2

and  

 y1 (x) 0 

F (x)
y10 (x)
 
 
0
 a0 (x)  F (x) y1 (x)
v1 (x) =   = .
 y1 (x) y2 (x)  a0 (x) W [y1 (x) , y2 (x)]
y10 (x) y20 (x)
 
 

Thus we obtain the functions v1 and v2 defined by

Z x
F (x) y2 (x)
v1 (x) = − dt, (4.38)
x0 a0 (x) W [y1 (x) , y2 (x)]

Z x
F (x) y1 (x)
v2 (x) = dt. (4.39)
x0 a0 (x) W [y1 (x) , y2 (x)]
Therefore the particular solution of (4.28) is

yp (x) = v1 (x) y1 (x) + v2 (x) y2 (x) ,

where v1 and v2 are defined by (4.38) and (4.39).

 66

B. Examples

Example 4.36. Solve the differential equation

d2 y
+ y = tan x. (4.40)
dx2

Solution: The complementary solution is

yc (x) = c1 sin x + c2 cos x.

We assume
yp (x) = v1 (x) sin x + v2 (x) cos x. (4.41)
Differentiating the last equation, we obtain

yp0 (x) = v1 (x) cos x − v2 (x) sin x + v10 (x) sin x + v20 (x) cos x.

We impose the condition

v10 (x) sin x + v20 (x) cos x = 0.
So
yp0 (x) = v1 (x) cos x − v2 (x) sin x. (4.42)
We take one more derivative from the last equation to get

yp00 (x) = −v1 (x) sin x − v2 (x) cos x + v10 (x) cos x − v20 (x) sin x. (4.43)

Substituting (4.41), (4.42) and (4.43) into (4.40) we obtain

v10 (x) cos x − v20 (x) sin x = tan x. (4.44)

Hence we have the following system

 0
v1 (x) sin x + v20 (x) cos x = 0,
v10 (x) cos x − v20 (x) sin x = tan x.

We will use here Crammer’s rule. The main determinant is

 
 sin x cos x 
∆=  = − sin2 x − cos2 x = −1.
cos x − sin x 

So  

 0 cos x 

 tan x − sin x  0 − cos x tan x
v10 (x) = = = sin x
−1 −1
and  
 sin x 0 
 
0
 cos x tan x  sin x tan x sin2 x
v2 (x) = = =− = cos x − sec x.
−1 −1 cos x
Integrating we find 
v1 (x) = − cos x + c3 ,
(4.45)
v2 (x) = sin x − ln |sec x + tan x| + c4 .
Substituting (4.45) into (4.41) we have

yp (x) = (− cos x + c3 ) sin x + (sin x − ln |sec x + tan x| + c4 ) cos x

 67

= − sin x cos x + c3 sin x + sin x cos x − (cos x) (ln |sec x + tan x|) + c4 cos x
= c3 sin x + c4 cos x − (cos x) (ln |sec x + tan x|) .
Since a particular solution is a solution free of arbitrary constants, we may assign any partic-
ular values A and B to c3 and c4 , respectively, and the result will be the particular solution

A sin x + B cos x − (cos x) (ln |sec x + tan x|) .

Thus y = yc + yp becomes

y = c1 sin x + c2 cos x + A sin x + B cos x − (cos x) (ln |sec x + tan x|) ,

which we may write as

y = C1 sin x + C2 cos x − (cos x) (ln |sec x + tan x|) , (4.46)

where C1 = c1 + A and C2 = c2 + B. Therefore (4.46) is general solution of (4.40).

Example 4.37. Solve the differential equation

d3 y d2 y dy
3
− 6 2
+ 11 − 6y = ex . (4.47)
dx dx dx

Solution: The complementary solution is

yc (x) = c1 ex + c2 e2x + c3 e3x . (4.48)

We assume as a particular solution

yp (x) = v1 (x) ex + v2 (x) e2x + v3 (x) e3x . (4.49)

Since we have three functions v1 , v2 , v3 at our disposal in this case, we can apply three condi-
tions. We have:

yp0 (x) = v1 (x) ex + 2v2 (x) e2x + 3v3 (x) e3x + v10 (x) ex + v20 (x) e2x + v30 (x) e3x .

Proceeding in a manner analogous to that of the second order case, we impose the condition

v10 (x) ex + v20 (x) e2x + v30 (x) e3x = 0, (4.50)

leaving
yp0 (x) = v1 (x) ex + 2v2 (x) e2x + 3v3 (x) e3x . (4.51)
Then

yp00 (x) = v1 (x) ex + 4v2 (x) e2x + 9v3 (x) e3x + v10 (x) ex + 2v20 (x) e2x + 3v30 (x) e3x .

We now impose the condition

v10 (x) ex + 2v20 (x) e2x + 3v30 (x) e3x = 0, (4.52)

leaving
yp00 (x) = v1 (x) ex + 4v2 (x) e2x + 9v3 (x) e3x . (4.53)
From this,

yp000 (x) = v1 (x) ex + 8v2 (x) e2x + 27v3 (x) e3x + v10 (x) ex + 4v20 (x) e2x + 9v30 (x) e3x . (4.54)
 68

We substitute (4.49), (4.51), (4.53) and (4.54) into the differential equation (4.47), obtaining:

v1 (x) ex + 8v2 (x) e2x + 27v3 (x) e3x + v10 (x) ex + 4v20 (x) e2x + 9v30 (x) e3x

−6v1 (x) ex − 24v2 (x) e2x − 54v3 (x) e3x

+11v1 (x) ex + 22v2 (x) e2x + 33v3 (x) e3x
−6v1 (x) ex − 6v2 (x) e2x − 6v3 (x) e3x = ex
or
v10 (x) ex + 4v20 (x) e2x + 9v30 (x) e3x = ex . (4.55)
Thus we have  0
 v1 (x) ex + v20 (x) e2x + v30 (x) e3x = 0,
v 0 (x) ex + 2v20 (x) e2x + 3v30 (x) e3x = 0,
 10
v1 (x) ex + 4v20 (x) e2x + 9v30 (x) e3x = ex .
Solving, we find
 0 e2x e3x 
 
 
 0 2e2x 3e3x 
 
6x
 1 1 

 ex 4e2x 9e3x 
 e   
0 2 3  1
v1 (x) =  x 2x 3x
=  = ,
 e
 x e 2x e 3x 

 1 1 1 
  2
 e 2e 3e   6x
e  1 2 3 
 

 ex 4e2x 9e3x   1 4 9 
 x 3x 

 e
 x 0 e 3x   
 e 0 3e  5x  1 1 
 

 ex ex 9e3x  −e  1 3 
v20 (x) =  x 2x 3x 
= 6x
= −e−x ,
 e
 x e e  2e
 e 2e2x 3e3x 
 
 ex 4e2x 9e3x 
 x 2x

 e
 x e 2x 0 

 
 e 2e 0   1 1 
  4x
e   
 ex 4e2x ex  1 2  1
v30 (x) =  x 2x 3x
 = 6x
= e−2x .
 e
 x e 2x e 3x 
 2e 2
 e 2e 3e 
 
 ex 4e2x 9e3x 
We now integrate, choosing all the constants of integration to be zero (as the previous example
showed was possible). We find:

1

 v1 (x) = x,
2






v2 (x) = e−x ,




 v3 (x) = − 1 e−2x .



4
Thus
1 1 1 3
yp (x) = xex + e−x e2x − e−2x e3x = xex + ex .
2 4 2 4
Therefore, the general solution of (4.47) is
1 3
y = yc + yp = c1 ex + c2 e2x + c3 e3x + xex + ex
2 4
 69

or
1
y = C1 ex + c2 e2x + c3 e3x + xex ,
2
3
where C1 = c1 + .
4
Example 4.38. Find the general solution of

d2 y dy
2
x2 + 1 − 2x + 2y = 6 x2
+ 1 . (4.56)
dx2 dx
given that y = x and y = x2 − 1 are linearly independent solutions of the corresponding
homogeneous equation.

Solution: We see that the complementary function of equation (4.56) is

yc (x) = c1 x + c2 x2 − 1 ,

where c1 and c2 are arbitrary constants. To find a particular solution of (4.56), we therefore
let
yp (x) = v1 (x) x + v2 (x) x2 − 1 .


(4.57)
Then
yp0 (x) = v1 (x) + 2xv2 (x) + v10 (x) x + v20 (x) x2 − 1 .

We impose the condition

v10 (x) x + v20 (x) x2 − 1 = 0,


(4.58)
leaving
yp0 (x) = v1 (x) + 2xv2 (x) . (4.59)
From this, we find
yp00 (x) = 2v2 (x) + v10 (x) + 2xv20 (x) . (4.60)
Substituting (4.57),(4.59) and (4.60) into (4.56) we obtain

x2 + 1 2v2 (x) + v10 (x) + 2xv20 (x) − 2x [v1 (x) + 2xv2 (x)]

 

2
+2 v1 (x) x + v2 (x) x2 − 1 = 6 x2 + 1



or
2
x2 + 1 v10 (x) + 2xv20 (x) = 6 x2 + 1 .

 
(4.61)
Thus we have two equations (4.58) and (4.61) from which to determine v10 (x) and v20 (x) ; that
is, v10 (x) and v20 (x) satisfy the system
 0
v1 (x) x + v20 (x) x2 − 1 = 0,

 

v10 (x) + 2xv20 (x) = 6 x2 + 1 .

Solving this system, we find

0
x2 − 1
 
 
 
 6 x2 + 1 −6 x2 + 1 x2 − 1



2x 
v10 (x) = = −6 x2 − 1 ,


=
 x x2 − 1  2
 
  x +1
 1 2x 
 70

 
 x 0

 
 1 6 x2 + 1 6x x2 + 1



v20 (x) = = = 6x.
x2 + 1 x2 + 1
Integrating, we obtain
v1 (x) = −2x3 + 6x, v2 (x) = 3x2 , (4.62)
where we have chosen both constants of integration to be zero. Substituting (4.62) into (4.57),
we have
yp (x) = −2x3 + 6x x + 3x2 x2 − 1 = x4 + 3x2 .

Therefore the general solution of (4.56) is

y = yc + yp = c1 x + c2 x2 − 1 + x4 + 3x2 .

4.5 THE CAUCHY-EULER EQUATION

A. The Equation and The Method of Solution

An equation with variable coefficients of the form

dn y dn−1 y dy
a0 xn n
+ a1 xn−1 n−1 + · · · + an−1 x + an y = F (x) , (4.63)
dx dx dx
where a0 , a1 , · · ·, an−1 , an are constants.

Theorem 4.15. The transformation x = et reduces the equation

dn y dn−1 y dy
a0 xn n
+ a1 xn−1 n−1 + · · · + an−1 x + an y = F (x) ,
dx dx dx
to a linear differential equation with constants coefficients.

Proof. This what we need! We will prove this theorem for the case of the second order
Cauchy-Euler differential equation

d2 y dy
a0 x2 2
+ a1 x + a2 y = F (x) . (4.64)
dx dx
The proof in the general n-th order case proceeds in a similar fashion. Letting x = et , assuming
x > 0, we have t = ln x. Then
dy dy dt 1 dy
= =
dx dt dx x dt
and
d2 y
     
d dy d 1 dy 1 dy 1 d dy
= = =− 2 +
dx2 dx dx dx x dt x dt x dx dt
 2   2 
1 dy 1 d y dt 1 dy 1 d y1
=− 2 + =− 2 +
x dt x dt2 dx x dt x dt2 x
1 d2 y 1 d2 y dy
 
1 dy
=− 2 + 2 2 = 2 − .
x dt x dt x dt2 dt
Thus
dy dy d2 y d2 y dy
x = and x2 2 = 2 − .
dx dt dx dt dt
 71

Substituting into (4.64), we obtain

 2   
d y dy dy t


a0 − + a1 + a2 y = F e
dt2 dt dt
or
d2 y dy
A0 2
+ A1 + A2 y = G (t) , (4.65)
dt dt
where A0 = a0 , A1 = a1 −a0 , A2 = a2 , G (t) = F et . This is a second order linear differential

equation with constants coefficients, which was what we wished to show.

Remark 4.1. Note that the leading coefficient a0 xn in (4.63) is zero for x = 0. Thus the basic
interval a ≤ x ≤ b, referred to in the general theorems of Section 4.1, does not include x = 0.
Remark 4.2. Observe that in the above proof we assumed that x > 0. If x < 0, the substitution
x = −et is actually correct one. Unless the contrary is explicitly stated, we shall assume x > 0
when finding the general solution of a Cauchy-Euler differential equation.

B. Examples

Example 4.39. Find the general solution of

d2 y dy
x2 − 2x + 2y = x3 . (4.66)
dx2 dx

Solution: Let x = et . Then, assuming x > 0, we have t = ln x, and

dy dy dt 1 dy
= = ,
dx dt dx x dt
d2 y 1 d2 y dy
 
= 2 − .
dx2 x dt2 dt
Thus equation (4.66) becomes
d2 y dy
2
− 3 + 2y = e3t . (4.67)
dt dt
The complementary solution of (4.67) is yc = c1 et + c2 e2t . We find a particular solution by
the method of UC. We assume yp = Ae3t . Then yp0 = 3Ae3t , yp00 = 9Ae3t , and substituting into
(4.67), we obtain
2Ae3t = e3t .
1 1
Thus A = and we have yp = e3t . The general solution of (4.67) is then
2 2
1
y = c1 et + c2 e2t + e3t .
2
But we are not yet finished! we must return to the original independent variable x. Since
et = x, we find
1
y = c1 x + c2 x2 + x3 .
2
This is the general solution of (4.66).
Example 4.40. Find the general solution of
d3 y 2
2d y dy
x3 3
− 4x 2
+ 8x − 8y = 4 ln x. (4.68)
dx dx dx
 72

Solution: Assuming x > 0, we let x = et . Then t = ln x, and

dy 1 dy
= ,
dx x dt
d2 y 1 d2 y dy
 
= 2 − ,
dx2 x dt2 dt
d3 y 1 d3 y d2 y
 
dy
= 3 −3 2 +2 .
dx3 x dt3 dt dt
Thus, substituting into (4.68), we obtain

d3 y d2 y dy
− 7 + 14 − 8y = 4t. (4.69)
dt3 dt2 dt
the complementary function of the transformed equation (4.69) is

yc = c1 et + c2 e2t + c3 e4t .

We proceed to obtain a particular solution of (4.69) by the method of UC. We assume yp =

At + B. Then yp0 = A, yp00 = 0, yp000 = 0. Substituting into (4.69), we find

14A − 8At − 8B = 4t.

Thus
A = − 21 ,

−8A = 4, 14A − 8B = 0 ⇒
B = − 78 .
Thus the general solution of (4.69)
1 7
y = c1 et + c2 e2t + c3 e4t − t − ,
2 8
and so the general solution of (4.68) is
1 7
y = c1 x + c2 x2 + c3 x4 − ln x − .
2 8
Chapter 5

APPLICATIONS OF SECOND
ORDER LINEAR DIFFERENTIAL
EQUATIONS WITH CONSTANT
COEFFICIENTS

5.1 THE DIFFERENTIAL EQUATION OF THE VIBRATIONS OF A

MASS ON A SPRING

The Basic Problem

A coil spring is suspended vertically from a fixed point on a ceiling, beam, or other similar
object. A mass is attached to its lower end and allowed to come to rest in an equilibrium
position. The system is then set in motion either (1) by pulling the mass down a distance
below its equilibrium position (or pushing it up a distance above it) and subsequently releasing
it with an initial velocity (zero or nonzero, downward or upward) at t = 0; or (2) by forcing
the mass out of its equilibrium position by giving it a nonzero initial velocity (downward or
upward) at t = 0. Our problem is to determine the resulting motion of the mass on the spring.
In order to set up the differential equation for this problem we need two laws of physics:
Newton’s second law and Hooke’s law, namely; F = ma and |F | = kx, respectively.
We take as
d2 x dx
m 2
+a + kx = F (t) , (5.1)
dt dt
the differential equation for the motion of the mass on the spring. Here, m is the mass, a
is the damping constant, k is the spring constant and F (t) is any external impressed forces
that act on the mass. Observe that it is a nonhomogeneous second-order linear differential
equation with constant coefficients. If a = 0 the motion is called undamped ; otherwise it is
called damped. If F (t) = 0 for all t and the motion is called free; otherwise it is called forced.

5.2 FREE, UNDAMPED MOTION

This is a special case such that a = 0 and F (t) = 0 for all t. Then the differential equation
(5.1) reduces to
d2 x
m 2 + kx = 0, (5.2)
dt

73
 74

where m > 0 is the mass and k > 0 is the spring constant. Dividing through by m and letting
k/m = λ2 , we write (5.2) in the form
d2 x
+ λ2 x = 0. (5.3)
dt2
The auxiliary equation is
r 2 + λ2 = 0
has roots r = ±λi and hence the general solution of (5.2) can be written
x = c1 sin λt + c2 cos λt, (5.4)
where c1 and c2 are arbitrary constants.
Let us now assume that the mass was initially displaced a distance x0 from its equilib-
rium position and released from that point with initial velocity v0 . Then, we have the initial
conditions
x (0) = x0 , (5.5)

x0 (0) = v0 . (5.6)
Differentiating (5.4) with respect to t, we have
dx
= c1 λ cos λt − c2 λ sin λt. (5.7)
dt
Applying the conditions (5.5) and (5.6) to equations (5.4) and (5.7), respectively, we see at
once that
c2 = x0 ,
c1 λ = v0 .
Substituting the values c1 and c2 so determined in to equation (5.4) gives the particular
solution of the differential equation (5.2) satisfying the conditions (5.5) and (5.6) in the form
v0
x= sin λt + x0 cos λt.
λ
We put this in an alternative form by first writing it as
 
(v0 /λ) x0
x=c sin λt + cos λt , (5.8)
c c
where r
v0 2
c= + x20 > 0. (5.9)
λ
Then, letting
(v0 /λ)
= − sin φ, (5.10)
c
x0
= cos φ, (5.11)
c
equation (5.8) reduces at once to
x = c cos (λt + φ) , (5.12)
where c is p
given by the equation (5.9) and φ is determined by the equations (5.10) and (5.11).
Since λ = k/m, we now write the solution (5.12) in the form
r !
k
x = c cos t+φ . (5.13)
m
Chapter 6

THE LAPLACE TRANSFORM

6.1 DEFINITION, EXISTENCE, AND BASIC PROPERTIES OF THE

LAPLACE TRANSFORM

A. Definition and Existence

Definition 6.1. Let f be a real valued function of the real variable t, defined for t > 0. Let
s be a variable that we will assume to be real, and consider the function F defined by
Z ∞
F (s) = e−st f (t) dt, (6.1)
0

for all values of s for which this integral exists. The function F defined by the integral (6.1)
is called the Laplace transform of the function f. We will denote the Laplace transform F of
f by L {f } and will denote F (s) by L {f (t)} .
Example 6.1. Consider the function f defined by
f (t) = 1, for t > 0.
Then  −st R
Z ∞ Z ∞
−st −st e
L {1} = e · 1dt = lim e dt = lim −
0 R→∞ 0 R→∞ s 0
1 e−sR
 
1
= lim − =
R→∞ s s s
for all s > 0. Thus we have
1
L {1} =
(s > 0) .
s
Example 6.2. Consider the function f defined by
f (t) = t, for t > 0.
Then R
Z ∞ Z ∞  −st
−st e
−st
L {t} = e tdt = lim e tdt = lim − 2 (st + 1)
0 R→∞ 0 R→∞ s 0
−sR
 
1 e 1
= lim 2
− 2 (sR + 1) = 2
R→∞ s s s
for all s > 0. Thus
1
L {t} = (s > 0) .
s2

75
 76

Example 6.3. Consider the function f defined by

f (t) = eat , for t > 0.

Then " #R
∞ ∞
e(a−s)t
Z Z
at −st at (a−s)t

L e = e e dt = lim e dt = lim −
0 R→∞ 0 R→∞ a−s
0
" #
e(a−s)R 1 1 1
= lim − =− =
R→∞ a−s a−s a−s s−a
for all s > a. Thus
1
L eat =

(s > a) .
s−a
Example 6.4. Consider the function f defined by

f (t) = sin bt, for t > 0.

Then Z ∞ Z ∞
−st
L {sin bt} = e sin btdt = lim e−st sin btdt
0 R→∞ 0
R
e−st

= lim − 2 (s sin bt + b cos bt)
R→∞ s + b2 0

e−sR
 
b b
= lim − (s sin bR + b cos bR) = 2
R→∞ s2 + b2 s2 + b2 s + b2
for all s > 0. Thus
b
L {sin bt} = (s > 0) .
s2 + b2
Example 6.5. Consider the function f defined by

f (t) = cos bt, for t > 0.

Then
" #R
∞ ∞
e(a−s)t
Z Z
−st −st
L {cos bt} = e cos btdt = lim e cos btdt = lim −
0 R→∞ 0 R→∞ a−s
0
R
e−st

= lim (−s cos bt + b sin bt)
R→∞ s2 + b2 0
 −sR 
e s s
= lim 2 2
(−s cos bR + b sin bR) + 2 2
= 2
R→∞ s + b s +b s + b2
for all s > 0. Thus
s
L {cos bt} = (s > 0) .
s2 + b2
 77

B. Basic Properties of The Laplace Transform

Theorem 6.1. The Linearity Property

Let f1 and f2 be functions whose Laplace transform exist, and let c1 and c2 be constants.
Then
L {c1 f1 (t) + c2 f2 (t)} = c1 L {f1 (t)} + c2 L {f2 (t)} . (6.2)

Example 6.6. Use the Theorem 6.1 to find L sin at . Since sin2 at = (1 − cos 2at) /2, we
 2

have  
 2 1 1
L sin at = L − cos 2at .
2 2
By Theorem 6.1,  
1 1 1 1
L − cos 2at = L {1} − L {cos 2at} .
2 2 2 2
By Example 6.1, L {1} = 1/s and by Example 6.5, L {cos 2at} = s/ s2 + 4a2 . Thus

1 1 1 s 2a2
L sin2 at = · − · 2

= .
2 s 2 s + 4a2 s (s2 + 4a2 )

Theorem 6.2. Let f be a real function that is continuous for t ≥ 0 and of exponential order
eαt . Let f 0 be piecewise continuous in every finite closed interval 0 ≤ t ≤ b. Then L {f 0 } exists
for s > α; and
L f 0 (t) = sL {f (t)} − f (0) .


Theorem 6.3. (Generalization) Let f be a real function having a continuous (n − 1)st

derivative f (n−1) (and hence f, f 0 , · · ·, f (n−2) are also continuous) for t ≥ 0; and assume that
f, f 0 , · · ·, f (n−1) are all of exponential order αt
 (n)e . Suppose f
(n) is piecewise continuous in every

finite closed interval 0 ≤ t ≤ b. Then L f exists for s > α; and

n o
L f (n) (t) = sn L {f (t)} − sn−1 f (0) − sn−2 f 0 (0) − sn−3 f 00 (0) − · · · − f (n−1) (0) .

Theorem 6.4. Translation Property

Suppose f is a function such that L {f } exists for s > α. For any constant a,

L eat f (t) = F (s − a)

(6.3)

for s > α + a, where F (s) denotes L {f (t)} .

R∞
Proof. F (s) = L {f (t)} = 0 e−st f (t) dt. Replacing s by s − a, we have
Z ∞ Z ∞
−(s−a)t
e−st eat f (t) dt = L eat f (t) .
  
F (s − a) = e f (t) dt =
0 0

6.2 THE INVERSE TRANSFORM AND THE CONVOLUTION

A. The Inverse Transform

Thus far in this chapter we have been concerned with the following problem: Given function f,
defined for t > 0, to find its Laplace transform, which we denoted by L {f } of F. Now consider
the inverse problem: Given a function F, to find a function f whose Laplace transform is the
 78

given F. We introduce the notation L−1 {F } to denote such a function f, L−1 {F (s)} by f (t) ,
and call such a function an inverse transform of F. That is,

f (t) = L−1 {F (s)}

means that f (t) is such that

L {f (t)} = F (s) .

TABLE 6.1 LAPLACE TRANSFORMS

f (t) = L−1 {F (s)} F (s) = L {f (t)}

1
1 1
s
1
2 eat
s−a

b
3 sin bt
s2 + b2
s
4 cos bt
s2 + b2

b
5 sin hbt
s2 − b2
s
6 cos hbt
s2 − b2

n!
7 tn (n = 1, 2, · · ·)
sn+1

2bs
8 t sin bt
(s2 + b2 )2

s2 − b2
9 t cos bt
(s2 + b2 )2

n!
10 tn eat (n = 1, 2, · · ·)
(s − a)n+1

b
11 e−at sin bt
(s + a)2 + b2

s+a
12 e−at cos bt
(s + a)2 + b2

sin bt−bt cos bt 1

13 2b3
(s2 + b2 )2

t sin bt s
14 2b
(s2 + b2 )2
 79

 
1
Example 6.7. Using the Table 6.1, find L−1 .
s2 + 6s + 13

Solution: Looking in the F (s) column of Table 6.1 we would first look for
1
F (s) = .
as2 + bs + c
However, we find no such F (s) ; but we do find

b
F (s) = .
(s + a)2 + b2

We can put the given expression

1 1 1 2
= 2 = .
s2 + 6s + 13 (s + 3) + 4 2 (s + 3)2 + 22

Thus, using number 11 of Table 6.1, we have

   
−1 1 1 −1 2 1
L 2
= L 2 = e−3t sin 2t.
s + 6s + 13 2 (s + 3) + 2 2 2
 
1
Example 6.8. Using the Table 6.1, find L−1 2
.
s (s + 1)

Solution: No entry of this form appears in the F (s) column of Table 6.1. We employ
the method of partial fractions. We have
1 A B
= + 2
s (s2+ 1) s s +1

and hence A = 1, B = −1, C = 0. Thus

     
−1 1 −1 1 −1 s
L =L −L .
s (s2 + 1) s s2 + 1
   
−1 1 −1 s
By the number 1 of Table 6.1, L = 1 and by the number 4, L = cos t.
s s2 + 1
Thus  
−1 1
L = 1 − cos t.
s (s2 + 1)
 
−1 s+2
Example 6.9. Using the Table 6.1, find L .
s2 + 4s + 7
 
−1 1
Example 6.10. Using the Table 6.1, find L .
s3 + 4s + 3s

B. Convolution

Another important procedure in connection with the use of tables of transforms is that fur-
nished by the so-called convolution theorem which we will state below. We first define the
convolution of two functions f and g.
 80

Definition 6.2. Let f and g are tow functions that are piecewise continuously on every
finite closed interval 0 ≤ t ≤ b and of exponential order. The function denoted by f ∗ g and
defined by Z t
f (t) ∗ g (t) = f (t) g (t − τ ) dτ (6.4)
0
is called the convolution of the functions f and g.
Let us change the variable of integration in (6.4) by means of the substitution u = t − τ .
We have Z t Z 0
f (t) ∗ g (t) = f (t) g (t − τ ) dτ = − f (t − u) g (u) du
0 t
Z t
= g (u) f (t − u) du = g (t) ∗ f (t) .
0
Thus we have shown that
f ∗ g = g ∗ f.
Theorem 6.5. Let the functions f and g be piecewise continuous on every finite closed interval
0 ≤ t ≤ b and of exponential order eat . Then
L {f ∗ g} = L {f } L {g}
for s > a.

