You are on page 1of 24

1

Chapter One
Finite Integral Transforms
1-1 Introduction
The problem of representing an arbitrary function in a given
interval as a linear combination of members of some infinite set
of functions, say, bears a strong formal
resemblance to the problem of expressing a vector in n-dimensional space
as a linear combination of n linearly independent vector.
Thus in the case where A is a vector, we represent A in the form
n
A   ak e k (1-1)
k 1

When A is a scalar function of x, the desired representation is



A (x )   ak e k (x ) (1-2)
k 1
In each case the problem is to calculate the expansion coefficients ak. In
the vector case (1-1) we have seen that the problem can be solved, when
is an orthonormal basis, by taking the scalar product of A with each ek
in succession. When A(x) is infinitely differentiable in the interval a≤x≤b,
one solution of case (1-2) is provided by choosing

e k (x )  x k 1 (1-3)
So that
e (x ),e (x ),.....,e (x ),.......  1, x , x
1 2 n
2
, x 3 ,......., x n 1,...... (1-4)
Then

A (x )   ak x k (1-5)
k 0
represents A( x) as a power series in x. The coefficients ak are then
obtained from the familiar Taylor formula
1 dk
ak  k
A( x) | x 0 )1 - 6 (
k! dx
One must still prove, however, that the series on the right-hand side of
Eq.(1-5) converges to the function on the left-hand side for each x in the
interval a≤x≤b
Frequently, the power-series expansion fails because A(x) or some of its
derivatives has a finite discontinuity at a finite number of points in the
interval a≤x≤b. In such a case it is often possible to divide up the interval
2

a≤x≤b into subintervals such that in each subinterval A(x) is continuous.


Such an A(x) is called piecewise-continuous, and if in addition A(x) has a
piecewise-continuous first derivative, then A(x) is called piecewise-
smooth. We can obtain a representation of a piecewise-continuous function
of the form (1-2) by employing the properties of an orthogonal set of
functions

1-2 Orthogonal Functions


Suppose that A(x) is a piecewise-continuous function in the interval
a≤x≤b. We then postulate a development of the form

A( x )   ak ek ( x ) )1-7)
k 1

where each ek( x) is in general a complex-valued function of the real


*
variable x. Next, we multiply both sides of Eq. (1-7) by e j (x ) and form the
integral with respect to x, obtaining
b  b

e A( x)dx   a k  e j ( x)ek ( x)dx


* *
j (1-8)
a k 1 a
*
Here e j (x ) is the complex conjugate of e j (x ) .We shall put off for the
time being the justification of interchanging the order of integration and
summation in going from Eq.(1-7) to Eq.(1-8). We can solve Eq.(1-8) for
aj if

e ( x)ek ( x)dx  k  jk
*
j )1-9)
a

for then
b 

e A( x)dx   a k k  jk  a j  j
*
j (1-10)
a k 1

so that
b
1
aj   e j * A( x)dx (1-11)
j a
It is customary to alter the definitions and notation slightly when
discussing integral transforms. Thus one writes
b
A ( j )   e j *A (x )dx (1-12)
a
3

and refers to Eq.(1-12) as the finite integral transform of A(x). The


postulated expansion (1-7) is called the "inversion theorem" corresponding
to the finite integral transform (1-12). Since
A( j )
aj  (1-13)
j
the inversion theorem for Eq.(1-12) becomes

A(k )
A( x)   ek ( x) (1-14)
k 1  k
Two complex-valued functions such as ej(x) and ek(x) are said to be
orthogonal over the interval a≤x≤b if the relations


b b

e (x )e k (x )dx   e j (x )e k * (x )dx  k  jk  0, j k
*
j  j , j k (1-15)
a a

hold for j≠k. Thus Eq.(1-9) specifies that ej(x) is orthogonal to ek(x) for j≠k.
Since j and k are arbitrary indices, it follows that {ek(x);k =1,2, . . .} is an
orthogonal set of functions but if
b b
e j (x)e k (x)w (x)dx = e j (x)e k * (x)w(x)dx =  jk
*
(1-16)
a a

then {ek( x) ; k = 1,2, . . .} is an orthogonal set of functions with respect to


the weight function w(x) where w (x )  0.
For example on the interval[-a,a]:
n x m x
a

 sin( )sin( )dx  a nm


a
a a
n x m x
a

 cos(
a
a
) cos(
a
)dx  a nm

n x m x
a

 sin(
a
a
) cos(
a
)dx  0

(n  m ) x
a

 Exp (i
a
a
)dx  2a nm

Interval[0,a]:
n x m x
a
a
 sin(
0
a
)sin(
a
)dx   nm
2
n x m x
a
a
 cos(
0
a
) cos(
a
)dx   nm
2
4

n x m x
a
an
 sin(
0
a
) cos(
a
)dx 
 (n  m )
2 2
[1  (1) n  m ]

