Intoduction To Greens Theorem

Download as pdf or txt
Download as pdf or txt
You are on page 1of 25

Green Function

Structure

3.1. Introduction.

3.2. Construction of Green function.

3.3. Construction of

3.4. Non homogeneous ordinary Equation.

3.5.

3.6.

3.7. Check Your Progress.

3.8. Summary.

3.1. Introduction. This chapter contains methods to obtain Green function for a given non-
homogeneous linear second order boundary value problem and reduction of boundary value problem to
Fredholm integral equation with Green function as kernel.
31.1. Objective. The objective of these contents is to provide some important results to the reader like:
(i) Construction of Green function.
(ii) Reduction of boundary value problem to Fredholm integral equation with Green function as kernel.
3.1.2. Keywords. Green function, Integral Equations, Boundary Conditions.
2. Construction of Green function. Consider a differential equation of order n
L(u) = p0 (x) u n p1 (x) u n 1
p2 (x) u n 2
....... p n (x) u = 0 (1)

where the functions p0 (x), p1 (x) , p 2 (x),......., p n (x) are continuous on [a, b], p0(x) 0 on [a, b], and the
boundary conditions
1 2 n 1 n 1
Vk(u) = k u(a) ku (a) ku (a) .......... k u (a)
1 2 n 1 n 1
k u(b) ku (b) ku (b) ............ k u (b) (2)

1, V2 n in u(a), u u n 1 (a), u(b), u u n 1 (b)


are linearly independent.
The homogeneous boundary value problem (1), (2) contains only a trivial solution u(x) 0.
, (2) is the function G (x, ) constructed for any
point ,a< < b satisfying the following properties :

1. G(x, ) is continuous in x for fixed and has continuous derivatives with regard to x upto order
(n 2) inclusive for a x b.
2. Its (n 1)th derivative with regard to x at the point x = has a discontinuity of first kind, the
1
jump being equal to , that is,
p0 (x) x=
n 1 n 1
1
n 1
G ( x, ) n 1
G( x, ) (3)
x x 0
x x 0
p0 ( )

where G x 0
defines the limit of G(x, ) as x from the right and G x 0
defines the limit
of G(x, ) as x from the left.

3. In each of the intervals [a, ) and ( , b] the function G(x, ), considered as a function of x, is a
solution of the equation (1)
L(G) = 0 (4)
4. The function G(x, ) satisfies the boundary conditions (2)

Vk (5)
If the boundary value problem (1), (2) contains only the trivial solution u(x) 0 then the operator L
).
Consider u1(x), u2 n(x) be linearly independent solutions of the equation L(u) = 0. From the
) must have the representation on the intervals [a, )
and ( , b]

G(x, ) = a1u1(x) + a2u2 nun(x), a x<

and G(x, ) = b1u1(x) + b2u2 nun(x), x <b,

where a1, a2 n, b1, b2 n are some functions of .


From the condition 1, the continuity of the function G(x, ) and of its first (n 2) derivatives with
regard to x at the point x = yields

[b1u1( ) + b2u2( n un ( )] [a1u1( ) + a2u2( n un ( )] = 0

[b1u1 ( ) b 2 u 2 ( ) ... b n u n ( )] [a1u1 ( ) a 2u 2 ( ) ... a n u n ( )] 0

[b1u1 ( ) b 2 u 2 ( ) ... b n u n ( )] [a1u1 ( ) a 2u 2 ( ) ... a n u n ( )] 0

... ... ...


[b1u1n 2 ( ) b2u 2n 2 ( ) ... bn u nn 2 ( )] [a1u1n 2 ( ) a 2u 2n 2 ( ) ... a n u nn 2 ( )] 0

