MA-207 Differential Equations II

Ronnie Sebastian

Department of Mathematics
Indian Institute of Technology Bombay
Powai, Mumbai - 76

1 / 47
References

Elementary differential equations with boundary value problems

by William F. Trench (available online)
Differential Equations with Applications and Historical Notes
by George F. Simmons

2 / 47
Welcome to MA 207, a sequel to MA 108. We begin by reviewing
elementary functions, which were discussed in MA 108.
A function f : R → R of the type

f (x) = an xn + an−1 xn−1 + · · · + a0 ai ∈ R

is called a polynomial function.

Example
x3 + 2x + 5

A rational function is a quotient of polynomial functions.

Example
x3 + 3x + 2
,
x5 + 2x3 + 5

3 / 47
A function y = f (x) is called algebraic if it satisfies an equation of
the form

Pn (x)y n + Pn−1 (x)y n−1 + . . . + P1 (x)y + P0 (x) = 0

for some n, where each Pi (x) is a polynomial.

Next we have
1 trigonometric functions, for example, sin x, cos x, tan x
2 inverse trigonometric functions, for example
sin−1 x, cos−1 x, tan−1 x
3 exponential functions, for example ex , log x

4 / 47
A elementary function is one which can be obtained by adding,
subtracting, multiplying, dividing and composing any of the above
functions.
Thus
" 2 √ #1/3
xe1/x + tan−1 (1 + x2 ) + x2 + 3
y = tan √
sin x cos 2x − log x + x3/2

is an elementary function.

5 / 47
Beyond elementary functions lie the special functions, for example,
Gamma function, Beta function, Riemann zeta function etc.

Definition
The Riemann zeta function is defined on the set
{s ∈ C | Re(s) > 1} by
X 1
ζ(s) :=
ns
n≥1

It is a non-trivial theorem that the zeta function extends to the

whole plane as a meromorphic function. (Explaining this term is
beyond the scope of this course)
The Riemann hypothesis states that all the nontrivial zeros of the
zeta function lie on the line Re(s) = 21 .
This is one of the millennium problems and has a prize of 1 million
US dollars.

6 / 47
Large number of special functions arise as solutions of 2nd order
linear ODE. Suppose we want to solve

y 00 + y = 0

Then elementary functions y = sin x and y = cos x are solutions.

Suppose we want to solve

xy 00 + y 0 + xy = 0

This equation can not be solved in terms of elementary functions.

7 / 47
Let y1 (x) be one solution of the ODE

y 00 + p(x)y 0 + q(x)y = 0

with p(x), q(x) continuous. Then we can try to use the method of
variation of parameters to find another linearly independent
solution, that is, put
y2 = u(x)y1 (x)
in the ODE and solve for u(x).
Question. How to find the 1st solution?
For this, we will solve our ODE in terms of power series.
Let us review power series, which is used throughout in this course.

8 / 47
Definition (Power series)
For real numbers x0 , a0 , a1 , a2 , . . ., an infinite series
∞
X
an (x − x0 )n := a0 + a1 (x − x0 ) + a2 (x − x0 )2 + . . . .
n=0

is called a power series in x − x0 with center x0 .

9 / 47
For a real number x1 , if the limit
N
X
lim an (x1 − x0 )n
N →∞
n=0

exists and is finite, then we say the power series converges at the
point x = x1 . In this case, the sum of the series is the value of the
limit.
If the series does not converge at x1 , that is, either limit does not
exist or it is ±∞, then we say the power series diverges at x1 .

10 / 47
Theorem
For any power series,
∞
X
an (x − x0 )n
n=0

exactly one of these statements is true.

1 The power series converges only for x = x0 .
2 The power series converges for all values of x.
3 There is a positive number 0 < R < ∞ such that the power
series converges if |x − x0 | < R and diverges if |x − x0 | > R.

R is called the radius of convergence of the power series.

We define R = 0 in case (i)
and R = ∞ in case (ii).
Question. How to compute the radius of convergence?
11 / 47
Theorem
(Ratio test) If an 6= 0 for all n and

an+1
lim =L
n→∞ an

(Root test) lim sup |an |1/n = L

n→∞
∞
X
Then radius of convergence of the power series an (x − x0 )n is
n=0
R = 1/L .
For L = 0, we get R = ∞ and for L = ∞, we get R = 0.

