Solutions of Nonhomogeneous Higher Order

Linear ODEs:
Variation of Parameters, Cauchy-Euler Equations
& Finding one Solution from a Known Solution

Department of Mathematics
IIT Guwahati

SHB/SU MA-102 (2020)

Variation of Parameters

• The variation of parameter is a more general method for

finding a particular solution (yp ). The method applies
even when the coefficients of the differential equation are
functions of x.

Consider L(y ) = g (x), where

L(y ) := y (n) + pn−1 (x)y (n−1) + · · · + p1 (x)y 0 + p0 (x)y ,
where pn−1 (x), . . . , p0 (x) ∈ C ((a, b)). We know the general
solution to L(y ) = g is given by
y (x) = yh (x) + yp (x),

where yh is the general solution to Ly = 0 and yp (x) is a

particular solution to L(y ) = g .

SHB/SU MA-102 (2020)

Suppose we know a fundamental solution set {y1 , . . . , yn } for
L(y ) = 0. Then
yh (x) = C1 y1 (x) + · · · + Cn yn (x).
In this method, seek a particular solution (yp ) of the form
yp (x) = v1 (x)y1 (x) + · · · + vn (x)yn (x),
and try to determine the functions v1 , . . . , vn .
Differentiating yp ,
n
X n
X
yp0 = vi yi0 + vi0 yi .
i=1 i=1

To avoid second and higher-order derivatives of vi ’s, we

impose the condition
n
X
vi0 yi = 0. (1)
i=1

SHB/SU MA-102 (2020)

Therefore,
n
X
yp0 = vi yi0 .
i=1
Again, differentiating yp0 , we obatin
n
X n
X
yp00 = vi yi00 + vi0 yi0 .
i=1 i=1

For the same reason, we impose the condition

n
X
vi0 yi0 = 0. (2)
i=1
(n−1)
Likewise, on computing yp000 , . . . , yp , we impose (n − 3)
additional conditions involving vi0 ’s as
n
X n
X (n−2)
vi0 yi00 = 0, · · · , vi0 yi = 0. (3)
i=1 i=1

SHB/SU MA-102 (2020)

Finally, for the nth condition, use the derived conditions and
the fact that L(yh ) = 0 and L(yp ) = g to obtain
n
X (n−1)
vi0 yi = g. (4)
i=1

Therefore, we seek v10 , . . . , vn0 that satisfy the system

y1 v10 + · · · + yn vn0 = 0,
y10 v10 + · · · + yn0 vn0 = 0,
.. . . .
. + .. + .. = ..
(n−1) 0
y1 v1 + · · · + yn(n−1) vn0 = g .

SHB/SU MA-102 (2020)

The existence of v10 , . . . , vn0 requires
 
 y1 ··· yn 
 .. .. 

 . . 
 (n−2) (n−2)  = W (y1 , . . . , yn )(x) 6= 0
 y1
 (n−1) · · · yn 
(n−1) 

 y
1 · · · yn
on (a, b), which is true as {y1 , . . . , yn } is a fundamental
solution set.
On solving for v10 , . . . , vn0 , we find that
g (x)Wk (x)
vk0 (x) = , k = 1, . . . , n,
W (y1 , . . . , yn )(x)
where the determinant Wk (x) is obtained from
W (y1 , . . . , yn )(x) by replacing kth column by [0, . . . , 0, 1]T .

SHB/SU MA-102 (2020)

We can express Wk (x) as

Wk (x) = (−1)(n−k) W (y1 , . . . , yk−1 , yk+1 , . . . , yn )(x)

for k = 1, . . . , n. Integrating vk0 (x) yields
Z
g (x)Wk (x)
vk (x) = dx, k = 1, . . . , n.
W (y1 , . . . , yn )(x)
Finally, substituting the vk ’s back into yp , we obtain
yp (x) = v1 (x)y1 (x) + · · · + vn (x)yn (x)
we obtain
n Z
X g (x)Wk (x)
yp (x) = yk (x) dx.
k=1
W (y1 , . . . , yn )(x)

SHB/SU MA-102 (2020)

For n = 2, v10 and v20 are given by
−g (x)y2 (x) g (x)y1 (x)
v10 (x) = , v20 (x) = ,
W (y1 , y2 )(x) W (y1 , y2 )(x)
where W (y1 , y2 )(x) 6= 0. Integrating these equations, we
obtain
−g (x)y2 (x)
Z Z
g (x)y1 (x)
v1 (x) = dx, v2 (x) = dx.
W (y1 , y2 )(x) W (y1 , y2 )(x)
Thus, the particular solution is given by
yp (x) = v1 (x)y1 (x) + v2 (x)y2 (x).