1
Also, [b1u1n 1 ( ) b 2 u n2 1 ( ) ... b n u nn 1 ( )] [a1u1n 1 ( ) a 2 u n2 1 ( ) ... a n u nn 1 ( )]
p0 ( )

Assume Ck( ) = bk( ) ak( linear equations in Ck( ) are


obtained
C1u1 ( ) C2 u 2 ( ) ... C n u n ( ) 0

C1u1 ( ) C2 u 2 ( ) ... C n u n ( ) 0

C1u1n 2 ( ) C2u 2n 2 ( ) ... Cn u nn 2 ( ) 0

1
C1u1n 1 ( ) C 2 u n2 1 ( ) ... C n u nn 1 ( ) (6)
p0 ( )

The determinant of the system is equal to the value of the Wronskian W(u1, u2 n) at the point x =
and is therefore different from zero.
From the boundary conditions (2), we have
Vk(u) = Ak(u) + Bk(u) (7)
1 2 n 1 n 1
where Ak(u) = k u(a) ku (a) ku (a) ........ k u (a)
1 2 n 1 n 1
Bk(u) = k u(b) ku (b) ku (b) ........ k u (b)

Using the condition 4, we have


Vk(G) = a1Ak(u1) + a2Ak(u2 nAk(un 1Bk(u1) + b2Bk(u2 nBk(un) = 0,

Since ak = bk ck, so we have


(b1 c1 )A k (u1 ) (b 2 c 2 )A k (u 2 ) ... (b n c n )A k (u n ) b1B k (u1 ) b 2B k (u 2 ) ... b n Bk (u n ) 0

b1Vk (u1 ) b 2 Vk (u 2 ) ... b n Vk (u n ) = c1A k (u1 ) c 2 A k (u 2 ) ... c n A k (u n ) (8)


which is a linear system in the quantities b1, b2 n. The determinant of the system is different from
zero, that is,
V1 (u1 ) V1 (u 2 ) V1 (u n )
V2 (u1 ) V2 (u 2 ) V2 (u n )
0

Vn (u1 ) Vn (u 2 ) Vn (u n )

The system of equations (8) contain a unique solution in b1( ), b2( ), ..., bn( ) and since

ak( ) = bk( ) ck( ), it follows that the quantities ak( ) are defined uniquely.

I. If the boundary value problem (1), (2) is self


G(x, ) = G( , x). The converse is true as well.
II. If at one of the extremities of an interval [a, b], the coefficient of the derivative vanishes. For
example, p0(a) = 0, then the natural boundary condition for the boundedness of the solution x = a is
imposed, and at the other extremity the ordinary boundary condition is specified.
2.1. Particular case. G x, for a given number , for the
second differential equation
Lu x 0 (1)

d d
where L p q (2)
dx dx

Together with the homogenous boundary conditions of the form


du
u 0 (3)
dx
G x, constructed for any point ,a b contains the following
properties:
1. G1 G2 ; it follows that the function G x, is continuous in x for fixed , in particular,
continuous at the point x = .
2. The derivatives of G(which are of finite magnitude) are continuous at every point within the range of
x except at x = where it is continuous so that
1
G2 G1
p

3. The functions G1 and G2 satisfy homogenous conditions at the end points x = a and x = b respectively.
4. The function G1 and G2 satisfy the homogenous equations LG = 0 in their defined intervals except at z
= , that is, LG1 0, x , L G2 0, x .
G x, exists, then the solution of the given differential equation can be
transformed to the relation
b
u x G x, d (4)
a

Consider two linearly independent solutions of the homogeneous equation


L u 0. Let u v1 x and u u2 x be the non-trivial solution of the equation, which satisfy the
homogenous conditions at x = a and x = b respectively.
problem from the conditions III and IV, in the form

C1u1 x , x
G x, (5)
C2 u 2 x , x

where the constant C1 and C2 are chosen in a manner that the conditions I and II are fulfilled. Thus, we
have
C2 u 2 C1u1 0

