GUIDELINES FOR THE
METHOD OF UNDETERMINED COEFFICIENTS
Given a constant coefficient linear differential equation
ay 00 + by 0 + cy = g(t),
where g is an exponential, a simple sinusoidal function, a polynomial, or a product
of these functions:
1. Find a pair of linearly independent solutions of the homogeneous problem:
{y1 , y2 }.
2. If g is NOT a solution of the homogeneous equation, take a trial solution
of the same type as g as suggested in the table below:
Forcing Function Trial Solution
rt
ae Aert
a sin (ωt) or a cos (ωt) A sin (ωt) + B cos (ωt)
n
at P (t)
n a positive integer P a general polynomial of degree n
n rt
at e P (t)ert
n a positive integer P a general polynomial of degree n
n
t [a sin (ωt) + b cos (ωt)] P (t)[A sin (ωt) + B cos (ωt)]
n a positive integer P a general polynomial of degree n
ert [a sin (ωt) + b cos (ωt)] ert [A sin (ωt) + B cos (ωt)]
3. If g is a solution of the homogeneous problem, take a trial solution of the
same type as g multiplied by the lowest power of t for which NO TERM
of the trial solution is a solution of the homogeneous equation.
4. Substitute the trial solution into the differential equation and solve for the
undetermined coefficients so that it is a particular solution yp .
5. Set y(t) = yp (t) + [c1 y1 (t) + c2 y2 (t)] where the constants c1 and c2 can be
determined if initial conditions are given.
6. If g is a sum of the type of forcing function described above, split the
problem into simpler parts. Find a particular solution for each of these,
then add particular solutions to obtain yp for the complete equation.
1
Examples: In most of the following examples, we discuss the differential equa-
tion
y 00 + y 0 − 6y = g(t),
for various choices of the forcing function g. Note that the associated homoge-
neous equation has characteristic polynomial λ2 + λ − 6 which two distinct real
roots, λ1 = −3, and λ2 = 2. Hence the two linearly independent solutions for
the homogeneous equation are y1 = e−3t and y2 = e2t .
(a) If g(t) = 5e4t , then we take a trial solution yp (t) := Ae4t , with yp0 = 4Ae4t
and yp00 = 16Ae4t . Then, we substitute these values into the differential
equation. One convenient way to organize our work is to write things down
as follows:
yp00 (t) = 16Ae4t
+yp0 (t) = 4e4t
−6yp (t) = −6Ae4t
yp00 + yp0 − 6y = 14Ae4t
or
4t 4t
14Ae = 5e
Hence, 14A = 5 or A = 5/14 and the particular solution for this problem
is
5 4t
yp (t) = e ,
14
with general solution
5 4t h −3t i
y(t) = e + c1 e + c2 e2t .
14
Suppose
√ that we are given initial conditions y(0) = 1/2 and y 0 (0) =
2π/2. Then, since
5 4t h i
y 0 (t) = e + −3c1 e−3t + 2c2 e2t ,
14
we can compute the constants c1 , c2 so that the solution satisfies the initial
conditions. Indeed,
5 1
y(0) = + c1 + c2 =
14 2 √
20 2π
y 0 (0) = − 3c1 + 2c2 =
14 2
or
2
1
c1 + c2 =
7√
7 2π − 20
−3c1 + 2c2 =
14
From the first, c1 = (1/7) − c2 so that, substituting into the second yields
√
1 7 2π − 20
−3 − c2 + 2c2 =
7 14
or √
7 2π − 20 3
5c2 = +
14 7
so that √ √
2π − 2 1 2π − 2
c2 = , and c1 = .
10 7 10
Thus we have the solution to the initial value problem:
√ ! √ !
5 4t 12 2π −3t 1 2π 2t
y(t) = e + − e + − + e .
14 35 10 5 10
(b) Let g(t) = 4e2t , and take a trial solution yp (t) = Ae2t . Then yp0 (t) = 2Ae2t
and yp00 (t) = 4Ae2t . Then,
yp00 (t) = 4Ae2t
+yp0 (t) = 2e2t
−6yp (t) = −6Ae2t
yp00 + yp0 − 6y = 0Ae4t
or
2t
0 = 5e
From this result, we conclued that we have chosen the wrong trial function.
The difficulty here is that Ae2t is a solution of the homogeneous problem,
so it cannot be used as a trial function because any constant multiple will
also be a solution of the homogeneous problem.
