Sturm-Liouville Eigenvalue Problems and Generalized Fourier Series

Examples of Regular Sturm-Liouville Eigenvalue Problems We will now look at examples of regular Sturm-Liouville differential equations with various combinations of the three types of boundary conditions Dirichlet, Neumann and Robin. All of the examples are special cases of the Sturm-Liouville differential equation L( y ) + w( x ) y = 0 where L is the Sturm-Liouville operator L( y ) = p( x ) + q( x ) y y x x We focus on three types of differential equations: Euler, Cauchy-Euler, and Bessel. Each one of these differential equations is characterized by a different set of Sturm-Liouville coefficients p( x ), q( x ), and w( x ). Subject to a particular set of boundary conditions, we will: 1. generate the eigenvalues, the corresponding eigenfunctions, and the statement of orthonormality. 2. then provide an example of a generalized Fourier series expansion of a given function in terms of the particular eigenfunctions. In solving for the allowed eigenvalues and corresponding eigenfunctions, we would ordinarily consider three possiblities for values of : < 0 , = 0 , and > 0. However, to make our task a little simpler, we will not consider the case for < 0 because it can be shown, by way of the Rayleigh quotient, that, for the particular Sturm-Liouville problems we will be considering, must be greater than or equal to zero.
2 In this worksheet we will look at the Euler operator Lu = u on the interval (0,b) and the BC x x are 1/ Dirichlet condition at both ends. 2/ Mixed BC: Dirichlet at 0 and Neumann at b.

EXAMPLE 2.5.1: Consider the Euler operator with Dirichlet conditions. We seek the eigenvalues and corresponding orthonormal eigenfunctions for the Euler differential equation [Sturm-Liouville type for p( x ) = 1 , q( x ) = 0, w( x ) = 1] over the interval I = { x | 0 < x < b }. The boundary conditions are type 1 at the left and type 1 at the right end points.

Page 1

Euler differential equation

Boundary conditions y( 0 ) = 0 and y( b ) = 0 SOLUTION: We consider two possibilities for values of . We first consider = 0. The system basis vectors (or fundamental solutions) are > restart:y1:= x->1;y2:=x->x; y1 := 1 y2 := x x General solution is then > y:=x->C1*y1(x)+C2*y2(x); y := x C1 y1( x ) + C2 y2( x ) Substitution into the boundary conditions yields the system to determine C1,C2 > {y(0)=0, y(b)=0}; solve(%,{C1,C2}); { C1 = 0, C1 + C2 b = 0 } { C2 = 0, C1 = 0 } We obtain the trivial solution y( x ) = 0, thus = 0 is not an eigenvalue. We next consider > 0. We set = . The system basis vectors are
2

> y1:=x->sin(mu*x);y2:=x->cos(mu*x); y1 := x sin( x ) y2 := x cos( x ) General solution > y:=x->C1*y1(x)+C2*y2(x); y := x C1 y1( x ) + C2 y2( x ) Substituting into the boundary conditions, we get > {y(0)=0, y(b)=0};
Page 2

{ C2 = 0, C1 sin( b ) + C2 cos( b ) = 0 } From the first equation, C2 = 0. We must look for such that y( x ) is not identically zero. The only nontrivial solutions to the above occur when C2 = 0, C1 is arbitrary and satisfies the following eigenvalue equation Let's ask Maple to find > sin(mu*b)=0; solve(%,mu); sin( b ) = 0 0 Maple did not give us all of possible solutions. In fact, sin( x ) = 0 if x = n for integers n . Thus, takes on the values > mu:=n->n*Pi/b; := n for n = 1, 2, 3, ... . Allowed eigenvalues are n = n
2

n b

> lambda[n]:=(n*Pi/b)^2; n := Non-normalized eigenfunctions are > Phi:=(n,x)->sin(mu(n)*x); := ( n, x ) sin( ( n ) x ) Normalization Evaluating the norm from the inner product of the eigenfunctions with respect to the weight function w( x ) = 1 over the interval yields > w(x):=1:sqrt(int(Phi(n,x)^2*w(x),x=0..b)); 1 2 b ( cos( n ) sin( n ) + n ) n n2 b2
2