12 / 47
Theorem
Let R > 0 be the radius of convergence of the power series
∞
X
an (x − x0 )n
n=0

Then the power series converges (absolutely) for all

x ∈ (x0 − R, x0 + R).

For R = ∞, we write (x0 − R, x0 + R) = (−∞, ∞) = R.

The open interval (x0 − R, x0 + R) is called the interval of
convergence of the power series.

13 / 47
Example
Find the radius of convergence and interval of convergence (if
R > 0) of the following three series
∞ ∞ n ∞
nx
X X X
n
(i) n!x (ii) (−1) (iii) 2n n3 (x − 1)n
nn
0 10 0

an+1 (n + 1)!
(i) lim = lim = lim (n + 1) = ∞
n→∞ an n→∞ n! n→∞

So R = 0 in case (i).
Similarly, in case (ii) R = ∞ and in case (iii) R = 1/2.
Interval of convergence : in case (ii) (−∞, ∞) and in case (iii)
(1/2, 3/2)

14 / 47
Theorem
Let R be the radius of convergence of the power series
X∞
an (x − x0 )n . We assume R > 0
n=0
• We can define a function f : (x0 − R, x0 + R) → R by
∞
X
f (x) = an (x − x0 )n
n=0

• f is infinitely differentiable ∀ x ∈ (x0 − R, x0 + R).

• The successive derivatives of f can be computed by
differentiating the power series term-by-term, that is
∞
X
f 0 (x) = nan (x − x0 )n−1 ...
n=0
∞
X
(k)
f (x) = n(n − 1) . . . (n − k + 1) an (x − x0 )n−k
n=0
15 / 47
Theorem (continued . . .)
• The power series representing the derivatives f (n) (x) have same
radius of convergence R.
• We can determine the coefficients an (in terms of derivatives of
f at x0 ) as

f (x0 ) = a0 , f 0 (x0 ) = a1 , f 00 (x0 ) = 2a2 , . . .

In general,
f (n) (x0 )
an =
n!
∞
X
• We can also integrate the function f (x) = an (x − x0 )n
0
term-wise that is if [a, b] ⊂ (x0 − R, x0 + R), then
Z b ∞ Z b ∞
X
n
X an
f (x) dx = an (x − x0 ) dx = (x − x0 )n+1
a a n+1
n=0 0
16 / 47
Example (Power series representation of elementary functions)
∞
X xn
(i) ex = −∞<x<∞
n!
0
∞
X x2n+1
(ii) sin x = (−1)n −∞<x<∞
(2n + 1)!
0
∞
1 X
(iii) = xn −1<x<1
1−x
0
∞
x2n+1
 
d X d
(iv) (sin x) = (−1)n
dx dx (2n + 1)!
0
∞
X x2n
= (−1)n = cos x
(2n)!
0

17 / 47
Theorem
(i) Power series representation of f in an open interval I
containing x0 is unique, that is, if
∞
X ∞
X
f (x) = an (x − x0 )n = bn (x − x0 )n
0 0

for all x ∈ I, then an = bn ∀ n.

(ii) If
∞
X
an (x − x0 )n = 0
0

for all x ∈ I, then an = 0 for all n.

Proof. (i)
f (n) (x0 )
an = = bn for all n.
n!
It is clear that (ii) follows from (i).
18 / 47
Algebraic operations on power series
Definition
∞
X ∞
X
If f (x) = an (x − x0 )n g(x) = bn (x − x0 )n
0 0
have radius of convergence R1 and R2 respectively, then
∞
X
c1 f (x) + c2 g(x) := (c1 an + c2 bn )(x − x0 )n
0

has radius of convergence R ≥ min {R1 , R2 } for c1 , c2 ∈ R.

Further, we can multiply the series as if they were polynomials,
that is
∞
X
f (x)g(x) = cn (x − x0 )n ; cn = a0 bn + a1 bn−1 + . . . + an b0
0

It also has radius of convergence R ≥ min {R1 , R2 }.

19 / 47
Example
Find the power series expansion for cosh x in terms of powers of
xn .
1 1
cosh x = ex + e−x
2 2
∞ n ∞
1 X x 1X xn
= + (−1)n
2 n! 2 n!
n=0 n=0
∞
X 1 xn
= [1 + (−1)n ]
2 n!
n=0
∞
X x2n
=
(2n)!
n=0

Since radius of convergence for Taylor series of ex and e−x are ∞,

the power series expansion of cosh x is valid on R.