Denoting L {f } by F and L {f g} by G, we may write

L {f (t) ∗ g (t)} = F (s) G (s) .
Hence, we have
Z t
−1
L {F (s) G (s)} = f (t) ∗ g (t) = f (t) g (t − τ ) dτ (6.5)
0
also Z t
−1
L {F (s) G (s)} = g (t) ∗ f (t) = g (τ ) f (t − τ ) dτ . (6.6)
0
 
1
Example 6.11. Find L−1 using the convolution.
s (s2 + 1)

Solution: We write s s2 + 1 as the product

  F (s) G (s) , where F (s) = 1/s and G (s) =
1
1/s2 + 1. By the Table 6.1, number 1, L−1 = 1, and by the number 3,
s
 
−1 1
g (t) = L = sin t.
s2 + 1
Thus by (6.5),   Z t
−1 1
L 2
= f (t) ∗ g (t) = 1 · sin (t − τ ) dτ ,
s (s + 1) 0
and by (6.6),   Z t
−1 1
L = g (t) ∗ f (t) = sin τ · 1dτ .
s (s2 + 1) 0
The second of these two integrals is slightly more simple. Evaluating it, we have
 
1
L−1 = 1 − cos t.
s (s2 + 1)
 81

 
1
Example 6.12. Find L−1 using the convolution.
s (s2 + 4s + 13)

6.3 LAPLACE TRANSFORM SOLUTION OF LINEAR DIFFERENTIAL

EQUATIONS WITH CONSTANT COEFFICIENTS

A. The Method

We now consider how the Laplace transform may be applied to solve the initial value problem
consisting of nth order linear differential equation with constant coefficients

dn y dn−1 y dy
a0 n
+ a1 n−1 + · · · + an−1 + an y = b (t) , (6.7)
dt dt dt
plus the initial conditions

y (0) = c0 , y 0 (0) = c1 , · · ·, y (n−1) (0) = cn−1 . (6.8)

By Theorem 4.1 this problem has a unique solution. We summarize the procedure as follows:
1. Take the Laplace transform of both sides of the differential equation (6.7), applying
Theorem 6.3 and using the initial conditions (6.8) in the process, and equate the results to
obtain an algebraic equation in the ”unknown” Y (s) .
2. Solve that algebraic equation thus obtained to determine Y (s).
3. Having found Y (s) , employ the table of transforms to determine the solution y (t) =
L−1 {Y (s)} of the given initial value problem.

Example 6.13. Solve the initial value problem

dy
− 2y = e5t , y (0) = 3.
dt

Solution. Taking the Laplace transform of both sides, we obtain

 
dy
− 2L {y} = L e5t

L
dt
or
1
sL {y (t)} − y (0) − 2L {y (t)} = .
s−5
Denoting L {y (t)} = Y (s) , we get
1 3s − 14
[s − 2] Y (s) = +3=
s−5 s−5
or
3s − 14
Y (s) = .
(s − 2) (s − 5)
Now, we must determine  
−1 3s − 14
L .
(s − 2) (s − 5)
We employ partial fractions. We have
3s − 14 A B
= + ,
(s − 2) (s − 5) s−2 s−5
 82

and so
3s − 14 = A (s − 5) + B (s − 2) .
From this we find that A = 8/3 and B = 1/3 and so
     
3s − 14 8 1 1 1
L−1 = L−1 + L−1 .
(s − 2) (s − 5) 3 s−2 3 s−5

Using Table 6.1, we obtain the solution of the given initial value problem
8 1
y (t) = e2t + e5t .
3 3
Example 6.14. Solve the initial value problem

d2 y dy
− 2 − 8y = 0, y (0) = 3, y 0 (0) = 6.
dt2 dt

Solution. Taking the Laplace transform of both sides of the given differential equation,
we have  2   
d y dy
L 2
− 2L − 8L {y} = L {0}
dt dt
or
s2 L {y (t)} − sy (0) − y 0 (0) − 2sL {y (t)} − 2y (0) − 8L {y (t)} = 0.
Denoting L {y (t)} = Y (s) , we get
 2 
s − 2s − 8 Y (s) = 3s

or
3s
Y (s) = .
s2 − 2s − 8
Now, we must determine  
−1 3s
L 2
.
s − 2s − 8
We employ partial fractions. We have
3s A B
= + ,
s2 − 2s − 8 s−4 s+2
and so
3s = A (s + 2) + B (s − 4) .
From this we find that A = 2 and B = 1 and so
     
−1 3s −1 1 −1 1
L = 2L +L .
s2 − 2s − 8 s−4 s+2

Using Table 6.1, we obtain the solution of the given initial value problem

y (t) = 2e4t + e−2t .

Example 6.15. Solve the initial value problem

d2 y
+ y = e−2t sin t, y (0) = 0, y 0 (0) = 0.
dt2
 83

Solution. Taking the Laplace transform of both sides of the given differential equation,
we have  2 
d y
+ L {y (t)} = L e−2t sin t .

L 2
dt
or
1
s2 L {y (t)} − sy (0) − y 0 (0) + L {y (t)} = .
(s + 2)2 + 1
Denoting L {y (t)} = Y (s) , we get
 2  1
s + 1 Y (s) =
(s + 2)2 + 1
or
1
Y (s) = h i.
(s2 + 1) (s + 2)2 + 1

Now, we must determine  

 1 
L−1 h i .
 (s2 + 1) (s + 2)2 + 1 

We employ partial fractions. We have

1 As + B Cs + D
h i= + 2 ,
(s2 + 1) (s + 2) + 12 s2 + 1 s + 4s + 5

and so

1 = (As + B) s2 + 4s + 5 + (Cs + D) s2 + 1

= (A + C) s3 + (4A + B + D) s2 + (5A + 4B + C) s + (5B + D) .

Thus we obtain the equations 


 A + C = 0,
4A + B + D = 0,


 5A + 4B + C = 0
5B + D = 1.


From these equations we find

1 1 1 3
A=− , B= , C= , D= ,
8 8 8 8
and so
     
−1 1 1 −1 s 1 −1 1
L = − L + L (6.9)
(s2 + 1) (s2 + 4s + 5) 8 s2 + 1 8 s2 + 1
 
1 s
+ L−1 2
8 s + 4s + 5
 
3 −1 1
+ L .
8 s2 + 4s + 5

In order to determine
   
1 −1 s 3 1
L 2
+ L−1 2
, (6.10)
8 s + 4s + 5 8 s + 4s + 5
 84

we write
s s+2 2
= − .
s2 + 4s + 5 (s + 2) + 1 (s + 2)2 + 1
2

Thus the expression (6.10) becomes

   
1 −1 s+2 1 −1 1
L + L ,
8 (s + 2)2 + 1 8 (s + 2)2 + 1

and so (6.9) may be written

     
−1 1 1 −1 s 1 −1 1
L = − L + L
(s2 + 1) (s2 + 4s + 5) 8 s2 + 1 8 s2 + 1
 
1 s+2
+ L−1
8 (s + 2)2 + 1
 
1 −1 1
+ L .
8 (s + 2)2 + 1

Using Table 6.1, we obtain the solution of the given initial value problem
1 1 1 1
y (t) = − cos t + sin t + e−2t cos t + e−2t sin t
8 8 8 8
or
1 1
y (t) = (sin t − cos t) + e−2t (sin t + cos t) .
8 8
Example 6.16. Solve the initial value problem

3 2
 d y + 4 d y + 5 dy + 2y = 10 cos t,


dt3 dt2 dt

 y (0) = 0, y 0 (0) = 0, y 00 (0) = 3.


Solution. Taking the Laplace transform of both sides of the given differential equation,
we have  3   2   
d y d y dy
L 3
+ 4L 2
+ 5L + 2L {y (t)} = L {10 cos t}
dt dt dt
or
s3 L {y (t)} − s2 y (0) − sy 0 (0) − y 00 (0)
10s
+4s2 L {y (t)} − 4sy (0) − 4y 0 (0) + 2sL {y (t)} − 2y (0) = .
s2 + 1
Denoting L {y (t)} = Y (s) , we get
 3 10s
s + 4s2 + 5s + 2 Y (s) − 3 = 2

s +1
or
3s2 + 10s + 3
Y (s) = .
(s2 + 1) (s3 + 4s2 + 5s + 2)
Now, we must determine

3s2 + 10s + 3
 
−1
L .
(s2 + 1) (s3 + 4s2 + 5s + 2)
 85

We employ partial fractions. We have

3s2 + 10s + 3 3s2 + 10s + 3

=
(s2 + 1) (s3 + 4s2 + 5s + 2) (s2 + 1) (s + 1)2 (s + 2)

A B C Ds + E
= + + + 2 .
s + 2 s + 1 (s + 1)2 s +1
From this we find

3s2 + 10s + 3 = A (s + 1)2 s2 + 1 + B (s + 2) (s + 1) s2 + 1

+C (s + 2) s2 + 1 + (Ds + E) (s + 2) (s + 1)2 ,

or

3s2 + 10s + 3 = (A + B + D) s4 + (2A + 3B + C + 4D + E) s3

+ (2A + 3B + 2C + 5D + 4E) s2
+ (2A + 3B + C + 2D + 5E) s + (A + 2B + 2C + 2E) .

From this we obtain the system of equations


 A + B + D = 0,

 2A + 3B + C + 4D + E = 0,


2A + 3B + 2C + 5D + 4E = 3,
 2A + 3B + C + 2D + 5E = 10,



A + 2B + 2C + 2E = 3.


From these equations we find

A = −1, B = 2, C = −2, D = −1, E=2

and so
3s2 + 10s + 3
     
−1 −1 1 −1 1
L = −L + 2L
(s2 + 1) (s3 + 4s2 + 5s + 2) s+2 s+1
   
−1 1 −1 s
−2L −L
(s + 1)2 s2 + 1
 
1
+2L−1 2
.
s +1

Using Table 6.1, we obtain the solution of the given initial value problem

y (t) = −e−2t + 2e−t − 2te−t − cos t + 2 sin t.

Chapter 7

SERIES SOLUTIONS OF LINEAR

DIFFERENTIAL EQUATIONS

7.1 POWER SERIES SOLUTIONS ABOUT AN ORDINARY POINT

A. Basic Concepts and Results

Consider the second oreide homogeneous linear differential equation

d2 y dy
a0 (x) 2
+ a1 (x) + a2 (x) y = 0. (7.1)
dx dx
We assume that (7.1) has a solution in the form
∞
X
c0 + c1 (x − x0 ) + · · · = cn (x − x0 )n , (7.2)
n=0

where c0 , c1 , · · ·, cn are constants. An expression of the form (7.2) is called a power series in
x − x0 . Let us write the differential equation (7.1) in the equivalent normalized form

d2 y dy
2
+ P1 (x) + P2 (x) y = 0, (7.3)
dx dx
where
a1 (x) a2 (x)
P1 (x) = and P2 (x) = .
a0 (x) a0 (x)

Definition 5.1. A function f said to be analytic at x0 if its Taylor series about x0 ,

∞
X f (n) (x0 )
(x − x0 )n ,
n!
n=0

exists and converges to f (x) for all x in some open interval including x0 .
For example, all polynomial functions, ex , sin x and cos x are analytic everywhere.
Definition 5.2. The point x0 is called an ordinary point of the differential equation (7.1)
if both of the functions P1 and P2 in the equivalent normalized equation (7.3) are analytic at
x0 . If either (or both) of these functions is of analytic at x0 , then x0 is called a singular point
of the differential equation (7.1).

86
 87

Example 7.1. Consider the differential equation

d2 y dy
+ x2 + 2 y = 0.


2
+x (7.4)
dx dx
Here P1 (x) = x and P2 (x) = x2 +2. Both of the functions P1 and P2 are polynomial functions
and so they are analytic everywhere. Thus all points are ordinary points of this differential
equation.

Example 7.2. Consider the differential equation

d2 y dy 1
(x − 1) +x + y = 0. (7.5)
dx2 dx x
Equation (7.5) has not been written in the normalized for (7.3). We must first express (7.5)
in the normalized form, thereby obtaining

d2 y x dy 1
2
+ + y = 0.
dx x − 1 dx x (x − 1)
Here
x 1
P1 (x) = and P2 (x) = .
x−1 x (x − 1)
The function P1 is analytic except at x = 1, and P2 is analytic except at x = 0 and x = 1.
Thus x = 0 and x = 1 are singular points of the differential equation under consideration. All
other points are ordinary points. Note clearly that x = 0 is a singular point, even though P1
is analytic at x = 0.

Theorem 7.1. Let the point x0 be an ordinary point of the differential equation (7.1). Then
the differential equation has two nontrivial linearly independent power series solutions of the
form
X∞
cn (x − x0 )n ,
n=0

and these power series converge in some interval |x − x0 | < R (where R > 0) about x0 .

B. The Method of Solution

Now that we are assured that under appropriate hypotheses equation (7.1) actually does have
power series solution of the form (7.2), how do we proceed to find these solutions? In other
words, how do we determine the coefficients c0 , c1 , c2 , · · · in the expression
∞
X
cn (x − x0 )n
n=0

so that this expression actually does satisfy equation (7.1)? We will first give a brief outline
of the procedure for finding these coefficients and then illustrate the procedure in detail by
considering specific examples.
Assume that x0 is an ordinary point of the differential equation (7.1), so that solution in
powers od x − x0 actually exist, we denote such a solution by
∞
X
2
y = c0 + c1 (x − x0 ) + c2 (x − x0 ) + · · · = cn (x − x0 )n . (7.6)
n=0
 88

Since the series in (7.6) converges on an interval |x − x0 | < R about x0 , it may be differentiated
term on this interval twice in succession to obtain
∞
dy X
= c1 + 2c2 (x − x0 ) + 3c3 (x − x0 )2 + · · · = ncn (x − x0 )n−1 (7.7)
dx
n=1

and
∞
d2 y X
= 2c2 + 6c3 (x − x0 ) + · · · = n (n − 1) cn (x − x0 )n−2 , (7.8)
dx2
n=2

respectively. We now substitute the series in the right members of (7.6), (7.7) and (7.8) for
y, y 0 and y 00 , respectively, in the differential equation (7.1). We then simplify the resulting
expression so that it takes the form

K0 + K1 (x − x0 ) + K2 (x − x0 )2 + · · · = 0, (7.9)

where the coefficients Ki (i = 0, 1, 2, · · ·) are functions of certain coefficients cn of the solution

(7.6). In order that (7.9) be valid for all x in the interval of convergence |x − x0 | < R, we
must set
K0 = K1 = K2 = · · · = 0.
In other words, we must equate to zero the coefficient of each power of x − x0 in the left
member of (7.9). This leads to a set of conditions that must be satisfied by the various
equation (7.1). If the cn are chosen to satisfy the set of conditions that thus occurs, then the
resulting series (7.6) is the desired solution of the differential equation (7.1).

Example 7.3. Find the power series solution of the differential equation

d2 y dy 2


+ x + x + 2 y=0 (7.10)
dx2 dx
in powers of x (that is, about x0 = 0).

Solution: We assume that

∞
X
y= cn xn (7.11)
n=0

is a solution of (7.10). Differentiating term by term, we obtain

∞
dy X
= ncn xn−1 (7.12)
dx
n=1

and
∞
d2 y X
= n (n − 1) cn xn−2 . (7.13)
dx2
n=2

Substituting the series (7.11), (7.12) and (7.13) into the differential equation (7.10), we obtain
∞
X ∞
X ∞
X ∞
X
n (n − 1) cn xn−2 + x ncn xn−1 + x2 cn xn + 2 cn xn = 0
n=2 n=1 n=0 n=0

or
∞
X ∞
X ∞
X ∞
X
n−2 n n+2
n (n − 1) cn x + ncn x + cn x +2 cn xn = 0.
n=2 n=1 n=0 n=0
 89

Using Abel’s formula, we get

∞
X ∞
X ∞
X ∞
X
(n + 2) (n + 1) cn+2 xn + ncn xn + cn−2 xn + 2 cn xn = 0.
n=0 n=1 n=2 n=0

From that it follows

∞
X ∞
X
2c2 + 6c3 x + (n + 2) (n + 1) cn+2 xn + c1 x + ncn xn
n=2 n=2

∞
X ∞
X
+ cn−2 xn + 2c0 + 2c1 x + 2 cn xn = 0
n=2 n=0
or
∞
X
(2c0 + 2c2 ) + (3c1 + 6c3 ) x + [(n + 2) (n + 1) cn+2 + (n + 2) cn + cn−2 ] xn = 0.
n=2

Hence  
2c0 + 2c2 = 0 c2 = −c0
⇒
3c1 + 6c3 = 0 c3 = − 21 c1
and
(n + 2) (n + 1) cn+2 + (n + 2) cn + cn−2 = 0, n ≥ 2.
(n + 2) cn + cn−2
cn+2 = − , n ≥ 2.
(n + 2) (n + 1)
For n = 2, we have that
4c2 + c0 1
c4 = − = c0 .
12 4
For n = 3, we have that
5c3 + c1 3
c5 = − = c1 .
20 40
Therefore
1 1 3
y = c0 + c1 x − c0 x2 − c1 x3 + c0 x4 + c1 x5 + · · ·
2 4 40
or    
2 1 4 1 3 3 5
y = c0 1 − x + x + · · · + c1 x − x + x + · · ·
4 2 40
is the general solution of the given differential equation in powers of x through terms in x5 .

Example 7.4. Find a power series solution of the initial-value problem

d2 y dy
x2 − 1 2
+ 3x + xy = 0, y (0) = 4, y 0 (0) = 6. (7.14)
dx dx

Solution: We first observe that all points except x = ±1 are ordinary points for the
differential equation (7.14). We assume that
∞
X
y= cn xn (7.15)
n=0
 90

is a solution of (7.14). Differentiating term by term, we obtain

∞
dy X
= ncn xn−1 (7.16)
dx
n=1

and
∞
d2 y X
= n (n − 1) cn xn−2 . (7.17)
dx2
n=2

Substituting the series (7.15), (7.16) and (7.17) into the differential equation (7.14), we obtain
∞ ∞ ∞

X X X
x2 − 1 n (n − 1) cn xn−2 + 3x ncn xn−1 + x cn xn = 0. (7.18)
n=2 n=1 n=0

We now rewrite second and fourth summations in (7.18) so that x in each of these summations
has the exponent n. Doing this, equation (7.18) takes the form
∞
X ∞
X ∞
X ∞
X
n (n − 1) cn xn − (n + 2) (n + 1) cn+2 xn + 3 ncn xn + cn−1 xn = 0. (7.19)
n=2 n=0 n=1 n=1

The common range of the four summations in (7.19) is from 2 to ∞. We can write out the
individual terms in each summation that do not belong to this common range and thus express
(7.19) in the form
∞
X ∞
X
n (n − 1) cn xn − 2c2 − 6c3 x − (n + 2) (n + 1) cn+2 xn
n=2 n=2

∞
X ∞
X
+3c1 x + 3 ncn xn + c0 x + cn−1 xn = 0.
n=2 n=2
Combining the powers of x, this takes the form

−2c2 + (c0 + 3c1 − 6c3 ) x

∞
X
+ [− (n + 2) (n + 1) cn+2 + n (n + 2) cn + cn−1 ] xn = 0. (7.20)
n=2

For (7.20) to be valid for all x in the interval of convergence |x − x0 | < R, the coefficient of
each power of x in the left member of (7.20) must be equated to zero. In doing this, we are
led to the relations 
−2c2 = 0,
c0 + 3c1 − 6c3 = 0,
and
− (n + 2) (n + 1) cn+2 + n (n + 2) cn + cn−1 = 0, n ≥ 2. (7.21)
From that it follows 
c2 = 0,
c3 = 16 c0 + 12 c1 .
The recurrence formula (7.21) gives

n (n + 2) cn + cn−1
cn+2 = , n ≥ 2.
(n + 2) (n + 1)
 91

Using this, we find successively

8c2 + c1 1
c4 = = c1 ,
12 12
15c3 + c2 1 3
c5 = = c0 + c1 .
20 8 8
Substituting these values of c2 , c3 , c4 , c5 , · · · into assumed solution (7.15), we have
c  
0 c1  3 c1 4 1 3
y = c0 + c1 x + + x + x + c0 + c1 x5 + · · ·
6 2 12 8 8
or    
1 3 1 5 1 3 1 4 3 5
y = c0 1 + x + x + · · · + c1 x + x + x + x + · · · . (7.22)
6 8 2 12 8
The solution (7.22) is the general solution of the differential equation (7.14) in powers of x
(through terms in x5 ).
We must now apply the given initial conditions. Applying them, we obtain

c0 = 4 and c1 = 6.

Thus the exact solution of the given initial value problem is

11 3 1 4 11 5
y = 4 + 6x + x + x + x +···
3 2 4
Remark 7.3. Suppose the initial values of y and its first derivative in conditions (7.14) of last
example are prescribed at x = 2, instead x = 0. Then we have the initial value problem

d2 y dy
x2 − 1 2
+ 3x + xy = 0, y (2) = 4, y 0 (2) = 6. (7.23)
dx dx
Since the initial values in this problem are prescribed at x = 2, we would seek solutions in
powers of x − 2. That is, in this case we would seek solutions of the form
∞
X
cn (x − 2)n . (7.24)
n=0

The simplest procedure for obtaining a solution of the form (7.24) is first to make the sub-
stitution t = x − 2. This replaces the initial value problem (7.23) by the equivalent problem

d2 y dy
t2 + 4t + 3 2
+ (3t + 6) + (t + 2) y = 0, y (0) = 4, y 0 (0) = 6, (7.25)
dt dt
in which t is the independent variable and the initial values are prescribed at x = 0. One then
seek a solution of the problem (7.25) in powers of t,
∞
X
y= cn tn . (7.26)
n=0

Differentiating (7.26) and substituting into the differential equation (7.25), one determines the
cn as in last two examples. The initial conditions in (7.25) are then applied. Replacing t by
x − 2 in the resulting solution (7.26), one obtains the desired solution (7.24) of the original
problem (7.23).
 92

Remark 7.4. In the last two examples we obtained power series solutions of the differential
equations under consideration but made no attempt to discuss the convergence of these so-
lutions. According to Theorem 5.1, if x0 is an ordinary point of the differential equation

d2 y dy
a0 (x) 2
+ a1 (x) + a0 (x) y = 0, (7.27)
dx dx
then the power series solutions of the form (7.2) converge in some interval |x − x0 | < R (where
R > 0) about x0 . Let us again write (7.27) in the normalized form

d2 y dy
2
+ P1 (x) + P2 (x) y = 0, (7.28)
dx dx
where
a1 (x) a2 (x)
P1 (x) = and P2 (x) = .
a0 (x) a0 (x)
If x0 is an ordinary point of (7.27), the functions P1 and P2 have Taylor series expansions
about x0 that converge in intervals |x − x0 | < R1 and |x − x0 | < R2 , respectively, about x0 .
It can be proved that the interval of convergence |x − x0 | < R of a series solution (7.2) of
(7.27) is at least as great as the smaller of the intervals |x − x0 | < R1 and |x − x0 | < R2 .
 Lecture Notes on

Differential Equations

Emre Sermutlu
Contents

Preface ix

1 First Order ODE 1

1.1 Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Mathematical Modeling . . . . . . . . . . . . . . . . . . . . . 3
1.3 Separable Equations . . . . . . . . . . . . . . . . . . . . . . . 3
1.4 Transformations . . . . . . . . . . . . . . . . . . . . . . . . . . 5

2 Exact Equations 9
2.1 Exact Equations . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.2 Integrating Factors . . . . . . . . . . . . . . . . . . . . . . . . 11
2.3 Linear First Order Equations . . . . . . . . . . . . . . . . . . 13
2.4 Bernoulli Equation . . . . . . . . . . . . . . . . . . . . . . . . 14

3 Second Order Equations 17

3.1 Linear Differential Equations . . . . . . . . . . . . . . . . . . . 17
3.2 Reduction of Order . . . . . . . . . . . . . . . . . . . . . . . . 19
3.3 Constant Coefficients . . . . . . . . . . . . . . . . . . . . . . . 20
3.4 Cauchy-Euler Equation . . . . . . . . . . . . . . . . . . . . . . 22

4 Nonhomogeneous Equations 25
4.1 General and Particular Solutions . . . . . . . . . . . . . . . . 25
4.2 Method of Undetermined Coefficients . . . . . . . . . . . . . . 27
4.3 Method of Variation of Parameters . . . . . . . . . . . . . . . 29

5 Higher Order Equations 33

5.1 Linear Equations of Order n . . . . . . . . . . . . . . . . . . . 33

v
vi CONTENTS

5.2 Differential Operators . . . . . . . . . . . . . . . . . . . . . . . 34

5.3 Homogeneous Equations . . . . . . . . . . . . . . . . . . . . . 35
5.4 Nonhomogeneous Equations . . . . . . . . . . . . . . . . . . . 37

6 Series Solutions 41
6.1 Power Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
6.2 Classification of Points . . . . . . . . . . . . . . . . . . . . . . 43
6.3 Power Series Method . . . . . . . . . . . . . . . . . . . . . . . 43

7 Frobenius’ Method 49
7.1 An Extension of Power Series Method . . . . . . . . . . . . . . 49
7.2 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

8 Laplace Transform I 57
8.1 Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
8.2 Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
8.3 Initial Value Problems . . . . . . . . . . . . . . . . . . . . . . 61

9 Laplace Transform II 69
9.1 Convolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
9.2 Unit Step Function . . . . . . . . . . . . . . . . . . . . . . . . 72
9.3 Differentiation of Transforms . . . . . . . . . . . . . . . . . . . 73
9.4 Partial Fractions Expansion . . . . . . . . . . . . . . . . . . . 74
9.5 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75

10 Fourier Analysis I 81
10.1 Fourier Series . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
10.2 Convergence of Fourier Series . . . . . . . . . . . . . . . . . . 84
10.3 Parseval’s Identity . . . . . . . . . . . . . . . . . . . . . . . . 85

11 Fourier Analysis II 91
11.1 Fourier Cosine and Sine Series . . . . . . . . . . . . . . . . . . 91
11.2 Complex Fourier Series . . . . . . . . . . . . . . . . . . . . . . 94
11.3 Fourier Integral Representation . . . . . . . . . . . . . . . . . 96
CONTENTS vii

12 Partial Differential Equations 101

12.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
12.2 Modeling a Vibrating String . . . . . . . . . . . . . . . . . . . 103
12.3 Method of Separation of Variables . . . . . . . . . . . . . . . . 104

13 Heat Equation 111

13.1 Modeling Heat Flow . . . . . . . . . . . . . . . . . . . . . . . 111
13.2 Homogeneous Boundary Conditions . . . . . . . . . . . . . . . 113
13.3 Nonzero Boundary Conditions . . . . . . . . . . . . . . . . . . 115
13.4 Two Dimensional Problems . . . . . . . . . . . . . . . . . . . 117

14 Laplace Equation 121

14.1 Rectangular Coordinates . . . . . . . . . . . . . . . . . . . . . 121
14.2 Polar Coordinates . . . . . . . . . . . . . . . . . . . . . . . . . 126

To the Student 133

References 135

Index 137
Chapter 1

First Order Differential

Equations

The subject of differential equations is an important part of applied mathe-

matics. Many real life problems can be formulated as differential equations.
In this chapter we will first learn the basic concepts and classification of
differential equations, then we will see where they come from and how the
simplest ones are solved. The concepts and techniques of calculus, especially
integration, will be necessary to understand differential equations.

1.1 Definitions
Ordinary Differential Equation: An ordinary differential equation is an
equation that contains derivatives of an unknown function y(x).
Partial Differential Equation: A partial differential equation is a differ-
ential equation involving an unknown function of two or more variables, like
u(x, y).
For example,
y 00 − 4y 0 + y = 0
p
y 2 + 1 = x2 y + sin x
are ordinary differential equations.
uxx + uyy = 0
u2x + u2y = ln u

1
2 CHAPTER 1. FIRST ORDER ODE

are partial differential equations. (Partial Differential Equations are usually

much more difficult)
Order: The order of a differential equation is the order of the highest deriva-
tive that occurs in the equation.
A first order differential equation contains y 0 , y and x so it is of the form
F (x, y, y 0 ) = 0 or y 0 = f (x, y).
For example, the following differential equations are first order:

y 0 + x2 y = ex
xy 0 = (1 + y 2 )
y 02 = 4xy

While these are second order:

y 00 − x2 y 0 + y = 1 + sin x
y 00 + 6yy 0 = x3

General and Particular Solutions: A general solution of a differential

equation involves arbitrary constants. In a particular solution, these con-
stants are determined using initial values.
As an example, consider the differential equation y 0 = 2x.
y = x2 + c is a general solution ,
y = x2 + 4 is a particular solution .

