MATH 219

Fall 2017
Lecture 20
Lecture notes by Özgür Kişisel

Content: Series Solutions Near A Regular Singular Point

Suggested Problems:
§5.5: 4,10,11,12,13
Suppose that x0 is a regular singular point for the equation

p(x)y 00 + q(x)y 0 + r(x)y = 0 (∗)

Suppose that

(x − x0 )q(x) (x − x0 )2 r(x)
lim = α, lim = β.
x→x0 p(x) x→x0 p(x)

Then we regard the Cauchy-Euler equation

(x − x0 )2 y 00 + α(x − x0 )y 0 + βy = 0 (∗∗)

as being “close” to the equation (∗) in the sense that the first Taylor series terms
of the coefficients of the ODE’s agree. We have seen how to solve Cauchy-Euler
equations. A reasonable guess is that if we perturb a solution of the Cauchy-Euler
equation (∗∗) by multiplying it with an appropriate power series, then we can obtain
a solution for (∗). This strategy turns out to be reasonably succesful, as detailed
below:
Strategy for solving and ODE near a regular singular point

• Check that x0 is a regular singular point and find the limits α, β above.

• Find the roots r1 , r2 of the indicial equation r2 + (α − 1)r + β = 0. We

will assume the roots are real for simplicity, but the complex case is also
manageable.

1
 • Say r1 ≥ r2 . If r1 − r2 is not an integer, then one can obtain two linearly
independent power series solutions for (*)
∞
X ∞
X
y1 = |x − x0 |r1 an (x − x0 )n , y2 = |x − x0 |r2 bn (x − x0 )n
n=0 n=0

• If r1 − r2 is an integer, then a solution y1 as above still exists, but y2 above

need not exist. Instead, usually one needs to have a logarithmic term in the
second solution (we will not cover the details for such a second solution here).

• In the solutions above, the coefficients a0 and b0 will be free, so they can be
taken to be 1 without loss of generality.

Example 0.1 Solve the equation 2x2 y 00 + 3xy 0 + (2x2 − 1)y = 0, centered at x0 = 0.
3x
Solution: The function 2x 2 is not analytic at 0, therefore x0 = 0 is not an ordinary
2 2 −1)
point. The functions 2x2 and x ·(2x
x·3x
2x2
are both analytic near 0, so the singularity
is regular. The limits of the two functions are α = 23 and β = − 12 respectively.
The indicial equation is
1 1
r2 + r − = 0
2 2
1
The two roots of this equation are r1 = 2 and r2 = −1. Their difference is not an
integer, so we should have two solutions of the form
∞ ∞
1
X X
y1 = |x| 2 an x n , y2 = |x|−1 bn x n
n=0 n=0

Let us assume from now on that x > 0, so that we can remove the absolute values.
The case x < 0 is similar.
∞
1
X
y1 = an xn+ 2
n=0
∞
X 1 1
y10 = (n + )an xn− 2
n=0
2
∞
X 1 1 3
y100 = (n + )(n − )an xn− 2
n=0
2 2

2
Notice that the initial terms in the sums are non-constant, hence they should be still
kept after taking derivatives. Putting these terms in the ODE, we get
∞ ∞ ∞
2
X 1 1 n− 32
X 1 n− 12 2
X 1
2x (n + )(n − )an x + 3x (n + )an x + (2x − 1) an xn+ 2 = 0
n=0
2 2 n=0
2 n=0
∞ ∞ ∞ ∞
X 1 1 1
X 1 1
X 5
X 1
2(n + )(n − )an xn+ 2 + 3(n + )an xn+ 2 + 2an xn+ 2 − an xn+ 2 = 0
n=0
2 2 n=0
2 n=0 n=0
∞ ∞ ∞ ∞
X 1 1 1
X 1 1
X 1
X 1
2(n + )(n − )an xn+ 2 + 3(n + )an xn+ 2 + 2an−2 xn+ 2 − an xn+ 2 = 0
n=0
2 2 n=0
2 n=2 n=0
1 1 1 1 3 1 3 3
(2 · · (− ) · a0 + 3 · · a0 − a0 )x 2 + (2 · · a1 + 3 · a1 − a1 )x 2 +
2 2 2 2 2 2
∞
X 1 1 1 1
[2(n + )(n − )an + 3(n + )an + 2an−2 − an ]xn+ 2 = 0
n=2
2 2 2

We now equate the coefficient of each power of x in the above expression to 0. First
1 3
of all, the coefficient of x 2 is 0, therefore a0 is free. Next, the coefficient of x 2 shows
a1 = 0. The coefficient in the final summand gives, for n ≥ 2,

(2n2 + 3n)an + 2an−2 = 0.

Therefore, we obtain the recursion relation

2
an = − an−2 .
2n2 + 3n
We immediately get a1 = a3 = a5 = . . . = 0. The first few of the even indexed terms
are
a0 a2 a0 1 a0
a2 = − , a4 = − = , a6 = − a4 = −
7 22 154 45 6930
Taking a0 = 1, we get

x2 x4 x6
 
1
y1 = x 2 1 − + − + ...
7 154 6930

3
We will now carry out the same steps for the second solution y2 :
∞
X
y2 = bn xn−1
n=0
∞
X
y20 = (n − 1)bn xn−2
n=0
∞
X
y200 = (n − 1)(n − 2)bn xn−3
n=0

Put these terms in the ODE:

∞
X ∞
X ∞
X
2 n−3 n−2 2
2x (n − 1)(n − 2)bn x + 3x (n − 1)bn x + (2x − 1) bn xn−1 = 0
n=0 n=0 n=0
∞
X ∞
X ∞
X X∞
2(n − 1)(n − 2)bn xn−1 + 3(n − 1)bn xn−1 + 2bn xn+1 − bn xn−1 = 0
n=0 n=0 n=0 n=0
∞
X ∞
X ∞
X ∞
X
2(n − 1)(n − 2)bn xn−1 + 3(n − 1)bn xn−1 + 2bn−2 xn−1 − bn xn−1 = 0
n=0 n=0 n=2 n=0
(2 · (−1) · (−2)b0 + 3 · (−1)b0 − b0 )x−1 + (2 · 0 · (−1)b1 + 3 · 0 · b1 − b1 )x0 +
X∞
[2(n − 1)(n − 2)bn + 3(n − 1)bn + 2bn−2 − bn ]xn−1 = 0
n=2

Again, equate all coefficients of powers of x in the above expression to 0. We see

that b0 is free and b1 = 0. The coefficient in the last summand gives, for n ≥ 2,

(2n2 − 3n)bn + 2bn−2 = 0.

Therefore, the recursion relation is

2
bn = − bn−2 .
2n2 − 3n
We get b1 = b3 = b5 = . . . = 0. The first few even indexed terms are:
b2 b0 b4 b0
b2 = −b0 , b4 = − = , b6 = − =−
10 10 27 270

4
Taking b0 = 1, we get

x4 x6
 
−1 2
y2 = x 1−x + − + ...
10 270

Finally, the general solution is

y = c1 y 1 + c2 y 2

where c1 , c2 are arbitrary constants and y1 , y2 are as above.