SHB/SU MA-102 (2020)

Example: Consider y 00 + y = cosec x.
yh (x) = c1 sin x + c2 cos x.

The two linearly independent solutions are y1 (x) = sin x and

y2 (x) = cos x and W (y1 , y2 ) = −1 6= 0.
−g (x)y2 (x) − cos x cosec x
Z Z
v1 (x) = dx = dx = log(sin x).
W (y1 , y2 )(x) −1
Z Z
g (x)y1 (x) sin x cosec x
v2 (x) = dx = dx = −x.
W (y1 , y2 )(x) −1
yp = sin x log(sin x) − x cos x.
The general solution is
y (x) = c1 sin x + c2 cos x + sin x log(sin x) − x cos x.

SHB/SU MA-102 (2020)

Use of a known solution to find another
Assume that y1 (x) 6= 0 is a known solution of L(y ) = 0, where
L(y ) = y 00 + p(x)y 0 + q(x)y .

We know L((cy1 )) = 0, where c is any arbitrary constant. The

basic idea is to replace c by an unknown function v (x) so that
L(y2 ) = 0, where y2 = v (x)y1 (x) and {y1 , y2 } is a
fundamental solution set of Ly = 0.
Suppose L(y2 ) = L(vy1 ) = 0. Then, we have
v (y100 + py10 + qy1 ) + v 00 y1 + v 0 (2y10 + py1 ) = 0.
Since L(y1 ) = 0, we have
 0 
00 0 y1
v y1 + v (2y10 0
+ py1 ) = 0 ⇒ w + w 2 + p = 0,
y1
where w = v 0 .
SHB/SU MA-102 (2020)
Solving this homogeneous linear equation gives
Z
0 C − R pdx C − R pdx
w (= v ) = 2 e ⇒ v (x) = e dx.
y1 y12
Therefore the second solution is
Z 
1 − R pdx
y2 (x) = e dx y1 (x),
y12
as the constant of integration C may be ignored because
Ker L is a vector space.

Example: Given that y1 = e x is a solution to y 00 − 2y 0 + y = 0.

Determine the second linear independent solution y2 .
Note that v (x) = x. The second linearly dependent solution is
y2 (x) = vy1 = xe x .

SHB/SU MA-102 (2020)

Cauchy-Euler Equation

An equation of the form

an x n y (n) + an−1 x n−1 y (n−1) + · · · + a1 xy 0 + a0 y = g (x),
where ai ’s are constants is called Cauchy-Euler equation.
The substitution x = e t transform the above equation into an
equation with constant coefficients. For simplicity, take n = 2.
Assume that x > 0 and let x = e t . By the chain rule,
dy dy dx dy t dy
= = e =x ,
dt dx dt dx dx
hence
dy dy
x = .
dx dt

SHB/SU MA-102 (2020)

Differentiating x dy
dx
= dy
with respect to t, we find that
dt

d 2y
   
d dy dx dy d dy
= x = +x
dt 2 dt dx dt dx dt dx
2
dy d y dx dy d 2y
= +x 2 = + x 2 et
dt dx dt dt dx
2
dy d y
= + x2 2 .
dt dx
Thus
d 2y d 2y dy
x2 = − .
dx 2 dt 2 dt

SHB/SU MA-102 (2020)

Substituting into the equation we obtain the constant
coefficient ODE
 2 
d y dy dy
a2 2
− + a1 + a0 y = g (e t ),
dt dt dt
which may be written as

d 2y dy
a2 2
+ (a1 − a2 ) + a0 y = g (e t ).
dt dt

Note: Observe that in the proof it is assumed that x > 0. If

x < 0, the substitution x = −e t will reduced the Cauchy-Euler
equation to constant coefficients ODE. The method can be
applied to higher-order Cauchy-Euler equation.

SHB/SU MA-102 (2020)

Example: Consider x 2 y 00 − 2xy 0 + 2y = x 3 , x > 0.
Setting x = e t , we obtain
d 2y dy dy
2
− − 2 + 2y = e 3t ,
dt dt dt
or
d 2y dy
2
− 3 + 2y = e 3t .
dt dt
The GS to the homogeneous equation is
yh (x) = c1 e t + c2 e 2t = c1 x + c2 x 2 .
1
To find a particular solution, let yp = Ae 3t . Then, A = 2
hence, yp = 21 e 3t = 12 x 3 . The GS is
y (x) = yh (x) + yp (x)
1
= c1 x + c2 x 2 + x 3 , x > 0.
2

*** End ***

SHB/SU MA-102 (2020)