1
C2u2 C1u1 (6)
p

The determinant of the system (6) is the Wronskian W u1 , u2 evaluated at the point x = for
linearly independent solution u1(x) and u2(x), and, hence it is different from zero W 0

u1 u2
W u1 , u2 = u1 u2 u2 u1 (7)
u1 u2

)}, where C is a constant


independent of , that is,
C
u1 u2 u2 u1 (8)
p

From the system (6), we have


1 1
C1 u2 , C2 u1
C C
Thus the relation (5) reduces to
1
u1 x u2 , x
C
G x, (9)
1
u1 u2 x , x
C
This result breaks down iff C vanishes, so that u1 and u2 are linearly dependent, and hence are each
multiples of a certain non-trivia function U(x). In this case, the function u(x) satisfies the equation L(u)
= 0 together with the end conditions at x = a, x = b.
Converse. The integral equation
b
u(x) = G(x , ) ( ) d (10)
a

where G(x, ) are defined by the relation (9), satisfy the differential equation

L(u) + (x) = 0 (11)


together with the prescribed boundary condition.
We know that
1 x b
u(x) = u1( ) u2 (x) ( ) d u1(x) u2 ( ) ( ) d (12)
C a x

1 x b
u (x) = u2 (x) u1( ) ( ) d u1(x) u2 ( ) ( ) d (13)
C a x

1 x b
u (x) u2 (x) u1( ) ( ) d u1(x) u2 ( ) ( ) d
C a x

1
u2 (x) u1(x) u1(x) u2 (x) ( x) (14)
C
Since L(u) p(x)u (x) p(x)u (x) q(x) u(x)

Thus,

1 x b 1 C
Lu(x)= {Lu2 (x)} u1( ) ( ) d {Lu2 (x)} u2 ( ) ( ) d p(x). ( x)
C a x C p(x)

Again, u1(x) and u2(x) satisfy L (u) = 0, hence the first two terms vanish identically.
So, L u(x) = (x) L u(x) + (x) = 0

Therefore, a function u(x) satisfying (10) also satisfies the differential equation (11)
Again from (12) and (13), we have
u1 (a) b
u(a) = u2 ( ) ( ) d
C a

u1 (b) b
u (a) = u2 ( ) ( ) d
C a

which shows that the function u defined by (11) satisfies the same homogeneous condition at x = a as
the function u1.
Note. Let (x) = r(x) u(x) f(x).
From the differential equation (1), we have
Lu(x) + r(x) u(x) = f(x) (15)
The corresponding Fredholm integral equation becomes
b b
u(x) = G ( x , ) r( ) u( ) d G( x , ) f ( ) d (16)
a a

where G(x,

From (9), it follows that G(x, ) is symmetric but the kernel K(x, ) {= G(x, )r( )} is not
symmetric unless r(x) is a constant.

Consider {r(x)}u(x) V(x) with the assumption that r(x) is non negative over (a, b). This equation
(16) may be expressed as
V(x) b b
G (x , ) r( ) V( ) d G (x , ) f( ) d
r(x) a a

b b f( )
or V(x) = K (x , ) V( ) d K (x , ) d , (17)
a a r( )

where K(x, ) = {r(x)r( )} G(x , ) and hence possesses the same symmetry as G(x, ).

2.2. Example. Construct an integral equation corresponding to the boundary value problem.

d 2u du
x2 x ( x 2 1) u = 0 , (1)
dx 2 dx
u(0) = 0, u(1) = 0 (2)
Solution. The differential equation (1) may be written as
d du 1
x x u = 0.
dx dx x

d du u
or x xu = 0.
dx dx x

Comparing with the equation (15), we have


1
p = x, q = ,r=x (3)
x
The general solution of the homogeneous equation

d du u
L(u) = 0 x = 0 is given by
dx dx x
1
u(x) = C1x + C2
x

Consider u = u1(x) and u = u2(x) be the non trivial solutions of the equation, which satisfy the
conditions at x = 0 and x = 1 respectively then
1
u1(x) = x and u2(x) = x.
x
The Wronskian of u1(x) and u2(x) is given by

u1(x) u2 (x) 1 1 2
W[u1(x), u2(x)] = = x 1 x
u1(x) u2 (x) x2 x x

2
So, u1(x)u2 (x) u2 (x)u1(x) = C= 2
x
Thus from the relation (19), we have
1 x 2
(1 ), x< ,
2
G(x, )= (4)
1 2
(1 x ) , x > ,
2 x
Therefore, from (16), the corresponding Fredholm integral equation becomes
1
u(x) = G(x , ) u( ) d ) is defined by the relation (4).
0