Example 1.1 Find the general solution of the differential equation y 00 = 0.

Then find the particular solution that satisfies y(0) = 5, y 0 (0) = 3.

y 00 = 0 ⇒ y 0 = c ⇒ y = cx + d. This is the general solution.

y 0 (0) = 3 ⇒ c = 3, y(0) = 5 ⇒ d = 5
Therefore y = 3x + 5. This is the particular solution.
Explicit and Implicit Solutions: y = f (x) is an explicit solution,
F (x, y) = 0 is an implicit solution. We have to solve equations to obtain y
for a given x in implicit solutions, whereas it is straightforward for explicit
solutions.
For example, y = e4x is an explicit solution of the equation y 0 = 4y.
x3 + y 3 = 1 is an implicit solution of the equation y 2 y 0 + x2 = 0
1.2. MATHEMATICAL MODELING 3

1.2 Mathematical Modeling

Differential equations are the natural tools to formulate, solve and under-
stand many engineering and scientific systems. The mathematical models of
most of the simple systems are differential equations.

Example 1.2 The rate of growth of a population is proportional to itself.

Find the population as a function of time.

dP
= αP
dt

P = P0 eαt
where P0 = P (0)

Example 1.3 The downward acceleration of an object in free fall is g. Find

the height as a function of time if the initial height is y0 and initial speed is
v0 .

d2 y
= −g
dt2

dy
= −gt + v0
dt

1
y = − gt2 + v0 t + y0
2

1.3 Separable Equations

If we can separate x and y in a first order differential equation and put them
to different sides as g(y)dy = f (x)dx, it is called a separable equation. We
can find the solution by integrating both sides. (Don’t forget the integration
constant!)
Z Z
g(y)dy = f (x)dx + c (1.1)
4 CHAPTER 1. FIRST ORDER ODE

Example 1.4 Solve the initial value problem y 0 + y 2 xex = 0, y(0) = 2

dy
y 0 = −y 2 xex ⇒ − 2
= xex dx
y
Z Z
dy
− = xex dx
y2
Using integration by parts, we have u = x, dv = ex dx, du = dx, v = ex
therefore
Z
1 1
= xex − ex dx ⇒ = xex − ex + c
y y

1
y=
xex
− ex + c
This is the general solution. Now we will use the condition y(0) = 2 to
determine the constant c.

1 3
2= ⇒ c=
−1 + c 2

1
y= 3
xex − ex + 2

Example 1.5 Find the general solution of the differential equation

y 0 + y 2 = 1.

dy dy dy
+ y2 = 1 ⇒ = 1 − y2 ⇒ = dx
dx dx 1 − y2
Z Z
dy
= dx
1 − y2
Z   Z
1 1 1
+ dy = dx
2 1−y 1+y
 
1  1 + y 
ln =x+c
2 1 − y 
 
1 + y  2x+2c
1 − y  = e
 
1.4. TRANSFORMATIONS 5

After some algebra, we obtain

ke2x − 1
y=
ke2x + 1
where k = e2c

Example 1.6 Solve the initial value problem y 0 = x3 e−y , y(1) = 0.

Z Z
e dy = x3 dx
y

x4
ey = +c
4

1
y(1) = 0 ⇒ e0 = +c
4

3
c=
4

x4 3
 
y = ln +
4 4

1.4 Transformations
Sometimes a change of variables simplifies a differential
 y  equation just as
0
substitutions simplify integrals. For example if y = f , the substitution
x
y
u = will make the new equation separable.
x
r
0 y x
Example 1.7 Solve y = + 3 .
x y
r
0 0 0 1
If y = ux, then y = u x + u and u x + u = u + 3
u
r
1
u0 x = 3
u
√ 3dx
u du =
x
6 CHAPTER 1. FIRST ORDER ODE

u3/2
= 3 ln x + c
3/2
 2/3
9
u= ln x + c1
2
 2/3
9
y=x ln x + c1
2

Example 1.8 Solve the differential equation (x + y + 6)dx = (−x − y − 3)dy.

Let’s use the substitution u = x + y. Then,

y = u − x, ⇒ dy = du − dx

and the equation can be expressed in terms of u and x.

(u + 6)dx = (−u − 3)(du − dx)

3dx = (−u − 3)du

Z Z
3dx = (−u − 3) du

u2
3x = − − 3u + c
2

(x + y)2
3x = − − 3(x + y) + c
2
This is an implicit solution.
EXERCISES 7

Exercises
Solve the following differential equations.
1) y 3 y 0 + x3 = 0
2) y 0 + 4x3 y 2 = 0 y
0 0
3) xy = x + y Hint: y = f
x
y
4) (x2 + y 2 ) dx + xydy = 0 Hint: y 0 = f
x
2
5) y 0 = xey−x
1 + ln x
6) y 0 =
4y 3

7) y 0 = 3x2 sec2 y
8) y 0 = y(y + 1)
9) y 0 + 2y = y 2 + 1
10) (1 + y 2 )dx + x2 dy = 0

y
11) y 0 = a
x

12) y 0 = eax+by
13) y 0 = x2 y 2 − 2y 2 + x2 − 2

2x + y
14) y 0 = −
x

Solve the following initial value problems:

F15) (y 2 + 5xy + 9x2 )dx + x2 dy = 0, y(1) = −4
16) y 3 y 0 + x3 = 0, y(0) = 1
17) y 0 = −2xy, y(0) = 3
18) y 0 = 1 + 4y 2 , y(0) = 0
19) (x2 + 1)1/2 y 0 = xy 3 , y(0) = 2

dx x x2
F20) = − , x(0) = 1
dt 5 25
8 CHAPTER 1. FIRST ORDER ODE

Answers
1) x4 + y 4 = c
1
2) y = 4
x +c

3) y = x(ln |x| + c)
c x2
4) y 2 = 2 −
x 2
2
!
e−x
5) y = − ln c +
2
6) y 4 = x ln x + c
7) 2y + sin 2y = 4x3 + c
ex
8) y=
c − ex
1
9) y = 1 −
x+c
 
1
10) y = tan c +
x

11) y = cxa

eax e−by
12) + =c
a b
 3 
x
13) y = tan − 2x + c
3
c
14) y = −x +
x
x
15) y = − 3x
ln x − 1

16) x4 + y 4 = 1
2
17) y = 3e−x
18) y = 21 tan 2x
√
19) y = ( 94 − 2 x2 + 1)−1/2

5et/5
20) x =
4 + et/5
Chapter 2

Exact and Linear Differential

Equations

In this chapter, we will learn how to recognize and solve three different types
of equations: Exact, linear, and Bernoulli. All of them are first order equa-
tions, therefore we expect a single integration constant in the solution.
At this stage it seems like there’s a special trick for every different kind
of question. You will gain familiarity with exercise and experience.

2.1 Exact Equations

A first order differential equation of the form

M (x, y)dx + N (x, y)dy = 0 (2.1)

is called an exact differential equation if there exists a function u(x, y) such

that
∂u ∂u
= M, =N (2.2)
∂x ∂y
In other words, du = M dx + N dy, so M dx + N dy is a total differential.
For example, the equation

(4x3 + 2xy 2 )dx + (4y 3 + 2x2 y)dy = 0

is exact, and
u = x4 + x2 y 2 + y 4

9
10 CHAPTER 2. EXACT EQUATIONS

So, the solution of this equation is very simple, if du is zero, u must be a

constant, therefore
x4 + x2 y 2 + y 4 = c
∂M ∂N
Theorem 2.1: The condition = is necessary and sufficient for the
∂y ∂x
equation M (x, y)dx + N (x, y)dy = 0 to be exact.
Method of Solution: To solve M dx + N dy = 0,

• Check for Exactness

• If the equation is exact, find u by integrating either M or N .

Z Z
u = M dx + k(y) or u = N dy + l(x)

Note that we have arbitrary functions as integration constants.

• Determine the arbitrary functions using the original equation. The

solution is u(x, y) = c

Example 2.1 Solve the equation 3y 2 dx + (3y 2 + 6xy)dy = 0.

Let’s check for exactness first.

∂(3y 2 ) ∂(3y 2 + 6xy)
= 6y, = 6y
∂y ∂x
The equation is exact.
Z
u(x, y) = 3y 2 dx + k(y) = 3y 2 x + k(y)

∂u
= 6yx + k 0 (y) = 3y 2 + 6xy
∂y
k 0 (y) = 3y 2 ⇒ k(y) = y 3
We do not need an integration constant here because u(x, y) = c already
contains one
u(x, y) = 3y 2 x + y 3 = c
2.2. INTEGRATING FACTORS 11

2.2 Integrating Factors

Consider the equation

P dx + Qdy = 0 (2.3)

that is not exact. If it becomes exact after multiplying by F , i.e. if

F P dx + F Qdy = 0 (2.4)

is exact, then F is called an integrating factor. (Note that P, Q and F are

functions of x and y)
1
For example, ydx − xdy = 0 is not exact, but F = 2 is an integrating
x
factor.

Example 2.2 Solve (2xex − y 2 )dx + 2ydy = 0. Use F = e−x .

∂(2xex − y 2 ) ∂(2y)
= −2y, =0
∂y ∂x
The equation is not exact. Let’s multiply both sides by e−x . The new
equation is:

(2x − y 2 e−x )dx + 2ye−x dy = 0

∂(2x − y 2 e−x ) ∂(2ye−x )

= −2ye−x , = −2ye−x
∂y ∂x
Now the equation is exact. We can solve it as we did the previous example
and obtain the result

x2 + y 2 e−x = c

How To Find an Integrating Factor: Let P dx+Qdy = 0 be a differential

equation that is not exact, and let F = F (x, y) be an integrating factor. By
definition,

(F P )y = (F Q)x ⇒ Fy P + F P y = Fx Q + F Qx (2.5)
12 CHAPTER 2. EXACT EQUATIONS

But this equation is more difficult than the one we started with. If we make
a simplifying assumption that F is a function of one variable only, we can
solve for F and obtain the following theorem:
Theorem 2.2: Consider the equation P dx + Qdy = 0. Define
  
1 ∂P ∂Q 1
∂Q ∂P
R= − and R̃ = − (2.6)
Q ∂y ∂x ∂x P ∂y
R

a) If R depends only on x, then F (x) = exp R(x) dx is an integrating
factor. R 
b) If R̃ depends only on y, then F (y) = exp R̃(y)dy is an integrating
factor.

Example 2.3 Solve (4x2 y 2 + 2y)dx + (2x3 y + x)dy = 0

∂(4x2 y 2 + 2y) ∂(2x3 y + x)

= 8x2 y + 2, = 6x2 y + 1
∂y ∂x
The equation is not exact.

8x2 y + 2 − 6x2 y − 1 2x2 y + 1 1

R= 3
= 3
=
2x y + x 2x y + x x

R
R(x)dx
F (x) = e = eln x = x

Multiply the equation by x to obtain the exact equation

(4x3 y 2 + 2yx)dx + (2x4 y + x2 )dy = 0

Z
u(x, y) = (4x3 y 2 + 2yx) dx + k(y) = x4 y 2 + yx2 + k(y)

∂u
= 2x4 y + x2 + k 0 (y) = 2x4 y + x2 ⇒ k(y) = 0
∂y

u(x, y) = x4 y 2 + x2 y = c
2.3. LINEAR FIRST ORDER EQUATIONS 13

2.3 Linear First Order Equations

If a first order differential equation can be written in the form
y 0 + p(x)y = r(x) (2.7)
it is called a linear differential equation. If r(x) = 0, the equation is homo-
geneous, otherwise it is nonhomogeneous.
We can express the equation (2.7) as [p(x)y − r(x)]dx + dy = 0. This is
not exact but it has an integrating factor:
R
p dx
R = p(x), F = e (2.8)
Method of Solution:
• Given a first order linear equation, express it in the following form:
y 0 + p(x)y = r(x) (2.9)
Z
• Multiply both sides by the integrating factor F (x) = exp p(x) dx to
obtain
R R R
p dx 0 p dx p dx
e y +e py = re (2.10)

• Express the left hand side as a single parenthesis.

 R 0 R
e p dx y = re p dx (2.11)

• Integrate both sides. Don’t forget the integration constant. The solu-
tion is: Z 
−h h
y(x) = e e r dx + c (2.12)
R
where h = p dx.
Example 2.4 Solve y 0 + 4y = 1
R
4 dx
The integrating factor is F = e = e4x . Multiply both sides of the
equation by e4x to obtain
e4x y 0 + 4e4x y =e4x
(e4x y)0 =e4x
e4x 1
4x
e y= +c ⇒ y= + ce−4x
4 4
14 CHAPTER 2. EXACT EQUATIONS

2.4 Bernoulli Equation

The equation
y 0 + p(x)y = g(x)y a (2.13)
is called Bernoulli equation. It is nonlinear. Nonlinear equations are usually
much more difficult than linear ones, but Bernoulli equation is an exception.
It can be linearized by the substitution

u(x) = [y(x)]1−a (2.14)

Then, we can solve it as other linear equations.

Example 2.5 Solve the equation

2
2x 0ex
y − y=
3 3xy 2

Here a = −2 therefore u = y 1−(−2) = y 3 ⇒ u0 = 3y 2 y 0

Multiplying both sides of the equation by 3y 2 we obtain
2 2
2 0 3ex 0 ex
3y y − 2xy = ⇒ u − 2xu =
x x
This equation is linear. Its integrating factor is
2
R
−2x dx
e = e−x
2
Multiplying both sides by e−x , we get
2 2 1
e−x u0 − 2xe−x u =
x
2 1
(e−x u)0 =
x

2 2
e−x u = ln x + c ⇒ u = (ln x + c)ex

2 1/3
h i
y = (ln x + c)ex
EXERCISES 15

Exercises
Solve the following differential equations. (Find an integrating factor if
necessary)
1) (yex + xyex + 1)dx + xex dy = 0
2) (2r + 2 cos θ)dr − 2r sin θdθ = 0
3) (sin xy + xy cos xy)dx + (x2 cos xy)dy = 0
4) 2 cos ydx = sin ydy
5) 5dx − ey−x dy = 0
6) (2xy + 3x2 y 6 ) dx + (4x2 + 9x3 y 5 ) dy = 0
7) (3xey + 2y) dx + (x2 ey + x) dy = 0
5 1
8) y 0 + y =
x x
1 1
9) y 0 + y=
x ln x ln x

10) y 0 − y tan x = tan x

11) y 0 + y tan x = 4x3 cos x
12) y 0 + x3 y = 4x3 , y(0) = −1

Reduce to linear form and solve the following equations:

2 sin x 1/2
13) y 0 − 4y tan x = y
cos3 x
x
14) y 0 + y = −
y
25 5 ln x 4/5
15) y 0 + y= y
x x5
y 1
16) y 0 + =− 9 3
x xy
tan y
17) y 0 =
x−1
18) y 2 dx + (3xy − 1)dy = 0

F19) y 0 (sinh 3y − 2xy) = y 2 Hint: x ↔ y

F20) 2xyy 0 + (x − 1)y 2 = x2 ex Hint: z = y 2

16 CHAPTER 2. EXACT EQUATIONS

Answers
 c
1) y = − 1 e−x
x
2) r2 + 2r cos θ = c
3) x sin xy = c
4) F = e2x , e2x cos y = c
5) F = ex , 5ex − ey = c
6) F = y 3 , x 2 y 4 + x3 y 9 = c
7) F = x, x3 ey + x2 y = c
1 c
8) y= + 5
5 x
x+c
9) y =
ln x
c
10) y = −1 +
cos x

11) y = x4 cos x + c cos x

x4
12) y = 4 − 5e− 4
 2
c − ln cos x
13) y =
cos2 x
q
14) y = 12 − x + ce−2x
 5
x ln x − x + c
15) y =
x5
 1/4
1 c
16) y = +
x8 x4

17) y = arcsin[c(x − 1)]

1 c
18) F = y, x = + 3
2y y

19) x = y −2 1


3
cosh 3y + c
q
20) y = cxe−x + 12 xex
Chapter 3

Second Order Homogeneous

Differential Equations

For first order equations, concepts from calculus and some extensions were
sufficient. Now we are starting second order equations and we will learn many
new ideas, like reduction of order, linear independence and superposition of
solutions.
Many differential equations in applied science and engineering are second
order and linear. If in addition they have constant coefficients, we can solve
them easily, as explained in this chapter and the next. For nonconstant
coefficients, we will have limited success.

3.1 Linear Differential Equations

If we can express a second order differential equation in the form

y 00 + p(x)y 0 + q(x)y = r(x) (3.1)

it is called linear. Otherwise, it is nonlinear.

Consider a linear differential equation. If r(x) = 0 it is called homoge-
neous, otherwise it is called nonhomogeneous. Some examples are:
y 00 + y 2 = x2 y Nonlinear
00
sin xy + cos xy = 4 tan x Linear Nonhomogeneous
x2 y 00 + y = 0 Linear Homogeneous

17
18 CHAPTER 3. SECOND ORDER EQUATIONS

Linear Combination: A linear combination of y1 , y2 is y = c1 y1 + c2 y2 .

Theorem 3.1: For a homogeneous linear differential equation any linear
combination of solutions is again a solution.
The above result does NOT hold for nonhomogeneous equations.
For example, both y = sin x and y = cos x are solutions to y 00 + y = 0, so
is y = 2 sin x + 5 cos x.
Both y = sin x + x and y = cos x + x are solutions to y 00 + y = x, but
y = sin x + cos x + 2x is not.
This is a very important property of linear homogeneous equations, called
superposition. It means we can multiply a solution by any number, or add
two solutions, and obtain a new solution.
Linear Independence: Two functions y1 , y2 are linearly independent if
c1 y1 + c2 y2 = 0 ⇒ c1 = 0, c2 = 0. Otherwise they are linearly dependent.
(One is a multiple of the other).
For example, ex and e2x are linearly independent. ex and 2ex are linearly
dependent.
General Solution and Basis: Given a second order, linear, homogeneous
differential equation, the general solution is:

y = c1 y1 + c2 y2 (3.2)

where y1 , y2 are linearly independent. The set {y1 , y2 } is called a basis, or a

fundamental set of the differential equation.
As an illustration, consider the equation x2 y 00 − 5xy 0 + 8y = 0. You can
easily check that y = x2 is a solution. (We will see how to find it in the
last section) Therefore 2x2 , 7x2 or −x2 are also solutions. But all these are
linearly dependent.
We expect a second, linearly independent solution, and this is y = x4 . A
combination of solutions is also a solution, so y = x2 + x4 or y = 10x2 − 5x4
are also solutions. Therefore the general solution is

y = c1 x 2 + c2 x 4 (3.3)

and the basis of solutions is {x2 , x4 }.

3.2. REDUCTION OF ORDER 19

3.2 Reduction of Order

If we know one solution of a second order homogeneous differential equation,
we can find the second solution by the method of reduction of order.
Consider the differential equation
y 00 + py 0 + qy = 0 (3.4)
Suppose one solution y1 is known, then set y2 = uy1 and insert in the equa-
tion. The result will be
y1 u00 + (2y10 + py1 )u0 + (y100 + py10 + qy1 )u = 0 (3.5)
But y1 is a solution, so the last term is canceled. So we have
y1 u00 + (2y10 + py1 )u0 = 0 (3.6)
This is still second order, but if we set w = u0 , we will obtain a first order
equation:
y1 w0 + (2y10 + py1 )w = 0 (3.7)
Solving this, we can find w, then u and then y2 .
Example 3.1 Given that y1 = x2 is a solution of
x2 y 00 − 3xy 0 + 4y = 0
find a second linearly independent solution.
Let y2 = ux2 . Then
y20 = u0 x2 + 2xu
and
y200 = u00 x2 + 4xu0 + 2u
Inserting these in the equation, we obtain
x4 u00 + x3 u0 = 0
If w = u0 then
1
x4 w0 + x3 w = 0 or w0 + w = 0
x
1
This linear first order equation gives w = , therefore u = ln x and
x
2
y2 = x ln x
20 CHAPTER 3. SECOND ORDER EQUATIONS

3.3 Homogeneous Equations with Constant

Coefficients
Up to now we have studied the theoretical aspects of the solution of linear ho-
mogeneous differential equations. Now we will see how to solve the constant
coefficient equation y 00 + ay 0 + by = 0 in practice.
We have the sum of a function and its derivatives equal to zero, so the
derivatives must have the same form as the function. Therefore we expect
the function to be eλx . If we insert this in the equation, we obtain:

λ2 + aλ + b = 0 (3.8)

This is called the characteristic equation of the homogeneous differential

equation y 00 + ay 0 + by = 0.
If we solve the characteristic equation, we will see three different possibilities:
Two real roots, double real root and complex conjugate roots.
Two Real Roots: The general solution is

y = c1 eλ1 x + c2 eλ2 x (3.9)

Example 3.2 Solve y 00 − 3y 0 − 10y = 0

Try y = eλx . The characteristic equation is λ2 − 3λ − 10 = 0 with solution

λ1 = 5, λ = −2, so the general solution is

y = c1 e5x + c2 e−2x

Example 3.3 Solve the initial value problem y 00 −y = 0, y(0) = 2, y 0 (0) = 4

We start with y = eλx as usual. The characteristic equation is λ2 − 1 = 0.

Therefore λ = ±1. The general solution is: y = c1 ex + c2 e−x
Now, we have to use the initial values to determine the constants.
y(0) = 2 ⇒ c1 + c2 = 2 and y 0 (0) = 4 ⇒ c1 − c2 = 4.
By solving this system, we obtain c1 = 3, c2 = −1 so the particular solution
is:
y = 3ex − e−x
3.3. CONSTANT COEFFICIENTS 21

Double Real Root: One solution is eλx but we know that a second order
equation must have two independent solutions. Let’s use the method of
reduction of order to find the second solution.

y 00 − 2ay 0 + a2 y = 0 ⇒ y1 = eax (3.10)

Let’s insert y2 = ueax in the equation.

eax u00 + (2a − 2a)eax u0 = 0 (3.11)

Obviously, u00 = 0 therefore u = c1 + c2 x. The general solution is

y = c1 eλx + c2 xeλx (3.12)

Example 3.4 Solve y 00 + 2y 0 + y = 0

y = eλx . The characteristic equation is λ2 + 2λ + 1 = 0. Its solution is the
double root λ = −1, therefore the general solution is

y = c1 e−x + c2 xe−x

Complex Conjugate Roots: We need the complex exponentials for this

case. Euler’s formula is

eix = cos x + i sin x (3.13)

This can be proved using Taylor series expansions.

If the solution of the characteristic equation is

λ1 = α + iβ, λ2 = α − iβ (3.14)

then the general solution of the differential equation will be

y = c1 eαx (cos βx + i sin βx) + c2 eαx (cos βx − i sin βx) (3.15)

By choosing new constants A, B, we can express this as

y = eαx (A cos βx + B sin βx) (3.16)

Example 3.5 Solve y 00 − 4y 0 + 29y = 0.

y = eλx . The characteristic equation is λ2 −4λ+29 = 0. Therefore λ = 2±5i.
The general solution is

y = e2x (A cos 5x + B sin 5x)

22 CHAPTER 3. SECOND ORDER EQUATIONS

3.4 Cauchy-Euler Equation

The equation x2 y 00 + axy 0 + by = 0 is called the Cauchy-Euler equation. By
inspection, we can easily see that the solution must be a power of x. Let’s
substitute y = xr in the equation and try to determine r. We will obtain

r(r − 1)xr + arxr + bxr = 0 (3.17)

r2 + (a − 1)r + b = 0 (3.18)
This is called the auxiliary equation. Once again, we have three different
cases according to the types of roots. The general solution is given as follows:

• Two real roots

y = c1 xr1 + c2 xr2 (3.19)

• Double real root

y = c1 xr + c2 xr ln x (3.20)

• Complex conjugate roots where r1 , r2 = r ± si

y = xr [c1 cos(s ln x) + c2 sin(s ln x)] (3.21)

Example 3.6 Solve x2 y 00 + 2xy 0 − 6y = 0

Insert y = xr . Auxiliary equation is r2 + r − 6 = 0. The roots are

r = 2, r = −3 therefore

y = c1 x2 + c2 x−3

Example 3.7 Solve x2 y 00 − 9xy 0 + 25y = 0

Insert y = xr . Auxiliary equation is r2 − 10r + 25 = 0. Auxiliary equation

has the double root r = 5 therefore the general solution is

y = c1 x5 + c2 x5 ln x
EXERCISES 23

Exercises
Are the following sets linearly independent?
1) {x4 , x8 }
2) {sin x, sin2 x}
3) {ln(x5 ), ln x}

Use reduction of order to find a second linearly independent solution:

F4) x2 (ln x − 1) y 00 − xy 0 + y = 0, y1 = x
2 00 0 1
5) x ln x y + (2x ln x − x)y − y = 0, y1 =
x
6) y 00 + 3 tan x y 0 + (3 tan2 x + 1)y = 0, y1 = cos x

Solve the following equations:

7) y 00 + 2y 0 + y = 0, y(0) = 1, y 0 (0) = 0

5
8) y 00 + y 0 + y = 0
2

9) y 00 − 64y = 0, y(0) = 1, y 0 (0) = 8

10) y 00 + 24y 0 + 144y = 0
7
11) y 00 + 2y 0 + y = 0, y(−1) = e, y(1) =
e
12) 5y 00 − 8y 0 +5y = 0 
π2
13) y 00 + 2y 0 + 1 + y = 0, y(0) = 1, y 0 (0) = −1
4
14) y 00 − 2y 0 + 2y = 0, y(π) = 0, y(−π) = 0
15) xy 00 + y 0 = 0
16) x2 y 00 − 3xy 0 + 5y = 0
17) x2 y 00 − 10xy 0 + 18y = 0
18) x2 y 00 − 13xy 0 + 49y = 0
Z
19) Show that y1 = u and y2 = u vdx are solutions of the equation

v0 u0 v 0 u0 u02 u00
   
00 0
y − +2 y + +2 2 − y=0
v u vu u u
20) Show that y1 = u and y2 = v are solutions of the equation
(uv 0 − vu0 )y 00 + (vu00 − uv 00 )y 0 + (u0 v 00 − v 0 u00 )y = 0
24 CHAPTER 3. SECOND ORDER EQUATIONS

Answers
1) Yes
2) Yes
3) No
4) y2 = ln x
5) y2 = ln x − 1

6) y2 = sin x cos x

7) y = (1 + x)e−x

1
8) y = c1 e−2x + c2 e− 2 x

9) y = e8x

10) y = c1 e−12x + c2 xe−12x

11) y = 4e−x + 3xe−x

12) y = e0.8x [A cos(0.6x) + B sin(0.6x)]

π 
13) y = e−x cos x
2

14) y = ex sin x

15) y = c1 + c2 ln x

16) y = x2 [c1 cos(ln x) + c2 sin(ln x)]

17) y = c1 x2 + c2 x9

18) y = c1 x7 + c2 x7 ln x
Chapter 4

Second Order Nonhomogeneous

Equations

In this chapter we will start to solve the nonhomogeneous equations, and

see that we will need the homogeneous solutions we found in the previous
chapter.
Of the two methods we will learn, undetermined coefficients is simpler,
but it can be applied to a restricted class of problems. Variation of parameters
is more general but involves more calculations.