The correct choice is yp (t) = A t e2t , with yp0 = A[2 t e2t + e2t ] and yp00 =
A[4 t e2t + 4e2t ]. Then
3
2t
yp00 (t) = 4Ate2ta+4Ae
+yp0 (t) = 2Ate2t + Ae2t
−6yp (t) = −6Ate2t
yp00 + yp0 − 6y = 5Ae2t
or
2t 4t
5Ae = 4e ,
so that 5A = 4 or A = 4/5. It follows that a general solution is
4 2t h −3t i
y(t) = te + c1 e + c2 e2t .
5
(c) Suppose that g(t) = 5t. We take a trial solution in the form of a general
polynomial of degree one, yp (t) = At + B with yp0 = A and yp00 = 0. Hence
yp00 (t) = 0
+yp0 (t) = A
−6yp (t) = −6At − 6B
yp00 + yp0 − 6y
= −6At + (A − 6B)
or
−6At + (A − 6B) = 5t,
It follows that A − 6B = 0 and −6A = 5, and therefore we find that
A = −(5/6), B = −(5/36), which results in the particular solution
5 5
yp (t) = − t − .
6 36
CAUTIONARY NOTE: Suppose we had chosen the trial function simply
as yp (t) = At with no constant term B which has the same derivatives as
our first selection. Then substitution yields
yp00 (t) + yp0 (t) − 6yp (t) = A − 6At = 5t.
This last equation, A − 6At = 5t has no constant solution A and the
method fails.
4
In order to illustrate the next point, we will look at a different differential equa-
tion:
1 t
y 00 + y 0 + y = te− 2 .
4
(d) It is easy to check that the characteristic polynomial λ2 + λ + (1/4) has a
double root λ = −(1/2). While we may be tempted to take a trial solution
t
of the form yp (t) = (A + Bt)e− 2 , we see that such a choice is doomed as it
is simply a combination of the solutions to the homogeneous problem. We
t t
then try the form yp (t) = t(A + Bt)e− 2 = (Bt2 + At)e− 2 . Then a tedious
calculation leads to:
1 t t
yp0 = − (At + Bt2 )e− 2 + (A + 2Bt)e− 2
2
1 1
t
= − Bt2 + (2B − A)t) + A e− 2
2 2
and
1 t t t
yp00 = (At + Bt2 )e− 2 + (A + 2Bt)e− 2 + 2Be− 2
4
1 2 1
t
= Bt + (−2B + A)t) + (2B − A) e− 2
4 4
Using these results, we have:
h i t
yp00 (t) = 1
4
Bt2 + (−2B + 41 A)t) + (2B − A) e− 2
h i t
+yp0 (t) = − 21 Bt2 + (2B − 12 A)t) + A e− 2
t
1
y (t)
4 p
= 1
4
Bt2 + 14 At e− 2
t
yp00 + yp0 − 6y = 0t2 + 0t + 2Be 2
or
t t
2Be− 2 = te− 2 ,
Again, the resulting equation 2B = t has no constant solution. In this
case our error is that, while we have avoided a solution of the homogeneous
equation by multiplying the polynomial by t, we still have that ONE of
the terms in the new trial solution is a solution of the homogeneous equa-
t
tion, namely the term A t e− 2 . We must multiply by a high enough power
of t so that NO PART of the resulting trial solution is a solution of the
homogeneous problem. In this case, we need a trial solution of the form
t
yp (t) = t2 (A + Bt)e− 2 .
(e) Now we return to the equation y 00 + y 0 − 6y = g(t) and consider the case
when g(t) = 5t + 4e2t . Here we split the problem into two parts, which
we first solve separately and then add to get the partucular solution of the
given non-homogeneous equation. So we consider separately
5
(e1) y 00 + y 0 − 6y = 5t
(e2) y 00 + y 0 − 6y = 4e2t .
Since this example is meant only to illustrate the point, we have chosen
two problems which we have already solved above in parts (c) and (b)
respectively. Thus a particular solution to (e1) is yp (t) = −(5/6)t + (5/36)
while a solution of (e2) is yp (t) = (4/5) t e2t . Hence a particular solution of
the full problem corresponding to the forcing function g(t) = 5t + 4e2t is
5 5 4
yp (t) = − t − + te2t .
6 36 5
