SOLVING DIFFERENTIAL EQUATIONS

Amir Asif

Department of Computer Science and Engineering

York University, Toronto, ON M3J1P3

ABSTRACT
This article reviews a direct method for solving linear, constant-coefficient differential equations without using
transforms. The method that we use, offers a general characterization of solutions that provide insight into the
behaviour of the modeled system.

1. INTRODUCTION

It is convenient to express the output y(t) of the continuous-time (CT) system described by a differential equation
as the sum of two components:
Homogeneous component yh (t) associated with the initial conditions.
Particular component yp (t) associated with the input applied to the system.
The homogeneous component is the output of the system when the input is set to zero. The homogeneous compo-
nent, therefore, describes the manner in which the system dissipates any energy or memory of the past. The past
information is specified as initial conditions to the differential equation.
The particular component is the output of the system with the initial conditions set to zero. It describes the
system behaviour that is forced by the input.

2. HOMOGENEOUS COMPONENT

Assume that the differential equation modeling the CT system is given by

N
X M
X
dk y(t) dk x(t)
ak = bk . (1)
dtk dtk
k=0 k=0

with the following initial conditions at time t = t0 known

dy(t) dk1 y(t)

y(t), , ... , . (2)
dt dtk1
In most cases, t0 equals 0. The highest derivative of y(t) gives the order of the differential equation. In (1), the
order is N .
The homogeneous component yh (t) is the output of the system when the input is zero. Hence, yh (t) is the solution
to the homogeneous differential equation
N
X dk y(t)
ak = 0. (3)
dtk
k=0

In order to determine yh (t), we hypothesize a solution of the form

yh (t) = Aest , (4)

substitute in the differential equation, (1), and solve the resulting equation. We illustrate the procedure for
calculating the homogeneous solution by considering an example.
Example 1: Solving for the homogeneous solution. Consider a CT system modeled by the following
differential equation

d2 y(t) dy
2
+ 3 + 2y(t) = 2x(t).
dt dt
Substituting yh (t) = Aest into the homogeneous equation,

d2 y(t) dy
+3 + 2y(t) = 0
dt2 dt
gives

Aest (s2 + 3s + 2) = 0.

Ignoring the trivial solution, i.e., assuming Aest 6= 0, the above equality condition gives the following quadratic
equation, called the characteristic equation, in s,

s2 + 3s + 2 = 0

which has two roots at s = 1 and 2.

The homogeneous solution is, therefore, given by

yh (t) = A1 et + A2 e2t

where A1 and A2 are constants determined from the initial conditions, as will be explained later.
Repeated roots: The form of the homogeneous solution changes slightly when the characteristic equation has
repeated roots. If a root s = a is repeated J times, then we include J distinct terms in the homogeneous solution
associated with a, i.e., the homogeneous solution involves the following J functions

eat , teat , t2 eat , . . . , tJ1 eat (5)

and is given by

yh (t) = A1 eat + A2 teat + A3 t2 eat + . . . + AJ tJ1 eat . (6)

The procedure for calculating the homogeneous solution for differential equations with repeated roots is illustrated
in example 2.
Example 2: Characteristic equation with repeated roots. Consider the differential equation

d3 y(t) d2 y(t) dy
3
+4 2
+5 + 2y(t) = x(t).
dt dt dt
Its characteristic equation is given by

s3 + 4s2 + 5s + 2 = 0

which has two repeated roots at s = 1 and one root at s = 2. The homogeneous solution is given by

yh (t) = A1 et + A2 tet + A3 e2t (7)

where A1 , A2 and A3 are constants.

Complex roots: Solving a characteristic equation in general may give rise to complex roots of the form,

s =a+b
b. (8)
 Input Particular Solution
Constant A C
Exponential Aebt Cebt
Sinusoidal A cos(o t + ) C1 cos(o t) + C2 sin(o t)

Table 1: Particular solution corresponding to several common inputs.

For differential equations with real coefficients, it can be proven that the roots occurs in conjugate pairs, i.e., if
s = a+b b is a root then s = a b
b is also a root to the same characteristic equation. For such complex roots, the
homogeneous solution can be modified to the following form

yh (t) = A1 eat cos(bt) + A2 eat sin(bt). (9)

Example 3: Characteristic equation with complex roots. Assume a fourth order differential equation of
the form
d4 (t) d2 (t)
4
2 2 +1 = 0 (10)
dt dt
which is represented by the following characteristic equation

s4 2s2 + 1 = 0. (11)

The roots of the characteristic equation are given by s = b

,b
, b
, b
. Note that the roots are not only complex
but repeated. The homogenous solution is therefore given by

yh (t) = A1 cos(t) + A2 sin(t) + A3 t cos(t) + A4 t sin(t) (12)

where A1 , A2 , A3 , and A4 are constants.