2.3. Example.

d 4u
= 0 with the conditions u(0) = u (0) = 0, u(1) = u (1) = 0.
dx 4

d 4u
=0 (1)
dx 4
We notice that the boundary value problem contains only a trivial solution. The fundamental system of
solutions for the differential equation (1) is
u1(x) = 1, u2(x) = x, u3(x) = x2, u4(x) = x3 (2)
Its general solution is of the form
u(x) = A + Bx + Cx2 + Dx3,
where A, B, C, D are arbitrary constants. The boundary conditions give the relations for determining the
constants A, B, C, D :
u(0) = 0 A = 0, u (0) = 0 B=0
u(1) = 0 A + B + C + D = 0, u (1) = 0 B + 2C + 3D = 0
A = B = C = D = 0.
Thus the boundary value problem has only a zero solution u(x) 0 and hence we can construct a

C ) must have the


representation on the interval [0, ) and ( , 1].

a1 .1 a2 .x a3 .x 2 a4 .x3 , 0 x
G (x, )= (3)
b1 .1 b2 .x b3 .x 2 b4 .x 3 , x 1

where a1, a2, a3, a4, b1, b2, b3, b4 are the unknown functions of .

Consider Ck = bk( ) ak( (4)

The system of linear equations for determining the functions Ck( ) become

2 3
C1 + C2 + C3 + C4 =0
Solution. The differential equation is given by
2
C2 + 2C3 + 3C4 =0

2C3 + 6C4 =0

6C4 = 1
1 1 1 2 1 3
C4( ) = , C3( ) = , C2( ) = , C 1( ) = (5)
6 2 2 6
:
G(0, ) = 0, G x (0, )=0

G(1, ) = 0, G x (1, )=0

The relations reduce to


a1 = 0, a2 = 0
b1 + b2 + b3 + b4 = 0
b2 + 2b3 + 3b4 = 0 (6)
From the relation (4), (5) and (6), we have
1 3
C1 = b1( ) a1( ) b1 ( ) =
6
1 2
or C2 = b2( ) a2( ) b2 ( ) =
2
1 3 1 2 1 2
or b3 + b 4 = , , 2b3 +3b4 =
6 2 2
1 2 1 3 1 3 2
b4 ( ) = and b3( ) =
2 3 2
or C3( ) = b3( ) a3( )

1 3 2 1
a3( ) = b3( ) C3( ) = +
2 2
and C4( ) = b4( ) a4( )

1 2 1 3 1
a4( ) = b4( ) C4( ) =
2 3 6
Substituting the value of the constants a1, a2, a3, a4, b1, b2, C3, C4
G(x, ) is obtained as

1 2 1 3 1 1 1
x2 2 3
x3 , 0 x
2 2 6 2 3
G(x, )=
1 3 1 2 1 3 2 1 1
x x2 2 3
x3 , x 1
6 2 2 2 3

The expression G(x, ) may be transformed to

1 1 3 1 1 2 1 3
G(x, )= x x2 x 2
x x 3
, x 1
2 2 6 2 3

G(x, ) = G(

d 2u du
2.4. Example. = 0 with the conditions u(x) is
dx 2 dx
bounded as x 0, u(1) = u (1) , 0.

d 2u du
Solution. The differential equation is given by x =0 (1)
dx 2 dx

d 2u dx 2 1
or dx dx
du dx x

du
or log = logx + logA
dx
du A
or =
dx x
or u(x) = Alogx + B (2)
The conditions u(x) is bounded as x 0 and u(1) = u (1) , 0 has only a trivial solution u(x) 0,
)

Consider the function G x, as:

a1 a2 log x , 0 < x
G(x, )= (3)
b1 b2 log x , x 1

where a1, a2, b1, b2 are unknown functions of .

Consider Ck = bk( ) ak(

From the continuity of G(x, ) for x = , we obtain

b1 + b2log a1 a2 log =0

1
and the jump G x (x, ) at the point x = is equal to so that

1 1 1
b2 . a2. =

Putting C1 = b1 a1, C2 = b2 a2 (4)


C1 + C2 log = 0, C2 = 1.