Remarks
1- The series on the right-hand of Eq.(1-14) converges to A(x) in the mean-
square sense under the conditions
(a) A(x) must be "square or quadratically integrable" i.e. A(x) satisfies the
inequality 0  a  A (x ) dx  M .
b 2

(b) the set of orthogonal functions{ek( x) ; k = 1,2, . . .} is complete i.e. if


there is no nontrivial function B(x) which is orthogonal to each ek (x). In
other words,
{ek( x) ; k = 1,2, . . .}
is complete set of functions if
b
 a
e k (x )B (x )dx  0.
(c) In general, the orthogonal expansion of a piecewise-continuous
function A(x) in a complete set of functions {ek(x)} can always be
integrated term by term. The integrated series will converge in the mean-
square sense to the integral of A(x). Term-by-term differentiation,
however, is a much more precarious matter. In general, term-by-term
differentiation of a convergent orthogonal expansion is permissible when it
can be shown that differentiated expansion converges uniformly in the
interval considered

1-3 The Finite Sine Transform


We shall now consider some special examples of orthogonal
expansions. Let us examine the integral
1 a
cos(r  s )x  cos(r  s )x dx
a
 0
sin rx sin sxdx 
2 0
(1-17)
Simple integration gives
a 1  sin(r  s )a sin(r  s )a 
 0
sin rx sin sxdx 
2  r  s

r  s 
(1-18)

If we now expand the right-hand side of Eq.(1-18), using the well-known


trigonometric identities for sin (r-s) a and sin (r+s) a, we obtain
sin ra sin sa
s cot sa  r cot ra 
a
 0
sin rx sin sxdx 
r2 s2
(1-19)

Then the functions sin rx, for different values of r, will form an orthogonal
set if either
1- r=nπ/a, where n is an integer, or
5

2- r=pn, where p1, p2, . . . , pn, ... are positive roots of the transcendental
equation
p n cot p n a  h  0 (1-20)
The validity of Eq. (1-20) can readily be checked. For example, let p1 and
p2 be different positive roots of (1-20) Then
p1 cot p1a  h  0 (1-21)
p 2 cot p 2a  h  0 (1-22)
Subtracting Eq.(1-21) from Eq.(1-22) gives
p 2 cot p 2a  p1 cot p1a  0
Now let r = p1 and s = p2, so that Eq.(1-18) becomes
a
 0
sin rx sin sxdx  0 (1-23)

Normalization
The normalization constants are computed, using Eq.(1-9), which in this
case reduces to
a

 sin ( rx )dx  r
2

To evaluate λr notice that


1
sin 2 rx  1  cos 2rx 
2
then
a  sin 2ra 
a

0  1 
2
sin rxdx 
2 2ra 
For r=nπ/a, we find
a
n 
2
When r = pn

a sin 2 p n a 
n  1   (1-24)
2 2 pna 
In order to simplify Eq.(1-24) we make use of Eq.(1-20)
h
cot p n a 
pn
pn
sin p n a 
pn 2  h 2
and
6

h
cos p n a 
pn 2  h 2
However
sin 2 p n a  2sin p n a cos p n a
then
2hp n
sin 2 p n a 
pn 2  h 2
then
a h  a  pn 2  h 2   h
n  1  = (1-25)
2  a  pn 2  h 2   2 pn 2  h 2 
 
Hence, if we define a finite sine transform by
n x
a
A s (n )   A (x )sin dx (1-26)
0
a
the corresponding inversion theorem (1-14) gives an orthogonal expansion
of the form

2  n x
A (x )  
a n 1
As (n )sin
a
(1-27)