20 / 47
Shifting the summation index
∞
X ∞
X
If f (x) = an (x − x0 )n =⇒ f 0 (x) = nan (x − x0 )n−1
n=0 n=1
Let us rewrite the series for f 0 (x) in powers of (x − x0 )n . Put
r = n − 1, we get
X∞
f 0 (x) = (r + 1)ar+1 (x − x0 )r
r=0
Similarly,
∞
X
(k)
f (x) = n(n − 1) . . . (n − k + 1)an (x − x0 )n−k
n=k
X∞
= (n + k)(n + k − 1) . . . (n + 1)an+k (x − x0 )n
n=0
 
∞
X X
In general,  bn (x − x0 )n−k = bn+k (x − x0 )n 
n=n0 n=n0 −k
21 / 47
Example
∞
X
Let f (x) = an xn . Write (x − 1)f 00 as a power series around 0.
n=0

(x − 1)f 00 = xf 00 − f 00
∞ ∞
!
X X
n−2
=x n(n − 1)an x − n(n − 1)an xn−2
n=2 n=2
∞
X ∞
X
= n(n − 1)an xn−1 − n(n − 1)an xn−2
n=2 n=2
∞
X ∞
X
= (n + 1)nan+1 xn − (n + 2)(n + 1)an+2 xn
n=1 n=0
∞
X
= [(n + 1)nan+1 − (n + 2)(n + 1)an+2 ] xn
n=0

22 / 47
Example (Solving ODE)
Suppose
∞
X
y(x) = an (x − 1)n
n=0

for all x in an open interval I containing x0 = 1.

Find the power series of y 0 and y 00 in terms of x − 1 in the
interval I. Use these to express the function

(1 + x)y 00 + 2(x − 1)2 y 0 + 3y

as a power series in x − 1 on I.
Find necessary and sufficient conditions on the coefficients
an ’s, so that y(x) is a solution of the ODE

(1 + x)y 00 + 2(x − 1)2 y 0 + 3y = 0

23 / 47
Example (Continue . . .)
Solution. Write the ODE in (x − 1), that is

(1 + x)y 00 + 2(x − 1)2 y 0 + 3y = (x − 1)y 00 + 2y 00 + 2(x − 1)2 y 0 + 3y

Express each of (x − 1)y 00 , 2y 00 , 2(x − 1)2 y 0 and 3y as a power

series in powers of (x − 1) and add them.
∞
X
(x − 1)y 00 = (x − 1) n(n − 1)an (x − 1)n−2
n=2
∞
X
= n(n − 1)an (x − 1)n−1
n=2
∞
X
= (n + 1)nan+1 (x − 1)n
n=1
∞
X
= (n + 1)nan+1 (x − 1)n
n=0
24 / 47
Example (Continue . . .)
∞
X
2y 00 = 2n(n − 1)an (x − 1)n−2
n=2
∞
X
= 2(n + 2)(n + 1)an+2 (x − 1)n
n=0
∞
X
2 0 2
2(x − 1) y = 2(x − 1) nan (x − 1)n−1
n=1
∞
X
= 2nan (x − 1)n+1
n=1
∞
X
= 2(n − 1)an−1 (x − 1)n
n=2
∞
X
= 2(n − 1)an−1 (x − 1)n (a−1 = 0)
n=0
25 / 47
Example (Continue . . .)
We have
∞
X
(x − 1)y 00 = (n + 1)nan+1 (x − 1)n
n=0
∞
X
2y 00 = 2(n + 2)(n + 1)an+2 (x − 1)n
n=0
∞
X
2 0
2(x − 1) y = 2(n − 1)an−1 (x − 1)n (a−1 = 0)
n=0
Now we get
∞
X
(x − 1)y 00 + 2y 00 + 2(x − 1)2 y 0 + 3y = bn (x − 1)n
n=0
where
bn = (n + 1)nan+1 + 2(n + 2)(n + 1)an+2 + 2(n − 1)an−1 + 3an

26 / 47
Example (Continue . . .)
∞
X
y(x) = an (x − 1)n
0
is the solution of the ODE

(x − 1)y 00 + 2y 00 + 2(x − 1)2 y 0 + 3y = 0

on the open interval I containing 1 if and only if

∞
X
bn (x − 1)n = 0 on I ⇐⇒ bn = 0 for all n
n=0

that is an ’s satisfy the following recursive relation

(n + 1)nan+1 + 2(n + 2)(n + 1)an+2 + 2(n − 1)an−1 + 3an = 0

for all n.