Hence C1 = log and C2 = 1 (5)

The boundedness of the function G(x, ) as x 0 gives a2 = 0

Also, G(x, )= G x (x, ), b1 = b2

a1 = ( +log ), a2 = 0, b1 = 1, b2 =

Substituting the value of the constants a1, a2, b1, b2


( log ) , 0 < x
G(x, )= .
(1 log x) , x 1
2.6. Article. If u(x) has continuous first and second derivatives, and satisfies the boundary value
d 2u
problem u = 0 with u(0) = u(l) = 0 then u(x) is continuous and satisfies the homogeneous linear
dx 2
1
integral equation u(x) = G(x , ) u( ) d .
0

Solution : The differential equation may be written as

d 2u d 2u
u=0 u (1)
dx 2 dx 2
By integrating with regard to x over the interval (0, x) two times, we obtain
du x
= u( ) d C
dx 0

x
or u(x) = (x ) u( ) d Cx D (2)
0

where C and D are the integration constants, to be determined by the boundary conditions.
u(0) = 0 D=0
l
u(l) = 0 (l ) u( ) d Cl = 0
0

l
C = (l ) u( ) d
l 0

Substituting the value of the constants C and D in (2), we have


x l
u(x) = (x ) u( ) d x(l ) u( ) d
0 l 0

x x l
or u(x) = (x ) u( ) d x(l ) u( ) d x(l ) u( ) d
0 l 0 l x

x l x
or u(x) = (l x) u( ) d (l ) u( ) d
0 l x l

l
or u(x) = G(x , ) u( ) d
0

(l x) , x >
l
where G(x, )= .
x
(l ) , x <
l
Consider a differential equation of order n
L(u) h = h(x) (1)
with Vk (2)

where L(u) p0 ( x)u n ( x) p1( x)u n 1( x) ... pn ( x)u( x) (3)


1 n 1 n 1 1 n 1 n 1
and Vk (u) k u ( a) k u '(a) ... k u (a) ... k u (b) k u '(b) ... k u (b) ... (4)

where the linear forms V1, V2 n in u(a), u u n 1 (a), u(b), u u n 1 (b) are linearly
independent, h(x) is a given continuous function of x, is some non zero numerical parameter.

For h(x) 0, the equation (1) reduces to homogeneous boundary value problem
L(u) = u,
Vk (5)
Those values of for which the boundary value problem (5) has non trivial solutions u(x) are called the
eigenvalues. The non trivial solutions are called the associated eigen functions.
If the boundary value problem
L(u) = 0,
Vk (6)
), then the boundary value problem (1) and (2) is equivalent to the
Fredholm integral equation
b
u(x) = G(x , ) u( ) d f(x) (7)
a

b
where f(x) = G(x , ) h( ) d (8)
a

In particular, the homogeneous boundary value problem (5) is equivalent to the homogeneous integral
equation
b
u(x) = G(x , ) u( ) d (9)
a

Since G(x, ) is a continuous kernel, therefore the Fredholm homogeneous integral equation of second
kind (9) can have at most a countable number of eigen values 1, 2 n which do not have a finite
limit point. For all values of different from the eigen values, the non-homogeneous equation (7) has a
solution for any continuous function f(x). Thus the solution is given by
b
u(x) = R(x , ; ) f( ) d f( ) (10)
a

where R(x, ; ) is the resolvent kernel of the kernel G(x, ). The function R(x, ; ) is a
meromorphic function of for any fixed values of x and in [a, b]. The eigen values of the
homogeneous integral equation (9) may by the pole of this function.

d 2u
3.1. Example. Reduce the boundary value problem u = x, u(0) = u( 2) = 0, to an integral
dx 2
.
Solution. Consider the associated boundary value problem

d2u
0 (1)
dx 2
whose general solution is given by u(x) = Ax + B

The boundary conditions u(0) = 0, u( 2 ) = 0 yields only the trivial solution u(x) 0. Therefore, the
) exists for the associated boundary value problem

a 1x a 2 , 0 x<
G(x, )= (2)
b1x b 2 , x 2

) must satisfy the following properties :