4.1 General and Particular Solutions

Consider the nonhomogeneous equation

y 00 + p(x)y 0 + q(x)y = r(x) (4.1)

Let yp be a solution of this equation. Now consider the corresponding homo-

geneous equation
y 00 + p(x)y 0 + q(x)y = 0 (4.2)

Let yh be the general solution of this one. If we add yh and yp , the result
will still be a solution for the nonhomogeneous equation, and it must be the
general solution because yh contains two arbitrary constants. This interesting
property means that we need the homogeneous equation when we are solving

25
26 CHAPTER 4. NONHOMOGENEOUS EQUATIONS

the nonhomogeneous one. The general solution is of the form

y = yh + yp (4.3)
Example 4.1 Find the general solution of y 00 − 3y 0 + 2y = 2x − 3 using
yp = x.
Let’s solve y 00 − 3y 0 + 2y = 0 first. Let yh = eλx . Then
λ2 − 3λ + 2 = 0
which means λ = 2 or λ = 1. The homogenous solution is
yh = c1 ex + c2 e2x
therefore the general solution is:
y = x + c1 ex + c2 e2x
Example 4.2 Find the general solution of y 00 = cos x using yp = − cos x.
The solution of y 00 = 0 is simply yh = c1 x + c2 , therefore the general solution
must be

y = − cos x + c1 x + c2
As you can see, once we have a particular solution, the rest is straight-
forward, but how can we find yp for a given equation?
Example 4.3 Find a particular solution of the following differential equa-
tions. Try the suggested functions. (Success not guaranteed!)
a) y 00 + y = ex , Try yp = Aex
b) y 00 − y = ex , Try yp = Aex
c) y 00 + 2y 0 + y = x Try yp = Ax + B
00 0
d) y + 2y = x Try yp = Ax + B
00 0
e) y + 2y + y = 2 cos x Try yp = A cos x and yp = A cos x + B sin x
As you can see, some of the suggestions work and some do not.
yp is usually similar to r(x). We can summarize our findings as:
• Start with a set of functions that contains not only r(x), but also all
derivatives of r(x).

• If one of the terms of yp candidate occurs in yh , there is a problem.

4.2. METHOD OF UNDETERMINED COEFFICIENTS 27

4.2 Method of Undetermined Coefficients

To solve the constant coefficient equation

d2 y dy
+ a + by = r(x) (4.4)
dx2 dx
• Solve the corresponding homogeneous equation, find yh .

• Find a candidate for yp using the following table:

Term in r(x) Choice for yp

xn An xn + · · · + A1 x + A0
eax Aeax
cos bx or sin bx A cos bx + B sin bx
xn eax (An xn + · · · + A1 x + A0 )eax
xn cos bx or xn sin bx (An xn + · · · + A0 ) cos bx
+(Bn xn + · · · + B0 ) sin bx
eax cos bx or eax sin bx Aeax cos bx + Beax sin bx
xn eax cos bx or xn eax sin bx (An xn + · · · + A0 )eax cos bx
+(Bn xn + · · · + B0 )eax sin bx
(You don’t have to memorize the table. Just note that the choice
consists of r(x) and all its derivatives)

• If your choice for yp occurs in yh , you have to change it. Multiply it

by x if the solution corresponds to a single root, by x2 if it is a double
root.

• Find the constants in yp by inserting it in the equation.

• The general solution is y = yp + yh

Note that this method works only for constant coefficient equations, and
only when r(x) is relatively simple.

Example 4.4 Find the general solution of the equation

3y 00 + 10y 0 + 3y = 9x
28 CHAPTER 4. NONHOMOGENEOUS EQUATIONS

The homogeneous equation is

3y 00 + 10y 0 + 3y = 0

Its solution is
yh = c1 e−3x + c2 e−x/3
To find a particular solution, let’s try yp = Ax + B. Inserting this in the
equation, we obtain:
10A + 3Ax + 3B = 9x
Therefore, A = 3, B = −10. The particular solution is:

yp = 3x − 10

The general solution is:

y = c1 e−3x + c2 e−x/3 + 3x − 10

Example 4.5 Find the general solution of y 00 − 4y 0 + 4y = e2x

The solution of the associated homogeneous equation

y 00 − 4y 0 + 4y = 0

is
yh = c1 e2x + c2 xe2x
Our candidate for yp is yp = Ae2x . But this is already in the yh so we have
to change it. If we multiply by x, we will obtain Axe2x but this is also in yh .
Therefore we have to multiply by x2 . So our choice for yp is yp = Ax2 e2x .
Now we have to determine A by inserting in the equation.

yp0 = 2Ax2 e2x + 2Axe2x

yp00 = 4Ax2 e2x + 8Axe2x + 2Ae2x

4Ax2 e2x + 8Axe2x + 2Ae2x − 4(2Ax2 e2x + 2Axe2x ) + 4Ax2 e2x = e2x
4.3. METHOD OF VARIATION OF PARAMETERS 29

1 1
2Ae2x = e2x ⇒ A = , yp = x2 e2x
2 2

1
y = yh + yp = c1 e2x + c2 xe2x + x2 e2x
2

4.3 Method of Variation of Parameters

Consider the linear second order nonhomogeneous differential equation

a(x)y 00 + b(x)y 0 + c(x)y = r(x) (4.5)

If a(x), b(x) and c(x) are not constants, or if r(x) is not among the functions
given in the table, we can not use the method of undetermined coefficients. In
this case, the variation of parameters can be used if we know the homogeneous
solution.
Let yh = c1 y1 + c2 y2 be the solution of the associated homogeneous equa-
tion
a(x)y 00 + b(x)y 0 + c(x)y = 0 (4.6)

Let us express the particular solution as:

yp = v1 (x)y1 + v2 (x)y2 (4.7)

There are two unknowns, so we may impose an extra condition. Let’s choose
v10 y1 + v20 y2 = 0 for simplicity. Inserting yp in the equation, we obtain
r
v10 y10 + v20 y20 =
a (4.8)
v10 y1 + v20 y2 = 0

The solution to this linear system is

−y2 r y1 r
v10 = , v20 = (4.9)
aW aW
where W is the Wronskian
 
 y y 
 1 2 
W = 0  = y1 y20 − y2 y10 (4.10)
 y1 y20 
30 CHAPTER 4. NONHOMOGENEOUS EQUATIONS

Therefore the particular solution is

Z Z
y2 r y1 r
yp (x) = −y1 dx + y2 dx (4.11)
aW aW
e−x
Example 4.6 Find the general solution of y 00 + 2y 0 + y = √
x
yh = c1 e−x + c2 xe−x
 
 e−x xe−x 
W =  = e−2x
 
 −e −x −x −x
e − xe 
xe−x e−x
Z Z −x −x
−x −x e e
yp = −e −2x
√ dx + xe √ dx
e x e−2x x
√
Z Z
−x −x 1
yp = −e x dx + xe √ dx
x
x3/2 x1/2 4
yp = −e−x + xe−x = e−x x3/2
3/2 1/2 3
4
y = yh + yp = c1 e−x + c2 xe−x + e−x x3/2
3
Example 4.7 Find the general solution of x y − 5xy 0 + 8y = x5
2 00

We can find the homogeneous solution of the Cauchy-Euler equation as:

yh = c1 x4 + c2 x2
 
 x4 x 2 
W = 3  = −2x5
 
 4x 2x 
Therefore the particular solution is
x2 x5 x4 x5
Z Z
4 2
yp (x) = −x dx + x dx
x2 (−2x5 ) x2 (−2x5 )
Z Z
1 1
yp (x) = x4 dx − x2 x2 dx
2 2
1 5
yp (x) = x
3
The general solution is

1
y = c1 x 4 + c2 x 2 + x 5
3
EXERCISES 31

Exercises
Find the general solution of the following differential equations
1) y 00 + 4y = x cos x
2) y 00 − 18y 0 + 81y = e9x
3) y 00 = −4x cos 2x − 4 cos 2x − 8x sin 2x − 8 sin 2x
4) y 00 + 3y 0 − 18y = 9 sinh 3x
5) y 00 + 16y = x2 + 2x
6) y 00 − 2y 0 + y = x2 ex

1
7) 2x2 y 00 − xy 0 + y =
x

F8) x2 y 00 + xy 0 − 4y = x2 ln x
9) y 00 − 8y 0 + 16y = 16x
10) y 00 = x3
11) y 00 + 7y 0 + 12y = e2x + x

12) y 00 + 12y 0 + 36y = 100 cos 2x

F13) y 00 + 9y = ex + cos 3x + 2 sin 3x

14) y 00 + 10y 0 + 16y = e−2x

15) y 00 − 4y 0 + 53y = (53x)2

16) y 00 + y = (x2 + 1)e3x

17) y 00 + y = csc x

18) y 00 + y = csc x sec x

00 e2x ln x
0
19) y − 4y + 4y =
x
00 e2x
0
F20) y − 2y + y = x
(e + 1)2
32 CHAPTER 4. NONHOMOGENEOUS EQUATIONS

Answers
1) y = c1 sin 2x + c2 cos 2x + 13 x cos x + 29 sin x
1
2) y = c1 e9x + c2 xe9x + x2 e9x
2
3) y = c1 + c2 x + x cos 2x + 3 cos 2x + 2x sin 2x + sin 2x
1 1
4) y = c1 e3x + c2 e−6x + e−3x + xe3x
4 2
1 2 1 1
5) y = c1 sin 4x + c2 cos 4x + x + x −
16 8 128
x x 1 4 x
6) y = c1 e + c2 xe + x e
12
√ 1
7) y = c1 x + c2 x +
6x
1 1 x2
8) y = c1 x2 + c2 x−2 + x2 ln2 x − x2 ln x +
8 16 64
4x 4x 1
9) y = c1 e + c2 xe + x +
2
x5
10) y = + c1 + c2 x
20
1 2x 1 7
11) y = c1 e−3x + c2 e−4x + e + x−
30 12 144
3
12) y = c1 e−6x + c2 xe−6x + 2 cos 2x + sin 2x
2
1 x 1 1
13) y = c1 cos 3x + c2 sin 3x + e − x cos 3x + x sin 3x
10 3 6
1
14) y = c1 e−2x + c2 e−8x + xe−2x
6
74
15) y = e2x (c1 cos 7x + c2 sin 7x) + 53x2 + 8x −
53

16) y = e3x (0.1x2 − 0.12x + 0.152) + c1 sin x + c2 cos x

17) y = c1 sin x + c2 cos x − x cos x + sin x ln | sin x|
18) y = c1 sin x + c2 cos x − cos x ln | sec x + tan x| − sin x ln | csc x + cot x|

(ln x)2
 
2x 2x 2x
19) y = c1 e + c2 xe + xe − ln x + 1
2

20) y = c1 ex + c2 xex + ex ln(1 + ex )

Chapter 5

Higher Order Equations

In this chapter, we will generalize our results about second order equations to
higher orders. The basic ideas are the same. We still need the homogeneous
solution to find the general nonhomogeneous solution. We will extend the two
methods, undetermined coefficients and variation of parameters, to higher
dimensions and this will naturally involve many more terms and constants
in the solution. We also need some new notation to express nth derivatives
easily.

5.1 Linear Equations of Order n

An nth order differential equation is called linear if it can be written in the
form
dn y dn−1 y dy
a0 (x) + a 1 (x) + · · · + a n−1 (x) + an (x)y = r(x) (5.1)
dxn dxn−1 dx
and nonlinear if it is not linear.(Note that a0 6= 0)
If the coefficients a0 (x), a1 (x), . . . an (x) are continuous, then the equation has
exactly n linearly independent solutions. The general solution is

y = c1 y1 + c2 y2 + · · · + cn yn (5.2)

Linear Independence: If

c1 y1 + c2 y2 + · · · + cn yn = 0 (5.3)

33
34 CHAPTER 5. HIGHER ORDER EQUATIONS

means that all the constants c1 , c2 , . . . , cn are zero, then this set of functions
is linearly independent. Otherwise, they are dependent.
For example, the functions x, x2 , x3 are linearly independent. The func-
tions cos2 x, sin2 x, cos 2x are not.
Given n functions, we can check their linear dependence by calculating
the Wronskian. The Wronskian is defined as
 
 y ... yn 
 1 
 y10 ... yn0
 

W (y1 , y2 , . . . , yn ) =  .. ..  (5.4)
. .

 
 (n−1) (n−1) 
 y1 . . . yn 
and the functions are linearly dependent if and only if W = 0 at some point.

5.2 Differential Operators

We can denote differentiation with respect to x by the symbol D

dy d2 y
Dy = = y0, D2 y = = y 00 (5.5)
dx dx2
etc. A differential operator is

L = a0 Dn + a1 Dn−1 + · · · + an−1 D + an (5.6)

We will only work with operators where coefficients are constant.

We can add, multiply, expand and factor constant coefficient differen-
tial operators using common rules of algebra. In this respect, they are like
polynomials. So, the following expressions are all equivalent:

(D − 2)(D − 3)y = (D − 3)(D − 2)y

= (D2 − 5D + 6)y
= y 00 − 5y 0 + 6y
Let’s apply some simple operators to selected functions:
(D − 2)ex = Dex − 2ex
= ex − 2ex = −ex
5.3. HOMOGENEOUS EQUATIONS 35

(D − 2)e2x = De2x − 2e2x

= 2e2x − 2e2x = 0
(D − 2)2 xe2x = (D − 2)(D − 2)xe2x
= (D − 2)(e2x + 2xe2x − 2xe2x )
= (D − 2)e2x = 0
(D2 − 4) sin(2x) = (D − 2)(D + 2) sin(2x)
= (D − 2)(2 cos(2x) + 2 sin(2x))
= −4 sin(2x) + 4 cos(2x) − 4 cos(2x) − 4 sin(2x)
= −8 sin 2x

5.3 Homogeneous Equations

Based on the examples in the previous section, we can easily see that:
The general solution of the equation (D − a)n y = 0 is

y = eax (c0 + c1 x + . . . + cn−1 xn−1 ) (5.7)

if a is real.
Some special cases are:

Dn y = 0 ⇒ y = c0 + c1 x + . . . + cn−1 xn−1
(D − a)y = 0 ⇒ y = eax (5.8)
(D − a)2 y = 0 ⇒ y = c1 eax + c2 xeax

We can extend these results to the case of complex roots. If z = a + ib is a

root of the characteristic polynomial, then so is z = a − ib. (Why?)
Consider the equation

(D − a − ib)n (D − a + ib)n y = (D2 − 2aD + a2 + b2 )n y = 0 (5.9)

The solution is

y = eax cos bx(c0 + c1 x + . . . + cn−1 xn−1 )

(5.10)
+eax sin bx(k0 + k1 x + . . . + kn−1 xn−1 )

A special case is obtained if a = 0.

(D2 + b2 )y = 0 ⇒ y = c1 cos bx + c2 sin bx (5.11)

36 CHAPTER 5. HIGHER ORDER EQUATIONS

Now we are in a position to solve very complicated-looking homogeneous

equations.

Method of Solution:

• Express the given equation using operator notation (D notation).

• Factor the polynomial.

• Find the solution for each component.

• Add the components to obtain the general solution.

Example 5.1 Find the general solution of y (4) − 7y 000 + y 00 − 7y 0 = 0.

In operator notation, we have

(D4 − 7D3 + D2 − 7D)y = 0

Factoring this, we obtain

D(D − 7)(D2 + 1)y = 0

We know that
Dy = 0 ⇒ y = c
(D − 7)y = 0 ⇒ y = ce7x
(D2 + 1)y = 0 ⇒ y = c1 sin x + c2 cos x
Therefore the general solution is

y = c1 + c2 e7x + c3 sin x + c4 cos x

Note that the equation is fourth order and the solution has four arbitrary
constants.

Example 5.2 Solve D3 (D − 2)(D − 3)2 (D2 + 4)y = 0.

Using the same method, we find:

y = c1 + c2 x + c3 x2 + c4 e2x + c5 e3x + c6 e3x x + c7 cos 2x + c8 sin 2x

5.4. NONHOMOGENEOUS EQUATIONS 37

5.4 Nonhomogeneous Equations

In this section, we will generalize the methods of undetermined coefficients
and variation of parameters to nth order equations.
Undetermined Coefficients: Method of undetermined coefficients is the
same as given on page 27. We will use the same table, but this time the
modification rule is more general. It should be:

• In case one of the terms of yp occurs in yh , multiply it by xk where k

is the smallest integer which will eliminate any duplication between yp
and yh .

Example 5.3 Solve the equation (D − 1)4 y = xex .

The homogeneous solution is yh = (c0 + c1 x + c2 x2 + c3 x3 )ex . According to

the table, we should choose yp as Aex + Bxex , but this already occurs in
the homogeneous solution. Multiplying by x, x2 , x3 are not enough, so, we
should multiply by x4 .
yp = Ax4 ex + Bx5 ex
Inserting this in the equation, we obtain:

24Aex + 120Bxex = xex

Therefore A = 0, B = 1/120 and the general solution is

1 5 x
y = (c0 + c1 x + c2 x2 + c3 x3 )ex + xe
120
Variation of Parameters: The idea is the same as in second order equa-
tions, but there are more unknowns to find and more integrals to evaluate.
Consider
dn y dn−1 y dy
a0 (x) + a 1 (x) + · · · + a n−1 (x) + an (x)y = r(x) (5.12)
dxn dxn−1 dx
Let the homogeneous solution be yh = c1 y1 + · · · + cn yn
Then the particular solution is yp = v1 y1 + · · · + vn yn
Here, vi are functions of x. Since we have n functions, we can impose n − 1
conditions on them. The first condition will be

v10 y1 + · · · + vn0 yn = 0 (5.13)

38 CHAPTER 5. HIGHER ORDER EQUATIONS

Then we will proceed similarly to simplify the steps. Eventually, we will

obtain the system

v10 y1 + ··· + vn0 yn = 0

v10 y10 + ··· + vn0 yn0 = 0
.. .. ..
. . . (5.14)
(n−1) (n−1)
v10 y1 + ··· + vn0 yn = 0
(n) (n) r(x)
v10 y1 + ··· + vn0 yn = a0 (x)

Then, we will solve this linear system to find vi0 , and integrate them to
obtain yp .
Z Z
yp = y1 v1 dx + · · · + yn vn0 dx
0
(5.15)

Example 5.4 Find the general solution of

x3 y 000 − 6x2 y 00 + 15xy 0 − 15y = 8x6

We can find the homogeneous solution yh = c1 x + c2 x3 + c3 x5 using our

method for Cauchy-Euler equations. Then, the particular solution will be
yp = xv1 + x3 v2 + x5 v3 . Using the above equations, we obtain the system

xv10 + x3 v20 + x5 v30 = 0

v10 + 3x2 v20 + 5x4 v30 = 0
6xv20 + 20x3 v30 = 8x3
The solution of this system is v10 = x4 , v20 = −2x2 , v30 = 1 therefore the
particular solution is
Z Z Z
4 3 2 5 8 6
yp = x x dx + x (−2x ) dx + x dx = x
15
and the general solution is

8 6
y = c1 x + c2 x 3 + c3 x 5 + x
15
EXERCISES 39

Exercises
1) D5 y = 0
2) (D − 1)3 y = 0
3) y 000 − 4y 00 + 13y 0 = 0
4) (D − 2)2 (D + 3)3 y = 0
5) (D2 + 2)3 y = 0
d4 y d2 y
6) + 5 + 4y = 0
dx4 dx2
7) (D2 + 9)2 (D2 − 9)2 y = 0
d4 y d3 y d2 y
8) − 2 + 2 =0
dx4 dx3 dx2
9) y 000 − 3y 00 + 12y 0 − 10y = 0
10) (D2 + 2D + 17)2 y = 0
11) (D4 + 2D2 + 1)y = x2
12) (D3 + 2D2 − D − 2)y = 1 − 4x3
√ √ √
F13) (2D4 + 4D3 + 8D2 )y = 40e−x [ 3 sin( 3x) + 3 cos( 3x)]
14) (D3 − 4D2 + 5D − 2)y = 4 cos x + sin x
15) (D3 − 9D)y = 8xex
40 CHAPTER 5. HIGHER ORDER EQUATIONS

Answers
1) y = c0 + c1 x + c2 x2 + c3 x3 + c4 x4
2) y = c1 ex + c2 xex + c3 x2 ex
3) y = c1 e2x cos 3x + c2 e2x sin 3x + c3
4) y = c1 e2x + c2 xe2x + c3 e−3x + c4 xe−3x + c5 x2 e−3x
√ √ √ √
5) y = c1 cos 2x + c2 sin 2x + c3 x cos 2x + c4 x sin 2x
√ √
+ c5 x2 cos 2x + c6 x2 sin 2x
6) y = c1 cos 2x + c2 sin 2x + c3 cos x + c4 sin x
7) y = c1 e3x + c2 xe3x + c3 e−3x + c4 xe−3x + c5 cos 3x + c6 sin 3x
+ c7 x cos 3x + c8 x sin 3x
8) y = c1 + c2 x + c3 ex cos x + c4 ex sin x
9) y = c1 ex + c2 ex cos 3x + c3 ex sin 3x
10) y = c1 e−x sin 4x + c2 e−x cos 4x + c3 xe−x sin 4x + c4 xe−x cos 4x
11) y = c1 cos x + c2 sin x + c3 x cos x + c4 x sin x + x2 − 4
12) y = c1 ex + c2 e−x + c3 e−2x + 2x3 − 3x2 + 15x − 8
√ √ √
13) y = c1 + c2 x + c3 e−x cos 3x + c4 e−x sin 3x + 5xe−x cos 3x
14) y = c1 ex + c2 xex + c3 e2x + 0.2 cos x + 0.9 sin x
3
15) y = c1 + c2 e3x + c3 e−3x + ex − xex
4
Chapter 6

Series Solutions

If none of the methods we have studied up to now works for a differential

equation, we may use power series. This is usually the only choice if the
solution cannot be expressed in terms of the elementary functions. (That
is, exponential, logarithmic, trigonometric and polynomial functions). If the
solution can be expressed as a power series, in other words, if it is analytic,
this method will work. But it takes time and patience to reach the solution.
Remember, we are dealing with infinitely many coefficients!

6.1 Power Series

Let’s remember some facts about the series
∞
X
an (x − x0 )n = a0 + a1 (x − x0 ) + a2 (x − x0 )2 + · · · (6.1)
n=0

from calculus.

• There is a nonnegative number ρ, called the radius of convergence, such

that the series converges absolutely for |x − x0 | < ρ and diverges for
|x − x0 | > ρ . The series defines a function f (x) = ∞ n
P
n=0 an (x − x0 )
in its interval of convergence.

• In the interval of convergence, the series can be added or subtracted

41
42 CHAPTER 6. SERIES SOLUTIONS

term wise, i.e.

∞
X
f (x) ± g(x) = (an ± bn )(x − x0 )n
n=0

• In the interval of convergence, the series can be multiplied or divided

to give another power series.
∞
X
f (x)g(x) = cn (x − x0 )n
n=0

where
cn = a0 bn + a1 bn−1 + · · · + an b0

• In the interval of convergence, derivatives and integrals of f (x) can be

found by term wise differentiation and integration, for example
∞
X
0
f (x) = a1 + 2a2 (x − x0 ) + · · · = n an (x − x0 )n−1
n=1

f (n) (x0 )
• The series ∞ (x − x0 )n is called the Taylor Series of the func-
P
n=0 n!
tion f (x). The function f (x) is called analytic if its Taylor series
converges.

Examples of some common power series are:

∞
x
X xn x2
e = =1+x+ + ···
n=0
n! 2!
∞
X (−1)n x2n x2 x4
cos x = =1− + − ···
n=0
2n! 2! 4!
∞
X (−1)n x2n+1 x3 x5
sin x = =x− + − ···
n=0
(2n + 1)! 3! 5!
∞
1 X
= xn = 1 + x + x2 + · · ·
1−x n=0
∞
X (−1)n+1 xn x2 x3
ln(1 + x) = =x− + − ···
n=1
n 2 3
6.2. CLASSIFICATION OF POINTS 43

6.2 Classification of Points

Consider the equation

R(x)y 00 + P (x)y 0 + Q(x)y = 0 (6.2)

If both of the functions

P (x) Q(x)
, (6.3)
R(x) R(x)
are analytic at x = x0 , then the point x0 is an ordinary point. Otherwise, x0
is a singular point.
Suppose that x0 is a singular point of the above equation. If both of the
functions
P (x) Q(x)
(x − x0 ) , (x − x0 )2 (6.4)
R(x) R(x)
are analytic at x = x0 , then the point x0 is called a regular singular point.
Otherwise, x0 is an irregular singular point.
For example, the functions 1+x+x2 , sin x, ex (1+x4 ) cos x are all analytic
cos x 1 ex 1 + x2
at x = 0. But, the functions , , , are not.
x x x x3
We will use power series method around ordinary points and Frobenius’
method around regular singular points. We will not consider irregular singu-
lar points.

6.3 Power Series Method

If x0 is an ordinary point of the equation R(x)y 00 + P (x)y 0 + Q(x)y = 0, then
the general solution is
∞
X
y= an (x − x0 )n (6.5)
n=0

The coefficients an can be found by inserting y in the equation and setting

the coefficients of all powers to zero. Two coefficients (Usually a0 and a1 )
must be arbitrary, others must be defined in terms of them. We expect two
linearly independent solutions because the equation is second order linear.
44 CHAPTER 6. SERIES SOLUTIONS

Example 6.1 Solve y 00 + 2xy 0 + 2y = 0 around x0 = 0.

First we should classify the point. Obviously, x = 0 is an ordinary point, so

we can use power series method.
∞
X ∞
X ∞
X
n 0 n−1 00
y= an x , y = nan x , y = n(n − 1)an xn−2
n=0 n=1 n=2

Inserting these in the equation, we obtain

∞
X ∞
X ∞
X
n(n − 1)an xn−2 + 2x nan xn−1 + 2 an x n = 0
n=2 n=1 n=0
∞
X ∞
X ∞
X
n(n − 1)an xn−2 + 2nan xn + 2an xn = 0
n=2 n=1 n=0
To equate the powers of x, let us replace n by n + 2 in the first sigma.
(n → n + 2)
∞
X ∞
X ∞
X
n n
(n + 2)(n + 1)an+2 x + 2nan x + 2an xn = 0
n=0 n=1 n=0

Now we can express the equation using a single sigma, but we should start
the index from n = 1. Therefore we have to write n = 0 terms separately.
∞
X
2a2 + 2a0 + [(n + 2)(n + 1)an+2 + (2n + 2)an ] xn = 0
n=1

−2(n + 1) −2
a2 = −a0 , an+2 = an = an
(n + 2)(n + 1) (n + 2)
This is called the recursion relation. Using it, we can find all the constants
in terms of a0 and a1 .
2 1 2 1
a4 = − a2 = a0 a6 = − a4 = − a0
4 2 6 6

2 2 4
a3 = − a1 , a 5 = − a3 = a1
3 5 15
We can find as many coefficients as we want in this way. Collecting them
together, the solution is :
   
2 1 4 1 6 2 3 4 5
y = a0 1 − x + x − x + · · · + a1 x − x + x + · · ·
2 6 3 15
6.3. POWER SERIES METHOD 45

In most applications, we want a solution close to 0, therefore we can neglect

the higher order terms of the series.
Remark: Sometimes we can express the solution in closed form (in terms
of elementary functions rather than an infinite summation) as in the next
example:

Example 6.2 Solve (x − 1)y 00 + 2y 0 = 0 around x0 = 0.

2
Once again, first we should classify the given point. The function is
x−1
analytic at x = 0, therefore x = 0 is an ordinary point.
∞
X ∞
X ∞
X
n 0 n−1 00
y= an x , y = nan x , y = n(n − 1)an xn−2
n=0 n=1 n=2

Inserting these in the equation, we obtain

∞
X ∞
X
n−2
(x − 1) n(n − 1)an x +2 nan xn−1 = 0
n=2 n=1
∞
X ∞
X ∞
X
n(n − 1)an xn−1 − n(n − 1)an xn−2 + 2nan xn−1 = 0
n=2 n=2 n=1
To equate the powers of x, let us replace n by n+1 in the second summation.
∞
X ∞
X ∞
X
n−1 n−1
n(n − 1)an x − (n + 1)nan+1 x + 2nan xn−1 = 0
n=2 n=1 n=1

Now we can express the equation using a single sigma.