27 / 47
Definition
If a function f (x) is infinitely differentiable at x0 , then the Taylor
series of f at x0 is defined as the power series
∞
X f (n) (x0 )
T S f |x0 := (x − x0 )n
n!
0

When x0 = 0, the series is also called the Maclaurin series of f .

Example
( 2
e−1/x if x 6= 0
The function f (x) =
0 if x = 0
is infinitely differentiable at 0. But f (n) (0) = 0 for all n.
Hence the Taylor series of f at 0 is the constant function taking
value 0.
Therefore Taylor series of f at 0 does not converge to function
f (x) on any open interval around 0.
28 / 47
Definition
Suppose
f (x) is infinitely differentiable at x0 ; and
Taylor series of f at x0 converges to f (x) for all x in some
open interval around x0 ;
Then f is called analytic at x0 .

Example
( 2
e−1/x if x 6= 0
The function f (x) =
0 if x = 0
is not analytic at 0. Here 2nd condition fails.
However, f is analytic at all x 6= 0.

29 / 47
Theorem (Analytic functions)
1 If f (x) and g(x) are analytic at x0 , then f (x) ± g(x)
f (x)g(x) f (x)/g(x) (if g(x0 ) 6= 0) are analytic at x0 .
2 If f (x) is analytic at x0 and g(x) is analytic at f (x0 ), then
g(f (x)) := (g ◦ f )(x) is analytic at x0 .
X∞
3 If a power series an (x − x0 )n has radius of convergence
0
∞
X
R > 0, then the function f (x) := an (x − x0 )n is analytic
0
at all points x ∈ (x0 − R, x0 + R).

30 / 47
Example
The function f (x) = x2 + 1 is analytic everywhere. Since x2 + 1 is
never 0, the function h(x) := x21+1 is analytic everywhere.
However, there is no power series around 0 which represents h(x)
everywhere.
If there were such a power series, then by uniqueness, it has to be
the power series expansion of h(x) around 0, which is

1 − x2 + x4 − x6 + · · ·

However, the radius of convergence of this is R = 1.

In fact, for any x0 , there is no power series around x0 which
represents h(x) everywhere.

31 / 47
Theorem
Let
N (x)  x3 − 1 
F (x) = example F (x) =
D(x) x2 + 1
be a rational function, where N (x) and D(x) are polynomials
without any common factors, that is they do not have any
common (complex) zeros. Let α1 , . . . , αr be distinct complex zeros
of D(x).
Then F (x) is analytic at all x except at x ∈ {α1 , . . . , αr }.
If x0 is different from {α1 , . . . , αr }, then the radius of convergence
R of the Taylor series of F at x0
∞
X F (n) (x0 )
T S F x0 = (x − x0 )n
n!
0

is given by

R = min {|x0 − α1 |, |x0 − α2 |, . . . , |x0 − αr |}

32 / 47
Example
If
N (x) (2 + 3x)
F (x) = =
D(x) (4 + x)(9 + x2 )
√
then D(x) has zeros at −4 and ±3ι, where ι = −1.
Hence F is analytic at all x except at x ∈ {−4, ±3ι}.
If x = 2, then radius of convergence of Taylor series of F at x = 2
is √ √
min {|2 + 4|, |2 + 3ι|, |2 − 3ι|} = min {6, 13} = 13
If x = −6, then radius of convergence of Taylor series of F at
x = −6 is
√
min {| − 6 + 4|, | − 6 ± 3ι|} = min {2, 45} = 2

33 / 47
Power series solution of ODE

Theorem (Existence Theorem)

If p(x) and q(x) are analytic functions at x0 , then every solution of

y 00 + p(x)y 0 + q(x)y = 0

is also analytic at x0 ; and therefore any solution can be expressed

as
X∞
y(x) = an (x − x0 )n
0

If R1 = radius of convergence of Taylor series of p(x) at x0 ,

R2 = radius of convergence of Taylor series of q(x) at x0 ,
then radius of convergence of y(x) is at least min(R1 , R2 ) > 0.