The orthogonal expansion (1-27) is called a "Fourier sine series," and it


converges to A(x) in the mean-square sense for 0≤x≤ a.
We can even make a precise statement regarding the pointwise
convergence of Eq.(1-27).
The function A(x) is said to satisfy Dirichlet's conditions on
0≤x≤a provided:
1 - A(x) is continuous almost everywhere on [0, a].
2 - A(x) has only a finite number of finite discontinuities on [0, a] (and no
infinite discontinuities).
3 - A(x) has only a finite number of maxima and minima on [0, a].
4- If A(x) satisfies Dirichlet's conditions on [0, a], then
2  n x

a n 1
As (n )sin
a
converges at every point x in [0, a] to A(x)if x is a continuaty point and to
A (x  0)  A (x  0)
2
if x is discontinuity point on the interval [0,a]
The notation A(x±0) means
7

lim A (x   )
 0

In a similar fashion we can define the so-called finite radiation sine


transform by
a
Asr (n )   A (x )sin p n xdx (1-28)
0
where pn is a positive root of the transcendental equation
p n cot p n a  h  0 h  constant
for h= constant. The corresponding inversion theorem is

pn 2  h 2
A ( x )  2 Asr (n )sin p n x (1-29)
n 1 a  p n  h   h
2 2

Observe that the sum (1-29) is to be taken over all positive roots pn

1-4 The Finite Cosine Transform


Arguments similar to those used to obtain Eq.(1-19) yield
cos ra cos sa
r tan ra - s tan sa 
a
 0
cos rx cos sxdx  2
r -s 2
(1-30)
Therefore the functions cos rx will form an orthogonal set if either
1- r=nπ/a for n integer or
2- r=qn where qn is a positive root of the transcendental equation
q n tan q n a  h h  constant
The normalization integral becomes

a  sin 2ra 
a

 cos rxdx  1 
2

0
2 2ra 

so that, for r=nπ/a


a
n 
2
or, for r=qn
a q n 2  h 2   h
n =
2 q n 2  h 2 
We define the finite cosine transform by
n x
a
Ac (n )   A (x )cos dx (1-31)
0
a
8

and the corresponding inversion formula by


Ac (0) 2  n x
A (x )    Ac (n )cos (1-32)
a a n 1 a
The finite radiation cosine transform is defined by
a
Acr (n )   A (x ) cos q n xdx (1-33)
0

where qn is a positive root of q n tan q n a  h with h=constant


The corresponding inversion formula is
qn 2  h 2

A ( x )  2 Acr (n )cos q n x (1-34)
n 1  n   
2 2
a q h h
Observe that the sum (1-29) is to be taken over all positive roots qn
of the transcendental equation
q n tan q n a  h h  constant

1-5 Properties of Finite Fourier Transforms


Consider
n x
a
As (n )   A (x )sin dx (1-35)
0
a
and
n x
a
Ac (n )   A (x )cos dx (1-36)
0
a
We can write Eqs. (1-35) and (1-36)
T s A (x )  As (n ) (1-37)
T c A (x )  Ac (n ) (1-38)
Using this notation, many properties of finite Fourier sine and cosine
transforms can be exhibited in compact form. For example, if we integrate
A n x
a

 x sin
0
a
dx

by parts, we obtain
x a
n x n n x
a
A (x )sin
a x 0

a 
0
A ( x )cos
a
dx

The first term in this expression vanishes at the limits of integration x=0
and x=a

 A  n
T s     T c A (x ) (1-39)
 x  a
9

by a similar argument, one can prove


 A  n
T c     1 A (a )  A (0)  T s A (x )
n
(1-40)
 x  a
Further application of integration by parts gives
  2 A  n   n 
2

 1 A (a )  A (0)   
  a  s 
T A (x )
n 1
Ts  2   (1-41)
 x  a 
  2A   n 
2

T c  2    1 A (a )  A (0)    T c A (x )


n
(1-42)
 x   a 
Note that the sine Fourier transform of the second derivative is related
to the values of A(x) at the boundaries x=0, x=a while, cosine Fourier
transform of the second derivative is related to the first derivative of A(x)
at the boundaries x=0, x=a.
The radiation transforms give similar formulas
T sr A (x )  Asr (n ) (1-43)
T cr A (x )  Acr (n ) (1-44)
Thus (1-43) and (1-44) together with integration by parts, permit us to
write
  2A   A 
T sr  2   p n A (0)    hA  sin p n a - p n2T sr A (1-45)
 x   x x a
and
  2A   A 
T cr  2   A (0)    hA  cos q n a - q n2T cr A (1-46)
 x   x x a
1-6 Classical Fourier Series
Let A(x) be a periodic function defined on the interval [-a,a] with
period 2a i.e.
A (x )  A (x  2a )
then Fourier series corresponding to A(x) is defined as
a0   n x n x 
( x )     an cos  b n sin  (1-47)
2 n 1  a a 
where the coefficients of this expansion are given by
n x
a
1
a a
an  (x ) cos dx , n  0,1, 2,3,... (1-48)
a
n x
a
1
bn   A (x )sin dx , n  1, 2,3,... (1-49)
a a a
11

