Method of variation of parameters

Consider the following second order non-homogeneous linear equation

d2 y dy
2
+ a1 + a2 y = f (x) (1)
dx dx
Let y = c1 y1 + c2 y2 , with c1 and c2 as arbitrary constants, be the general solution
of the homogeneous equation
d2 y dy
2
+ a1 + a2 y = 0
dx dx
We assume that

y = C1 y 1 + C2 y 2 (2)

is the general solution of the non-homogeneous equation (1), where C1 and C2 are
functions of x to be so chosen that (1) is satisfied.

Differentiating (2) we get

′ ′ ′ ′ ′
y = C1 y 1 + C2 y 2 + C1 y 1 + C2 y 2 (3)
| {z }
=0

For simplicity, in order to find C1 and C2 we assume that

′ ′
C1 y 1 + C2 y 2 = 0 (4)

Differentiating (3) again,

′′ ′′ ′′ ′ ′ ′ ′
y = C1 y 1 + C2 y 2 + C1 y 1 + C2 y 2 (5)

Substituting y, y ′ and y ′′ in (1) we get

 ′′ ′
  ′′ ′
 ′ ′ ′ ′
C1 y1 + a1 y1 + a2 y1 + C2 y2 + a1 y2 + a2 y2 + C1 y1 + C2 y2 = f (x)

′ ′ ′ ′
=⇒ C1 y1 + C2 y2 = f (x) (6)

Solving the equations (4) and (6):

 
 0 y2 
 ′
f (x) y2  y f (x)
 =− 2
′
C1 = 
y1
 ′ y2  W
′
y y2 
1

1
Here W is called Wronskian. It is non-zero because y1 and y2 are linearly indepen-
dent. Similarly  
y1 0 
 ′
y f (x) y f (x)
1
 = 1
′
C2 = 
 y1 y2 
 ′  W
y y ′ 
1 2

After integrating:
Z Z
y2 f (x) y1 f (x)
C1 = − dx + d1 and C2 = dx + d2
W W
Hence the general solution of the non-homogeneous equation
Z Z
y2 f (x) y1 f (x)
y = d1 y1 + d2 y2 + y1 − dx + y2 dx
W W
Example: Apply the method of variation of parameters to solve

d2 y 2
2
−y = (7)
dx 1 + ex
Solution:
C.F. = c1 ex + c2 e−x
Let y = C1 ex + C2 e−x be the general solution of the given equation.
′ ′
y ′ = C1 ex − C2 e−x + C1 ex + C2 e−x
| {z }
=0

′ ′
y ′′ = C1 ex + C2 e−x + C1 ex − C2 e−x
Substituting in (7)
′ ′ 2
C1 ex − C2 e−x =
1 + ex
The Wronskian  x 
e e−x 
W = x
 = −2
e −e−x 
Hence Z Z
1 −x 2 e−x
C1 = − −e dx + d1 = dx + d1
2 1 + ex 1 + ex
Substitute ex = z ⇒ ex dx = dz
Z Z
1 1 1 1
C1 = dz + d 1 = − + dz + d1
z 2 (1 + z) z2 z 1 + z

2
 1
C1 = − − ln z + ln(1 + z) + d1 = −e−x − x + ln(1 + ex ) + d1
z
Similarly Z
1 2
C2 = − ex dx + d1 = − ln(1 + ex ) + d2
2 1 + ex
The general solution of the differential equation

y = d1 ex + d2 e−x − 1 − xex + (ex − e−x ) ln(1 + ex )

3
Cauchy-Euler Equations
A linear differential equation of the form

dn y n−1 d
n−1
y
xn n
+ a1 x n−1
+ · · · + an y = X (8)
dx dx
or

(xn D n + a1 D n−1 + · · · + an )y = X (9)

is called Euler-Cauchy equation.