In most applications, p(x) and q(x) are rational functions, that is

quotient of polynomial functions.
34 / 47
Series solution of ODE
Example
Let us solve y 00 + y = 0 (1) by power series method.
Compare with y 00 + p(x)y 0 + q(x)y = 0,
p(x) = 0 and q(x) = 1 are analytic at all x.
We can find power series solution in x − x0 for any x0 .
Let us assume x0 = 0 for simplicity.
By existence theorem, all solution of (1) can be found in the form
∞
X
y(x) = an xn
0
and the series will have ∞ radius of convergence.
Since
∞
X ∞
X
y 00 = n(n − 1)an xn−2 = (n + 2)(n + 1)an+2 xn
2 n=0

35 / 47
Example (Continue . . .)
∞
X
y 00 + y = ((n + 2)(n + 1)an+2 + an )xn = 0
0
By uniqueness of power series in x − x0 with positive radius of
convergence, we get the recursion formula

(n + 2)(n + 1)an+2 + an = 0
−1
=⇒ an+2 = an ∀n
(n + 2)(n + 1)
Therefore,
−1 −1 1 1
a2 = a0 , a4 = a2 = a0 . . . a2n = (−1)n a0
2.1 4.3 4! (2n)!
−1 −1 1 1
a3 = a1 , a 5 = a3 = a1 . . . a2n+1 = (−1)n a1
3.2 5.4 5! (2n + 1)!

36 / 47
Example (Continue . . .)
Define
1 2 1
y1 (x) = 1 − x + x4 − . . . (a0 = 1, a1 = 0)
2! 4!
1 3 1
y2 (x) = x − x + x5 − . . . (a0 = 0, a1 = 1)
3! 5!
Then
∞
X
y(x) = an xn = a0 y1 (x) + a1 y2 (x)
0

is a general solution of the ODE (1).

In this case, y1 (x) = cos x and y2 (x) = sin x. Thus, y(x) is an
elementary function. In general, however, the solution may not be
an elementary function.
We don’t need to check the series for converges, since the
existence theorem guarantees that the series converges for all x.
37 / 47
Steps for Series solution of linear ODE

1 Write ODE in standard form y 00 + p(x)y 0 + q(x)y = 0.

2 Choose x0 at which p(x) and q(x) are analytic. If boundary
conditions at x0 are given, choose the center of the power
series as x0 .
3 Find minimum of radius of convergence of Taylor series of
p(x) and q(x) at x0 .
X∞
4 Let y(x) = an (x − x0 )n , compute the power series for
0
y 0 (x) and y 00 (x) at x0 and substitute these into the ODE.
5 Set the coefficients of (x − x0 )n to zero and find recursion
formula.
6 From the recursion formula, obtain (linearly independent)
solutions y1 (x) and y2 (x). The general solution then looks
like y(x) = a1 y1 (x) + a2 y2 (x).

38 / 47
The following ODE’s are classical:
Bessel’s equation :

x2 y 00 + xy 0 + (x2 − ν 2 )y = 0

It occurs in problems displaying cylindrical symmetry,

example diffusion of light through a circular aperture,
vibration of a circular head drum, etc.
Airy’s equation :
y 00 − xy = 0
It occurs in astronomy and quantum physics.
Legendre’s equation :

(1 − x2 )y 00 − 2xy 0 + α(α + 1)y = 0

It occurs in problems displaying spherical symmetry,

particularly in electromagnetism.

39 / 47
In this course, we will consider ODE

P0 (x)y 00 + P1 (x)y 0 + P2 (x)y = 0

with Pi (x) polynomials for i = 0, 1, 2 without any common factor.