(I) The function G(x, ) is continuous at x = , that is,

b1 + b2 = a1 + a2

(b1 a1) + (b2 a2) = 0 (3)

1
(II) The derivative G(x, ) has a discontinuity of magnitude at the point x = ,
p0 ( )

G G
that is, 1 b1 a1 = 1 (4)
x x = +0 x x= 0

(III) The function G(x, ) must satisfy the boundary conditions

G(0, )=0 a2 = 0 (5)

G( 2, )=0 b1 + b2 = 0 (6)
2
Solving the equations (3), (4), (5) and (6), we have
2 2
a1 = 1 , a2 = 0, b2 = , b1 = .

) is obtained

2
1 x , 0 x<
G(x, )= (7)
2x
1 , x 2

) given by the relation (7) as the kernel of the integral equation,


we obtain the integral equation for u(x) :
2 2
u(x) = f(x) G(x , ) u( ) d , where f(x) = G(x , ) d
0 0

x 2x 2
2 2
or f(x) = 1 d 1 x d
0 x

2
1 2x 3 1 2 2 3
or f(x) = 1 x x
3 2 3 x

2
1 3 2 4 x 1 3 2 4
or f(x) = x x x x
3 3 24 2 3
2
x3
or f(x) = x
24 6
Thus, the given boundary value problem has been reduced to an integral equation
2
2 1 3
u(x) + G(x , ) u( ) d = x x .
0 24 6
4. Non homogeneous ordinary Equation. The boundary value problem associated with a non
homogenous ordinary differential equation of second order is

d2y dy
Ly A0(x) 2
+ A1(x) + A2(x) y = f(x), a < x < b (1)
dx dx

1 y(a) 2 y (a) = 0
with boundary conditions (2)
1 y(b) 2 y (b) = 0

where 1, 2, 1 and 2 are constants.

4.1. Self Adjoint Operator. The operator L is said to be self adjoint if for any two functions say
u(x) and v(x) operated on L, the expression (vLu uLv) dx is an exact differential that is, there exist a
function g such that dg = (vLu uLv) dx.

Function Method.
associated with non homogeneous ordinary or partial differential equation . Here we shall show that a
B.V.P. will be reduced to a Fredholm integral equation whose kernel is Gre
using a special type of B.V.P. namely Sturm
4.3. Theorem. Show that the differential operator L of the Sturm
problem (S.L.B.V.P.)
d dy
Ly = r(x) + q(x) p(x) y(x) = 0 (1)
dx dx

1 y(a) 2 y (a) = 0
with (2)
1 y(b) 2 y (b) = 0

where , , 2 and 2 are constants is self adjoint.

Proof. Let u and v be two solutions of the given S.L.B.V.P. then


d du
Lu = r(x) + g(x) p(x) u(x) = 0
dx dx

d dv
and Lv = r(x) + q(x) p(x) v(x) = 0
dx dx

So,

d du d dv
vLu uLv = v r(x) + q(x) p(x) u(x)v u r(x) q(x) p(x) v(x)u
dx dx dx dx

d du d dv
=v r(x) u r(x)
dx dx dx dx

d du du dv d dv dv du
= v r(x) r(x) u r(x) r(x)
dx dx dx dx dx dx dx dx

d du d dv
= r(x) v(x) r(x) u(x)
dx dx dx dx

d du dv d du dv dg
= r(x) v(x) r(x) u(x) = r(x) v(x) u(x) =
dx dx dx dx dx dx dx

du dv
where g = r(x) v(x) u(x) . Then, (vLu uLv) dx = dg
dx dx

Hence operator in equation (1) is self adjoint.


