MA-417-ODE

Lecture 20

Debanjana Mitra

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

September 30, 2021

Sturm-Liouville’s problem
Let −∞ < a < b < ∞ and let p ∈ C 1 ([a, b]) and q ∈ c([a, b]) be real
valued functions and p(t) > 0 for all t ∈ [a, b]. Consider the operator
d
L(y )(t) = − (p(t)y 0 (t)) + q(t)y (t), t ∈ (a, b),
dt
with boundary conditions:
U1 (y ) = α1 y (a) + α2 y 0 (a) = 0, U2 (y ) = β1 y (b) + β2 y 0 (b) = 0,
where α1 , α2 , β1 , β2 are given real numbers with α12 + α22 6= 0 and
β12 + β22 6= 0.
Recall that the operator L with boundary conditions U1 and U2 is
self-adjoint in L2 (a, b), i.e., it satisfies
hL(v ), uiL2 = hv , L(u)iL2 ,
for any u, v in C 2 ([a, b]) satisfying U1 (u) = 0 = U2 (u) and
U1 (v ) = 0 = U2 (v ).

Can any second order differential operator with boundary conditions be

transformed in self-adjoint form?
By multiplying with an integrating factor; the operator becomes
self-adjoint in weighted L2r (a, b) space where the weight r (t) is induced
by the integrating factor.
Eigenvalue problem-Self-adjoint form

Let −∞ < a < b < ∞ and let p0 ∈ C 1 ([a, b]) and p1 , p2 ∈ c([a, b]) be
real valued functions and p0 (t) > 0 for all t ∈ [a, b]. For any λ ∈ C,
consider the eigenvalue problem

p0 (t)y 00 (t) + p1 (t)y 0 (t) + p2 (t)y (t) + λy (t) = 0, t ∈ (a, b),

with boundary conditions:

U1 (y ) = α1 y (a) + α2 y 0 (a) = 0, U2 (y ) = β1 y (b) + β2 y 0 (b) = 0,

where α1 , α2 , β1 , β2 are given real numbers with α12 + α22 6= 0 and

β12 + β22 6= 0.
Rt p1 (s)
ds
I Set µ(t) = e a p0 (s)
.Note that µ(t) > 0 for all t ∈ [a, b].
 µ(t)
I Multiplying the ODE with p0 (t) , we get

p2 (t)µ(t) µ(t)
µ(t)y 00 (t)+µ0 (t)y 0 (t)+ y (t)+λ y (t) = 0, t ∈ (a, b).
p0 (t) p0 (t)

I Denoting p(t) = µ(t), q(t) = − p2 p(t)µ(t)

0 (t)
and r (t) = µ(t)
p0 (t) , we rewrite
the BVP as
d
− (p(t)y 0 (t)) + q(t)y (t) = λr (t)y (t), t ∈ (a, b),
dt
U1 (y ) = α1 y (a) + α2 y 0 (a) = 0, U2 (y ) = β1 y (b) + β2 y 0 (b) = 0.
I Note that p(t) > 0 and r (t) > 0 on [a, b].
I Due to the weight function r (t) in the RHS of the above eigenvalue
1
problem, set Lr (y )(t) = r (t) L(y )(t), where

d
L(y (t)) = − (p(t)y 0 (t)) + q(t)y (t), ∀ t ∈ (a, b).
dt
Then Eigenvalue problem: Lr (y )(t) = λy (t) on (a, b) with
boundary conditions U1 (y ) = 0 and U2 (y ) = 0.
I The operator Lr with boundary conditions U1 and U2 is self-adjoint
in L2r (a, b).
Examples.

Reduce into a self-adjoint form:

1. Hermite equation:
y 00 (t) − 2ty 0 (t) + λy (t) = 0, y (0) = 0, y (1) = 0.

2. Euler equation:
t 2 y 00 (t) + ty 0 (t) + λy (t) = 0, y (1) = 0, y (e) = 0.
Weighted L2 -space

Let r ∈ C ([a, b]) with r (t) > 0 for all t ∈ [a, b].