The formulas (1-48), (1-49) can be proved using the properties of


orthogonality of sine and cosine functions on [-a,a], for example, by
m x
multiplying both sides of equation (1-47) with cos( ) and integrating
a
from -a to a
m x
a

 A (x )cos
a
a
dx

m x 
m x n x m x n x
a a a
a0
  cos dx  {an  cos cos dx  b n  cos sin dx }
2 a a n 1 a
a a a
a a

  an  mn a  aam , m 0
n 1

n x
a
1
an   A (x ) cos dx
a a a
when m=0
a a
a 2aa
a A( x)dx  20 a dx  2 0  aa0
a
1
a a
a0  A( x)dx

Similarly, we can obtain b n by multiplying both sides of equation (1-47)


m x
with sin( ) and integrating from -a to a
a
m x m x 
m  x n x m  x n x
a a a a
a0
 A ( x )sin
a
dx 
2 a
sin
a
dx  
n 1
{an  asin cos
a
dx  b n  a sin a dx }  bm a
sin
a a a

n x
a
1
b n   A (x )sin dx
a a a
Parseval’s theorem for finite integral transform:
If f(x) satisfies Dirichlet's conditions on the interval -a≤x≤a, then:
(1-50)
Proof:
By multiplying both sides of Classical Fourier Series for f(x) with f(x) and
integrating from -a to a :
11

Where:

Then we get equation (1-50)

Remarks
1- In the same manner, the coefficients an ,bn can be written in the form
c  2a
1 n x
an 
a 
c
A (x )cos
a
dx , n  0,1,2,3,..... (1-51a)
c  2a
1 n x
bn 
a 
c
A (x )sin
a
dx , n  1,2,3,... (1-51b)

where c is any real number. In the special case c=-a we obtain the
equations (1-48), (1-49)
2- If A(x) satisfies Dirichlet's conditions on [-a, a], then the series on the
right-hand of equation (1- 47) converges at every point x
in [-a, a] to A(x)if x is a continuaty point and to
A (x  0)  A (x  0)
2
if x is discontinuity point on the interval [-a,a] where
A (x  0)  lim A (x   )  >0
 0

3- (a) If A(x) is even on the interval [-a,a] i.e.,


A ( x )  A ( x )
then, the classical Fourier series, which is called half-range cosine series,
will not contains sine terms i.e.
bn  0 (1-52a)
while the coefficients an are given by
2a n x
an  
a0
A (x )cos
a
dx , n  0,1,2,3,..... (1-52b)

and the classical Fourier series is given by


a0  n x
( x )    an cos (1-53)
2 n 1 a
which is exactly the same inversion of finite cosine transform (1-32)
(b)- If A(x) is odd on the interval [-a,a] i.e
A ( x )  A ( x )
then, the classical Fourier series, which is called half-range sine, series will
not contains cosine terms i.e.
12

an  0 (1-54a)
while the coefficients b n are given by
2a n x
bn  
a0
A (x )sin
a
dx , n  1,2,3,..... (1-54b)

and the classical Fourier series is given by



n x
(x )  bn sin (1-55)
n 1 a
which is exactly the same inversion of finite sine transform (1-27)

1-7 Representation of arbitrary Functions


Example 1
a) Find Fourier series corresponding to function
0, -5<x<0
f (x )  
3, 0<x<5
b) How should f(x) be defined at x=-5,x=0 and x=5 in order that Fourier
series will converge to f(x) for -5<x<5?
Solution
The graph of function f(x) is shown below