If we write ODE in the standard form
P1 (x) 0 P2 (x)
y 00 + y + y=0
P0 (x) P0 (x)

we see that if x0 is not a zero of P0 (x), then

P1 (x)/P0 (x) and P2 (x)/P0 (x) will be analytic at x0
hence we can find the series solution of ODE in the form
∞
X
y(x) = an (x − x0 )n
0

When x0 is a zero of P0 (x), then x0 is called a singular point of

ODE. This case will be considered later.
40 / 47
Example
Find the power series in x for the general solution of

(1 + 2x2 )y 00 + 6xy 0 + 2y = 0

Solution. Note that 0 is not a zero of P0 (x) = 1 + 2x2 , hence the

series solution in powers of x exists.
∞
X
Put y = an xn in the ODE, we get
0

(1 + 2x2 )y 00 + 6xy 0 + 2y
= y 00 + 2x2 y 00 + 6xy 0 + 2y
∞
X
= ((n + 2)(n + 1)an+2 + 2n(n − 1)an + 6nan + 2an )xn
0

=⇒ (n + 2)(n + 1)an+2 + [2n(n − 1) + 6n + 2]an = 0

41 / 47
Example (Continue . . .)

2n2 + 4n + 2 n+1
=⇒ an+2 = − an = −2 an n ≥ 0
(n + 2)(n + 1) (n + 2)

Since indices on left and right differ by 2, we write separately for

n = 2m and n = 2m + 1, m ≥ 0, so
2m + 1 2m + 1
a2m+2 = −2 a2m = − a2m
2m + 2 m+1
2m + 2 m+1
a2m+3 = −2 a2m+1 = −4 a2m+1
2m + 3 2m + 3
1
a2 = − a0
1
3 1.3
a4 = − a2 = a0
2 1.2
5 1.3.5
a6 = − a4 = − a0
3 1.2.3
42 / 47
Example (Continue . . .)
1.3.5. . . . (2m − 1)
a2m = (−1)m a0
m!
Qm
j=1 (2j − 1))
= (−1)m a0
m!
m+1
a2m+3 = −4 a2m+1
2m + 3
1
a3 = −4 a1
3
2 1.2
a5 = −4 a3 = 42 a1
5 3.5
3 1.2.3
a7 = −4 a5 = −43 a1
7 3.5.7
m!
a2m+1 = (−1)m 4m Qm a1
j=1 (2j + 1)

43 / 47
Example (Continue . . .)
We can write the solution
∞
X
y= an xn = a0 y1 (x) + a1 y2 (x)
0

where a0 and a1 are arbitrary scalars and

∞ Qm
j=1 (2j − 1) 2m
X
m
y1 (x) = (−1) x
m!
m=0

∞
X 4m m!
y2 (x) = (−1) Qm x2m+1
j=1 (2j + 1)
m=0
±ι
Since P0 (x) = 1 + 2x2 has complex zeros √ , the power series
 2
−1 1
solution converges in the interval √ , √ . 
2 2
44 / 47
Example
Find the coefficients a0 , . . . , a6 in the series solution
∞
X
y= an xn
0

of the IVP

(1 + x + 2x2 )y 00 + (1 + 7x)y 0 + 2y = 0

with
y(0) = −1, y 0 (0) = −2.

√
Zeros of P0 (x) = 1 + x + 2x2 are 14 (−1 ± ι 7) whose absolute
√
values are 1/ 2. Hence the
 series solution to the IVP converges
−1 1
on the interval √ , √ .
2 2

45 / 47
Example (Continue . . .)
∞
X
(1 + x + 2x2 )y 00 + (1 + 7x)y 0 + 2y = bn xn = 0
0

bn = (n + 2)(n + 1)an+2 + (n + 1)nan+1 + 2n(n − 1)an

+(n + 1)an+1 + 7nan + 2an = 0
that is

(n + 2)(n + 1)an+2 + (n + 1)2 an+1 + (2n2 + 5n + 2)an = 0

Since 2n2 + 5n + 2 = (n + 2)(2n + 1),

n+1 2n + 1
an+2 = − an+1 − an n≥0
n+2 n+1

46 / 47
Example (Continue . . .)
n+1 2n + 1
an+2 = − an+1 − an n≥0
n+2 n+1

From the initial conditions y(0) = −1, y 0 (0) = −2 we get

a0 = y(0) = −1, a1 = y 0 (0) = −2

1
a2 = − a1 − a0 = 2
2
2 3 5
a3 = − a2 − a1 =
3 2 3

Check that
5 55 3 61
y(x) = −1 − 2x + 2x2 + x3 − x4 + x5 + x6 + . . .
3 12 4 8

47 / 47