n f : (a, b) → C | f is integrable on (a, b) and o

L2r (a, b) = Rb ,
a
r (t)|f (t)|2 dt < ∞

with the norm

Z b  12
kf kL2r = r (t)|f (t)|2 dt , ∀ f ∈ L2r (a, b),
a

and the inner product

Z b
hf , g iL2r = r (t)f (t)ḡ (t) dt, ∀ f , g ∈ L2r (a, b).
a
Lemma
The operator Lr with boundary conditions U1 and U2 is self-adjoint in
L2r (a, b).
Proof. We need to show that L∗r = Lr in L2r (a, b) with B1 = U1 and
B2 = U2 . Using an integration by parts, for any u, v in C 2 ([a, b],
Rb 1
hv , Lr (u)iL2r = a r (t)v (t) r (t) d
[− dt (p(t)ū 0 (t)) + q(t)ū(t)] dt
Rb
= a v (t)[− dt (p(t)ū 0 (t)) + q(t)ū(t)] dt
d
Rb d
= a [− dt (p(t)v 0 (t)) + q(t)v (t)]ū(t) dt + [p(t)W (ū, v )(t)]bt=a
Rb 1
= a r (t) r (t) d
[− dt (p(t)v 0 (t)) + q(t)v (t)]ū(t) dt + [p(t)W (ū, v )(t)]bt=a

Setting L∗r (v )(t) = 1 d 0

r (t) [− dt (p(t)v (t)) + q(t)v (t)], from the above
identity, we get
hv , Lr (u)iL2r = hL∗r (v ), uiL2r + [p(t)W (ū, v )(t)]bt=a .
Since p(a) 6= 0 and p(b) 6= 0 and u satisfies U1 (u) = 0 = U2 (u),
[p(t)W (ū, v )(t)]bt=a = 0 if and only if U1 (v ) = 0 and U2 (v ) = 0. Thus,
the boundary conditions for the adjoint problem are determined by
B1 = U1 and B2 = U2 . From the expression of L∗r , we already have
L∗r = Lr . Hence, the operator Lr with boundary conditions U1 and U2 is
self-adjoint in L2r (a, b).
Sturm-Liouville’s problem
Regular Sturm-Liouville’s problem:
Under the following assumptions:
1. The interval (a, b) is open and bounded, i.e., a and b are distinct
and finite real numbers.
2. The functions p ∈ C 1 ([a, b]), q ∈ C ([a, b]) and r ∈ C ([a, b]) are real
valued. Further, p(t) > 0 and r (t) > 0 on [a, b].
3. The boundary condtions:

U1 (y ) = α1 y (a) + α2 y 0 (a) = 0, U2 (y ) = β1 y (b) + β2 y 0 (b) = 0,

where α1 , α2 , β1 , β2 are given real numbers with α12 + α22 6= 0 and

β12 + β22 6= 0,
the eigenvalue problem
d
L(y )(t) = − (p(t)y 0 (t)) + q(t)y (t) = λr (t)y (t), ∀t ∈ (a, b),
dt
with homogeneous boundary conditions:

U1 (y ) = 0, U2 (y ) = 0,

is called the Regular Sturm-Liouville’s problem.

Singular Sturm-Liouville’s problem: If the interval (a, b) is unbounded or
if the functions p(t) and r (t) vanish at the boundary t = a and t = b,
then the eigenvalue problem is called singular Sturm-Liouville’s problem.

In particular, we will study the Regular Sturm-Liouville’s problem to

characterize the eigenvalues and eigenfunctions of such problem.

Recall eigenvalue problem: Under the assumptions for Regular

Sturm-Liouville’s problem, if for any λ ∈ C , there exists a non-trivial
solution φ satisfying
1
Lr (φ)(t) = L(φ)(t) = λφ(t), on (a, b),
r (t)

with boundary conditions U1 (φ) = 0 and U2 (φ) = 0, then λ is called the

eigenvalue of the problem and the function φ is called an eigenfunction
corresponding to the eigenvalue λ.
Main Results
Theorem
All the eigenvalues of regular Sturm-Liouville’s problem are real.
Proof. Let λ be an eigenvalue and φ be the corresponding eigenfunction
for i.e., φ 6= 0 on [a, b] satisfies Lr (φ)(t) = λφ(t) on (a, b) and
U1 (φ) = 0 = U2 (φ). Then
hLr (φ), φiL2r = hλφ, φiL2r = λkφk2L2r ,
and
hφ, Lr (φ)iL2r = λ̄kφk2L2r ,
where λ̄ is the complex conjugate of λ.
Further, since Lr with boundary conditions U1 and U2 is self-adjoint in
L2r (a, b), we have hLr (φ), φiL2r = hφ, Lr (φ)iL2r .
Thus, using above results, we finally get
(λ − λ̄)kφk2L2r = 0.
Since kφk2L2 6= 0 (because non-zero function on [a, b]), we get that
r

