Problem 1a

If we set
∞
X
y= cn x n
n=0

then
∞
X
y′ = ncn xn−1
n=1

So, plugging in, we find

∞
X ∞
X
ncn xn−1 + 4cn xn = 0
n=1 n=0

On the left summation, set n → n + 1, which changes the lower limit as follows

1→n+1

n=0
So, we have that
∞
X ∞
X
n−1
ncn x = (n + 1)cn+1 xn
n=1 n=0

So, putting this back into our equation, we have

∞
X ∞
X
(n + 1)cn+1 xn + 4cn xn = 0
n=0 n=0

∞
X
((n + 1)cn+1 + 4cn ) xn = 0
n=0

Using the Identity Theorem, we find that all the coefficients on the left must be 0, so we have the
recurrence relation
(n + 1)cn+1 + 4cn = 0
−4
cn+1 = cn
n+1
So, we find that
−4 (−4)2 2 (−4)3 3 (−4)4 4
y = c0 (1 + x+ x + x + x + ···)
1 2 6 24
So,
∞
X (−4)n n
y = c0 x =
n=0
n!
∞
X (−4x)n
c0 =
n=0
n!

c0 e−4x

Problem 1b
If we set
∞
X
y= cn x n
n=0

then
∞
X
y ′′ = n(n − 1)cn xn−2
n=2
So, plugging in, we find
∞
X ∞
X
n(n − 1)cn xn−2 + 4cn xn = 0
n=2 n=0

On the left summation, set n → n + 2, which changes the lower limit as follows

2→n+2

n=0
So, we have that
∞
X ∞
X
n(n − 1)cn xn−2 = (n + 2)(n + 1)cn+2 xn
n=2 n=0

Putting this back into our equation, we have

∞
X ∞
X
(n + 2)(n + 1)cn+2 xn + 4cn xn = 0
n=0 n=0

∞
X
((n + 2)(n + 1)cn+2 + 4cn ) xn = 0
n=0

Using the Identity Theorem, we find that all the coefficients on the left must be 0, so we have the
recurrence relation
(n + 2)(n + 1)cn+2 + 4cn = 0
−4
cn+2 = cn
(n + 2)(n + 1)
So, we find that
−4
c2 = c0
2·1
(−4)2
c4 = c0
4·3·2·1
(−4)3
c6 = c0
6·5·4·3·2·1
and
−4
c3 = c1
3·2
(−4)2
c5 = c1
5·4·3·2
(−4)3
c7 = c1
7·6·5·4·3·2·1
So, splitting y up by even and odd powers of x, we find that
∞ ∞
X (−4)n 2n X (−4)n 2n+1
y = c0 x + c1 x
n=0
(2n)! n=0
(2n + 1)!

∞ ∞
X (−1)n (2x)2n X (−1)n (2)2n x2n+1
y = c0 + c1
n=0
(2n)! n=0
(2n + 1)!
∞
c1 X (−1)n (2x)2n+1
y = c0 cos(2x) +
2 n=0 (2n + 1)!

c1
y = c0 cos(2x) + sin(2x)
2
Problem 2
The advantage of using the power series method over others is it provides us with an easily approximable
solution. With Variation of Parameters or Laplace Transforms, we end up with some integral or inverse
Laplace we need to evaluate, and there isn’t a nice way to approximate that. However, with the power
series method, we can simply compute more coefficients through the recurrence relation, which will
give us a better and better polynomial approximation to the true solution.