∞
X
(−2a2 + 2a1 ) + [(n(n − 1) + 2n)an − n(n + 1)an+1 ] xn−1 = 0
n=2

n2 − n + 2n
a2 = a1 , an+1 = an for n > 2
n(n + 1)
So the recursion relation is:
an+1 = an
All the coefficients are equal to a1 , except a0 . We have no information about
it, so it must be arbitrary. Therefore, the solution is:

y = a0 + a1 x + x 2 + x 3 + · · ·

x
y = a0 + a1
1−x
46 CHAPTER 6. SERIES SOLUTIONS

Exercises
Find the general solution of the following differential equations in the
form of series. Find solutions around the origin (use x0 = 0). Write the
solution in closed form if possible.
1) (1 − x2 )y 00 − 2xy 0 = 0
2) y 00 + x4 y 0 + 4x3 y = 0
3) (2 + x3 )y 00 + 6x2 y 0 + 6xy = 0
4) (1 + x2 )y 00 − xy 0 − 3y = 0
5) (1 + 2x2 )y 00 + xy 0 + 2y = 0
6) y 00 − xy 0 + ky = 0
7) (1 + x2 )y 00 − 4xy 0 + 6y = 0
8) (1 − 2x2 )y 00 + (2x + 4x3 )y 0 − (2 + 4x2 )y = 0
9) (1 + 8x2 )y 00 − 16y = 0
10) y 00 + x2 y = 0
The following equations give certain special functions that are very im-
portant in applications. Solve them for n = 1, 2, 3 around origin. Find
polynomial solutions only.

11) (1 − x2 )y 00 − 2xy 0 + n(n + 1)y = 0 (Legendre’s Equation)

12) y 00 − 2xy 0 + 2ny = 0 (Hermite’s Equation)
13) xy 00 + (1 − x)y 0 + ny = 0 (Laguerre’s Equation)
14) (1 − x2 )y 00 − xy 0 + n2 y = 0 (Chebyshev’s Equation)

Solve the following initial value problems. Find the solution around the
point where initial conditions are given.

F15) xy 00 + (x + 1)y 0 − 2y = 0, x0 = −1, y(−1) = 1, y 0 (−1) = 0

16) y 00 + 2xy 0 − 4y = 0, x0 = 0, y(0) = 1, y 0 (0) = 0
17) 4y 00 + 3xy 0 − 6y = 0, x0 =0 y(0) = 4, y 0 (0) = 0
18) (x2 − 4x + 7)y 00 + y = 0, x0 =2 y(2) = 4, y 0 (2) = 10

F19) Find the recursion relation for (p + x2 )y 00 + (1 − q − r)xy 0 + qry = 0

around x = 0. (Here p, q, r are real numbers, p 6= 0)
F20) Solve (1 + ax2 )y 00 + bxy 0 + cy = 0 around x0 = 0
EXERCISES 47

Answers
x3 x5
   
1 1+x
1) y = a0 + a1 x + + + · · · OR y = a0 + a1 ln
3 5 2 1−x

x5 x10 x15
 
2) y = a0 1 − + − + ···
5 5 · 10 5 · 10 · 15
x6 x11 x16
 
+a1 x − + − + ···
6 6 · 11 6 · 11 · 16

x3 x6 x9 x4 x7 x10
   
3) y = a0 1 − + − + · · · + a1 x − + − + · · · OR
2 4 8 2 4 8
a0 a1 x
y= 3 + 3
1 + x2 1 + x2
   
3 2 3 4 1 6 2 3
4) y = a0 1 + x + x − x + · · · + a1 x + x
2 8 16 3
   
2 2 4 2 6 1 3 17 5
5) y = a0 1 − x + x − x + · · · + a1 x − x + x + · · ·
3 3 2 40
 
k 2 k(k − 2) 4 k(k − 2)(k − 4) 6
6) y = a0 1 − x + x − x + ···
2! 4! 6!
 
k − 1 3 (k − 1)(k − 3) 5 (k − 1)(k − 3)(k − 5) 7
+a1 x − x + x − x + ···
3! 5! 7!
x3
 
2
7) y = a0 (1 − 3x ) + a1 x −
3

x4 x6
 
2
8) y = a0 1 + x + + + · · · + a1 x
2 6
 
2 8 3 64 5
9) y = a0 (1 + 8x ) + a1 x + x − x + · · ·
3 15

x4 x8 x5 x9
   
10) y = a0 1 − + + · · · + a1 x − + + ···
12 672 20 1440
11) n = 1 ⇒ y = a1 x
n=2 ⇒ y = a0 (1 − 3x2 )
5
n=3 ⇒ y = a1 (x − x3 )
3
48 CHAPTER 6. SERIES SOLUTIONS

12) n = 1 ⇒ y = a1 x
n=2 ⇒ y = a0 (1 − 2x2 )
2
n=3 ⇒ y = a1 (x − x3 )
3

13) n = 1 ⇒ y = a0 (1 − x)
1
n=2 ⇒ y = a0 (1 − 2x + x2 )
2
3 1
n=3 ⇒ y = a0 (1 − 3x + x2 − x3 )
2 6

14) n = 1 ⇒ y = a1 x
n=2 ⇒ y = a0 (1 − 2x2 )
4
n=3 ⇒ y = a1 (x − x3 )
3
1 1
15) y = 1 − (x + 1)2 − (x + 1)3 − (x + 1)4 − · · ·
3 6

16) y = 1 + 2x2

17) y = 4 + 3x2
 
1 2 1 4
18) y = 4 1 − (x − 2) + (x − 2) + · · ·
 6 72 
1 3 7 5
+ 10 (x − 2) − (x − 2) + (x − 2) + · · ·
18 1080
(n − q)(n − r)
19) an+2 = − an
p(n + 2)(n + 1)
x2 x4

20) y = a0 1 − c + c(2a + 2b + c)
2 4!
x6

− c(2a + 2b + c)(12a + 4b + c) + ···
6!
x3 x5

+ a1 x − (b + c) + (b + c)(6a + 3b + c)
3! 5!
x7
−(b + c)(6a + 3b + c)(20a + 5b + c) + ···
7!
Chapter 7

Frobenius’ Method

In this chapter, we will extend the methods of the previous chapter to regular
singular points. The calculations will be considerably longer, but the basic
ideas are the same. The classification of the given point is necessary to make
a choice of methods.

7.1 An Extension of Power Series Method

Suppose x0 is a regular singular point. For simplicity, assume x0 = 0. Then
p(x) 0 q(x)
the differential equation can be written as y 00 + y + 2 y = 0 where
x x
p(x) and q(x) are analytic. We can try a solution of the form
∞
X
y = xr an x n (7.1)
n=0

The equation corresponding to the lowest power xr−2 , in other words

r(r − 1) + p0 r + q0 = 0 is called the indicial equation, where p0 = p(0), and
q0 = q(0). Now we can find r, insert it in the series formula, and proceed as
we did in the previous chapter.
We can classify the solutions according to the roots of the indicial equa-
tion.
Case 1 - Distinct roots not differing by an integer: A basis of solutions
is ∞ ∞
X X
y 1 = xr 1 an xn , y2 = xr2 bn x n (7.2)
n=0 n=0

49
50 CHAPTER 7. FROBENIUS’ METHOD

Case 2 - Equal roots: A basis of solutions is

∞
X ∞
X
r n r
y1 = x an x , y2 = y1 ln x + x bn x n (7.3)
n=0 n=1

Case 3 - Roots differing by an integer: A basis of solutions is

∞
X ∞
X
y 1 = xr 1 an xn , y2 = ky1 ln x + xr2 bn x n (7.4)
n=1 n=0

where r1 − r2 = N > 0 (r1 is the greater root) and k may or may not be zero.
In all three cases, there is at least one relatively simple solution of the
form y = xr ∞ n
P
n=0 an x . The equation is second order, so there must be a
second linearly independent solution. In Cases 2 and 3, it may be difficult
to find the second solution. You may use the method of reduction of order.
This is convenient especially if y1 is simple enough. Alternatively, you may
use the above formulas directly, and determine bn one by one using the an
and the equation.

7.2 Examples
Example 7.1 Solve 4xy 00 + 2y 0 + y = 0 around x0 = 0.
2
First we should classify the given point. The function 4x is not analytic at
x = 0 therefore x = 0 is a singular point. We should make a further test to
determine whether it is regular or not.
x2
The functions 2x
4x
and 4x are analytic therefore x = 0 is a R.S.P., we can
use the method of Frobenius.
∞
X ∞
X ∞
X
n+r 0 n+r−1 00
y= an x , y = (n + r)an x , y = (n + r)(n + r − 1)an xn+r−2
n=0 n=0 n=0

Note that the summation for the derivatives still starts from 0, because r
does not have to be an integer. This is an important difference between
methods of power series and Frobenius.
Inserting these in the equation, we obtain
∞
X ∞
X ∞
X
4x (n + r)(n + r − 1)an xn+r−2 + 2 (n + r)an xn+r−1 + an xn+r = 0
n=0 n=0 n=0
7.2. EXAMPLES 51

∞
X ∞
X ∞
X
n+r−1 n+r−1
4(n + r)(n + r − 1)an x + 2(n + r)an x + an xn+r = 0
n=0 n=0 n=0

We want to equate the powers of x, so n → n + 1 in the first two terms.

∞
X ∞
X ∞
X
n+r n+r
4(n + r + 1)(n + r)an+1 x + 2(n + r + 1)an+1 x + an xn+r = 0
n=−1 n=−1 n=0

Now we can express the equation using a single sigma, but the index of the
common sigma must start from n = 0. Therefore we have to write n = −1
terms separately.

X∞
r−1
[4r(r−1)+2r]a0 x + {[4(n + r + 1)(n + r) + 2(n + r + 1)]an+1 + an } xn+r = 0
n=0

We know that a0 6= 0, therefore 4r2 − 2r = 0. This is the indicial equation.

Its solutions are r = 0, r = 12 . Therefore this is Case 1.
If r = 0, the recursion relation is
−1
an+1 = an
4(n + 1)(n + 21 )
a0 a1 a0 a2 a0
a1 = − , a2 = − 3 = , a3 = − 5 = − ,...
2 4.2. 2 4! 4.3. 2 6!
For simplicity, we may choose a0 = 1. Then

(−1)n
an =
2n!
Therefore the first solution is:
∞
X (−1)n xn √
y1 = = cos x
n=0
2n!

1
If r = , the recursion relation is
2
−1 −an
an+1 = 3 an =
4(n + 2 )(n + 1) (2n + 3)(2n + 2)

a0 a1 a0 a2 a0
a1 = − , a2 = − = , a3 = − = − ,...
3.2 5.4 5! 7.6 7!
52 CHAPTER 7. FROBENIUS’ METHOD

For simplicity, we may choose a0 = 1. Then

(−1)n
an =
(2n + 1)!
Therefore the second solution is :
∞
1/2
X (−1)n xn √
y2 = x = sin x
n=0
(2n + 1)!

The general solution is y = c1 y1 + c2 y2

Example 7.2 Solve x2 y 00 + (x2 − x)y 0 + (1 + x)y = 0 around x0 = 0.

2
First we should classify the given point. The function x x−x
2 is not analytic
at x = 0 therefore x = 0 is a singular point. The functions x − 1 and
1 + x are analytic at x = 0 therefore x = 0 is a R.S.P., we can use the
method of Frobenius. Evaluating the derivatives of y and inserting them in
the equation, we obtain
∞
X ∞
X
n+r
(n + r)(n + r − 1)an x + (n + r)an xn+r+1
n=0 n=0
X∞ ∞
X ∞
X
− (n + r)an xn+r + an xn+r + an xn+r+1 = 0
n=0 n=0 n=0

Let’s replace n by n − 1 in the second and fifth terms.

∞
X ∞
X
n+r
(n + r)(n + r − 1)an x + (n + r − 1)an−1 xn+r
n=0 n=1
X∞ ∞
X ∞
X
n+r n+r
− (n + r)an x + an x + an−1 xn+r = 0
n=0 n=0 n=1

[r2 − 2r + 1]a0 xr +
X∞
{[(n + r)(n + r − 1) − (n + r) + 1]an + [(n + r − 1) + 1]an−1 } xn+r = 0
n=1

The indicial equation is r2 − 2r + 1 = 0 ⇒ r = 1 (double root). Therefore

this is Case 2. The recursion relation is
n+1
an = − an−1
n2
7.2. EXAMPLES 53

For simplicity, let a0 = 1. Then

3 3 4 2
a1 = −2, a2 = − a1 = , a3 = − a2 = −
4 2 9 3
Therefore the first solution is :
 
3 2 2 3
y1 = x 1 − 2x + x − x + · · ·
2 3

To find the second solution, we will use reduction of order. Let y2 = uy1 .
Inserting y2 in the equation, we obtain

x2 y1 u00 + (2x2 y10 − xy1 + x2 y1 )u0 = 0

Let w = u0 then
y10
 
0 1
w + 2 − +1 w =0
y1 x
y10
 
dw 1
= −2 + − 1 dx
w y1 x
xe−x
ln w = −2 ln y1 + ln x − x ⇒ w= 2
y1
Z
1
To evaluate the integral u = w dx we need to find . This is also a series.
y12
 −2  
1 1 3 2 2 3 1 2 46 3
2
= 2 1 − 2x + x − x + · · · = 2 1 + 4x + 9x + x + · · ·
y1 x 2 3 x 3

xe−x x2 x3
   
1 2 46 3
w = 2 =x 1−x+ − + ··· 1 + 4x + 9x + x + · · ·
y1 2! 3! x2 3
 
1 11 13
w= 1 + 3x + x2 + x3 + · · ·
x 2 6
Z
11 13
u = w dx = ln x + 3x + x2 + x3 + · · ·
4 18
 
13 2 3 3
y2 = uy1 = y1 ln x + x 3x − x + x + · · ·
4 2
54 CHAPTER 7. FROBENIUS’ METHOD

Exercises
Find two linearly independent solutions of the following differential equa-
tions in the form of series. Find solutions around the origin (use x0 = 0).
Write the solution in closed form if possible.
1) 2x2 y 00 − xy 0 + (1 + x)y = 0

2) 2xy 00 + (1 + x)y 0 − 2y = 0

3) (x2 + 2x)y 00 + (3x + 1)y 0 + y = 0

4) xy 00 − y 0 − 4x3 y = 0

5) xy 00 + y 0 − xy = 0

6) 3x2 y 00 + (−10x − 3x2 )y 0 + (14 + 4x)y = 0

7) x2 y 00 + (x2 − x)y 0 + y = 0

8) (2x2 + 2x)y 00 − y 0 − 4y = 0

9) 2x2 y 00 + (2x2 − x)y 0 + y = 0

10) 4x2 y 00 + (2x2 − 10x)y 0 + (12 − x)y = 0

11) (x2 + 2x)y 00 + (4x + 1)y 0 + 2y = 0

Use Frobenius’ method to solve the following differential equations around

origin. Find the roots of the indicial equation, find the recursion relation,
and two linearly independent solutions.
12) (x2 + cx)y 00 + [(2 + b)x + c(1 − d)]y 0 + by = 0
(b 6= 0, c 6= 0, d is not an integer).

13) x2 y 00 + [(1 − b − d)x + cx2 ]y 0 + [bd + (1 − b)cx]y = 0

(c 6= 0, b − d is not an integer).

14) x2 y 00 + [(1 − 2d)x + cx2 ]y 0 + (d2 + (1 − d)cx)y = 0

(c 6= 0)

15) xy 00 + [1 − d + cx2 ]y 0 + 2cxy = 0

(c 6= 0, d is not an integer).
EXERCISES 55

Answers
∞
!
X (−1)n xn
1) y = c1 x 1 +
n=1
n! · 3 · 5 · 7 · · · (2n + 1)
∞
!
1
X (−1)n xn
+ c2 x 2 1+
n=1
n! · 1 · 3 · 5 · · · (2n − 1)
∞
!
(−1)n 3xn
 
1 2 1
X
2) y = c1 1 + 2x + x + c2 x 2 1+
3 n=1
2n n!(2n − 3)(2n − 1)(2n + 1)
 
2 6 3 15 2 35 3
3) y1 = 1−x+ x2 − x3 +· · · , y2 = x 1/2
1− x+ x − x + ···
3 15 4 32 128
∞ ∞
X x4n X x4n+2 2 2
4) y = a0 + a2 , OR y = c1 ex + c2 e−x
n=0
(2n)! n=0
(2n + 1)!
x2 x4 x6
5) y1 = 1 + + + + ···
22 (2 · 4)2 (2 · 4 · 6)2
x2 3x4 11x6
y2 = y1 ln x − − − − ···
4 8 · 16 64 · 6 · 36
 
7/3 3 9 2 27 3
6) y1 = x 1+ x+ x + x + ···
4 28 280
x2 x3
 
2
y2 = x 1 + x + + + · · · = x2 e x
2! 3!
x2 x3
 
7) y1 = x 1 − x + − + · · · = xe−x
2! 3!
x2 x3
 
−x −x
y2 = xe ln x + xe x+ + + ···
2 · 2! 3 · 3!
 
2 3/2 1 1 2 1 3
8) y1 = 1 − 4x − 8x , y2 = x 1 + x − x + x − ···
2 8 16
" ∞
#
n n
X (−1) (2x)
9) y1 = x1/2 e−x , y2 = x 1 +
1 · 3 · 5 · · · (2n + 1)
" n=1 ∞ #
n n
X (−1) x
10) y1 = x2 e−x/2 , y2 = x3/2 1 +
n=1
1 · 3 · 5 · · · (2n − 1)
56 CHAPTER 7. FROBENIUS’ METHOD

8
11) y1 = 1 − 2x + 2x2 − x3 + · · ·
 5 
1/2 5 35 2 105 3
y2 = x 1− x+ x − x + ···
4 32 128
n+b
12) r = 0 ⇒ an+1 = − an
c (n + 1 − d)
b b(b + 1)
y1 = 1 − x+ 2 x2 − · · ·
c(1 − d) c (1 − d)(2 − d)
n+b+d
r=d ⇒ an+1 = − an
c (n + 1)
 
d d+b (d + b)(d + b + 1) 2
y2 = x 1 − x+ x − ···
c 2! c2
c
13) r = b ⇒ an = − an−1
n+b−d
c2
 
b c 2
y1 = x 1 − x+ x − ···
1+b−d (1 + b − d)(2 + b − d)
c
r = d ⇒ an = − an−1
n
c 2 c3 3
2
 
y2 = x 1 − c x + x − x + · · · = xd e−cx
d
2! 3!
c
14) r = d (double root) an = − an−1
n
c 2 c3 3
2
 
y1 = x 1 − c x + x − x + · · · = xd e−cx
d
2! 3!
Z cx
e
y2 = xd e−cx dx
x
c2 2 c3 3
 
d −cx d −cx
y2 = x e ln x + x e cx + x + x + ···
2 · 2! 3 · 3!
c
15) r = 0 ⇒ an+2 = − an
(n + 2 − d)
c c2 c3
y1 = 1 − x2 + x4 − x6 + · · ·
2−d (2 − d)(4 − d) (2 − d)(4 − d)(6 − d)
c
r = d ⇒ an+2 = − an
n+2
c2 4 c3
 
d c 2 6
y2 = x 1 − x + x − x + ···
2 2·4 2·4·6
Chapter 8

Laplace Transform I

Laplace transform provides an alternative method for many equations. We

first transform the differential equation to an algebraic equation, then solve
it, and then make an inverse transform. Laplace transform has a lot of
interesting properties that make these operations easy. In this chapter, we
will see the definition and the basic properties. We will also compare this
method to the method of undetermined coefficients, and see in what ways
Laplace transform is more convenient.

8.1 Definition, Existence and Inverse of Laplace

Transform
The Laplace transform of a function f (t) is defined as:
Z ∞
F (s) = L {f (t)} = e−st f (t)dt (8.1)
0

then, the inverse transform will be

f (t) = L−1 {F (s)} (8.2)

Note that we use lowercase letters for functions and capital letters for their
transforms.

57
58 CHAPTER 8. LAPLACE TRANSFORM I

Example 8.1 Evaluate the Laplace transform of the following functions:

a) f (t) = 1 ∞
∞
e−st 
Z
−st 1
L {1} = e dt = = , s>0
0 −s 0  s
b) f (t) = eat
∞
∞
e(a−s)t 
Z
 at
at −st 1
L e = e e dt = = , s>a
0 a−s 0  s−a
(
0 if 0<t<1
c) f (t) =
1 if 16t

∞
∞
e−st  e−s
Z
−st
L {f } = e dt = = , s>0
1 −s 1 s

d) f (t) = t Z ∞
L {t} = te−st dt
0

Using integration by parts, we obtain

∞ Z ∞ −st
e−st  e
L {t} = −t + dt
s 0 
0 s
∞
e−st  1
L {t} = − 2  = 2 , s > 0
s 0 s
The integral that defines the Laplace transform is an improper integral,
it may or may not converge. In the above examples, the transform is defined
for a certain range of s.
In practice, we can use Laplace transform on most of the functions we
encounter in differential equations. The following definitions and the theorem
answer the question Which functions have a Laplace transform?
Piecewise Continuous Functions: A function f (t) is piecewise continuous
on [a, b] if the interval can be subdivided into subintervals [ti , tj ],
a = t0 < t1 < t2 · · · < tn = b such that f (t) is continuous on each interval
and has finite one-sided limits at the endpoints (from the interior).
An example can be seen on Figure 8.1.
8.2. PROPERTIES 59

Figure 8.1: A piecewise continuous function

Exponential Order: f (t) is of exponential order as t → ∞ if there exist

real constants M, c, T such that |f (t)| 6 M ect for all t > T . In other words,
a function is of exponential order if it does not grow faster than ect .
Theorem 8.1: If f (t) is of exponential order and piecewise continuous on
[0, k] for all k > 0, then its Laplace transform exists for all s > c.
For example, all the polynomials have a Laplace transform. The function
2
et does NOT have a Laplace transform.

8.2 Basic Properties of Laplace Transforms

It is difficult to evaluate the Laplace transform of each function by perform-
ing an integration. Instead of this, we use various properties of Laplace
transform.
Let L {f (t)} = F (s), then, some basic properties are: (assuming these
transforms exists)

• Linearity
L {af + bg} = aL {f } + bL {g}

• Shifting
L eat f (t) = F (s − a)


L−1 {F (s − a)} = eat f (t)

60 CHAPTER 8. LAPLACE TRANSFORM I

• Transform of Derivatives
L {f 0 } = sL {f } − f (0)
00
L {f } = s2 L {f } − sf (0) − f 0 (0)
= sn L {f } − sn−1 f (0) − sn−2 f 0 (0) − · · · − f (n−1) (0)
 (n)
L f

• Transform of Integrals
Z t 
F (s)
L f (x) dx =
0 s
Example 8.2 Find the Laplace transform of sin at and cos at. Hint: Use
Euler’s formula eix = cos x + i sin x and linearity.
eiat − e−iat L {eiat } − L {e−iat }
sin at = ⇒ L {sin at} =
2i   2i
1 1 1 a
L {sin at} = − = 2
2i s − ia s + ia s + a2
Similarly, we can show that the transform of f (t) = cos at is
s
F (s) = 2
s + a2
1
Example 8.3 Find the inverse Laplace transform of F (s) = .
(s + 5)2
Hint: Use shifting.
 
−1 1
We know that L = t. Therefore
s2
 
−1 1
L = te−5t
(s + 5)2
Example 8.4 Find the Laplace transform of f (t) = t2 . Hint: Use Deriva-
tives.
Using L {f 0 } = sL {f } − f (0), we obtain
 L {2t} 2
L {2t} = sL t2 − 0 ⇒ L t2 =

= 3
s s
3
Example 8.5 Find the Laplace transform of f (t) = t . Hint: Use Integrals.
Using the integral rule, we see that
 3
t L {t2 } 2
L = = 4
3 s s
6
L t3 = 4

s
8.3. INITIAL VALUE PROBLEMS 61

8.3 Initial Value Problems

Consider the constant-coefficient equation

y 00 + ay 0 + by = r(t) (8.3)

with initial values

y(0) = p, y 0 (0) = q (8.4)

Here y is a function of t (y = y(t)). We can solve it by the method of undeter-

mined coefficients. The method of Laplace transform will be an alternative
that is more efficient in certain cases. It also works for discontinuous r(t).
Let us evaluate the Laplace transform of both sides.

L {y 00 } + aL {y 0 } + bL {y} = L {r(t)} (8.5)

Using L {y} = Y (s) and L {r(t)} = R(s)

s2 Y − sp − q + a(sY − p) + bY = R (8.6)

(s2 + as + b)Y = R + (s + a)p + q (8.7)

R + (s + a)p + q
Y = (8.8)
s2 + as + b
 
−1 R + sp + ap + q
y=L (8.9)
s2 + as + b
Note that this method can be generalized to higher order equations. The
advantages compared to the method of undetermined coefficients are:

• The initial conditions are built in the solution, we don’t need to deter-
mine constants after obtaining the general solution.

• There is no distinction between homogeneous and nonhomogeneous

equations, or single and multiple roots. The same method works in all
cases the same way.

• The function on the right hand side r(t) belongs to a wider class. For
example, it can be discontinuous.
62 CHAPTER 8. LAPLACE TRANSFORM I

The only disadvantage is that, sometimes finding the inverse Laplace

transform is too difficult.
We have to find roots of the polynomial s2 + as + b, which is the same as
the characteristic polynomial we would encounter if we were using method
of undetermined coefficients.

Example 8.6 Solve the initial value problem

y 00 + 4y = 0, y(0) = 5, y 0 (0) = 3.

Let’s start by finding the transform of the equation.

L {y 00 } + 4L {y} = 0

s2 Y − 5s − 3 + 4Y = 0 ⇒ (s2 + 4)Y = 5s + 3
5s + 3
Y = 2
s +4
Now, we have to find the inverse transform of Y to obtain y(t).
5s 3 2
Y = +
s2 + 4 2 s2 + 4
3
y(t) = L−1 {Y } = 5 cos 2t + sin 2t
2
Note that we did not first find the general solution containing arbitrary con-
stants. We directly found the result.

Example 8.7 Solve the initial value problem

1
y 00 − 4y 0 + 3y = 1, y(0) = 0, y 0 (0) = −
3
Transform both sides:

L {y 00 − 4y 0 + 3y} = L {1}

Use the derivative rule

1 1
s2 Y − s.0 + − 4(sY − 0) + 3Y =
3 s
8.3. INITIAL VALUE PROBLEMS 63

Isolate Y
1 1 3−s
(s2 − 4s + 3)Y = − =
s 3 3s
s−3
(s − 1)(s − 3)Y = −
3s
 
1 1 1 1
Y =− = −
3s(s − 1) 3 s s−1
Find the inverse transform
1 1 t
y(t) = L−1 {Y } = − e
3 3
As you can see, there’s no difference between homogeneous and nonhomoge-
neous equations. Laplace transform works for both types in the same way.