3. PARTICULAR COMPONENT

The particular component is denoted by yp (t) and represents the solution to the differential equation for a given
input. It is usually obtained by assuming the system output has the same general form as the input. For example,
if the input is x(t) = ebt , then we assume the output is of the form yp (t) = Cebt and find the constant C so
that yp (t) is a solution to the systems differential equation. Assuming an output of the same form as the input is
consistent with our intuition that the output of the system be directly related to the input.
The forms of the particular solutions associated with a few common input signals is given in table 1. The procedure
for identifying a particular solution is illustrated in example 3.
Example 4: Solving for the Particular solution. Consider the system of example 1
d2 y(t) dy
+ 3 + 2y(t) = 2x(t).
dt2 dt
except now we are interested in calculating the particular solution of the differential equation. Assume that the
input x(t) is given by

x(t) = cos t U (t).

Using table 1, the particular solution may be assumed of the form

yp (t)(t) = C1 cos t + C2 sin t

By substitution in the differential equation, we have

(C1 cos t C2 sin t) + 3(C1 sin t + C2 cos t) + 2(C1 cos t + C2 sin t) = 2 cos t
or (C1 + 3C2 ) cos t + (3C1 + C2 ) sin t = 2 cos t.
Equating the cosine and the sine terms on the left and right hand sides of the equation, we get a set of simultaneous
equations given by

C1 + 3C2 = 2
3C1 + C2 = 0

which has a solution at C1 = 1/5 and C2 = 3/5.

Substituting for C1 and C2 results in the following particular solution

1 3
yp (t) = cos t + sin t.
5 5

This approach for finding the particular solution is modified when the input is of the same form as one of the
components of the natural response. We illustrate by an example where we also outline the modified procedure for
calculating the particular solution.
Example 5: Particular solution with input of the same form as the natural response. In example 2,
assume the applied input is given by

x(t) = 2et

and we are interested in finding the particular solution. Recall that the homogeneous solution for the differential
equation is given by

yh (t) = A1 et + A2 tet + A3 e2t . (13)

The input x(t) = 2et is of the same form as the first two terms in the homogeneous solution. In this case, we must
assume a particular solution that is independent of all terms in the natural response in order to obtain the forced
response of the system. To achieve this, we multiply the form of the particular solution by the lowest power of t
that will give a response component not included in the natural response. In this example, the particular solution
is

yp (t) = Ct2 et . (14)

In order to evaluate the constant C, we substitute the value of yp (t) and x(t) in the systems differential equation
repeated below for convenience

d3 y(t) d2 y(t) dy
3
+ 4 2
+5 + 2y(t) = x(t).
dt dt dt

Substituting the value of the first, second and third derivatives of the output y(t) and the input x(t), we get




6Cet + 6Ctet Ct2 et + 4 2Cet 4Ctet + Ct2 et + 5 2Ctet Ct2 et + 2Ct2 et = 2et

which simplifies to

2Cet = 2et . (15)

From the above expression, it is straightforward to verify that C = 1. The particular component is, therefore,
given by

yp (t) = t2 et . (16)
 4. COMPLETE RESPONSE

The complete response of the system is the sum of the homogeneous component and the particular component
using the specified initial conditions. Assuming the input is applied at t = 0, the procedure for evaluating the
complete response is enlisted below:
1. Find the form of the homogeneous component yh (t) from the roots of the characteristic equation.
2. Find the particular solution yp (t) by assuming it of the same form as the input yet independent of all terms
in the homogeneous component.
3. Determine the coefficients in the homogeneous component so that the overall response
y(t) = yh (t) + yp (t) (17)
satisfies the actual initial condition specified with the differential equation.

Example 6: Solve the differential equation

d2 y(t) dy
+ 3 + 2y(t) = 2x(t).
dt2 dt
with input x(t) given by
x(t) = cos t U (t)
and the initial conditions specified by
dy
y(0) = 1 and |t=0 = 1.
dt
From example 1, the homogeneous solution is given by
yh (t) = A1 et + A2 e2t
and from the results in example 2, the particular solution is given by
1 3
yp (t) =
cos t + sin t.
5 5
Following item 3 in the listed algorithm, the complete solution is given by
1 3
y(t) = A1 et + A2 e2t + cos t + sin t.
5 5
dy
Substituting the initial conditions y(t) = 1 and dt = 1 for t = 0 in
1 3
y(t) = A1 et + A2 e2t + cos t + sin t
5 5
dy 1 3
and = A1 e 2A2 e
t 2t
sin t + cos t
dt 5 5
gives
6
A1 + A2 =
5
2
2A1 A2 =
5
which has the solution, A1 = 1/5 and A2 = 4/5. The complete solution to the differential equation for t 0 is
therefore given by
1

y(t) = et + 4e2t + cos t + 3 sin t .
5
 5. SUMMARY

A direct method for solving linear, constant coefficient differential equations was presented.