Substitution of the eigenvalue equation simplifies the norm > e_norm[n]:=radsimp(subs({sin(n*Pi)=0,cos(n*Pi)=(-1)^n},%));

Page 3

2 b 2 Orthonormal eigenfunctions is then obtained by dividing the non-normalized eigenfunctions by their norm > phi:=(n,x)->Phi(n,x)/e_norm[n]; ( n, x ) e_normn

e_normn :=

:= ( n, x ) Statement of orthonormality

> Int(phi(n,x)*phi(m,x)*w(x),x=0..b)=delta(n,m); n x m x sin 2 sin b b dx = ( n, m ) b e_norm m

Fourier coefficients > F[n]:=int(f(x)*phi(n,x)*w(x),x=0..b); n x f( x ) sin 2 b Fn := dx b

Generalized Fourier series expansion > f_series:=x->Sum(F[n]*phi(n,x),n=1..infinity);

f_series := x

n=1

Fn ( n, x )

This is the generalized series expansion of f( x ) in terms of the "complete" set of eigenfunctions for the particular Sturm-Liouville operator and given boundary conditions over the interval. DEMONSTRATION: Develop the generalized series expansion for f( x ) = x over the interval I = { x |
Page 4

0 < x < 1 } in terms of the above eigenfunctions. We assign the system values > a:=0;b:=1;f:=x->x; a := 0 b := 1 f := x x SOLUTION: We evaluate the Fourier coefficients > eval(int(f(x)*phi(n,x)*w(x),x=a..b)); 2 ( sin( n ) + n cos( n ) ) n2
2

> F[n]:=subs({sin(n*Pi)=0,cos(n*Pi)=(-1)^n},%); 2 ( -1 )n n

Fn :=

> Series:=x->sum(F[n]*phi(n,x),n=1..infinity);

Series := x First five terms of expansion

n=1

Fn ( n, x )

> Part_Series:=(m,x)->sum(F[n]*phi(n,x),n=1..m); Part_Series := ( m, x ) > Part_Series(2,.5); .6366197722 > plot({Part_Series(5,x),f(x)},x=0..b,thickness=3);

n=1

Fn ( n, x )

Page 5

Figure 2.4 The curves of Figure 2.4 depict the function f( x ) and its Fourier series approximation in terms of the orthonormal eigenfunctions for the particular operator and boundary conditions given earlier. Note that f( x ) satisfies the given boundary conditions at the left but fails to do so at the right end point. The convergence is pointwise. EXAMPLE 2.5.2: Consider the Euler operator with Dirichlet and Neumann conditions. We seek the eigenvalues and corresponding orthonormal eigenfunctions for the Euler differential equation [Sturm-Liouville type for p( x ) = 1 , q( x ) = 0, w( x ) = 1] over the interval I = { x | 0 < x < b }. The boundary conditions are type 1 at the left and type 2 at the right. Euler differential equation
2 y( x ) + y( x ) = 0 x x

Boundary conditions
Page 6

y( 0 ) = 0 and yx( b ) = 0 SOLUTION: We consider two possibilities for values of . We first consider = 0. The system basis vectors are > restart:y1:=x->1;y2:=x->x; y1 := 1 y2 := x x General solution > y:=x->C1*y1(x)+C2*y2(x); y := x C1 y1( x ) + C2 y2( x ) Substituting the boundary conditions yields > subs(x=0,y(x))=0; C1 = 0 > subs(x=b,diff(y(x),x))=0; C2 = 0 The only solution to the above is the trivial solution. We next consider > 0. We set = .
2

The system basis vectors are > y1:=x->sin(mu*x);y2:=x->cos(mu*x); y1 := x sin( x ) y2 := x cos( x ) General solution > y:=x->C1*y1(x)+C2*y2(x); y := x C1 y1( x ) + C2 y2( x ) Substituting the boundary conditions yields > eval(subs(x=0,y(x)))=0; C2 = 0 > eval(subs(x=b,diff(y(x),x)))=0;