Example 8.8 Solve the initial value problem

y 00 + 4y 0 + 4y = 42te−2t , y(0) = 0, y 0 (0) = 0

L {y 00 } + 4L {y 0 } + 4L {y} = 42L te−2t


1
s2 Y + 4sY + 4Y = 42 ·
(s + 2)2
42
(s2 + 4s + 4)Y =
(s + 2)2
42
Y =
(s + 2)4
42 3 −2t
y(t) = L−1 {Y (s)} = te
3!
y(t) = 7t3 e−2t
If you try the method of undetermined coefficients on this problem, you will
appreciate the efficiency of Laplace transforms better.
64 CHAPTER 8. LAPLACE TRANSFORM I

f (t) F (s) f (t) F (s)

1 eat − ebt 1
1
s a−b (s − a)(s − b)
1 aeat − bebt s
t
s2 a−b (s − a)(s − b)
n! b
tn eat sin bt
sn+1 (s − a)2 + b2
1 s−a
eat eat cos bt
s−a (s − a)2 + b2
1 n!
teat tn eat
(s − a)2 (s − a)n+1
a a
sin at sinh at
s + a2
2 s − a2
2

s s
cos at cosh at
s 2 + a2 s 2 − a2
2as 2as
t sin at t sinh at
(s2 + a2 )2 (s2 − a2 )2
s 2 − a2 s 2 + a2
t cos at t cosh at
(s2 + a2 )2 (s2 − a2 )2
2a3 2as2
sin at − at cos at sin at + at cos at
(s2 + a2 )2 (s2 + a2 )2

Table 8.1: A Table of Laplace Transforms

EXERCISES 65

Exercises
Find the Laplace transform of the following functions:
1) f (t) = cos2 2t 2) f (t) = et sin 3t

3) f (t) = 2e−t cos2 t 4) f (t) = (t + 1)2 et

(
1 0<t<a
5) f (t) = t3 e3t 6) f (t) =
0 a<t

 t 0<t<a
( 
t 0<t<a
7) f (t) = 8) f (t) = 1 a<t<b
0 a<t 
0 b<t


Find the inverse Laplace transform of the following functions:

s−4
9) F (s) = 2
s −4
3
10) F (s) =
(s − 2)2
6
11) F (s) =
s(s + 4)
1
12) F (s) = 2
s(s + 9)
1
13) F (s) = 2
s (s + 1)
5s + 1
14) F (s) = 2
s +4
1
15) F (s) =
s+8
1
16) F (s) =
(s − a)n
Solve the following initial value problems using Laplace transform:
17) y 00 − 2y 0 + y = 0, y(0) = 4, y 0 (0) = −3
18) y 00 − 2y 0 + 2y = 0, y(0) = 0, y 0 (0) = 1
19) y 00 + 2y = 4t2 + 12, y(0) = 4, y 0 (0) = 0
20) y 00 + 6y 0 + 9y = e−3t , y(0) = 0, y 0 (0) = 0
66 CHAPTER 8. LAPLACE TRANSFORM I

Answers

1 s
1) F (s) = + 2
2s 2s + 2
3
2) F (s) =
(s − 1)2 + 9
1 s+1
3) F (s) = + 2
s + 1 s + 2s + 5
2 2 1
4) F (s) = 3
+ 2
+
(s − 1) (s − 1) s−1
6
5) F (s) =
(s − 3)4
1 − e−as
6) F (s) =
s
1 ae−as e−as
7) F (s) = 2 − − 2
s s s
−as −as
1−e e − ae−as − e−bs
8) F (s) = +
s2 s
9) f (t) = cosh 2t − 2 sinh 2t
10) f (t) = 3te2t
11) f (t) = (3 − 3e−4t )/2
12) f (t) = (1 − cos 3t)/9
13) f (t) = e−t + t − 1
1
14) f (t) = 5 cos 2t + sin 2t
2
−8t
15) f (t) = e
tn−1 eat
16) f (t) =
(n − 1)!
17) y(t) = 4et − 7tet
18) y(t) = et sin t
19) y(t) = 4 + 2t2
1
20) y(t) = e−3t t2
2
Chapter 9

Laplace Transform II

In this chapter, we will study more advanced properties of Laplace transform.

At the end, we will be able to find transform and inverse transform of a wider
range of functions. This will enable us to solve almost any linear constant
coefficient equation, including discontinuous inputs.

9.1 Convolution
The convolution of two functions f and g is defined as
Z t
h(t) = (f ∗ g)(t) = f (x)g(t − x) dx (9.1)
0

The convolution operation is commutative, in other words f ∗ g = g ∗ f

Theorem 9.1: The transform of convolution of two functions is equal to
the product of their transforms, i.e.

L {f ∗ g} = F (s) · G(s) (9.2)

L−1 {F (s) · G(s)} = f ∗ g (9.3)

where L {f } = F (s) and L {g} = G(s).
Proof: Using the definitions of convolution and Laplace transform,
Z t 
L {f ∗ g} =L f (x) g(t − x) dx
0
Z ∞Z t
= f (x) g(t − x)e−st dx dt
0 0

67
68 CHAPTER 9. LAPLACE TRANSFORM II

Reversing the order of integration, we obtain:

Z ∞Z ∞
= f (x) g(t − x)e−st dt dx
0 x
Making the substitution u = t − x, we obtain:
Z ∞Z ∞
L {f ∗ g} = f (x) g(u)e−su−sx dudx
Z0 ∞ 0 Z ∞
−sx
= f (x)e dx g(u)e−su du
0 0

=F (s) G(s)
1
Example 9.1 Find the inverse Laplace transform of F (s) = 3 .
s + 4s2
     
−1 1 −1 1 −4t −1 1 1
L = t, L =e ⇒ L · = t ∗ e−4t
s2 s+4 s2 s + 4
Z t
−4t
f (t) = t ∗ e = xe−4(t−x) dx
0
 4x t
−4t xe e4x 
=e −
4 16 0
t 1 e−4t
= − +
4 16 16
s
Example 9.2 Find the inverse Laplace transform of F (s) = 2 .
(s + 1)2
s 1
If we express F as F (s) = 2 · 2 = L {cos t} · L {sin t},
(s + 1) (s + 1)
we will see that f (t) = L−1 {F } = cos t ∗ sin t.
Z t
f (t) = cos(x) sin(t − x) dx
0
Z t
1
= [sin(t − x + x) + sin(t − x − x)] dx
0 2
1 t
Z
= [sin(t) + sin(t − 2x)] dx
2 0
 t
1 cos(t − 2x) 
= x sin t +
2 2 
0
 
1 1
= t sin t + (cos t − cos t)
2 2
1
= t sin t
2
9.2. UNIT STEP FUNCTION 69

9.2 Unit Step Function

The Heaviside step function (or unit step function) is defined as
(
0 if t<a
ua (t) = u(t − a) = (9.4)
1 if t>a
This is a simple on off function. It is especially useful to express discon-
tinuous inputs.

Figure 9.1: u(t − a) and its effect on f (t)

Theorem 9.2: [t−shifting] Let L {f (t)} = F (s), then

L {f (t − a) u(t − a)} = e−as F (s) (9.5)

Proof: Using the definition,

Z ∞
L {f (t − a) u(t − a)} = e−st f (t − a) u(t − a) dt
Z0 ∞
= e−st f (t − a) dt
Za ∞
= e−sa−sx f (x) dx ( where x = t − a)
0
−as
=e F (s)
(
0 if t<5
Example 9.3 Find the Laplace transform of g(t) =
t if t>5

We can express g(t) as g(t) = u(t − 5)f (t − 5) where f (t) = (t + 5). Then
 
1 5 −5s 1 5
F (s) = L {f (t)} = 2 + ⇒ L {g(t)} = e +
s s s2 s

9.3 Differentiation of Transforms

If f (t) is piecewise continuous and of exponential order, then we can differ-
entiate its Laplace transform integral.
70 CHAPTER 9. LAPLACE TRANSFORM II

Z ∞
F (s) = e−st f (t)dt
Z0 ∞ (9.6)
0 −st
F (s) = (−t)e f (t)dt
0
In other words
L {tf (t)} = −F 0 (s) (9.7)
Repeating this procedure n times, we obtain:

dn
L {tn f (t)} = (−1)n F (s) (9.8)
dsn

Example 9.4 Find the Laplace transform of f (t) = t sin t.

Using the derivative formula, we find

 
d 1 2s
L {t sin t} = − =
ds 1 + s2 (1 + s2 )2
9.4. PARTIAL FRACTIONS EXPANSION 71

9.4 Partial Fractions Expansion

In many applications of Laplace transform, we need to expand a rational
function in partial fractions. Here, we will review this technique by examples.

2x + 1 A B C
= + +
(x − 2)(x + 3)(x − 1) x−2 x+3 x−1
2
x + 4x − 5 A B C D
= + + +
(x − 2)(x − 1)3 x − 2 x − 1 (x − 1)2 (x − 1)3
x3 + 1 A Bx + C Dx + E
2 2
= + 2 + 2
x(x + 4) x x +4 (x + 4)2
x3 − 4x2 + x + 9 3 A B
= x + 1 + = x + 1 + +
x2 − 5x + 6 x2 − 5x + 6 x−2 x−3
• We can express any polynomial as a product of first and second order
polynomials.

• For second order polynomials in the expansion, we have to use Ax + B

(not simply a constant) in the numerator.

• If numerator’s degree is greater or equal to the denominator, we should

first divide them using polynomial division.
−s2 + 7s − 1
Example 9.5 Find the inverse Laplace transform of F (s) = .
(s − 2)(s − 5)2
First, we have to express F (s) in terms of simpler fractions:
−s2 + 7s − 1 A B C
2
= + +
(s − 2)(s − 5) s − 2 s − 5 (s − 5)2

−s2 + 7s − 1 = A(s − 5)2 + B(s − 2)(s − 5) + C(s − 2)

Inserting s = 2, we see that 9 = 9A ⇒ A = 1.
Inserting s = 5, we see that 9 = 3C ⇒ C = 3.
The coefficient of s2 : A + B = −1 therefore B = −2. So
−s2 + 7s − 1 1 2 3
2
= − +
(s − 2)(s − 5) s − 2 s − 5 (s − 5)2
Now we can easily find the inverse Laplace transform:
L−1 {F (s)} = e2t − 2e5t + 3te5t
72 CHAPTER 9. LAPLACE TRANSFORM II

9.5 Applications
Now we are in a position to solve a wider class of differential equations using
Laplace transform.

Example 9.6 Solve the initial value problem

y 00 − 6y 0 + 8y = 2e2t , y(0) = 11, y 0 (0) = 37

We will first find the Laplace transform of both sides, then find Y (s)

L {y 00 } − 6L {y 0 } + 8L {y} = L 2e2t


2
s2 Y − 11s − 37 − 6(sY − 11) + 8Y =
s−2

2
(s2 − 6s + 8)Y = + 11s − 29
s−2
The factors of s2 − 6s + 8 are (s − 2) and (s − 4), so

2 11s − 29
Y = +
(s − 2)(s − 2)(s − 4) (s − 2)(s − 4)

11s2 − 51s + 60
Y =
(s − 2)2 (s − 4)
Now we need to find the inverse Laplace transform. Using partial fractions
expansion

A B C
Y = + 2
+
s − 2 (s − 2) s−4
After some algebra we find that A = 3, B = −1, C = 8 so

3 1 8
Y (s) = − +
s − 2 (s − 2)2 s − 4

y(t) = L−1 {Y (s)} = 3e2t − te2t + 8e4t

9.5. APPLICATIONS 73

Example 9.7 Solve the initial value problem

y 00 + y = f (t), y(0) = 0, y 0 (0) = 3

(
0 if 0 < t < 5π
where f (t) =
2 cos t if 5π < t

As you can see, the input function is discontinuous, but this makes no
difference for Laplace transform.

L {y 00 } + L {y} = L {f }
s2 Y − 3 + Y = F
F +3
Y = 2
s +1
1
Using the fact that L {sin t} = 2 , we can obtain y(t) by convolution:
s +1
y(t) = L−1 {Y } = f (t) ∗ sin t + 3 sin t

Using the definition of convolution,

Z t
f ∗ sin t = f (x) sin(t − x) dx
0

If t < 5π, f = 0 therefore this integral is also zero. If t > 5π we have

Z t
f ∗ sin t = 2 cos x sin(t − x) dx
5π

Using the trigonometric identity 2 sin A cos B = sin(A + B) + sin(A − B) we

obtain Z t
f ∗ sin t = sin t + sin(t − 2x) dx
5π
 t
cos(t − 2x) 
= x sin t +
2 
5π
= (t − 5π) sin t
Therefore
(
3 sin t if 0 < t < 5π
y(t) =
(t − 5π + 3) sin t if 5π < t
74 CHAPTER 9. LAPLACE TRANSFORM II

Example 9.8 Solve the initial value problem

y 00 + 2y 0 + y = r(t), y(0) = 0, y 0 (0) = 0

(
t if 0<t<1
where r(t) =
0 if 1<t

Once again we have a discontinuous input. This time we will use unit
step function. First, we have to express r(t) with a single formula.

r(t) = t − u(t − 1)t = t − u(t − 1)(t − 1) − u(t − 1)

Its Laplace transform is

1 e−s e−s
R(s) = L {r(t)} = − −
s2 s2 s
Finding the Laplace transform of the equation, we obtain

(s2 + 2s + 1)Y = R

R
Y =
(s + 1)2
1 e−s
Y = −
s2 (s + 1)2 s2 (s + 1)
Using partial fractions expansion
 
2 1 2 1 −s 1 1 1
Y =− + 2 + + −e − + 2+
s s s + 1 (s + 1)2 s s s+1

Using the fact that L−1 {e−as F (s)} = f (t − a)u(t − a), we obtain

y(t) = −2 + t + 2e−t + te−t − u(t − 1) −1 + (t − 1) + e−(t−1)

We know that u(t − 1) = 0 for t > 1 and u(t − 1) = 1 for t > 1 so

(
−2 + t + 2e−t + te−t if 0 < t < 1
y(t) =
(2 − e)e−t + te−t if 1 < t
EXERCISES 75

Exercises

Find the Laplace transform transform of the following functions:

1) f (t) = te−t cos t 2) f (t) = t2 sin 2t

Find the inverse Laplace transform transform of the following functions:

e−3s se−s
3) F (s) = 2 4) F (s) = 2
s +1 s +4
1 1
5) F (s) = 2 2
6) F (s) = 3
(s + 16) s + 4s2 + 3s
s+3 s3
7) F (s) = F8) F (s) =
(s2 + 4)2 s4 + 4a4
s2 3s2 − 2s + 5
9) F (s) = 10) F (s) =
(s2 + 4)2 (s − 2)(s2 + 9)

Solve the following initial value problems : (where y = y(t))

11) y 00 − y 0 − 2y = 0, y(0) = 8, y 0 (0) = 7

12) y 00 + y = 2 cos t, y(0) = 3, y 0 (0) = 4

13) y 00 + 0.64y = 5.12t2 , y(0) = −25, y 0 (0) = 0

14) y 00 − 2y 0 + 2y = e−t , y(0) = 0, y 0 (0) = 1

15) y 00 + y = t, y(0) = 0, y 0 (0) = 0

(
1 if 0 < t < 2π
16) y 00 + y = r(t), y(0) = 1, y 0 (0) = 0 where r(t) =
0 if 2π < t

17) y 00 + y = e−2t sin t, y(0) = 0, y 0 (0) = 0

(
5 if 0<t<π
18) y 00 +2y 0 +5y = r(t), y(0) = 0, y 0 (0) = 0 where r(t) =
0 if π<t

19) 4y 00 + 4y 0 + 17y = g(t), y(0) = 0, y 0 (0) = 0

(
sin t if 0 < t < 3π 1 7
20) y 00 − y 0 − 6y = , y(0) = , y 0 (0) = −
0 if 3π < t 50 50
76 CHAPTER 9. LAPLACE TRANSFORM II

Answers
s2 + 2s 12s2 − 16
1) F (s) = 2) F (s) =
(s2 + 2s + 2)2 (s2 + 4)3
3) f (t) = u(t − 3) sin(t − 3) 4) f (t) = u(t − 1) cos(2t − 2)
sin 4t − 4t cos 4t 1 e−t e−3t
5) f (t) = 6) f (t) = − +
128 3 2 6
4t sin 2t + 3 sin 2t − 6t cos 2t
7) f (t) = 8) f (t) = cosh at cos at
16
1 t 2
9) f (t) = sin 2t + cos 2t 10) f (t) = e2t + 2 cos 3t + sin 3t
4 2 3

11) y = 3e−t + 5e2t

12) y = 3 cos t + (4 + t) sin t

13) y = −25 + 8t2

1 −t
e − et cos t + 7et sin t


14) y =
5
15) y = t − sin t
(
1 0 < t < 2π
16) y =
cos t 2π < t
1 1
17) y = (sin t − cos t) + e−2t (sin t + cos t)
8 8
  
−t sin 2t
 1−e cos 2t + , 0<t<π


2
18) y =  
sin 2t
 e−t (eπ − 1) cos 2t + π<t


2
1 t − 1 (t−x)
Z
19) y = e 2 sin 2(t − x)g(x) dx
8 0

1
(cos t − 7 sin t) if 0 < t < 3π


20) y = 50
1 −9π 3t 2

 e e − e6π e−2t if 3π < t
50 50
Chapter 10

Fourier Analysis I

The trigonometric functions sine and cosine are the simplest periodic func-
tions. If we can express an arbitrary periodic function in terms of these,
many problems would be simplified. In this chapter, we will see how to
find the Fourier series of a periodic function. Fourier series is important in
many applications. We will also need them when we solve partial differential
equations.

10.1 Fourier Series

Let f (x) be a periodic function with period 2L. It is sufficient that f be
defined on [−L, L]. Is it possible to express f as a linear combination of sine
and cosine functions?
∞ ∞
X nπx X nπx
f (x) = a0 + an cos + bn sin (10.1)
n=1
L n=1
L
If possible, this expansion would be very useful in all kinds of applications.
Once we solve a question for sine and cosine functions, we will be able to
solve it for any periodic f . Here, an and bn are the coordinates of f in the
space of sine and cosine functions. But then how can we find an and bn ? The
following identities will help us:
Z L
nπx mπx
cos sin dx = 0 (for all m, n) (10.2)
−L L L

77
78 CHAPTER 10. FOURIER ANALYSIS I

Z L
nπx mπx
cos cos dx = 0 (m 6= n) (10.3)
−L L L

Z L
nπx mπx
sin sin dx = 0 (m 6= n) (10.4)
−L L L

Z L Z L
nπx 2 nπx
cos dx = sin2 dx = L (10.5)
−L L −L L
In the terminology of linear algebra, the trigonometric functions form
an orthogonal coordinate basis. We can easily prove these formulas if we
remember the following trigonometric identities:

2 cos A cos B = cos(A − B) + cos(A + B)

2 sin A sin B = cos(A − B) − cos(A + B) (10.6)
2 cos A sin B = sin(A + B) − sin(A − B)

Now, suppose the expansion (10.1) exists. To find ak , we will multiply

both sides by cos kπx
L
and then integrate from −L to L.

Z L Z L
kπx kπx
f (x) cos dx = a0 cos dx
−L L −L L
∞ Z L
X nπx kπx
+ an cos cos dx (10.7)
n=1 −L L L
∞ Z L
X nπx kπx
+ bn sin cos dx
n=1 −L L L

Using the property of orthogonality, we can see that all those integrals
are zero, except the kth one. Therefore

Z L Z L
kπx 1 kπx
f (x) cos dx = ak L ⇒ ak = f (x) cos dx (10.8)
−L L L −L L

We can apply the same procedure to find a0 and bn . In the end, we will
obtain the following formulas for a function f defined on [−L, L].
10.1. FOURIER SERIES 79

Z L
1
a0 = f (x) dx
2L −L
1 L
Z
nπx
Fourier coefficients: an = f (x) cos dx (10.9)
L −L L
1 L
Z
nπx
bn = f (x) sin dx
L −L L

∞ ∞
X nπx X nπx
Fourier series: f (x) = a0 + an cos + bn sin (10.10)
n=1
L n=1
L

Example 10.1 Find the Fourier series of the periodic function

f (x) = x2 , −L 6 x 6 L having period= 2L.

Z L
1
a0 = x2 dx
2L −L
3 L
L2

1 x 
= =
2L 3 −L 3
Using integration by parts two times we find:
Z L
1 nπx
an = x2 cos dx
L −L L

4L2 cos nπ
=
n2 π 2
1 L 2
Z
nπx
bn = x sin dx = 0
L −L L
Therefore the Fourier series is:
∞
L2 X 4L2 nπx
x2 = + (−1)n 2 2 cos
3 n=1
nπ L
The plot of the Fourier series up to n = 1, 2 and 3 is given in Figure 10.1.
80 CHAPTER 10. FOURIER ANALYSIS I

Figure 10.1: Fourier Series of f = x2 for n = 1, 2, 3

10.2. CONVERGENCE OF FOURIER SERIES 81

10.2 Convergence of Fourier Series

Like any infinite series, Fourier series is of no use if it is divergent. But
most functions that we are interested in have Fourier series that converge
and converge to the function.
Theorem 10.1: Let f be periodic with period 2L and let f and f 0 be
piecewise continuous on the interval [−L, L]. Then the Fourier expansion of
f converges to:
• f (x) if f is continuous at x.
f (x+ ) + f (x− )
• if f is discontinuous at x.
2
Example 10.2 Find the Fourier series of the periodic function
(
a if −L < x < 0
f (x) =
b if 0<x<L
having period= 2L. Then evaluate the series at x = L.

Z 0 Z L
1 1 a+b
a0 = a dx + b dx =
2L −L 2L 0 2
Z 0 Z L
1 nπx 1 nπx
an = a cos dx + b cos dx = 0
L −L L L 0 L
Z 0 Z L
1 nπx 1 nπx
bn = a sin dx + b sin dx
L −L L L 0 L
0 L
a L nπx  b L nπx  b−a
=− cos − cos = (1 − (−1)n )
L nπ L −L L nπ
 L 0  nπ
Therefore the Fourier series is:
∞
a+b Xb−a nπx
f (x) = + [1 − (−1)n ] sin
2 n=1
nπ L
 
a + b 2(b − a) πx 1 3πx 1 5πx
= + sin + sin + sin + ···
2 π L 3 L 5 L
a+b
If we insert x = L in that series, we obtain f (L) = . Thus the value at
2
discontinuity is the average of left and right limits. The summation of the
series up to n = 1, 5 and 9 is plotted on Figure 10.2.
82 CHAPTER 10. FOURIER ANALYSIS I

Figure 10.2: Convergence at a discontinuity

10.3 Parseval’s Identity

Theorem 10.2: Let f be continuous on [−L, L], f (L) = f (−L) and let f 0
be piecewise continuous. Then the Fourier coefficients of f satisfy:

∞
1 L
X Z
2a20 + 2 2
(an + bn ) = f (x)2 dx (10.11)
n=1
L −L

Proof: We can express f (x) as f (x) = a0 + ∞

P nπx
P∞ nπx
n=1 an cos L + n=1 bn sin L .
Now multiply both sides by f and integrate
∞ ∞
2
X nπx X nπx
f (x) = a0 f (x) + an f (x) cos + bn f (x) sin
n=1
L n=1
L
Z L Z L ∞ Z L ∞ Z L
2
X nπx X nπx
f (x) dx = a0 f (x) dx + an f (x) cos dx + bn f (x) sin dx
−L −L n=1 −L L n=1 −L L
Using equation (10.9) to evaluate these integrals, we can obtain the result.
∞
X 1 1 1
Example 10.3 Find the sum of the series S = 4
= 1 + 4 + 4 + ···
n=1
n 2 3
(Hint: Use the Fourier series of f (x) = x2 on the interval −π < x < π)

Evaluating the integrals in (10.9) for f (x) = x2 we obtain

π2 4(−1)n
a0 = , a n = and bn = 0 so
3 n2
π2
 
1 1
f (x) = − 4 cos x − cos 2x + cos 3x − · · ·
3 4 9
Using Parseval’s theorem, we have

2π 4 1 π 4
  Z
1 1
+ 16 1 + 4 + 4 + · · · = x dx
9 2 3 π −π
2
= π4
5
Therefore
10.3. PARSEVAL’S IDENTITY 83

   
1 1 4 2 2
16 1 + 4 + 4 + · · · = π −
2 3 5 9
1 1 π4
S = 1 + 4 + 4 + ··· =
2 3 90
84 CHAPTER 10. FOURIER ANALYSIS I

Exercises
Find the Fourier series of the periodic function f (x) defined on the given
interval

1) f (x) = x, −π < x < π 2) f (x) = x, 0 < x < 2π

(
0 if −π < x < 0
3) f (x) = 4) f (x) = x2 , 0 < x < 2π
1 if 0 < x < π
5) f (x) = sin2 x, −π < x < π 6) f (x) = x + |x|, −π < x < π
( (
−π/4 if −1 < x < 0 π if −π < x < 0
7) f (x) = 8) f (x) =
π/4 if 0 < x < 1 x if 0 < x < π
9) f (x) = |x|, −2 < x < 2 10) f (x) = | sin x|, −π < x < π
( (
x if 0 < x < 1 −a if −L < x < 0
11) f (x) = 12) f (x) =
1 − x if 1 < x < 2 a if 0 < x < L
13) f (x) = ax + b, −L < x < L 14) f (x) = 1 − x2 , −1 < x < 1
F15) f (x) = x3 , −π < x < π 16) f (x) = ex , −π < x < π

17) Using integration by parts, show that:

Z
x sin ax cos ax
x cos ax dx = +
a a2
Z
x cos ax sin ax
x sin ax dx = − +
a a2
x2 sin ax 2x cos ax 2 sin ax
Z
x2 cos ax dx = + −
a a2 a3
x2 cos ax 2x sin ax 2 cos ax
Z
x2 sin ax dx = − + +
a a2 a3

1 1 π2
18) Show that 1 + + + ··· = .
9 25 8
EXERCISES 85

Answers

 
1 1
1) f (x) = 2π sin x − sin 2x + sin 3x − · · ·
2 3
 
1 1
2) f (x) = π − 2 sin x + sin 2x + sin 3x + · · ·
2 3
∞
1 X 1 − (−1)n
 
1 2 1 1
3) f (x) = + sin nx = + sin x + sin 3x + sin 5x + · · ·
2 n=1 nπ 2 π 3 5

4π 2
 
1 1
4) f (x) = + 4 cos x + cos 2x + cos 3x + · · ·
3 4 9
 
1 1
−4π sin x + sin 2x + sin 3x + · · ·
2 3
1 1
5) f (x) = − cos 2x
2 2
 
π 4 1 1
6) f (x) = − cos x + cos 3x + cos 5x + · · ·
2 π 9 25
 
1 1 1
+2 sin x − sin 2x + sin 3x − sin 4x + · · ·
2 3 4
1 1
7) f (x) = sin πx + sin 3πx + sin 5πx + · · ·
3 5
∞
3π X (−1)n − 1
 
1
8) f (x) = + 2
cos nx − sin nx
4 n=1
πn n
 
8 πx 1 3πx 1 5πx
9) f (x) = 1 − 2 cos + cos + cos + ···
π 2 9 2 25 2
∞
2 4 X cos 2nx
10) f (x) = −
π π n=1 4n2 − 1

 
4 1 1
11) f (x) = − 2 cos πx + cos 3πx + cos 5πx + · · ·
π 9 25
 
2 1
+ sin πx + sin 3πx + · · ·
π 3
86 CHAPTER 10. FOURIER ANALYSIS I
 
4a πx 1 3πx 1 5πx
12) f (x) = sin + sin + sin + ···
π L 3 L 5 L
 
2aL πx 1 2πx 1 3πx
13) f (x) = b + sin − sin + sin − ···
π L 2 L 3 L
 
2 4 1 1
14) f (x) = + 2 cos πx − cos 2πx + cos 3πx + · · ·
3 π 4 9
∞  2

n+1 (nπ) − 6
X
15) f (x) = 2 (−1) 3
sin nx
n=1
n
" ∞
#
n
2 sinh π 1 X (−1)
16) f (x) = + (cos nx − n sin nx)
π 2 n=1 1 + n2

18) Use the function in exercise 12 in Parseval’s identity

Chapter 11

Fourier Analysis II

In this chapter, we will study more advanced properties of Fourier series. We

will find the even and odd periodic extensions of a given function, we will
express the series using complex notation and finally, we will extend the idea
of Fourier series to nonperiodic functions in the form of a Fourier integral.

11.1 Fourier Cosine and Sine Series

If f (−x) = f (x), f is an even function. If f (−x) = −f (x), f is an odd
function. We can easily see that, for functions:

even × even = even, odd × odd = even, even × odd = odd

For example |x|, x2 , x4 , cos x, cos nx, cosh x are even functions. x, x3 , sin x, sin nx, sinh x
are odd functions. ex is neither even nor odd.
Z L Z L
If f is even: f (x) dx = 2 f (x) dx (11.1)
−L 0
Z L
If f is odd: f (x) dx = 0 (11.2)
−L

Using the above equations, we can see that in the Fourier expansion of an
even function, bn = 0, and in the expansion of an odd function, an = 0. This
will cut our work in half if we can recognize the given function as odd or
even.