Working Rule: To solve equation (8) we change the variable from x to z by

putting x = ez i.e. z = ln(x).
dz 1
z = ln(x) ⇒ =
dx x
dy dy dz 1 dy dy dy
= = ⇒ x =
dx dz dx x dz dx dz
We define a new operator
d d
x ≡ ≡ D1
dx dz
Again
   
d2 y d 1 dy 1 dy 1 d dy 1 dy 1 d2 y
2
= =− 2 + =− 2 +
dx dx x dz x dz x dx dz x dz x2 dz 2
d2 y d2 y dy
⇒ x2 = − = D1 (D1 − 1)y
dx2 dz 2 dz
d d
Thus we have the following formulas for D ≡ dx and D1 ≡ dz

xD = D1
x2 D 2 = D1 (D1 − 1)
x3 D 3 = D1 (D1 − 1)(D1 − 2)
..
.
xn D n = D1 (D1 − 1)(D1 − 2) . . . (D1 − n + 1)

Substituting these operator relations in the equation (9), we obtain a linear differ-
ential equation with constant coefficient

f (D1 )y = Z, where Z becomes a function of z only

4
Example 1.

(x2 D 2 − xD + 2)y = x ln x (10)

d
Let x = ez so that z = ln x and D1 ≡ dz
then the equation (10) becomes

[D1 (D1 − 1) − D1 + 2] y = zez

Auxiliary equation m2 − 2m + 2 = 0 and its roots are m = 1 ± i

Hence

C.F. = ez [c1 cos(z) + c2 sin(z)] = x [c1 cos(ln(x)) + c2 sin(ln(x))]

1 1
P.I. = zez = ez z
D12
− 2D1 + 2 2
(D1 + 1) − 2(D1 + 1) + 2
1
= ez 2 z = ez (1 + D12 )−1 z = ez z = x ln(x)
D1 + 1

General solution

y = x [c1 cos(ln(x)) + c2 sin(ln(x))] + x ln(x)

Equations reducible to Euler-Cauchy form There can be several forms of

equation which can be reduced to Euler-Cauchy form

Example 1: Solve
d3 y 4 d2 y 5 dy 2y
3
− 2
+ 2 − =1
dx x dx x dx x3

√ √
Solution: y = c1 x2 + c2 x(5+ 21)/2
+ c3 x(5− 21)/2
− x3 /5

Example 2:  2
d2 y dy dy
2x y 2 + 4y 2 = x2
2
+ 2xy
dx dx dx
Hint: y = z 2

Solution:
dy dz
y = z2 ⇒ = 2z
dx dx

5
  2
d2 y dz d2 z
⇒ 2 =2 + 2z
dx dx dx2
Substituting these values in the differential equation we get

d2 z dz
x2 − x +z =0
dx2 dx
or

[x2 D 2 − xD + 1]z = 0
Substitute x = et ⇔ ln x = t
dz dz
⇒x = ⇒ xD ≡ D1
dx dt
Similarly
x2 D 2 = D1 (D1 − 1)
Then the equation becomes

[D12 − 2D1 + 1]z = 0 ⇒ z = [c1 + c2 t]et

⇒ z = [c1 + c2 ln(x)]x

⇒ y = (c1 + c2 ln(x))2 x2

Example 3: A differential equation of the form

dn y n−1 d
n−1
y dy
(a + bx)n n
+ a1 (a + bx) n−1
+ · · · + an−1 (a + bx) + an y = X
dx dx dx
can be reduced to Euler-Cauchy equation by putting
dv
a + bx = v ⇒ =b
dx
dy dy dv dy
= =b
dx dv dx dv
Again
d2 y 2
2d y dn y n
nd y
= b or in general = b
dx2 dv 2 dxn dv n

6
Substituting these derivatives in the equation, we get

dn y a1 n−1 dn−1 y an−1 dy an X

vn + v + · · · + v + y =
dv n b dv n−1 bn−1 dv bn bn
which is an standard Euler-Cauchy equation.

Example 4: solve

d2 y dy
(1 + x)2 2
+ (1 + x) + y = 4 cos ln(1 + x)
dx dx
dv
Solution: Let (1 + x) = v ⇒ dx = 1.
dy dy 2
d y 2
d y
Hence dx = dv and dx2 = dv2 and the differential equation becomes

d2 y dy
2
v2
+ v + y = 4 cos ln v
dv dv
z d
Put v = e ⇒ ln(v) = z and let D1 ≡ dz

[D1 (D1 − 1) + D1 + 1] y = 4 cos z

D12 + 1 y = 4 cos z

C.F. = c1 cos(z) + c2 sin(z) = c1 cos(ln v) + c2 sin(ln v)

= c1 cos(ln(1 + x)) + c2 sin(ln(1 + x))

P.I. = 2z sin z = 2 ln(v) sin(ln(v)) = 2 ln(1 + x) sin(ln(1 + x)).

The general solution

y = c1 cos(ln(1 + x)) + c2 sin(ln(1 + x)) + 2 ln(1 + x) sin(ln(1 + x)).