Page 7

C1 cos( b ) C2 sin( b ) = 0 The only nontrivial solutions occur when C2 = 0, C1 is arbitrary, and satisfies the following eigenvalue equation > cos(mu*b)=0; cos( b ) = 0 Thus, takes on values > mu[n]:=(2*n-1)*Pi/(2*b); n := for n = 1, 2, 3, ... Allowed eigenvalues are n = n > lambda[n]:=mu[n]^2; n := Non-normalized eigenfunctions are > Phi:=(n,x)->sin(mu[n]*x); := ( n, x ) sin( n x ) Normalization Evaluating the norm from the inner product of the eigenfunctions with respect to the weight function w( x ) = 1 over the interval yields > w(x):=1:e_norm[n]:=sqrt(int(Phi(n,x)^2*w(x),x=0..b)); 1 2 b ( 2 sin( n ) cos( n ) + 2 n ) (2 n 1) 1 ( 2 n 1 )2 4 b2
2 2

1 (2 n 1) 2 b

e_normn :=

Substitution of the eigenvalue equation simplifies the norm > e_norm[n]:=radsimp(subs({sin(n*Pi)=0,cos(n*Pi)=(-1)^n},%)); 1 2

e_normn :=

Page 8

Orthonormal eigenfunctions > phi:=(n,x)->Phi(n,x)/e_norm[n]; ( n, x ) e_normn

:= ( n, x ) Statement of orthonormality

> Int(phi(n,x)*phi(m,x)*w(x),x=0..b)=delta(n,m); 1 ( 2 n 1 ) x sin 2 sin( m x ) 2 b dx = ( n, m ) b e_norm m

Fourier coefficients > F[n]:=int(f(x)*phi(n,x)*w(x),x=0..b); 1 ( 2 n 1 ) x f( x ) sin 2 2 b Fn := dx b

Generalized Fourier series expansion > f_Series:=x->Sum(F[n]*phi(n,x),n=1..infinity); This is the generalized series expansion of f( x ) in terms of the "complete" set of eigenfunctions for the particular Sturm-Liouville operator and boundary conditions over the interval. f_Series := x

n=1

Fn ( n, x )

DEMONSTRATION : Develop the generalized series expansion for f( x ) = x over the interval I = { x | 0 < x < 1 } in terms of the preceding eigenfunctions. We assign the system values

Page 9

> a:=0;b:=1;f:=x->x; a := 0 b := 1 f := x x SOLUTION: We evaluate the Fourier coefficients > F[n]:=int(f(x)*phi(n,x)*w(x),x=a..b); 2 ( 2 cos( n ) + 2 sin( n ) n sin( n ) ) ( 2 n 1 )2
2

Fn := 2

> F[n]:=subs({sin(n*Pi)=0,cos(n*Pi)=(-1)^n,sin((2*n+1)/2*Pi)=(-1)^n, cos((2*n+1)/2*Pi)=0},F[n]); 2 ( -1 )n ( 2 n 1 )2

Fn := 4

> f_Series:=Sum(F[n]*phi(n,x),n=1..infinity); 1 ( -1 )n sin ( 2 n 1 ) x 2 8 f_Series := 2 2 (2 n 1) n=1

Partial expansion > Part_Series:=(m,x)->sum(F[n]*phi(n,x),n=1..m); Part_Series := ( m, x )

n=1

Fn ( n, x )

> plot({Part_Series(3,x),f(x)},x=0..b,thickness=3);

Page 10

Figure 2.5 The two curves of Figure 2.5 depict the function f( x ) and its Fourier series approximation in terms of the orthonormal eigenfunctions for the particular operator and boundary conditions given earlier. Note that f( x )satisfies the given boundary conditions at the left but fails to do so at the right end point. The convergence is pointwise. >

Page 11