87
88 CHAPTER 11. FOURIER ANALYSIS II

Figure 11.1: Plots of Some Even and Odd Functions

As you can see in Figure 11.1, an even function is symmetric with respect
to y−axis, an odd function is symmetric with respect to origin.
Half Range Extensions: Let f be a function defined on [0, L]. If we want
to expand it in terms of sine and cosine functions, we can think of it as
periodic with period 2L. Now we need to define f on the interval [−L, 0].
There are infinitely many possibilities, but for simplicity, we are interested
in making f an even or an odd function. If we define f for negative x
values as f (x) = f (−x), we obtain the even periodic extension of f , which
is represented by a Fourier cosine series. If we define f for negative x values
as f (x) = −f (−x), we obtain the odd periodic extension of f , which is
represented by a Fourier sine series.
Half-Range Cosine Expansion: (or Fourier cosine series)
∞
X nπx
f (x) = a0 + an cos , (0 < x < L) (11.3)
n=1
L

Z L Z L
1 2 nπx
where a0 = f (x) dx, an = f (x) cos dx (11.4)
L 0 L 0 L
11.1. FOURIER COSINE AND SINE SERIES 89

Half-Range Sine Expansion: (or Fourier sine series)

∞
X nπx
f (x) = bn sin , (0 < x < L) (11.5)
n=1
L
where Z L
2 nπx
bn = f (x) sin dx (11.6)
L 0 L

Example 11.1 Find the half-range cosine and sine expansions of

(
0 if 0 < x < π2
f (x) = π
2
if π2 < x < π

Here, L = π, therefore
Z π
1 π π
a0 = dx =
π π 2 4
Z 2π
2 π
an = cos nx dx
π
π
2
2
π
sin nx  sin nπ
2
= = −
n π n
2

Therefore half-range cosine series of f is

∞
π X sin nπ
 
2 π 1 1
f (x) = − cos nx = − cos x − cos 3x + cos 5x − · · ·
4 n=1 n 4 3 5

On the other hand,

Z π
2 π
bn = sin nx dx
π π
2
2
π
− cos nx  cos nπ
2
− cos nπ
= =
n  π n
2

Therefore half-range sine series of f is

∞
X cos nπ
2
− cos nπ 1 1
f (x) = sin nx = sin x − sin 2x + sin 3x + sin 5x + · · ·
n=1
n 3 5
90 CHAPTER 11. FOURIER ANALYSIS II

11.2 Complex Fourier Series

Consider the Fourier series of f (x):
∞
X ∞
X
f (x) = a0 + an cos nx + bn sin nx (11.7)
n=1 n=1

Using Euler’s formula eix = cos x + i sin x we can express the sine and cosine
functions as:
einx + e−inx einx − e−inx
cos nx = , sin nx = (11.8)
2 2i
Therefore
   
an − ibn an + ibn
an cos nx + bn sin nx = e inx
+ e−inx (11.9)
2 2

If we define c0 = a0 and
an − ibn an + ibn
cn = , c−n = , n = 1, 2, 3, . . . (11.10)
2 2
We will obtain
∞
X
f (x) = cn einx (11.11)
n=−∞

where Z π
1
cn = f (x)e−inx dx n = 0, ±1, ±2, . . . (11.12)
2π −π

For a function of period 2L we have

∞ Z L
X 1
f (x) = cn e inπx/L
, cn = f (x)e−inπx/L dx (11.13)
n=−∞
2L −L

Example 11.2 Find the complex Fourier series of

f (x) = x if −π < x < π and f (x + 2π) = f (x).

We have to evaluate the integral

Z π
1
cn = xe−inx dx
2π −π
11.2. COMPLEX FOURIER SERIES 91

For n = 0 this integral is zero, so we have c0 = 0. For n 6= 0

π Z π −inx 
1 xe−inx 

e
cn = − dx
2π −in −π 
−π −in
1 πe−inπ + πeinπ
 
= −0
2π −in
1 einπ + e−inπ cos nπ
=− =−
in 2 in
i
= (−1)n
n
Therefore ∞
X i
x= (−1)n einx , n 6= 0
n=−∞
n
Note that we can obtain the real Fourier series from the complex one. If we
add nth and −nth terms we get
cos nx + i sin nx cos(−nx) + i sin(−nx) sin nx
i(−1)n + i(−1)−n = (−1)n+1
n −n n
∞
X sin nx
x= (−1)n+1
n=1
n
This is the real Fourier series.

Example 11.3 Find the complex Fourier series of f (x) = k

Z π
1
cn = ke−inx dx
2π −π
π
k e−inx 
= (n 6= 0)
2π −in −π
k einπ − e−inπ
=
nπ 2i
k
= sin nπ
nπ
=0
If n = 0 we have Z π
1
c0 = k dx
2π −π

=k
92 CHAPTER 11. FOURIER ANALYSIS II

11.3 Fourier Integral Representation

In this section, we will apply the basic idea of the Fourier series to non-
periodic functions.
Consider a periodic function with period= 2L and its Fourier series. In
the limit L → ∞, the summation will be an integral, and f will be a non-
periodic function. Then we will obtain the Fourier integral representation:
Z ∞
f (x) = [A(u) cos ux + B(u) sin ux] du (11.14)
0

where Z ∞
1
A(u) = f (x) cos ux dx (11.15)
π −∞

Z ∞
1
B(u) = f (x) sin ux dx (11.16)
π −∞

Like the Fourier series, we have A(u) = 0 for odd functions and B(u) = 0 for
even functions.
0
TheoremZ 11.1: If f and f are piecewise continuous in every finite interval
∞
and if |f | dx is convergent, then the Fourier integral of f converges to:
−∞

• f (x) if f is continuous at x.

f (x+) + f (x−)
• if f is discontinuous at x.
2

Example 11.4 Find the Fourier integral representation of

(
π/2 if |x| < 1
f (x) =
0 if 1 < |x|

Note that f is even therefore B(u) = 0

1 ∞ 1 1 π
Z Z
A(u) = f (x) cos ux dx = cos ux dx
π −∞ π −1 2
Z 1 1
sin ux  sin u
= cos ux dx = =
0 u 0  u
11.3. FOURIER INTEGRAL REPRESENTATION 93

Therefore, Fourier integral representation of f is

Z ∞
sin u
f (x) = cos ux du
0 u
Example 11.5 Prove the following formulas using two different methods:

eax
Z
eax cos bx dx = (a cos bx + b sin bx)
a2 + b 2
eax
Z
eax sin bx dx = 2 (a sin bx − b cos bx)
a + b2
We can obtain the formulas using integration by parts, but this is the
long way. A better method is to express the integrals as a single complex
integral using eibx = cos bx + i sin bx, then evaluate it at one step, and then
separate the real and imaginary parts.

Example 11.6 Find the Fourier integral representation of

(
−ex cos x if x < 0
f (x) =
e−x cos x if 0 < x

This function is odd therefore A(u) = 0.

1 ∞ 2 ∞ −x
Z Z
B(u) = f (x) sin ux dx = e cos x sin ux dx
π −∞ π 0
2 ∞ −x sin(ux + x) + sin(ux − x)
Z
= e dx
π 0 2
∞
1 e−x 
= 2
[− sin(u + 1)x − (u + 1) cos(u + 1)x]
π 1 + (u + 1)
−x
0∞
1 e 
+ [− sin(u − 1)x − (u − 1) cos(u − 1)x] 
π 1 + (u − 1)2 
0
 
1 u+1 u−1
= +
π 1 + (u + 1)2 1 + (u − 1)2
2 u3
=
π u4 + 4
So ∞
u3
Z
2
f (x) = sin ux du
π 0 u4 + 4
94 CHAPTER 11. FOURIER ANALYSIS II

Exercises
For the following functions defined on 0 < x < L, find the half-range
cosine and half-range sine expansions:

(
2kx/L if 0 < x < L/2
1) f (x) = 2) f (x) = ex
2k(L − x)/L if L/2 < x < L

3) f (x) = k 4) f (x) = x4
(
0 if 0 < x < L/2
5) f (x) = cos 2x 0 < x < π 6) f (x) =
k if L/2 < x < L

Find the complex

( Fourier series of the following functions:
0 if −π < x < 0
7) f (x) = 8) f (x) = x2 , −L < x < L
1 if 0 < x < π
9) f (x) = sin x 10) f (x) = cos 2x
Find the Fourier integral representations of the following functions:
 π π
(  cos x, |x| <
π − x, 0 < x < π 
2 2
11) f (x) = (f odd) 12) f (x) =
0, π<x π
0, |x| >


2
( (
−x
e , 0<x π if 0 < x < 1
13) f (x) = 14) f (x) =
ex , x < 0 0 if Otherwise

Prove the following formulas. (Hint: Define a suitable function f and

then find its Fourier integral representation.)
πx2 /2,


 06x<1
Z ∞    
2 2 cos ux 
15) 1 − 2 sin u + cos u du = π/4, x=1
0 u u u 


0, 1<x


Z ∞
cos ux + u sin ux  0,
 x<0
16) du = π/2, x = 0
0 1 + u2 
 −x
πe , x > 0
EXERCISES 95

Answers
 
k 16k 1 2πx 1 6πx
1) f (x) = − 2 cos + 2 cos + ···
2 π 22 L 6 L
 
8k 1 πx 1 3πx 1 5πx
f (x) = 2 sin − 2 sin + 2 sin − ···
π 12 L 3 L 5 L
∞
1 L X 2L n L nπx
2) f (x) = (e − 1) + [(−1) e − 1] cos
L n=1
L2 + n2 π 2 L
∞
X 2nπ nπx
f (x) = [1 − (−1)n eL ] sin
n=1
L2 2
+n π 2 L

3) f (x) = k
 
4k πx 1 3πx 1 5πx
f (x) = sin + sin + sin + ···
π L 3 L 5 L
∞
L4
 
4
X
n 1 6 nπx
4) f (x) = + 8L (−1) − cos
5 n=1
n2 π 2 n4 π 4 L
∞    
4
X
n+1 1 12 24 24 nπx
f (x) = 2L (−1) − 3 3+ 5 5 + 5 5 sin
n=1
nπ nπ nπ nπ L

5) f (x) = cos 2x
∞
4 2X n
f (x) = − sin x + [1 − (−1)n ] 2 sin nx
3π π n=3 n −4
∞
k 2k X sin nπ
2 nπx
6) f (x) = − cos
2 π n=1 n L
∞
2k X cos nπ
2
− cos nπ nπx
f (x) = sin
π n=1 n L
∞
1 X i
7) f (x) = + [(−1)n − 1]einx , n 6= 0
2 n=−∞ 2πn
∞
L2 2L2 X (−1)n inπx/L
8) f (x) = + 2 e , n 6= 0
3 π n=−∞ n2
96 CHAPTER 11. FOURIER ANALYSIS II

i i
9) f (x) = − eix + e−ix
2 2
1 1
10) f (x) = e2ix + e−2ix
2 2
Z ∞
2 πu − sin πu
11) f (x) = sin xu du
π 0 u2
Z ∞
cos πu


2
cos xu
12) f (x) = du
0 1 − u2
2 ∞ cos xu
Z
13) f (x) = du
π 0 1 + u2
Z ∞   
1 − cos u sin u
14) f (x) = sin ux + cos ux du
0 u u
Chapter 12

Partial Differential Equations,

Wave Equation

All the differential equations we have seen up to now were ordinary, that is,
they had one independent variable. In real life, almost any problem has more
than one independent variables. Therefore the subject of partial differential
equations is vast and complicated. In this chapter we will see how to model
a physical situation to set up an equation. We will obtain a solution using
the method of separation of variables. Fourier series and ODE solutions will
be necessary in this process.

12.1 Introduction
An equation involving partial derivatives of an unknown function is called a
partial differential equation, or PDE for short. Mathematical formula-
tion of problems where there are more than one independent variables require
PDE’s and they are usually much more complicated than ODE’s. (Ordinary
Differential Equations)
The definition of linear, nonlinear, homogeneous and nonhomogeneous
equations are similar to that of ODE’s. So, a general second order linear
partial differential equation is:

∂ 2u ∂ 2u ∂ 2u ∂u ∂u
A + B + C + D + E + Fu = G (12.1)
∂x2 ∂x∂y ∂y 2 ∂x ∂y

97
98 CHAPTER 12. PARTIAL DIFFERENTIAL EQUATIONS

where the unknown function is u and the two independent variables are x and
y. Here A, B, . . . , G are functions that may depend on x and y but not on u.
If G is zero, the equation is homogeneous, otherwise it is nonhomogeneous.
We can generalize these concepts into higher order PDE’s, but we will
work with second order equations in the remainder of this book. A lot of
problems in elastic vibrations, heat conduction, potential theory, wave prop-
agation and quantum mechanics can be formulated by second order linear
PDE’s.
Examples: All of the following are linear and homogeneous equations:

Wave equation in one dimension utt − c2 uxx = 0 (12.2)

Wave equation in three dimensions utt − c2 ∇2 u = 0 (12.3)
Heat equation in one dimension ut − κuxx = 0 (12.4)

Laplace equation in Cartesian coordinates:

∇2 u = uxx + uyy + uzz = 0 (12.5)

Laplace equation in cylindrical coordinates: (x = ρ cos θ, y = ρ sin θ)

uρ uθθ
uρρ + + 2 + uzz = 0 (12.6)
ρ ρ
Solutions: Many different functions may solve a given PDE, for example
the functions
u(x, t) = cos ct sin x
u(x, t) = 4ect e−x
(12.7)
u(x, t) = (4x − 6)(10t + 1)
u(x, t) = (x − ct)5
are all solutions to equation 12.2. (Please verify.)
Initial and Boundary Conditions: If the unknown function is specified
at a certain time, this is called an Initial Condition (IC). If it is specified at
the boundary of a region, it is called a Boundary Condition (BC).
Superposition of Solutions: If u1 and u2 satisfy a linear homogeneous
PDE, then a linear combination of them (i.e. c1 u1 + c2 u2 ) also satisfies the
same equation.
12.2. MODELING A VIBRATING STRING 99

12.2 Modeling a Vibrating String

Figure 12.1: A piece of a vibrating string

Consider a small part of a string with linear mass density ρ and the
length of the undeflected string ∆x. (Figure 12.1) There’s no motion in the
horizontal direction, so the net force must be zero in this direction:

T1 cos θ1 = T2 cos θ2 = T (12.8)

Here T denotes the horizontal component of tension. The net force is mass
times acceleration by Newton’s second law, so

T2 sin θ2 − T1 sin θ1 = ρ∆x utt

(12.9)
T (tan θ2 − tan θ1 ) = ρ∆x utt

We know that tan θ is the same thing as the value of the derivative at that
point, therefore:  
∂u  ∂u 
−
∂x x+∆x ∂x x ρ
= utt (12.10)
∆x T
In the limit ∆x → 0 the expression on the left becomes the second derivative
at x. Using c2 = Tρ we obtain the one-dimensional wave equation:

utt = c2 uxx (12.11)

Here c is the wave velocity. As you can see, the velocity depends on tension
and linear density of the string.

12.3 Method of Separation of Variables

This is the basic method we will use in the solution of PDE’s. The idea is as
follows:

• Assume that the solution u(x, t) is u(x, t) = F (x)G(t).

• Insert this in the equation. Transform the PDE into two ODE’s.
100 CHAPTER 12. PARTIAL DIFFERENTIAL EQUATIONS

• Solve the ODE’s. Then, superpose all the solutions.

• Find the solutions that satisfy the given boundary and initial conditions

There are a lot of tricks and details in the process that are best explained
on an example:

Example 12.1 Formulate and solve the problem of motion of a guitar string
that is initially given a shape as seen in Figure 12.2 and no initial velocity.

Figure 12.2: The initial shape of a guitar string

We know that the PDE satisfied by a vibrating string is:

utt = c2 uxx

The string is fixed at the points x = 0 and x = L therefore the Boundary

Conditions are
u(0, t) = 0, u(L, t) = 0
The initial displacement is given in the figure, and the initial velocity is zero,
therefore 
2h L
x if 0 < x <


u(x, 0) = L 2
2h L

 (L − x) if <x<L
 L 2
∂u(x, t) 
=0
∂t  t=0
This is the typical formulation of a PDE together with BC and IC. Now we
start the method of separation of variables by assuming u(x, t) = F (x)G(t),
then
utt = F G00 , uxx = F 00 G ⇒ F G00 = c2 F 00 G
G00 F 00
=
c2 G F
Note that the left hand side depends on t only and the right hand side
depends on x only, so this equality is possible only if both are equal to a
constant. Therefore
G00 F 00
= =k
c2 G F
12.3. METHOD OF SEPARATION OF VARIABLES 101

Case 1) k > 0, k = p2 , F = Aepx + Be−px , using the BC we find

A + B = 0, AepL + Be−pL = 0

Inserting B = −A in the second equation, we get

A(epL − e−pL ) = 0, p 6= 0 ⇒ A = 0, B = 0

therefore F = 0 and u = F G = 0 so the solution is trivial.

Case 2) k = 0, F 00 = 0, F = Ax + B, using the BC we find

B = 0, AL + B = 0

therefore A = 0 and F = 0, u = F G = 0 so the solution is again trivial.

Case 3) k < 0, k = −p2 , F = A cos px + B sin px, using the BC we find

A = 0, A cos pL + B sin pL = 0

Therefore B sin pL = 0.
At this point, one possibility is to choose B = 0, but this would again give
the trivial solution u = 0. An alternative is to make sin pL = 0, which is
possible if pL = nπ. Therefore
nπ
p= , (n = 1, 2, 3 . . .)
L
Now we have infinitely many different F 0 s, so let’s denote them by Fn .
nπx
Fn = Bn sin
L
00 n 2 π 2 c2 nπct nπct
G =− 2
G ⇒ Gn = Kn cos + Ln sin
L L L

∂u(x,t) 
The IC ∂t  = 0 gives Ln = 0 so un can be written as
t=0

nπx nπct
un (x, t) = Bn Kn sin cos
L L
Without loss of generality, we can choose Kn = 1, because we do not need
two arbitrary constants. Using the superposition principle, we have to add
all the solutions to obtain the general solution:
∞ ∞
X X nπx nπct
u(x, t) = un (x, t) = Bn sin cos
n=1 n=1
L L
102 CHAPTER 12. PARTIAL DIFFERENTIAL EQUATIONS

Figure 12.3: The vibrating string

The only condition we did not use is the IC u(x, 0) = f (x). This gives
∞
X nπx
Bn sin = f (x)
n=1
L

Therefore Bn are the Fourier sine coefficients of f (x). So

2 L
Z
nπx
Bn = f (x) sin dx
L 0 L
2 L/2 2hx 2 L
Z Z  
nπx 2hx nπx
= sin dx + 2h − sin dx
L 0 L L L L/2 L L

Performing the integration, we find

8h nπ
Bn = sin
n2 π 2 2
So the solution is
 
8h πx πct 1 3πx 3πct
u(x, t) = 2 sin cos − 2 sin cos + ···
π L L 3 L L

The plot of the solution u(x, t) for selected times is given in Figure 12.3.

Example 12.2 Solve the PDE utt = c2 uxx , with

BC: u(0, t) = u(L, t) = 0
∂u(x, 0)
IC: u(x, 0) = 0, = g(x)
∂t

This question is very similar to the previous one, but this time initial
deflection is zero and the initial velocity is nonzero.
Following the same steps as we did, we obtain

nπx
Fn = Bn sin
L

n 2 π 2 c2
G00 = − G
L2
12.3. METHOD OF SEPARATION OF VARIABLES 103

nπct nπct
Gn = Kn cos + Ln sin
L L
The IC u(x, 0) = 0 gives Kn = 0 so un can be written as

nπx nπct
un (x, t) = Bn Ln sin
sin
L L
We choose Ln = 1 and superpose all the solutions to obtain

∞ ∞
X X nπx nπct
u(x, t) = un (x, t) = Bn sin sin
n=1 n=1
L L

The only condition we did not use is the IC

∂u(x, 0)
= g(x)
∂t
This gives
∞
X nπc nπx
Bn sin = g(x)
n=1
L L
nπc
Therefore Bn are the Fourier sine coefficients of g(x), so
L
Z L
2 nπx
Bn = g(x) sin dx
nπc 0 L
104 CHAPTER 12. PARTIAL DIFFERENTIAL EQUATIONS

Exercises

1) Solve the PDE utt = 4uxx on 0 < x < π, 0 < t, with

BC: u(0, t) = u(π, t) = 0
∂u(x, 0)
IC: u(x, 0) = sin(2x), =0
∂t
2) Solve the PDE utt = uxx on 0 < x < 1, 0 < t, with
BC: u(0, t) = u(1, t) = 0
∂u(x, 0)
IC: u(x, 0) = x(1 − x), =0
∂t
3) Solve the PDE utt = 91 uxx on 0 < x < 2, 0 < t, with
BC: u(0, t) = u(2, t) = 0
∂u(x, 0)
IC: u(x, 0) = 5 sin(πx) − 3 sin(2πx), =0
∂t
4) Solve the PDE utt = c2 uxx on 0 < x < L, 0 < t, with
BC: u(0, t) = u(L,
( t) = 0
hx
a
if 0<x<a ∂u(x, 0)
IC: u(x, 0) = h(L−x) , =0
L−a
if a<x<L ∂t

5) Solve the PDE utt = uxx on 0 < x < π, 0 < t, with

BC: u(0, t) = u(π, t) = 0
∂u(x, 0)
IC: u(x, 0) = 0, = x(π − x)
∂t
6) Solve the PDE utt = 12uxx on 0 < x < 3, 0 < t, with
BC: u(0, t) = u(3, t) = 0
∂u(x, 0)
IC: u(x, 0) = 0, = sin(πx)
∂t
7) Solve the PDE utt = uxx on 0 < x < π, 0 < t, with
BC: u(0, t) = u(π, t) = 0 (
∂u(x, 0) 0.1x if 0 < x < π/2
IC: u(x, 0) = 0, =
∂t 0.1(π − x) if π/2 < x < π
8) Solve the PDE utt = 4uxx on 0 < x < 5, 0 < t, with
BC: u(0, t) = u(5, t) = 0
∂u(x, 0)
IC: u(x, 0) = 0, =1
∂t
EXERCISES 105

Answers
1) u(x, t) = sin(2x) cos(4t)

∞
X 4
2) u(x, t) = [1 − (−1)n ] sin(nπx) cos(nπt)
n=1
n3 π 3
 
8 1
= 3 sin(πx) cos(πt) + sin(3πx) cos(3πt) + · · ·
π 27
   
πt 2πt
3) u(x, t) = 5 sin(πx) cos − 3 sin(2πx) cos
3 3
∞
2hL2  nπa   nπx   
X nπct
4) u(x, t) = sin sin cos
n=1
n2 π 2 a(L − a) L L L
∞
X 4
5) u(x, t) = 4π
[1 − (−1)n ] sin(nx) sin(nt)
n=1
n
 
8 1
= sin(πx) sin(πt) + sin(3πx) sin(3πt) + · · ·
π 81

1 √
6) u(x, t) = √ sin(πx) sin(2π 3t)
2π 3
∞
X 0.4  nπ 
7) u(x, t) = 3π
sin sin(nx) sin(nt)
n=1
n 2
∞  
X 5 n
 nπx  2nπt
8) u(x, t) = [1 − (−1) ] sin sin
n=1
n2 π 2 5 5
106 CHAPTER 12. PARTIAL DIFFERENTIAL EQUATIONS
Chapter 13

Heat Equation

In this chapter, we will set up and solve heat equation. Although it is very
similar to wave equation in form, the solutions will be quite different. We will
generalize our methods to nonzero boundary conditions and two-dimensional
problems.

13.1 Modeling Heat Flow

Figure 13.1: Heat Flow in One Dimension

Consider a long thin bar of length L on x-axis. It has uniform density

and cross section. The lateral surface is perfectly isolated, so the heat flow
is in x-direction only. Experiments show that the amount of heat flow is
proportional to the temperature gradient:

dQ du
= −KS (13.1)
dt dx

where Q is the heat, u is the temperature, S is the cross sectional area and
K is the thermal conductivity. The minus sign means that heat flows from
higher to lower temperatures as we expect. A piece of the material of length
∆x has two neighbours, so the change in its temperature is determined by
the net difference of heat flows:

107
108 CHAPTER 13. HEAT EQUATION

    
∂u  ∂u 
∆Q = −KS − −KS ∆t
∂x x ∂x x+∆x
    (13.2)
∂u  ∂u 
= − KS∆t
∂x x+∆x ∂x x

We know that when a material receives heat, its temperature rises propor-
tionally:
∆Q = mµ∆u

(13.3)
= S∆xρµ u|t+∆t − u|t

where µ is the specific heat and ρ is the density of the material. If we set
these two ∆Q values equal to each other, and rearrange, we will obtain
 
∂u  ∂u 
−
∂x x+∆x ∂x x u| − u|t
KS = Sρµ t+∆t (13.4)
∆x ∆t

In the limit ∆x → 0 and ∆t → 0 we will obtain second and first partial

derivatives of u(x, t), so
K ∂ 2u ∂u
2
= (13.5)
ρµ ∂x ∂t
If we define the diffusivity as k = K/(ρµ)

ut = k uxx (13.6)

This is the heat equation in one dimension. Its form is remarkably similar
to wave equation, yet the solutions are different. This time, we will have
only one Initial Condition u(x, 0) = f (x) which is the initial temperature
distribution of the bar.
If the ends of the bar are kept at fixed temperatures, we have Boundary
Conditions u(0, t) = T1 , u(L, t) = T2 where L is the length of the bar.
If the ends of the bar are isolated, the BC will be ux (0, t) = ux (L, t) = 0
A similar analysis shows that, in 2-dimensions, the heat equation is:

ut = k(uxx + uyy ) (13.7)

13.2. HOMOGENEOUS BOUNDARY CONDITIONS 109

13.2 Homogeneous Boundary Conditions

Example 13.1 Solve the one dimensional heat equation ut = kuxx on a bar
of length L with:
BC: u(0, t) = u(L, t) =
0
 x L
if 0 < x <

IC: u(x, 0) = f (x) = 2
L
 L − x if
 <x<L
2

Using separation of variables, we may write u(x, t) as

u(x, t) = F (x)G(t)

Then F G0 = kF 00 G or
G0 F 00
=
kG F
This is possible only if both sides are equal to a constant. Therefore

G0 F 00
= =c
kG F
Once again we have three cases. If c > 0, or c = 0, the solution is trivial.
(Please verify!) Therefore

c < 0, c = −p2 , ⇒ F = A cos px + B sin px

Using the BC we find A = 0 and

nπ
p= , (n = 1, 2, 3 . . .)
L
So
nπx
Fn = Bn sin
L
n2 π 2 k
G0 = − G
L2
therefore
n2 π 2 k
Gn = e−λn t where λn =
L2
nπx −λn t
un (x, t) = Bn sin e
L
110 CHAPTER 13. HEAT EQUATION

and because of the superposition principle

∞ ∞
X X nπx −λn t
u(x, t) = un (x, t) = Bn sin e
n=1 n=1
L

Bn can be determined as the Fourier sine coefficients of f (x). So

2 L
Z
nπx
Bn = f (x) sin dx
L 0 L
Performing the integration, (Please verify) we find
4L nπ
Bn = 2 2
sin
nπ 2
So the solution is
 
4L πx −λ1 t 1 3πx −λ3 t
u(x, t) = 2 sin e − 2 sin e + ···
π L 3 L
Example 13.2 Solve the PDE ut = kuxx with:
BC: ux (0, t) = ux (L, t) = 0
πx
IC: u(x, 0) = cos
L
This is a bar with insulated ends. The solution is exactly the same as
before up to the step

F = A cos px + B sin px ⇒ F 0 = −Ap sin(px) + Bp cos(px)

nπ
Using the BC we find B = 0, Ap sin(pL) = 0 ⇒ p= L

n2 π 2 kt
 
nπx
Fn = An cos ⇒ Gn (x, t) = exp −
L L2
∞ ∞
n2 π 2 kt
 
X X nπx
u(x, t) = un (x, t) = An cos exp −
n=1 n=1
L L2
Using the IC we see that
Z L
2 πx nπx
An = cos cos dx
L 0 L L
Using the orthogonality of trigonometric functions, we see that A1 = 1 and
all others are zero, so
π 2 kt
 
