Professional Documents
Culture Documents
Lecture 10 Non-Homogeneous Linear ODE, Method of Variation of Parameters
Lecture 10 Non-Homogeneous Linear ODE, Method of Variation of Parameters
Ghorai
Lecture X
Non-homegeneous linear ODE, method of variation of parameters
0.1
Again we concentrate on 2nd order equation but it can be applied to higher order ODE.
This has much more applicability than the method of undetermined coefficeints. First,
the ODE need not be with constant coefficeints. Second, the nonhomogeneos part r(x)
can be a much more general function.
Theorem 1. A particular solution yp to the linear ODE
y 00 + p(x)y 0 + q(x)y = r(x)
(1)
is given by
Z
y1 (x)r(x)
y2 (x)r(x)
dx + y2 (x)
dx,
W (y1 , y2 )
W (y1 , y2 )
where y1 and y2 are basis solutions for the homogeneous counterpart
yp (x) = y1 (x)
(2)
y 00 + p(x)y 0 + q(x)y = 0.
Important note: The (leading) coefficeint of y 00 in (1) must be unity. If it is not
unity, then make it unity by dividing the ODE by the leading coefficeint.
Proof: We try for yp of the form
yp (x) = u(x)y1 (x) + v(x)y2 (x),
where u(x) and v(x) are unknown functins. Now yp should satisfy (1). First, we find
yp0 (x) = u0 (x)y1 (x) + v 0 (x)y2 (x) + u(x)y10 (x) + v(x)y20 (x). Now to make calculations
easier (!), we take
u0 (x)y1 (x) + v 0 (x)y2 (x) = 0.
(3)
Next we find yp00 (x) = u0 (x)y10 (x) + v 0 (x)y20 (x) + u(x)y100 (x) + v(x)y200 (x). Subtituting
yp (x), yp0 (x) and yp00 (x) into (1) (and using the fact that y1 and y2 are solution of the
homogeneos part), we get
u0 (x)y10 (x) + v 0 (x)y20 (x) = r(x).
(4)
r(x)y2 (x)
,
W (y1 , y2 )
v0 =
r(x)y1 (x)
W (y1 , y2 )
Integrating we find
u=
y2 (x)r(x)
dx,
W (y1 , y2 )
v(x) =
y1 (x)r(x)
dx.
W (y1 , y2 )
S. Ghorai
Example 1. Consider
y 00 2y 0 3y = xex .
(This has been solved before by the method of undetermined coefficeints.) The LI
solutions of the homogenous part are y1 (x) = ex and y2 (x) = e3x . Hence,
yp (x) = y1 (x)u(x) + y2 (x)v(x)
where
u(x) =
y2 (x)r(x)
dx,
W (y1 , y2 )
v(x) =
y1 (x)r(x)
dx.
W (y1 , y2 )
x
x2
dx =
4
8
xe4x
x
1
dx = e4x e4x
4
16
64
Thus,
yp (x) =
x2 x
1
x
e + e3x e4x e4x
8
16
64
y2 (x)r(x)
dx,
W (y1 , y2 )
v(x) =
y1 (x)r(x)
dx.
W (y1 , y2 )
sin x dx = cos x
Thus,
yp (x) = cos x ln | sec x + tan x|
Hence, the general solution is
y(x) = C1 cos x + C2 sin x + yp (x)
S. Ghorai
Example 3. Consider
y 00 + y = |x|,
x (1, 1)
Solution: You can find the general solution using either the method of undetermined
coefficients (tricky!) OR method of variation of parameters. Try yourself.
Example 4. Consider
xy 00 (1 + x)y 0 + y = x2 e2x ,
x>0
This is linear but the coefficients are not constants. Note that y1 (x) = ex is a solution
(by inspection!) of
xy 00 (1 + x)y 0 + y = 0.
Let us first divide by the leading coefficient to find
(1 + x) 0 1
y + y=0
x
x
y 00
Using reduction of order, we find
y2 (x) = y1 (x)
1 R (1+x)/x dx
e
dx = (x + 1)
y12
The LI solutions of the homogenous part are y1 (x) = ex and y2 (x) = (1 + x).
Divinding by the leading coefficient leads to
y 00
(1 + x) 0 1
y + y = xe2x
x
x
y2 (x)r(x)
dx,
W (y1 , y2 )
v(x) =
y1 (x)r(x)
dx.
W (y1 , y2 )
(x + 1)ex dx = xex
v(x) =
1
e2x dx = e2x
2
Thus,
yp (x) =
(x 1) 2x
e
2
S. Ghorai
0.2
(5)
Note that the leading coefficeint is again unity. Suppose the three LI solutions to (5)
are y1 , y2 and y3 . As before let
yp (x) = u(x)y1 (x) + v(x)y2 (x) + w(x)y3 (x).
(6)
We find
yp0 (x) = u0 (x)y1 (x) + v 0 (x)y2 (x) + w0 (x)y3 (x) + u(x)y10 (x) + v(x)y20 (x) + w(x)y30 (x)
As before for the ease of computation (!) we set
u0 (x)y1 (x) + v 0 (x)y2 (x) + w0 (x)y3 (x) = 0
(7)
Now
yp00 (x) = u0 (x)y10 (x) + v 0 (x)y20 (x) + w0 (x)y30 (x) + u(x)y100 (x) + v(x)y200 (x) + w(x)y300 (x)
Again for the ease of computation (!!), we set
u0 (x)y10 (x) + v 0 (x)y20 (x) + w0 (x)y30 (x) = 0
(8)
Further
yp000 = u0 (x)y100 (x) + v 0 (x)y200 (x) + w0 (x)y300 (x) + u(x)y1000 (x) + v(x)y2000 (x) + w(x)y3000 (x)
Subtituting yp (x), yp0 (x), yp00 (x) and yp000 (x) into (5) (and using the fact that y1 , y2 and y3
are solutions of the homogeneos part), we get
u0 (x)y100 (x) + v 0 (x)y200 (x) + w0 (x)y30 (x) = r(x).
(9)
y1 y2 y3
W (y1 , y2 , y3 ) = y10 y20 y30
y100 y200 y300
be the Wronskian of three LI solutions. (Wronskian can similarly defined for n LI
solutions). Then
u0 =
W1
W2
W3
, v0 =
w0 =
.
W (y1 , y2 , y3 )
W (y1 , y2 , y3 )
W (y1 , y2 , y3 )
Here Wi is the determinate obtained from W (y1 , y2 , y3 ) by replacing the i-th column
by the column vector (0, 0, r(x))T . Hence,
u=
Z
Z
W1
W2
W3
dx, v =
dx, w =
dx.
W (y1 , y2 , y3 )
W (y1 , y2 , y3 )
W (y1 , y2 , y3 )