-10 -5 0 5 10
By testing the given function we find it even is-15not
-10 nor odd and the given
interval is symmetric, then all Fourier coefficients must be calculated, in
this case the period is 2a  10  a  5 , then
n x
5
1
an   f (x )cos dx
5 5 5
1 
5
n x n x
0
   (0)cos dx   (3)cos dx  
5  5 5 0
5 
3 5 n x  x  5
 sin   0, n 0
5  nx 5 x 0
when n  0
0 x
5 5
3 3
a0   cos dx   dx  3
50 a 50
13

and
1 n x n x 
0 5
b n    0sin dx   3sin dx 
5  5 5 0
5 
3  5 n x  5 3(1  cos(n )) 3(1  (1) n )
  cos   
5  n 5 0 n n
then
3  3(1  (1) n ) n x
f (x )    sin( )
2 n 1 n 5
3 6 x 1 3 x 1 5 x 
   sin( )  sin( )  sin( )  ....... 
2  5 3 5 5 5 
b) Since f(x) satisfies the Dirichlet conditions, we can say that the series
converges to f(x) at all points of continuity and to f (x  0)  f (x  0) at the
2
points of discontinuity, then, the series will converge to f(x) for -5<x<5 if
we redefine f(x) as follows
3
2 x  5

0 5  x  0
 3
f (x )   x 0
2
3 0x 5

3 x 5
 2

Example 2
Expand f (x )  x 2 ,0  x  2 , in a Fourier series if the period is 2
Solution
The graph of f(x) is shown below

The period is 2a  2  a   ,in this case we should use the formula in


remark-1with c=0,then
14

1 2
 
an  x 2
cos nxdx
0

 sin nx    cos nx    sin nx  2


x
1 4
 2
   2 x    2    , n 0
  n   n
2
  n
3
 0 n2

when n  0
1 2 8 2
a0   x dx  2

 0 3
1 2
4
b n   x 2 sin nxdx 
 0 n
then
4 2  4 4
f (x )  x    2 cos nx 
2
sin nx , 0  x  2
3 n 1 n n

Example 3
Use the results of example 2 to prove that
1 1 1 2
   ..... 
12 22 32 6
Solution
4 2  4
At x=0 the Fourier series of example 2 reduces to   2 , but by the
3 n 1 n

Dirichlet conditions, the series converges at x=0 to 2 2  1 (0  4 2 ) as it is


2
discontinuity point, then
4 2  4 2  1
2 2   2  
3 n 1 n 6 n1 n 2
Example 4
Expand the function A (x )  sin x , 0  x   in finite cosine Fourier series
Solution
we can find Fourier coefficients from
 
1
Ac (n )   sin x cos nxdx =  sin( x  nx )  sin( x  nx ) dx
0
20
1  1  cos n 1  cos n  (1  ( 1) n )
=   = , n 1
2 n 1 n  1  n 2 1
when n=1
 
sin 2 x
Ac (1)   sin x cos xdx  0
0
2 0
15

But Ac (0) can be calculated directly from the formula of Ac (n ) by putting



n=0 to get Ac (0)  2 or by the integral  sin xdx . Then the required series
0

using the inversion theorem (1-32)is given as


2 (1  (1) n )
2 
A (x )    cos nx
  n 2 n 2  1

Note: The same solution can be obtained using classical


half-range cosine Fourier series
a0 
(x )    an cos nx
2 n 1
The Fourier coefficients are
2 2(1  (1) n )
an   sin x cos nxd x =  , n  1, n  0,2,3,...
0  (n 2  1)
4
a0 

 
2 sin 2 x
a1   sin x cos xdx  0
0  0

which give the same solution


Example 5
Expand the function
A (x )  x , x a
in Fourier series.
Solution
The fundamental interval is 0 ≤ x ≤ a so that
A (x )  x , 0 x a
Notice that
A ( x )   x  x  A (x )
Therefore we employ the finite cosine transform,
n x n x n x
a a a
ax a
Ac (n )   x cos dx = sin   sin dx
0
a n a 0 n 0 a
2
 a 
   (-1)  1 , n 0
n