λ = λ̄,
and so, λ is a real number.
Theorem
Let n 6= m and let λn and λm be two distinct eigenvalues of the Regular
Sturm-Liouville’s problem and φn and φm be the corresponding
eigenfunctions of the problem. Then φn and φm are orthogonal with
respect to L2r (a, b), i.e.,
Z b
r (t)φn (t)φ̄m (t) dt = 0.
a

Proof Using L(φn )(t) = λn φn (t) with U1 (φn ) = 0 = U2 (φn ), we get

hLr (φn ), φm iL2r = λn hφn , φm iL2r .
Further, using Lr (φm )(t) = λm φm (t) with U1 (φm ) = 0 = U2 (φm ), we get
hφn , Lr (φm )iL2r = λ̄m hφn , φm iL2r = λm hφn , φm iL2r ,
since λm is real.
Since Lr with boundary conditions U1 and U2 is self-adjoint in L2r (a, b),
we have hLr (φn ), φm iL2r = hφn , Lr (φm )iL2r .
Using above results, we get that
(λn − λm )hφn , φm iL2r = 0,
and hence hφn , φm iL2r = 0 because λn 6= λm .
Main theorem: Regular Sturm-Liouville’s problem

Let −∞ < a < b < ∞. Let the functions p ∈ C 1 ([a, b]), q ∈ C ([a, b])
and r ∈ C ([a, b]) be real valued. Further, p(t) > 0 and r (t) > 0 on
[a, b]. Consider the regular Sturm-Liouville’s problem, i.e., an eigenvalue
problem
d
L(y )(t) = − (p(t)y 0 (t)) + q(t)y (t) = λr (t)y (t), ∀t ∈ (a, b),
dt
with homogeneous boundary conditions:

U1 (y ) = α1 y (a) + α2 y 0 (a) = 0, U2 (y ) = β1 y (b) + β2 y 0 (b) = 0,

where α1 , α2 , β1 , β2 are given real numbers with α12 + α22 6= 0 and

β12 + β22 6= 0. Then the following properties hold:
1. All the eigenvalues of the Regular Sturm-Liouville’s problem are real.
2. Eigenfunctions corresponding to distinct eigenvalues are mutually
orthogonal in L2r (a, b).
3 Each eigenvalue is simple, i.e., to each eigenvalue, there corresponds
one and only one eigenfunction, up to a constant factor, i.e., the
dimension of the eigenspace corresponding to each eigenvalue is 1.
4 The regular Sturm-Liouville’s problem has an infinite sequence of
real eigenvalues
λ1 < λ2 < · · · < λn < · · · ,
with limn→∞ λn = ∞.
5 The set of normalized eigenfunctions {φn }∞ n=1 , where the
eigenfunction φn corresponds to the eigenvalue λn , form an
orthonormal basis in L2r (a, b). That is any f ∈ L2r (a, b) can be
represented by
∞
X
f = hf , φn iL2r φn , in L2r (a, b).
n=1

Further, if f ∈ C 2 ([a, b]) satisfying U1 (f ) = 0 = U2 (f ), the series

converges absolutely and uniformly on [a, b].
Comments

In the above theorem, Properties 1, 2, 3 can be shown easily.

Properties 4 (existence of eigenvalues) and Properties 5 (orthonormal

basis) can be shown using Green’s function and the spectral theory of
compact operators (Fredholm alternative theorem). We omit their proofs
for this course. However, for particular examples, the existence of the
sequence of eigenvalues can be shown easily. See Assignment problems.

References:
1 E.A. Coddington and N. Levinsons, Theory of ordinary differential
equations, McGraw-Hill, 1955.
2 W.E. Boyce and R.C. Diprima, Elementary differential equations and
boundary value problems, John Wiley, 2003.