Consider the non homogeneous differential equation
d du
Lu = r(x) + q(x) p(x) u(x) = f(x) (1)
dx dx

subject to boundary condition:

1 u(a) 2 u (a) = 0
(*)
1u(b) 2 u (b) = 0

b
u(x) = G(x , ) f( ) d (**)
a

where G(x,

Solution. Let v1(x) and v2(x) be two linearly independent solution of the homogeneous differential
equation.
d du
Lu = r(x) + q(x) p(x) u(x) = 0 (2)
dx dx

Then the general solution of (2) by the method of variation of parameters is


u(x) = a1(x) v1(x) + a2(x) v2(x) (3)
where the unknown variables a1(x) and a2(x) are to be determined. We assume that neither the solution
v1(x) nor v2(x) satisfy both the boundary conditions at x = a and x = b but the general solution u(x)
satisfies these conditions.
Now, we differentiate (3) w.r.t. x and obtain
u (x) = a1v1 a1v1 a 2 v 2 a 2 v 2 (4)

Let us equate to zero the terms involving derivatives of parameter, that is,
a1 (x)v1 (x) a 2 (x)v 2 (x) = 0 (5)

which yields
u (x) = a1 (x) v1 (x) a 2 (x) v 2 (x) (6)

Putting the values of u(x) and u (x) from (3) and (6) respectively in equation (1), we obtain

d
Lu = r(x) a 1v1 a 2 v 2 + q(x) p(x) (a1v1 + a2v2) = f(x)
dx
d d
or a1 (r v1 ) v1 (q p) a2 (r v 2 ) v 2 (q p) + a1v1 a 2 v2 r(x) = f(x) (7)
dx dx

Since v1 and v2 are solutions of homogeneous equation (2), so by (7), we get


a1v1 a 2 v2 r(x) = f(x)

f(x)
a1 (x) v1 (x) a 2 (x) v 2 (x) = (8)
r(x)
Equations (5) and equation (8) can be solved to get
f(x) v2 (x) f(x) v1 (x)
a1 (x) and a 2 (x) (9)
r(x) v2 v1 v1v2 r(x) v2 v1 v1v2

Now the operator L is exact and we have proved that


d
v2 L v 1 v1 L v 2 = r(x) v 2 v1 v1v 2 (10)
dx
Since v1 and v2 are solutions of Sturm Liouville homogeneous differential equation so that Lv1 = 0 and
Lv2 = 0 and thus equation (10) gives
d
r(x) v 2 v1 v1v 2 =0
dx
r(x) v2 v1 v1v2 = constant = (say) (11)

Thus, equation (9) becomes

f(x) v 2 (x) f(x) v1 (x)


a 1 (x) = and a 2 (x) = (12)

Integrating (12), we get


x
1
a1(x) = f( ) v 2 ( ) d (13)
c1

x
1
and a2(x) = f( ) v1 ( ) d (14)
c2

where c1 and c2 are arbitrary constants to be determined from the boundary condition on a1(x) and a2(x).
These conditions are to be imposed in accordance with our earlier assumption that v 1(x) and v2(x) does
not satisfy boundary conditions but the final solution u(x) satisfies boundary conditions in equation (*).
So, that

1 u(a) 2 u (a) = 0 (15)

1 u(b) 2 u (b) = 0 (16)


Using (3) and (6) in equation (15), we obtain
0= 1 u(a) 2 u (a)

= 1 a1 (a)v1 (a) a 2 (a)v 2 (a) + 2 a1 (a)v1 (a) a 2 (a)v2 (a)

= a1 (a) 1 1 v (a) v (a) + a 2 (a)


2 1 v (a)
1 2 2 v2 (a)

Let us now assume that v2(x) satisfies first boundary condition of (*) but v1(x) does not satisfy it, then
v (a)
1 2 2 v 2 (a) = 0
v (a)
1 1 2 v 2 (a) 0

so that a1(a) v (a)


1 1 v (a) = 0
2 1 a1(a) = 0

Using this condition in (13), we get


a
1
0 = a1(a) = f( ) v 2 ( ) d which is satisfied when c1 = a
c1

Thus, the solution in (13) is :


x
1
a1(x) = f( ) v 2 ( ) d (17)
a

Similarly, using (3) and (6) in (16), we obtain c2 = b and the solution in (14) is :
x b
1 1
a2(x) = f( ) v1 ( ) d = f( ) v1 ( ) d (18)
b x

The final solution of the non homogeneous B.V.P. is


u(x) = a1(x) v1(x) + a2(x) v2(x)
x b
1 1
= v1 (x) f( ) v2 ( ) d v 2 (x) f( ) v1 ( ) d
a x

x b b
v1 (x) v2 ( ) v 2 (x) v1 ( )
= f( ) d f( ) d = G(x , ) f ( ) d
a x a

1
v1 (x) v 2 ( ) x b
where G(x, )=
1
v 2 (x) v1 ( ) a x
5.1. Theorem. The Green function G(x, ) is symmetric in x and , that is, G(x, ) = G( , x).