πx
u(x, t) = cos exp − 2
L L
13.3. NONZERO BOUNDARY CONDITIONS 111

13.3 Nonzero Boundary Conditions

Steady State Solution: The temperature distribution we get as t → ∞
must be time independent. So we call it steady state solution.
∂u d2 u
We expect = 0 which means = 0 therefore the steady state
∂t dx2
solution must be
u(x) = Ax + B

Example 13.3 Solve the steady state heat equation ut = kuxx on 0 < x < L
with BC: u(0) = T1 , u(L) = T2

We know that u(x) = Ax + B so

B = T1 , AL + T1 = T2

T2 − T1
u(x) = x + T1
L
Example 13.4 Solve the heat equation

∂u ∂ 2u
= , 0 < x < π, t > 0
∂t ∂x2
with BC: u(0, t) = 0, u(π, t) = 40, t > 0
and IC: u(x, 0) = 40, 0 < x < π

First, we will find the steady state solution u1 . Obviously,

40
u1 (x) = x
π
Now we will express the solution u as a combination of two functions u1 , u2 .
Here, u1 is the steady state solution, and u2 is the answer to a homogeneous
BC problem:

u(x, t) = u1 (x) + u2 (x, t)

Let’s obtain the BC and IC for u2

112 CHAPTER 13. HEAT EQUATION

BC: u2 (0, t) = 0, u2 (π, t) = 0, t > 0

 x
IC: u2 (x, 0) = 40 1 − , 0<x<π
π
This is a new problem with homogeneous BC, so we can solve it as before.

u2 (x, t) = F (x)G(t)
After similar steps,
Fn = Bn sin nx
and
2t
Gn = e−n

∞
2t
X
u2 (x, t) = Bn sin nx e−n
n=1
If we insert t = 0, we see that
∞
X  x
u2 (x, 0) = Bn sin nx = 40 1 −
n=1
π
So, we can obtain Bn as the Fourier sine coefficients of the right hand side.
2 π 
Z
x
Bn = 40 1 − sin nx dx
π 0 π
  π
2 40 cos nx 40 x cos nx sin nx 
Bn = − − − +
π n π n n2 
0
   
2 1 cos nπ 40  π cos nπ 
= 40 − − − +0
π n n π n
 
2 40 n 40 n
= (1 − (−1) ) + (−1)
π n π
80
=
nπ
∞
80 X sin nx −n2 t
u2 (x, t) = e
π n=1 n
Therefore the solution is
∞
40 80 X sin nx −n2 t
u(x, t) = x+ e
π π n=1 n
13.4. TWO DIMENSIONAL PROBLEMS 113

13.4 Two Dimensional Problems

We can generalize these methods to higher dimensions. Consider the tem-
perature distribution on a rectangular plate of dimensions 2 × 3.
Example 13.5 Solve the PDE ut = k(uxx + uyy ) where u = u(x, y, t) with:
BC: u(0, y, t) = u(2, y, t) = 0
u(x, 0, t) = u(x, 3, t) = 0
IC: u(x, y, 0) = x(4 − x2 )y(9 − y 2 )
This time we will apply the method of separation of variables to a three-
variable function u(x, y, t), therefore

u(x, y, t) = F (x)G(y)H(t)
After the usual steps, we obtain

  2 2
m2 π 2
 
nπx mπy nπ
Fn (x) = sin , Gm = sin , Hnm = Anm exp − + kt
2 3 4 9
Therefore

∞ X
∞   2 2
m2 π 2
 
X nπx mπy nπ
u(x, y, t) = Anm sin sin exp − + kt
n=1 m=1
2 3 4 9
Using the initial condition

∞ X
∞
2 2
X nπx mπy
u(x, y, 0) = x(4 − x )y(9 − y ) = Anm sin sin
n=1 m=1
2 3
Z 2  Z 3 
2 2 nπx 2 2 mπy
Anm = x(4 − x ) sin dx y(9 − y ) sin dx
2 0 2 3 0 3
96(−1)n+1 324(−1)m+1
  
=
n3 π 3 m3 π 3
∞ ∞
31104 X X (−1)n+m
  2 2
m2 π 2
 
nπx mπy nπ
u(x, y, t) = sin sin exp − + kt
π 6 n=1 m=1 n3 m3 2 3 4 9
The results are plotted on Figure 13.2 for three different t values. We can
easily see that u → 0 as time increases.
114 CHAPTER 13. HEAT EQUATION

Figure 13.2: The Temperature on a Rectangular Plate

Exercises

1) Solve the PDE ut = uxx on 0 < x < π, 0 < t, with

BC: u(0, t) = u(π, t) = 0, IC: u(x, 0) = sin 2x

2) Solve the PDE ut = 5uxx on 0 < x < 4, 0 < t, with

πx
BC: u(0, t) = u(4, t) = 0, IC: u(x, 0) = sin − sin πx
2
3) Solve the PDE ut = kuxx on 0 < x < L, 0 < t, with
BC: u(0, t) = u(L, t) = 0, IC: u(x, 0) = x(L − x)

4) Solve the PDE ut = uxx on 0 < x < π, 0 < t, with

BC: ux (0, t) = ux (π, t) = 0, IC: u(x, 0) = x

5) Solve the PDE ut = 3uxx on 0 < x < 10, 0 < t, with

BC: ux (0, t) = ux (10, t) = 0, IC: u(x, 0) = cos 0.3πx

6) Solve the PDE ut = kuxx on 0 < x < L, 0 < t, with

x
BC: ux (0, t) = ux (L, t) = 0, IC: u(x, 0) = 1 −
L
7) Solve the PDE ut = uxx with nonhomogeneous boundary conditions
BC : u(0, t) = 1, u(1, t) = 0, IC: u(x, 0) = sin(πx)

8) Solve the PDE ut = kuxx with nonhomogeneous ( boundary conditions

0 if 0 < x < L2
BC : u(0, t) = 0, u(L, t) = T , IC: u(x, 0) =
T if L2 < x < L

9) Solve the PDE ut = 8 (uxx + uyy ) on 0 < x < 2, 0 < y < 5, 0 < t, with
BC: u(0, y, t) = u(2, y, t) = 0, u(x, 0, t) = u(x, 5, t) = 0
πx πy
IC: u(x, y, 0) = sin sin
2 5
10) Solve the PDE ut = k (uxx + uyy ) on 0 < x < a, 0 < y < b, 0 < t, with
BC: u(0, y, t) = u(a, y, t) = 0, u(x, 0, t) = u(x, b, t) = 0
IC: u(x, y, 0) = T
EXERCISES 115

Answers

1) u(x, t) = sin 2x e−4t

πx − 5 π2 t 2
2) u(x, t) = sin e 4 − sin(πx) e−5π t
2
∞
4L2 n2 π 2 kt
 
X
n nπx
3) u(x, t) = 3π3
[1 − (−1) ] sin exp −
n=1
n L L2
∞
π X 2 2
4) u(x, t) = + 2
[(−1)n − 1] cos nx e−n t
2 n=1 n π

2t
5) u(x, t) = cos(0.3πx) e−0.27π

∞
n2 π 2 kt
 
1 X 2 n nπx
6) u(x, t) = + [1 − (−1) ] cos exp −
2 n=1 n2 π 2 L L2
∞
−π 2 t 2 X sin nπx −n2 π2 t
7) u(x, t) = 1 − x + e sin πx − e
π n=1 n
∞
T x X 2T nπ nπx −n2 π2 kt/L2
8) u(x, t) = + cos sin e
L n=1
nπ 2 L
 
Tx 2T 1 2πx −4π2 kt/L2 1 4πx −16π2 kt/L2
= − sin e − sin e + ···
L π 2 L 4 L

πx πy −2.32π2 t
9) u(x, y, t) = sin sin e
2 5
∞ ∞
4T X X nπx mπy −kπ2 na22 + mb22 t
 
10) u(x, y, t) = 2 Anm sin sin e
π n=1 m=1 a b

(1 − (−1)n ) (1 − (−1)m )
Where Anm =
nm
116 CHAPTER 13. HEAT EQUATION
Chapter 14

Laplace Equation

Laplace equation is the last PDE we will consider. It is different from the
wave and heat equations in that, time is not a variable. We can also think
of Laplace equation as the equilibrium configuration of heat and wave equa-
tions. It is possible to express these equations in any coordinate system that
suits the geometry of the problem. As an example, we will consider polar
coordinates in this chapter.

14.1 Rectangular Coordinates

Laplace equation in two dimensions is

uxx + uyy = 0 (14.1)

where u = u(x, y). The potential function for gravitational force in free space
satisfies Laplace equation. Similarly, the electrostatic potential also satisfies
the same equation. Therefore Laplace equation is sometimes called Potential
Equation.
There are no time derivatives in Laplace Equation, therefore there are no
initial conditions. We just have the boundary conditions. If the values of u
are given on the boundary, the problem is called a Dirichlet problem, if the
values of the normal derivative are given on boundary, it is called a Neumann
problem. It is also possible to set up mixed problems. In this book, we will
only consider Dirichlet problems.

117
118 CHAPTER 14. LAPLACE EQUATION

Figure 14.1: Laplace Equation on a rectangle

Let’s consider a Dirichlet problem on the rectangle shown in Figure 14.1.

uxx + uyy = 0 on 0 < x < a, 0 < y < b (14.2)

with BC:

u(0, y) = 0, u(a, y) = 0, u(x, 0) = 0, u(x, b) = f (x) (14.3)

Using the method of separation of variables, we start with the assumption

u(x, y) = F (x)G(y) and inserting in equation, we obtain

F 00 G00
=− =k (14.4)
F G
Depending on the sign of k, we have three different cases:
Case 1) k = 0, u = (Ax + B)(Cy + D),
Case 2) k > 0, k = p2 , u = (Aepx + Be−px )(C cos py + D sin py),
Case 3) k < 0, k = −p2 , u = (A cos px + B sin px)(Cepy + De−py ),
Using the BC x = 0 ⇒ u = 0 and x = a ⇒ u = 0 we can easily see that
the first two cases give trivial solutions. Using the same conditions on the
third case, we obtain A = 0, p = nπ a
as we did in the previous chapters.

nπx nπy nπy

un (x, y) = Bn sin (Ce a + De− a ) (14.5)
a
The third BC y = 0 ⇒ u = 0 gives

C + D = 0 ⇒ D = −C (14.6)

Remember the hyperbolic sine function, which is defined as

ey − e−y
sinh y = (14.7)
2
Now we can express the solution in terms of trigonometric and hyperbolic
functions as:

nπx nπy
un (x, y) = Bn sin sinh (14.8)
a a
14.1. RECTANGULAR COORDINATES 119

Superposition of these solutions give

∞
X nπx nπy
u(x, y) = Bn sin sinh (14.9)
n=1
a a
We have only the fourth boundary condition left: y = b ⇒ u = f (x)
∞
X nπx nπb
u(x, b) = Bn sin sinh = f (x) (14.10)
n=1
a a

Obviously, Bn sinh nπb

a
are the Fourier sine coefficients of f (x), so
Z a
nπb 2 nπx
Bn sinh = f (x) sin dx (14.11)
a a 0 a
Remark: If two sides have nonzero BC, we can consider them as two sep-
arate problems having zero BC on 3 sides, find the solutions and then add
them to obtain the result, as you can see on Figure 14.2.

Figure 14.2: Nonzero Boundary Conditions on two sides

120 CHAPTER 14. LAPLACE EQUATION

Example 14.1 Solve uxx + uyy = 0 on 0 < x < 2, 0 < y < 1, with
BC: u(0, y) = 0, u(2, y) = 0, u(x, 0) = 0, u(x, 1) = 1

Using the steps above, we find

∞
X nπx nπy
u(x, y) = Bn sin sinh
n=1
2 2
where
Z 2
nπ nπx
Bn sinh = dx sin
2 0 2
2[1 − (−1)n ]
Bn =
nπ sinh nπ
2
∞
X 2[1 − (−1)n ] nπx nπy
u(x, y) = nπ sin sinh
n=1
nπ sinh 2
2 2
You can see the solution on Figure 14.3 (up).

Example 14.2 Solve uxx + uyy = 0 on 0 < x < 1, 0 < y < 1, with
BC: u(x, 0) = 0, u(x, 1) = 0, u(0, y) = 0, u(1, y) = 3y(1 − y)

The solution satisfying the first three boundary conditions is:

∞
X
u(x, y) = cn sinh(nπx) sin(nπy)
n=1
Inserting x = 1 and using the fourth boundary condition, we obtain
Z 1
sinh(nπ) cn = 2 3y(1 − y) sin(nπy) dy
0

1
y cos nπy sin nπy y 2 cos nπy

2y sin nπy 2 cos nπy 
sinh(nπ) cn = 6 − + 2 2 + − −
nπ nπ nπ n2 π 2 n3 π 3 
0

12[1 − (−1)n ]
cn =
n3 π 3 sinh(nπ)
∞
12 X [1 − (−1)n ]
u(x, y) = 3 sinh(nπx) sin(nπy)
π n=1 n3 sinh(nπ)
Figure 14.3 (down) gives the plot.
14.1. RECTANGULAR COORDINATES 121

Figure 14.3: Solution of the Dirichlet Problem

122 CHAPTER 14. LAPLACE EQUATION

Figure 14.4: Polar Coordinates

14.2 Polar Coordinates

If the region of interest is circular, we have to express the Laplace Equation
in polar coordinates to be able to use the boundary conditions.
We will start with x = r cos θ, y = r sin θ and use chain rule to express
the derivatives of u with respect to r and θ.

∂u ∂u ∂r ∂u ∂θ
= + (14.12)
∂x ∂r ∂x ∂θ ∂x

r 2 = x2 + y 2 (14.13)

∂r ∂r x
2r = 2x ⇒ = (14.14)
∂x ∂x r
If you complete this derivation, (which is a nice exercise in calculus) you will
obtain the Laplace equation in polar coordinates:

ur uθθ
uxx + uyy = urr + + 2 =0 (14.15)
r r
To solve the Laplace equation inside a circle of radius a together with the
boundary condition u(a, θ) = f (θ), we start the method of separation of
variables with the assumption u(r, θ) = F (r)G(θ).
Inserting this in (14.15) we obtain

F 0 G F G00
F 00 G + + 2 =0 (14.16)
r r
r2 F 00 rF 0 G00
+ =− =k (14.17)
F F G
where k is the separation constant. Once again we have three possibilities:
Case 1) k = 0, u = (A ln r + B)(Cθ + D),
Case 2) k > 0, k = p2 , u = (Arp + Br−p )(C cos pθ + D sin pθ),
Case 3) k < 0, k = −p2 , u = [A cos(p ln r) + B sin(p ln r)](Cepθ + De−pθ )
We expect the solution to be periodic in θ with period 2π. Case 3 does
not satisfy this, so we eliminate this case.
14.2. POLAR COORDINATES 123

In Case 1, we have to choose C = 0 for periodicity. Besides, ln r is

undefined at r = 0. So A = 0. Therefore the contribution of Case 1 is only
a constant.
In Case 2, r−p is undefined at r = 0, so we choose B = 0. The resulting
separated solution is:

un (r, θ) = rn (Cn cos nθ + Dn sin nθ) (14.18)

Note that n must be an integer for periodicity.
After superposition, we obtain the general solution as
∞
X
u(r, θ) = C0 + rn (Cn cos nθ + Dn sin nθ) (14.19)
n=1

The boundary condition is: u(a, θ) = f (θ), we can find Cn and Dn using the
Fourier expansion of f .
Z π
1
C0 = f (θ) dθ
2π −π
Z π
1
Cn = n f (θ) cos nθ dθ (14.20)
a π −π
Z π
1
Dn = n f (θ) sin nθ dθ
a π −π
Remark: If the region is outside the circle, the same ideas apply. We have
to eliminate ln r because it is not finite at infinity. The only difference is that
we should have the negative powers of r, because they will be bounded as
r → ∞. So
∞
X
u(r, θ) = C0 + r−n (Cn cos nθ + Dn sin nθ) (14.21)
n=1

Remark: If we have a region between two circles as a < r < b, we need

both the positive and negative powers of r as well as the logarithmic term.
124 CHAPTER 14. LAPLACE EQUATION

Example 14.3 ( Solve Laplace equation in the region 0 6 r < 5, with

−1 if −π < θ < 0
BC: u(5, θ) =
1 if 0 < θ < π
We know that the general solution in this case is
∞
X
u(r, θ) = C0 + rn (Cn cos nθ + Dn sin nθ)
n=1
The boundary condition gives
∞
X
u(5, θ) = C0 + 5n (Cn cos nθ + Dn sin nθ) = f (θ)
n=1

The Fourier coefficients of f are

2
C0 = 0, Cn = 0, Dn = [1 − (−1)n ]
nπ5n
∞
2X  r n sin nθ
u(r, θ) = [1 − (−1)n ]
π n=1 5 n
The solution is plotted on Figure 14.5 (up).

Example 14.4 Solve Laplace equation in the region 0 6 r < 2, with

BC: u(2, θ) = sin(3θ)

Inserting r = 2 in the solution

∞
X
u(r, θ) = C0 + rn (Cn cos nθ + Dn sin nθ)
n=1
we obtain
∞
X
u(2, θ) = C0 + 2n (Cn cos nθ + Dn sin nθ) = sin 3θ
n=1

We can easily see that the only nonzero Fourier coefficient is D3

1
23 D3 = 1 ⇒ D3 =
8

1 3
u(r, θ) = r sin 3θ
8
The solution is plotted on Figure 14.5 (down).
14.2. POLAR COORDINATES 125

Figure 14.5: Potential on a Circle

126 CHAPTER 14. LAPLACE EQUATION

Example 14.5 Solve Laplace equation in the region 3 6 r, with

BC: u(3, θ) = cos2 θ
This time the region is outside the circle so the general solution is
∞
X
u(r, θ) = C0 + r−n (Cn cos nθ + Dn sin nθ)
n=1
The boundary condition gives
X∞
u(3, θ) = C0 + 3−n (Cn cos nθ + Dn sin nθ) = cos2 θ
n=1

1 + cos 2θ
We know that cos2 θ = , so
2
 
1 9
u(r, θ) = 1 + 2 cos 2θ
2 r
Example 14.6 Solve Laplace equation in the region 1 6 r 6 2, with
BC: u(1, θ) = 5 sin 3θ, u(2, θ) = 3 ln 2 + 40 sin 3θ
The region is between two circles, so the general solution is

X∞ X∞
u(r, θ) = A0 +B0 ln r+ n
r (An cos nθ+Bn sin nθ)+ r−n (Cn cos nθ+Dn sin nθ)
n=1 n=1

We can directly see that all the coefficients except A0 , B0 , B3 , D3 must be

zero, therefore
sin 3θ
u(r, θ) = A0 + B0 ln r + B3 sin 3θ r3 + D3 3
r
Using the boundary conditions at r = 1 and r = 2, we obtain
A0 = 0, B0 = 3, B3 = 5, D3 = 0, so

u(r, θ) = 3 log r + 5r3 sin 3θ

Remark: We will state without proof that if u satisfies Laplace equation in
a region, then its value at any point is equal to the average values around
any circle (within that region).
Using this principle, we can easily derive the result that maximum and
minimum values of u must occur on the boundary.
The given solution plots illustrate these principles.
EXERCISES 127

Exercises
1) Solve the PDE uxx + uyy = 0, on 0 < x < 2, 0 < y < 2, with
3πy
BC: u(x, 0) = 0, u(x, 2) = 0, u(0, y) = 0, u(2, y) = sin
2

2) Solve the PDE uxx + uyy = 0, on 0 < x < 5, 0 < y < 1, with
BC: u(x, 0) = sin πx, u(x, 1) = 0, u(0, y) = 0, u(5, y) = 0

3) Solve the PDE uxx + uyy = 0, on 0 < x < 2, 0 < y < 8, with
(
1 if 0<y<4
BC: u(x, 0) = 0, u(x, 8) = 0, u(0, y) = 0, u(2, y) =
−1 if 4<y<8

4) Solve the PDE uxx + uyy = 0, on 0 < x < 2, 0 < y < 2, with
πx πy
BC: u(x, 0) = 0, u(x, 2) = sin , u(0, y) = 0, u(2, y) = sin
2 2

5) Solve the PDE uxx + uyy = 0, on 0 < x < 3, 0 < y < 2, with
5πy 7πy
BC: u(x, 0) = 0, u(x, 2) = 0, u(0, y) = sin , u(3, y) = sin
2 2
ur uθθ
6) Solve the PDE urr + + 2 = 0 on 0 6 r < 1, with
r r
BC: u(1, θ) = cos 4θ

ur uθθ
7) Solve the PDE urr + + 2 = 0 on 0 6 r < 4, with
r r
BC: u(4, θ) = 2 sin 2θ − 7 cos 3θ

ur uθθ
8) Solve the PDE urr + + 2 = 0 on 3 < r, with
r r
BC: u(3, θ) = 5 − 5 cos 3θ

ur uθθ
9) Solve the PDE urr + + 2 = 0 on 3 < r < 5, with
r r
BC: u(3, θ) = 4, u(5, θ) = 12

ur uθθ
10) Solve the PDE urr + + 2 = 0 on 2 < r < 3, with
r r
BC: u(2, θ) = −5 sin 2θ, u(3, θ) = 10 cos 2θ
128 CHAPTER 14. LAPLACE EQUATION

Answers

1 3πx 3πy
1) u(x, y) = sinh sin
sinh 3π 2 2

1
2) u(x, y) = sin πx sinh π(1 − y)
sinh π
∞
2 X 1 + (−1)n − 2 cos nπ
2 nπx nπy
3) u(x, y) = nπ sinh sin
π n=1 n sinh 4 8 8

1  πx πy πy πx 
4) u(x, y) = sin sinh + sin sinh
sinh π 2 2 2 2

1 7πy 7πx 1 5πy 5π(x − 3)

5) u(x, y) = 21π sin sinh − 15π sin sinh
sinh 2 2 2 sinh 2 2 2

6) u(r, θ) = r4 cos 4θ

 r 2  r 3
7) u(r, θ) = 2 sin 2θ − 7 cos 3θ
4 4
 3
3
8) u(r, θ) = 5 − 5 cos 3θ
r

4 ln 5 − 12 ln 3 + 8 ln r
9) u(r, θ) =
ln 5 − ln 3
   
9 2 32 4 2 81
10) u(r, θ) = 2r − 2 cos 2θ + r − 2 sin 2θ
13 r 13 r
To the Student

If you have reached this point after solving all (or most) of the exercises,
you must have covered a lot of ground. But there’s no end to differential
equations. This was just a brief introduction. For further study, you may
consult the books listed in the references.
[6, 8] and [9] are big and useful books that contain all topics covered here
and many other ones besides.
For ordinary differential equations, [2, 11, 12, 14] give a complete treat-
ment with a large number of exercises.
For partial differential equations, [1] and [7] are good introductory books
that illustrate main ideas.
Detailed information on Fourier Series can be found on [3].
There are many aspects of differential equations that we did not even
touch in this book.
For a history of this subject, you may consult [13].
For nonlinear equations and dynamical systems, which is a vast subject
requiring another book even for the introduction, [10] and [15] will be a good
starting point.
For numerical methods, you may read the relevant chapters of [4] and [5].

129
References

[1] Asmar, N.H. Partial Differential Equations and Boundary Value Prob-
lems. Prentice Hall, 2000.

[2] Boyce, W.E. and DiPrima, R.C. Elementary Differential Equations and
Boundary Value Problems, 6th edition. Wiley, 1997.

[3] Churchill, R.V. and Brown, J.W. Fourier Series and Boundary Value
Problems, 6th edition. McGraw–Hill, 2000.

[4] Fausett, L.V. Numerical Methods: Algorithms and Applications. Pren-

tice Hall, 2003.

[5] Gerald, C.F. and Wheatley, P.O. Applied Numerical Analysis, 7th edi-
tion. Prentice Hall, 2004.

[6] Greenberg, M.D. Advanced Engineering Mathematics, 2nd edition. Pren-

tice Hall, 1998.

[7] Keane, M.K. A Very Applied First Course in Partial Differential Equa-
tions. Prentice Hall, 2002.

[8] Kreyszig, E. Advanced Engineering Mathematics, 8th edition. Wiley,

1998.

[9] O’Neil, P.V. Advanced Engineering Mathematics, 5th edition. Thomson,

2003.

[10] Perko, L. Differential Equations and Dynamical Systems, 3rd edition.

Springer, 2001.

131
132 REFERENCES

[11] Rainville, E.D., Bedient, P.E. and Bedient, R.E. Elementary Differential
Equations, 8th edition. Prentice Hall, 1997.

[12] Ross, S.L. Introduction to Ordinary Differential Equations, 4th edition.

Wiley, 1989.

[13] Simmons, G.F. Differential Equations with Applications and Historical

Notes, 2nd edition. McGraw–Hill, 1991.

[14] Trench, W.F. Elementary Differential Equations with Boundary Value

Problems. Brooks/Cole, 2001.

[15] Williamson, R.E. Introduction to Differential Equations and Dynamical

Systems, 2nd edition. McGraw–Hill, 2000.
Index

Analytic function, 42 Fourier series, 77–91

Auxiliary equation, 22 coefficients, 79
complex form of, 90
Basis of solutions, 18
convergence, 81
Bernoulli equation, 14
half range extensions, 87
Boundary condition, 98
Frobenius’ method, 49–56
Boundary value problem, 117
General solution, 2, 18, 25, 33
Cauchy-Euler equation, 22
Characteristic equation, 20 Half-range cosine expansion, 88
Chebyshev equation, 46 Half-range sine expansion, 89
Closed form, 45 Heat equation, 107–116
Complex conjugate roots, 21, 22 nonzero boundary conditions, 111
Complex exponentials, 21 two dimensional problems, 113
Complex Fourier series, 90 Heaviside step function, 69
Constant coefficient equations, 20 Hermite equation, 46
Convergence of Fourier series, 81 Higher order equations, 33–40
Convolution, 67 Homogeneous differential equations, 17,
35
Differential operator, 34
Discontinuous input, 73, 74 Implicit solution, 2
Dirichlet problem, 117, 120 Indicial equation, 49
Initial condition, 98
Euler’s formula, 21 Initial value problems, 61
Even function, 87 Integrating factor, 11
Exact equation, 9
Laguerre equation, 46
Explicit solution, 2
Laplace equation, 117–128
Exponential order, 58
rectangular coordinates, 117
Fourier integral, 92 polar coordinates, 122

133
134 INDEX

Laplace transform, 57–76 Second order homogeneous equations,

convolution, 67 17–24
definition, 57 Second order nonhomogeneous equa-
existence, 59 tions, 25–32
initial value problems, 61 Separable equations, 3
table of, 64 Separation of variables, 100
Legendre equation, 46 Series solutions, 43, 49
Linear differential equations, 17, 33 Shifting, 59
Linear first order equations, 13 Singular point, 43
Linear independence, 18, 33 Steady state solution, 111
Substitution, 5
Modeling, 3, 99, 107 Superposition, 18, 98

Neumann problem, 117 Table of Laplace transforms, 64

Nonhomogeneous differential equations,Taylor series, 42
25, 37 Transformations, 5

Undetermined coefficients, 27, 37

Odd functions, 87
Unit step function, 69
Order, 2
Ordinary point, 43 Variation of parameters, 29, 37
Vibrating string, 99, 103
Parseval’s identity, 83
Partial differential equations, 97 Wave equation, 99–106
Partial fractions, 71 Wronskian, 34
Particular solution, 2, 25
Piecewise continuous functions, 58
Polar coordinates, 122
Potential equation, 117
Power series, 41, 42
Power series method, 43

Rectangular coordinates, 117

Recursion relation, 44
Reduction of order, 19
Regular singular point, 43
Corrections of
Lecture Notes on Differential Equations
by Emre Sermutlu

Page Question Error Correction

23 14 y(π) = 0, y(−π) = 0 y(0) = 0, y 0 (0) = 1

85 1 2π(sin x − · · · ) 2(sin x − · · · )

∞ ∞
X sin nx X sin nx
91 11.2 Result (−1)n+1 2 (−1)n+1
n=1
n n=1
n