 n 
when n= 0
a2
Ac (0)  ,
2
16

Using the inversion theorem


Ac (0) 2  n x
A (x )    Ac (n )cos
a a n 1 a
yield,
a 2a   (-1)  1 n x
n

A (x )  x   2  cos (1-55)
2  n 1 n2 a
for 0 ≤ x ≤ a. However, since the right side of Eq (1-55) is even in x, we
see that (1-55) actually represents lxl for lxl ≤ a
Example 6
Expand the function
A (x )  x , a  x a
Solution
Again we take as the fundamental interval 0 ≤ x ≤ a, so that
A (x )  x , 0 x a
then
A (x )  x  A (x ),
which means that A(x) is odd. Hence, we employ the finite sine transform
to represent A( x) for 0 ≤ x ≤ a
n x ax n x n x
a a a
a
As (n )   x sin dx = cos   cos dx
0
a n a 0 n  0
a
a2
 (-1)n+1 , n 0
n
Using the inversion theorem
2   n x 
A (x )   A s (n )sin  
a n 1  a 
yield,
2a 
(1) n 1  n x 
A (x )  x  
 n 1 n
sin   (1  56)
 a 
for 0≤x≤ a. However, since the right side of Eq. (1-56) is odd in x, we see
.that (1-56) represents A(x)=x for the entire interval a≤x≤a
Example 7
Expand the function
A (x )  b , a  x  0
A (x )  x , 0x a
In this case the function A(x) is neither even nor odd. Two choices are
open to us. We can either decompose A(x) into even and odd parts and
then generate the corresponding Fourier cosine and sine series separately,
17

or we can use the classical Fourier series which we will use now. Let
a0   n x n x 
( x )     an cos  b n sin 
2 n 1  a a 
where
n x
a
1
an   (x ) cos dx , n  0,1, 2,3,...
a a a

n x
a
1
b n   A (x )sin dx , n  1, 2,3,...
a a a
which give
a  (1) n  1 
an  2  n 0
  n 2 
,

a  2b
a0 
2
b a b
bn    1
n

n n
then
a  2b   a  (1) n  1  n x  b a  b n n x 
( x )    2   cos      
1 sin 
4 n 1    n2  a  n n  a 

Remark: For more representations of the functions see tables (1).

1-8 Solution of Differential Equations with Boundary Conditions


The finite integral transforms can be used to find the solution of an
ordinary differential equation when the imposed boundary conditions are
given at the ends of finite interval say, [a.b]. These conditions are of three
types
1- Dirichlet boundary conditions: in this type the values of the unknown
function "the solution" are given at the boundary points
2- Neumann boundary conditions: in this type the values of the derivative
of the unknown function "the solution" are given at the boundary points
3- Radiation boundary conditions: in this type the condition
A
 hA  c , where h and c are given constants, is imposed at least at one
x
of the boundary points
18

Example 8
The finite Fourier transform can be used to solve certain inhomogeneous
boundary-value problem. For example solve
d 2A
2
 k 2 A  f (x )
dx
in the interval 0  x  a subjected to the boundary conditions
A (0)  A (a )  0
Solution
The Given boundary conditions are of Neumann type then, the suitable
integral transform is the finite cosine transform becuase in finding the
cosine transform of the second derivative we need the values of the
derivative of the solution at the boundary points, see equation (1-42). Now
take the finite cosine transform to all terms of the given ordinary
differential equation to get
d 2A 
T c  2   k 2T c A   T c f (x )
 dx 
using equation (1-42) in the first term, yield
n
2

 1 A (a )  A (0)    Ac (n )  k 2 Ac (n )  f c (n )


n

 a 
then
 2  n 2 
k     Ac (n )  f c (n )
  a  
then
f c (n )
Ac (n ) 
k  (n a ) 2
2

Applying the inversion theorem yield


Ac (0) 2  f (n ) n x
A (x )    2 c cos
a a n 1 k  (n a )2 a
Example 9
Solve the following boundary-value problem
d 2Y
2
  2Y  G (x ), 0x L
dx
dY
0 at x  0
dx
dY
k  HY  0, at x  L
dx
19

Solution
The Given boundary conditions are of radiation type then, the suitable
integral transform is the finite radiation cosine transform-becuase in
finding the finite radiation cosine transform of the second derivative we
need the values of the derivative of the solution at one boundary point and
the value of the radiation condition on the other boundary point, see
equation (1-45). Now take the finite radiation cosine transform to all terms
of the given ordinary differential equation to get
 d 2Y 
T cr  2    2T cr Y   T cr G (x )
 dx 
 dY 
Y (0)    hY
1  cos q n L  q nY cr (n )   Y cr (n )  G cr (n )
2 2

 dx x L
H
where h1  . then by using the given boundary conditions we find
k
G (n )
Y cr (n )  2cr 2
  qn
Applying the inversion theorem yield

q n 2  h12 Gcr (n )
Y ( x )  2 cos q n x
n 1 L  n 1 1
q 2
 h 2
 h  2
 q 2
n

where the sum is to be taken over all positive roots qn


of the transcendental equation
q n tan q n L  h1 h1  constant
21

Table (1)
21
22
23
24

Kraut, Edgar A. Fundamentals of Mathematical Physics (Dover Books


on Physics) . Dover Publication.

You might also like