Proof. Interchanging x and in G(x, ) defined above :

1
v1 ( ) v2 (x) x
G(x, )= = G( , x).
1
v1 (x) v2 ( ) x

5.2. Theorem. The function G(x, ) satisfies the boundary conditions given in equation (*).
Proof. Consider

v 2 (a)v1 ( ) v2 (a)v1 ( )
1 G(a, )+ 2 G (a, ) = 1 + 2

1
= v (a)
1 2 2 v 2 (a) v1 ( )

1
= 0 v1 ( ) = 0 x b

v1 (b)v2 ( ) v1 (b)v2 ( )
Again, 1G(b, )+ 2G (b, ) = 1 + 2

1
= v (b)
1 1 2 1 v (b) v 2 ( )

1
= 0 v2 ( ) = 0 ,a x.

G
5.3. Theorem. has a jump discontinuity at x = , given by
x
G G 1
=
x x= x x= r( )

where r(x) is the co efficient of u (x) in equation (1).

G G 1 1
Proof. We have = v1 (x) v 2 ( ) x=
v 2 (x) v1 ( ) x=
x x= x x=
(x > ) (x < )
1
= v1 ( ) v 2 ( ) v 2 ( ) v1 ( )

1 1
= = [By equation (11)]
r( ) r( )

Function. Consider the general boundary value


problem

d2y dy
A0(x) 2
+ A1(x) + A2(x) y + p(x) y = h(x) (1)
dx dx
with boundary conditions: y(a) = 0, y(b) = 0. (2)
its kernel.
To make the above operator in (1) as a self adjoint operator, we shift the term p(x)y to the right side
r(x)
and then divide it by .
A 0 (x)

b
y(x) = G(x , ) f( ) d where f(x) = h(x) p(x) y(x)
a

b
or y(x) = G(x , ) h( ) p( ) y d
a

b b
= G(x , ) h( ) d G(x , ) p( ) y( ) d
a a

b
= K(x) + G(x , ) p( ) y( ) d (3)
a

b
where K(x) = G(x , ) h( ) d (4)
a

This is a Fredholm integral equation of the second kind with kernel K(x, ) = G(x, ) p( ) and a non
homogeneous term K(x).

Now, multiplying equation (3) by p(x) , we get


b
p(x) y(x) = p(x) K(x) + p(x) p( ) G(x, ) p( ) y( ) d
a

Let us use, u(x) = p(x) y(x) and g(x) = p(x) K(x)


b
Then, u(x) = g(x) + p(x) p( ) G(x, ) u( ) d (5)
a

Here the kernel of Fredholm integral equation of second kind is symmetric that is,

K(x, )= p(x) p( ) G(x, ) (6)

is symmetric, since G(x, ) is symmetric.

Remark : We had obtained the required result in equation (3). We had proceed to obtain equation (5)
just to get the kernel in more symmetric form.

3.8. Summary. In this chapter, we discussed various methods to construct Green function for a given
non-homogeneous linear second order boundary value problem and then boundary value problem can be
reduced to Fredholm integral equation with Green function as kernel and hence can be solbed using the
methods studied in the previous chapter.
Books Suggested:
1. Jerri, A.J., Introduction to Integral Equations with Applications, A Wiley-Interscience
Publication, 1999.
2. Kanwal, R.P., Linear Integral Equations, Theory and Techniques, Academic Press, New York.
3. Lovitt, W.V., Linear Integral Equations, McGraw Hill, New York.
4. Hilderbrand, F.B., Methods of Applied Mathematics, Dover Publications.

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy