~tiucture Page No.

It 7.1 Introduction 51
Objectives ,
7.'2 Fourier Integral 52
b 7.3 Fourier Transforms 59
Properties of Fourier Transforms
Finite Fourier Transforms
7.4 Applications of Fourier Transforms to Boundary Value Problems 79
7.5 Summary 88
7.6 Solutions/Answers 90
Appendix 100

7.1 INTRODUCTION
You know from your knowledge of Real Analysis course that Fourier series are
powerful tools in treating various problems involving periodic functions (ref. MMT-
004). Fourier series are used to represent a function f defined on a finite interval such
as. [-L, L] or [ 0, L] . When f and f ' are piecewise continuous on such an interval, a
Fourier series represents the function on the interval and converges to a periodic
extension of f outside the interval. However, in many practical problems, the
functions involved are non-periodic for which Fourier series representations are not
1 possible. A suitable representation for non-periodic h c t i o n s can be obtained by
considering the limiting fonn of Fourier series when the fundamental period is made
infinite. In such cases, the Fourier series becomes a Fourier integral. Using
symmetry, Fourier integral can conveniently be expressed as a pair of formulas the
first giving the Fourier transform and the second giving its inverse. Like Laplace
transform, the Fourier integrals and transforms which we shall be discussing in this
unit, are useful in solving initial boundary value problems arising in science and
engineering, for example, conduction of heat, wave propagation, theory of
communication etc.
To begin with, in Sec.7.2, we have derived the Fourier integral representation of a
function f(x) defined qver .the interval - oo < x < a . We have also given the Fourier
cosine and sine integral re6esentation of f (x) ,when f (x) is even and odd function
b re~pectively.Complex form of Fourier integral representation is also given in this
section. In Sec.7.3, we have represented the Fourier integral representation of a
function f (x) as a pair of formu'ms, the first giving the Fourier transform of f (x) and
the second giving the inverse of that transform. We have also obtained the Fourier
transforms and their inverses for a number of functions amhiiscussed a number of
useful properties admitted by the Fourier transform in this section. AppEcatians of
Fourier transforms to boundary value p~oblemsare discussed in Sec.7.4. An
Appendix is attached at the end of the unit which gives a table of Fourier transforms of
the commonly used functions for ready reference.
Objectives
After studying this unit you should be able to
obtain the Fourier integral representation of a given non-periodic function over
the given interval;
obtain the Fourier transform and the inverse of that transform for a given
function over the given interval;
use various properties of Fourier transforms to obtain the Fourier transforms and
their inverses in various applications;
select the appropriate Fourier transform to be employed for the solution of a
given boundary value problem;
obtain the solution of a given boundary value problem using appropriate Fourier
transform.
Partial Differentlal
Equations 7.2 FOURIER INTEGRAL
We shall now derive, a means of representing certain kinds of non-periodic functions
that are defined oneither an infinite interval ] - a,a[or, semi-infinite interval
10, a[.
Let us consider f (x) having the following properties.
1) f(x) is piecewise continuous on every finite interval.

2 ) f(x) is absolutely integrable on the interval -m < x <m .that is. 7 if(x)ldx
-w

converges.
3') at every x on the real line, f(x) has left and right hand derivatives.
Consider the Fourier series representation of f(x) on the interval - L < x i!L. We
have

where. a, =-
L
1
5 f(t)dt, a,, =-L 1
f ( t ) c o s ( y t)dt
-L -L

nlr
Note that -L
in Eqn.(l) are the frequencies - spatial or temporal, depending on
whether x is a space variable or time.
nn . R
If we let a n=-, Aa = an+, -a, =-, then ~ ~ n : (becomes
l)
L L

We now expand the interval - L < x < L by lettlng L +a, . Since L -+ a,implies -
Aa + 0, the limit of Eqn.(2) has the fonn of a Riemann sum of a definite integral, i.e.
4,

of the form lim F(an ) Aa ,which i s suggestive of the definition of the integral
Aa-sO
n=l

I F(a) d a . Thus if I If (x) /dxconverges, the limit of the first term in Egn.(2) is
m m

0 -OD

zero and by the definition of the integral as the limit of the sum, Eqn.(2) becomes

The result given in Eqn.(3) is called the Fourier integral of f(x) or Fourier integral
representation of f(x) on the interval - a, < x < a. You may notice that the basic
structure of the Fourier integral is analogous to that of a Fourier series. We can sum
up as follows: , ,
The Fourier Integral of a function f(x) defined on the interval - oo < x < a, is given
1
by
 Fourier Transform
1
7c
I ) a x + ~ ( asin
f(x) = - [ ~ ( acos ) axIda (4) Method
0
w

where, ~ ( a=) If(x)cos a x dx (5)

-w

Eqn.(4) is called the Fourier integral formula for the function f(x) .
b

Sufficient conditions under which a Fourier integral converges to f(x) are similar to,
but are slightly more restrictive than, the conditions for a Fourier series. We state here
without proof, the following convergence theorem.

Theorem 1: Let f(x) be a piecewise continuous function on every finite interval and
absolutely integrable on - oo < x < co . Then the Fourier integral of f (x) on the
interval converges to f(x) at a point of continuity. At a point of discontinuity, the
Fourier integral will converge to the average
f(x+) + f(x-)
2
where f(x+)and f (x-) denote the limit of f (x) at x from the right and left,
respectively.
I Cosine a n d Sine Integrals
t When f(x) is an even function on the interval - co < x < oo ,then the product
f(x) cos a x is also an even'function, whereas f(x) sin a x is an odd function. Using
L
the property of definite integral, we obtain B(a) = 0 and so Eqn.(4) becomes

where in the above equation we have again used the property of definite integrals to
write

-w 0
Similarly, when f(x) is an odd function on the interval - oo < x < oo , the products
f (x) cos a x and f (x) sin a x are odd and even functions respectively. Therefore
A(a) = 0 and

f ix) =l]
II [ t ( t ) ,in at dt) sin a x d a
0
We summarize:

The Fourier integral of an even function on the interval ]oo,oo[ is the cosine integral
2
I
f(x) = - ~ ( acos
rt
) ax d a

where. A(a)= If(x) eos ax dx

0
The Fourier integral of an odd function on the interval ]a, oo[ is the sine integral
Partial Differential
Equations where, I
m

~ ( a=) f(x) sin ax dx

Let us consider the following examples.

Example 1: Find the Fourier integral representation of the function
0 , x<O

0, x>2
Solution: The graph of the function is shown in Fig. 1. It satisfies the hypothesis of
Theorem-1 . Hence from Eqns.(S) and (6), we have
m

A(a) = I f (x) cos a x dx

0 2
I f (x) cos a x dx + I f (x) cos a x dx + I f (x) cos a x dx
m

sin 2 a
dx=-

B(a) = I f (x) sin a x dx

1- cos 2 a
= Isin a x dx =
a

Substituting these coefficients in Eqn.(4) we obtain

1 [(sin 2 a )
f(x)=-r - cos a x +
7 f1-cos2a)sinaxl d a

'2 sin a
-- ; IT ( C O S ~ C O+Ssinasinax)da
~X
0

This is the Fourier integral representation of the given function.

The above Fourier integral representation can be used to evaluate integrals. For
example at x = 1 , it follows from Theorem-1 that integral (I 1 ) converges to f(l) ,that
is,
-
X
j
2 sina
-da
a
= f(i)
0

The above result is worthy of special note since it cannot be obtained in the usual
manner as the integral (sin x ) / x does not possess an anti-derivative that is an
elementary function.
- ~ - ~ ~ .
Example 2: Find the Fourier integral representation of the function Fourler Tnnsform
Method
1, Ixl<a
f (x) =
0 , lxl>a

Solution: The graph of the given function is shown in Fig.2. Clearly, the given
function f(x) is an even function. We represent f(x) by the Fourier cosine integral
(7). From Eqn.(8), we obtain,

I
00

A(a) = f (x) cos a x dx

= J' f(x) cos a x dx + I f(x) cos a x dx

= j c o s a x dx =
sinax
T I :=a.
sinaa
0

2
and thus. f(x)=- f sin a a cos a x d a

Example 3: Represent f(x) = e-\ x 2 0, k a positive constant by (a) a cosine

integral (b) a sine integral.

Solution: (a) Using Eqn.(8), we obtain

m

Using integration by parts, we find

k
or, A(a) =
k2 +,a2
Therefore, the cosine integral of f (x) is
2k " cos a x
f (x) = e-" = - da .
n oI k 2 + a 2

(b) Similarly, we have

m
a
B(a) = e-" sin a x dx =
0
k2 + a 2
The sine integral of f(x) is then
2" asinax
f(x) = e-" = -
n k2+a2
I
da .
0

Example 4: Find the Fourier cosine integral representation for the function
Partiai Dkfferential Solution: Using Eqn.(8), we obtain
Equations
I
m

A(a) = f (x) cosax dx = x 2 cos a x dx

i
= [x 2 [si y] 2x[- c y ) + ( - sin a x
a3 )I; [;;;yion]

-a
- sin a a
+ 2a cos a a --2 sin a a

Thus, the cosine integral representation of f(x) is gven by

2
I
f (x) = - ~ ( acos
X
) a xda
0

[I';
xo a + a
- -)
a sin a a 2a cos a a 2 sin a a
a 3 cos a x d a

Example 5: Find the solution of the integral equation

I f (x) cos
m

ax dx = e-. ,where a is a constant.

0

Solution: The given integral is a Fourier cosine integral representation

I I
m
2
Let g(x) = - ~ ( acos
) a x d a ,where A(a) = g(x) cos a x dx
X
0 0

X
Comparing the above equation with the giyen equation, we get x = a, -g(a) = e-",
2
f (x) = A(x) .

I
m
2 e -,cos a x dx =
m
)
cos a x dx = - - (using Example 3
Therefore, A(a) = -
0
X X
0
2 l +1a (
w"hk=l)

Hence, f (x) = A(x) =

Example 6: Using Fourier integral representation, show that

Solution: To prove the result consider right hand side which defines the function

1:
-cosx , ) x 1 < x / 2
f (x) = 2
I X ( >XI2

Here f (x) is an even function of x defined in the interval - 03 < x < 03 and since the
integral on the left hand side involves cosine terms, it indicates that we are required tr
find the Fourier cosine representation of f(x) .
Thus from Eqn.(8), we have
I

m nil Fourler Transform

A(a) = I f (x) cos a x dx = Method
0 0

Using Eqn.(7), the Fourier cosine representation of f(x) is given by

m m

0 0

0
Therefore, we have
-COSX, ) x ~ S R / ~

which is the desired result.

***
Complex form of Fourier integral representation
The Fourier integral (4) also possesses an equivalent complex form or, exponential
form, that is analogous to the complex form of a Fourier series. Substituting the
expressions for A(a) and B(a) from Eqns.(S) and (6) into Eqn.(4), we have
l m
f ( x ) = -xI ~f(t)[cosatcosax+sinatsinax]dtda
0 -m

=I]
n jf(t)cosa (t-x)dtda
0 -m

=l
2n
nf(t)cosa (t - x ) d t d a
-m -m
m m
=-
1 J J f(tl (ela([-x) + e-la(t-x) )dt d a .
2n
0 -m

To combine the two terms on the right hand side of Eqn.(l3), let us change the dummy
integration variable from a to -a In the first. Thus

0 0

17
-m -a,

= 1
2n
-m -m
f(t)e""('-"~d~da + -
2n
"I
0 -m
f(t)e-"(""'dtda
Partial Differentlpl
Equations

-m L -m J
The integral in Eqn.(l4) can be expressed as

-m
m

where c(a)= J f(x)e-"dx

-m

Eqns.(l5) and (16) give the complex form of Fourier integral representation. We
now take up an example to illustrate its application

Example 7: Find the complex fonn of Fourier integral representation for the
function
1x1, - n < x < n
f(x) =
0 , elsewhere.
Solution: We have
m n
C ( a ) = I f ( ~ ) e - ' ~ ~ dJ x~=x l e - ~ ~ ~ d ~

Therefore,
1" 1
f (x) = - +cos
n a
1 n u - n a sin n a - 1] eiaxd a

You may now try the following exercises.

El) Find the Fourier integral representation of the following functions.

0 , x<O -(l+x), -l<x<O
, OSx'SJ
0 , x>3 1x1 > l 9

sinx, - 2 1 x 5 0
cosx. 0 < x 1 2
0 , lx1>2
E2) Represent the following function by an appropriate cosine or sine integral
( 0 , x<-1
 sinx, O S x S x sinhx, O < x < 3 Fourier Transform
C) f(x) = d) f(x)= Method
0 x>X: 0 , x > 3

E3) Using Fourier integral representation, show that

m
a 3sin a x X: -
d a = - e "osx, x > O
2
0

E4) Solve the following integral equation.

ES) a) Use Eqn.(l 1) to show that 7-

0
sin 2x
X
n
dx=-.
2
b) Show in general that for k > 0

E6) For the following functions, find the complex form of the Fourier integral
representation
(0, - a , < x < - 1

In the discussion so far, we obtained the Fourier integral representation of non-

pe'iiodic functions f(x) defined on - a,< x < a,by taking the limit of the Fourier series
formula as the period tends to infinity. The Fourier integral representation of a
function f (x) , can be represented as a p a r of formulas, the hrst giving the Fourier
transform of f(x) and the second giving the inverse of that transform. Like the
Laplace transform, the Fourier transform, which we shall be discussing in the next
section, is highly developed as a methodology.

7.3 FOURIER TRANSFORMS

Consider the Fourier integral formula

0
00

where, A(?) = f (x) cos a x dx

-00
m

B(a) = f (x) sin a x dx

-m
Partial Differential You know that this formula possesses an equivalent complex form or an exponential
Equations form given by Eqns.(l5) and (l6), that is,

where, C ( a ) = I
m

f (x) e - ' ~ 'dx

-m
From your knowledge of the Laplace transforms which we discussed in Unit-6, you
may recall that integral transforms occur in transform pairs - that IS,if
b
F(a) = f(x) K(a, x) dx (17)
a
is an integral that transforms f(x) into F(a) ,then the function f(x) is recovered by
means of another integral transform

j
f(x) = F(a) H(a,x) d a (1 8)
c
called the inversion integral or inverse transform, the functions K and H being the
kernels of transforms (17) and (18), respectively.

Thus, rather than thinking of Eqn.(l5) as the Fourier integral of f(x) and Eqn.(l6) as
giving the Fourier coefkcients C ( a ) , we can think of Eqns.(l5) and (16) as a
transform pair. Eqn.(l6) defines the Fourier transform C ( a ) of the given function
f (x) and Eqn.(l5) is the inversion formula, as putting C ( a ) in it and integrating we
get back f (x) . It is standard to use the notation F(a) in place of C ( a ) for the
transform.

The Fourier transform of f(x) denoted by F [ f (x)] is defined as

-a
and the inverse Fourier transform is defined as

! F - ' [ ~ ( a )2n~ =
j Fl( a ) e h d a = f ( x ) (20)
-m
Thus, the Fourier transform and inversion formula together amounts simply to the
I
Fourier integral representation, expressed in complex exponential form. The
1 conditions imposed'on f(x) are the same as in Theorem-1 .

Cosine and sine transforms

When the function f(x) defined in the interval - oo < x < oo is an even function, then
from Fourier cosine integral, we have

If we write
m

F , [ ~ ( x ) I =[ r ( x ) ~ ~ s a ~ d x = F , ( a )
0

then fiom Eqn.(21), it follows that

 We call Fc (a) the Fourier cosine transform of f(x) ,while f (x) is the inverse Fourier Fourier Transform
Method
I
cosine transform of Fc ( a ) . Similarly, when f(x) defined in the interval - co < x < oo is
an odd function, then from Fourier sine integral, we have

"C
f (x) = -

If we write
X 0
j j f (t) sin a t dt 1
sin a x da

m
b
Fs[ f (x) ] = f (x) sin a x dx = Fs ( a )
0

then from Eqn.(22), it follows that

We call F, ( a ) the Fourier sine transform of f(x) , while f(x) is the inverse Fourier
sine transform of F, ( a ) .
We summarise:
The Fourier Cosine transform of even function f(x) on the interval -a, < x < oo
is defined as
w
F, 1f(x) 1 = F, (a)= j f(x) cos a x dx
0

and the inverse Fourier cosine tranqform of F, ( a ) is given by

F - ' [F, (a)] = -X2 * F, (a) cos ax d a = f(x) (24)

0

The Fourier sine transform of odd function f (x) on the interval - oo < x < m is
('
defined as \
00

F, [ f(x) 1 = F, (a)= j f(x) sin a x dx (25)

0

and the inverse Fourier sine transform of F, ( a ) is given by

2*
F-' IF, (a)] = - j F,(a)sin ax d a = f(x) (26)
n
0

We now illustrate the calculation of the transform F(a) of f(x) and its inverse
through the following examples.

Example 8: Find the Fourier transform of

f (x) =
i
1 , lxl<a
0 , Ixl>a
Also draw the graph of f(x) and its Fourier transform for a = 3

Solution: Using Eqn.(l9), the Fourier transform of f (x) is

iaa - -iaa
2 sin a a , a;cO
-- =-
ia a
or a = 0 , we obtain F ( a ) = 2a .

The graphs of f(x) and F(a) for a = 3 are shown in Fig.3 and Fig.4 respectively.
Partla1 Differentl~l
Equations

~ x a m ~9:l e Find the Fourier transform of the function -

0 , x20
e-ax, x 2 0 , a > o .

Solution: We have

+ [f(x)] = F(a) = I f(x)e-'~'dx I e-"xe-axdx

m

=
m

-m 0

ante [
lim e -(a+ia)x ]= lim e-ax[cos a x - i sin a x ] = 0 .
x+m A -*a

1
Hence -and. f(x) = e-ax"uo(x)where uo(x) is the unit step funciion, form a
a+ia
transform pair.
1
Therefore, +
[ e-" u, (x)] =
a + ia
-

Example 10: Find the Fourier transform of e-""- , a > 0 .

Solution: From the definition, we have

+fe-ax2 I=
m
Je-de-iax
m
-a
[ X
2 iax
+-
'1,
 i
Example 11 : Find the Fourier transform of Fourier Transform
-- Method

i
2
1 - x , IxISl
f (x) =
0 , Ixl>l
i and hence evaluate

1 Solution: The function f(x) is given by

I
f (x) =
0i 1-x2, -1SxSl
Ixl>l 9

This shows that f(-x) = f(x) i.e. f(x) is an even function in the interval
- co < x < co . Hence by result (23), the Fourier cosine transform of f(x) is
rn

F C [ f ( ~ ) lI=f ( x ) c o s a x d x = i m

I
( I - X ~ ) C O S ~+ X 0.cosaxdx
~X

= 2 ( ~ i na - a cos a
a3
By using inverse transform [result (24) 1, we have
1
f(x) = z b ( s i n a - a c o s a cos a x d a
7C
0
a'
=2](sina )
- a c o s q cos a x d a
X
0
a3
The result (27) can be expressed as
;
;cos a c o s a x d a = - f ( x ) =
j
0
(sin a

1
1 X
4

Putting x =-, which lies in - 1 S x S 1 ,i.e., in I x 111, we get

2

Changing the dummy variable a by x in the above integral, we obtain

b
x --: cosx X
COS-dx =-
3iI
0 1 2 16

which is the requ~redresult.

***
Example 12: Find the Fourier cosine transform of the function
cosx, O < x < a
f (x) =
0 , x>a

Solution: Using the definition, we have

m w

0
i
Fc[ f(x)] = If(x)cos a x dx = cos x cosax dx + O.cos a x dx
0 a
I
Partlal Differential
Equations

sin ( a + I) x
+ sin ( a - l)x
a-1

Example 13: Using inverse sine transform, find f (x) if

-
Solution: Using result (26), inverse sine transform of F, ( a ) is given by
2"
f(x)=- I ~ , ( a ) s i n a x d a=-
2 1 -,,sin a x d a = -2 I(x)
x II I-e
a x
0 0

I a xd a .
O0

where. I(x) = -sin

a
e-aa

0
a
IF(x)= I - -sin
e-au [Using the rule of differentiation
Now, axda
a ax under the integral sign]
0
m

= e-" cosax d a
0

=-
a2+x
a
(-a cos a x + x sin a x )
:I
a' + x' '
Integrating, we have
a
I ( x ) = I----?dx + A = tan-'L+ A
a' + x - a
Putting x = 0 , we get I(x)],=, = A
But we have
~(x)I,-o = o
X
therefore, A = 0and I (x) = tan-' - .
a
Hence, we obtain

sin a a
Example 14: Find the function f(x) , whose Fourier cosine transform is - .
a
sin a a
Solution: Given that F, ( a ) = - a
and we are required to find f(x) . Using result
(24), we get

R
j
f(x) =? F, (a)cosan d a = - cos a x d a

sin (a + x)a + sin (a - x ) a

da
a
0
 I- dx {
m
sin ax x/2 , a >0
Using the result = , we obtain
x -x/2, a <0
0 -
/ L [ E + E ] , a + x > O and a - x > O
f(x)= x 2 2
lo- , a + x < O and a - x < O

Example 15: Solve the following integral equation

m

I I
1, 01a<1
f(x)sinaxdx= 2 , l < a < 2

Solution: Since sin a x is present in the given integral, we use result (25) and write
co 1, 01a<1
F,(a)= j f ( x ) s i n a x d x =
o 0, a 2 2
Now to find f (x) we use result (26) and obtain the inverse Fourier sine transform.
Thus we have,

which is the required result.

***
As with the Laplace transform, for convenience, we have given in the Appendix at the
end of the unit, a table giving the Fourier transforms and their inverses of commonly
used functions.
You may now try the following exercises.
' E7) Find the Fourier transform of the following functions defined on ] - oo,a,[
a , -C<x<O, a>O a , -C<x<C, a>O
a) f(x) = b) f(x) =
0 , otherwise 0 , otherwise
a , -C<x<O
b , O < x < C a>O, b>O
0 , otherwise
Partial Differential
Equations E8) Find the Fourier transform of the function f (x) = e-"'I , - m a, x x m , a > 0 .
Write the inverse transform also.
e-ax
E9) Find the Fourier sine transform of the function f(x) = - and hence evaluate
X

I tan-' Ias i n x dx
w

E 10) Find the Fourier cosine transform of the function f (x) = e-' and .hence show that
m
COS mx
dx=-e
= -m
.
2
0

E 1 1) Solve the following integral equations

a) 7
0
f (x) sin a x dx =
1-a, OSaSl
0 , . a > l

w
1-a, O1a1l
E 12) Solve the integral equation
I f (x) cos a x dx
0
=
0 , a > l
and hence

m
sin2 z dz rr
show that =-
z2 2 '
0

E 13)- Find the Fourier cosine transform of the following function

I
f(x)= 1-x,
0
1
-<x<l
2
, x>l

E 14) Using inverse Fourier cosine transform, find f (x) , if

-

E 15) Find the function f (x) , satisfying the integral equation

I f (x) sin a x dx
m

=
a
a2+k2'
0

The Fourier transform admits a number of useful properties which we shall discuss
now. Our discussion will parallel our analogous discussion for the Laplace transform,
given in Unit 6.
7.3.1 Properties o f Fourier Transforms
We will assume without reiteration conditions given in Theorem-1 .
Linearity property: If 7 [ f (x)] = F(a) and 7 [g(x)] = G(a) and a and b are any
..
scalars then
!F[af(x)+bg(x)l =a!Flf(x)l+b!Flg(x)l= a ~ ( a ) + b ~ ( a ) (28)
where, F -' [F(a)] is f(x) and F - ' [ ~ ( a ) ]is g(x)
Fourter Trmdorm
Method

Proof: By definition of Fourier transform, we have

I f (x)e -'"'dx and F [g(x)]

m m

F [f (x)] = = g(x) e-iaxdx

Shlfting property: If F[f(x)] = F(a) and a is any real number then

F[f(x - a) j = e-la' ~ ( a )

Rroof: We have from the definition

m

F[f(x-a)] = lf(x-a)e-"'dx

Putting x - a = t or, x = t + a and d x = d t , w e h a v e

= I f(t)e-ia(l+a)dt=
cc
e-iaa
m
I f (t) e-"'dt
-m -w

Thus, F-I [e-""~(a)]= f(x - a) (31)

Modulation Theorem: If F [f (x)] = F(a) and a is any real number, then

1
F[f(x) cos a x] = IF(^ + a) + ~ ( -aa)] (32)
2
i
and [f(x) sin ax] = -[F(a+ a) - F (a - a)] (33)
2
Proof: We have
F [f (x) cos ax] = I f (x) cos ax e-"'dx
m

=
w

f (x)
[
elax + e-iax

)e-iaxdx
-m -m

1
+ a) + F(a - a) ]
= -[F(a
2
Similarly, result (33) can be proved.
-0-

Note: If Fs[f(x)] = F, ( a ) and Fc[f (x)] = F, (a)and a is any real number, then

-1
7; [f (x) sin ax] = [F, ( a - a) - F, ( a + a) ]
2
 i
Partial Differential 1
Equations Fc[f (x) cos ax] = -[Fc ( a + a ) + F, ( a - a)]
2

Convolution Theorem
Definition: The convolution of the functions f(x) and g(x) is denoted by f(x)*g(x)
and is defined by
m

f(x)*g(x) = Jf(tlg(x - t)dt (3 8)

-03

Theorem 2: The Fourier transform of the convolution of f (x) and g(x) is equal to the
prhduct of the Fourier transforms of f(x) and g(x) . In symbols,

Proof: We have

=
-m

-m
L -a,

-m
J

~ f ( ~ ) [ ] ~ ( ~ - t ) e - ' dt1
~~dx [changing the order of integration]

-t =u
[ ]
Putting x
dr = du

= s [f(x)l F [g(x)l= F(a) G ( a )

The inverse transform is given by
7 -' l~(a)G(a)l=
(f * g)(x)
-0-

Remark: The following important properties of convolution can be proved easily

(1) f(x)* g(x) = g(x) * f(x)
( 4 f(x) * [g(x) * h(x)l= [f(x) * * h(x)
(iii) f(x)*[g(x)+h(x)]=f(x)*g(x)+f(x)*h(x)
i.e., the convolution obeys the cpmmutative, associative and distributive laws of
algebra.
Fourier transform of nthderivative: If f (x), f '(x), ...,f ("-I' (x) all tend to zero as
m

x -t -tm and 1 f '" (x) 1 dx converges for each j = 0, I,.. .,n , then

Proof: Let us prove result (41) by ~nduction.First observe that (41) holds for n = 0 ,
by definition. We need to establish that if it holds for n = k ,then it also holds for
n = k + 1 . Intepating by parts, we get
m

r [ f ( k + o ( ~=) ] f(*+')(x)e-laXdx
-/
-0.)
 Fourler Tnnsform
Metbod

.. result (41) holds

[iorn-k ]
(42)

Now I f(k)(x)e-iaxI=( f ( k ) ( ~11 e-iax

) I=[ f ( k ) ( ~1,)and since f ( k ) ( ~+) 0 as x + +a,
by assumption, the first term on r.h.s. of Eqn.(42) vanishes. Thus,
~ [' kf+ " ( ~ )=] F(a) and result (41) holds for n = k + 1, which completes our

-0-

Differentiation with respect to frequency a : If F

! [f (x)] = F(a) then

~ [ x " f ( x )=] inF("'(a) (43)

Proof: From the definition, we have

~ ' ( a =) 4
da
>(x)
-m
m
e-jaxdx= I -[fdda
-m
(x) e-lax] dx
[
Differentiationunder
the integral sign. I
=-i f[xf(x)]e-""dx =-ig[xf(x)]
-a

I x f (x) e-"dx -i -[xdda f ( ~ ) e -]dx~ "

m
d
FR(a)= -i- =
da
-m -m

I x2f(x)e-iaxdx (-i)' ~ [ x ' f ( x ) ]

m

= (-i)' =
-a

Using induction result (43) can be proved easily. From result (43), we have
-'[F'" = (-i)" xnf(x)
(a)] (44)
-0-

Fourier transform of an integral: If g [f (x)] = F(a) and f (x) + 0 as x + a, and

F(a) satisfies F(0) = 0,then

(45)

-0-

Fourier transform of the Dirac-delta function: In Unit 6, we have shown that the
Laplace transform of the Dirac-delta function is one, i.e. 1[6(x)] = 1 . We also proved
there the filtering property of Dirac-delta functions as

If
m

(x) 6(x - a) dx = f (a) (46)

0
Similar results can be proved for the Fourier transform of the Dirac-delta function.
We define the delta function as
1
6(x) = lim-[H(x)
k+O k
- H(x - k)]
where H(x) is the Heaviside step function or unit step function.
Partlnt Differentlot
Equations

1 1- e - i ~ "
= lim -
k+ok[ L
ia
1=1 -I

Hence, 7 [6(x)] = 1 (47)

Therefore, Laplace transform and Fourier transform of the Diracdelta function are
both equal to 1.
Hence,
S -'[I] = 6(x) (48)
The filtering property given by Eqn.(46) gets modified as
41

j f (x)6(x - a)dx = f (a) (49)

-m

Further, using the shifting property given by Eqn.(30), we obtain

7 [6 (x - a)] = e -Iaa~ ( a=)e -Iaa (50)

Fourier s h e and cosine transform of derivatives: Let f(x) and f '(x) be continuous
on the interval [O,ao[ and f "(x) is piecewise continuous on every finite interval. If all
the three functions are abso)utely integrable on [ 0, ao [ ,then
!, [f (x)] + a f(0)
7, [f "(x)] = -a '!FC [f (x)] - f '(0) and !Fa[f '(x)] = -aZF (5 1)

Proof: From the definition we have

m

h [f '(x)] = f '(x) cos a x dx

0
(9
41
= [ f ' ( x ) c o ~ a x + a f ( x ) s i n a x ]-a2
~ [f(x)cosaxdx
0

= -a2FC[f(x)]-ff(0)=-a2~,(a)-fr(0) (52)
Similarly,
00

rS[fn(x)l= J fa(x)sin a x dx
0
m

=[fl(x)sinax - a f ( x ) c o s a x ] ~-a2jf(x)sinaxdx
0

=-a2~s[f(~)]+af(0)=-a2~s(a)+af(O) (53)
In the same way it can be verified that
FcIf'(x)l= aF,(a)- f(0) (54)
and 7, [f'(x) 1 = -aF, (a) (55)

Fourier Transform of partial derivatives: Let F(a, t) be the Fourier transform of the
function u(x, t) with respect to variable x ,then
00

F[u(x,t)] = F(a, t) = ju(x, t)e

-iax dX
(56)
 Fourlcr Transform
au
(i) Fourier transform of -: Suppose u -t 0 as x + fw ,then we have
Method
ax

Hence, =ia~[u(x,t)]=ia~(a,t) (57)

(ii) Fourier transform of

azu
-: If u and --,0
al
as x +iw ,then integrating by
axZ ax
parts as in (i) above, we have
I[$] =-az ~[u(x,t)]=-azs(a,t) (58)

In general, we have
I[$]= (ia)" rfu(x,t)l= (ia). ~(a,t)

where, u,-,T,..-,- an-'u+Oasx~*~.

ax ax ax "-'
au a l .
(iii) Fourier transform of - : By definition, Fourier transform of - w ~ t hrespect
dt at
to x is given by

d d
=--~[u(x,t)]=-F(a,t)
dt dt
Hence,

Note: that here we have written the total derivative, since fF[u(x, t)] depends only on
t and not on x .
In the same way it can be established that if F, (a, t) and F, (a, t) be the Fourier sine
au
and cosine transform of u(x, t) with respect to x ,then for u and - -t 0 as x -+oo ,
&
we have

and

We are leaving the proofs of the above results for you to do them yourself. Before
illustrating the properties discussed above we make the following remark.
Partial Ulfferentlal Remark: You may notice here that most of the properties (28), (29), (30), (31), (38),
Equations
(39), (40), (41) and (47) correspond to the analogous properties of the Laplace
transform. Properties (28) and (29) are identical to the corresponding properties of the
Laplace transform. The derivative property (41) IS similar to the property

but does not include any boundary values analogous to the initlal conditions
f (0), . ..,f ("-"(0) in Eqn.(63), because the boundaries are at x = fa,,and it is
assumed that f(x), f '(x), ...,f ("-" (x) all tend to zero as x -+ +_a
. The convolution
properties (39) and (40) are identical to the corresponding ones for the Laplace
transform, but keep in mind that the Fourier and Laplace convolutions are not quite the
same. In contrast with the Fourier convolution (38), the Laplace convolution was
defined as

(f * g)(x) = I f ( t - rIg(r)dr. (64)

0
That is, the integration limits are different. Let us now consider the following
examples where we have u'sed the properties of Fourier transforms and the table given
in the Appendix to solve the problems.

Example 16: Find F-' [c]

+
3 ia
1
Solution: From Example 9, we have ~ [ e - ~ k ~ (=-x ) ]
3 + ia

Using shift property (3 I), we get

Example 17: Given f(x) = 4e-I" - 5ee-3'"'21evaluate its transform F(a) .

Solution: Using the linearity property (28), we have

F[4e-IA1- 5e-31"+21] = 4F{e-IX1
) - 5~{e-'1x'2'
)
gntry 4 in Appendix gives
' ?

[also see E8) ]

and entry 11 (with a = 3, b = 6 ) gives

r J
~ - I x + ~ I ) =( ~ P x + ~ I

Thus, we have
 Fourier Trnnsform
Example 18: Find f [f (x)], where f (x) = x e4"' Method

Solution: We have -
&-4x2)=$e-a1/16
(see Example 10)

I Entry (17) with n = !gives

a -
da
[(".
2
- a 2)=-*-ae -a2/11

Example 19: Find 9 - ' [ ~ ( a] )= e-21u'.

Solution: Entry 1 with a = 2, gives

F-~{;e--~} == 1
7C
Using the linearity property (28) with a = -and b = 0 , the above equation can be
2
,,written as
"f-1 6-21.1)-
1
2 x2+ 4
or, f-l(e-4ai)=2-
1
It x 2 + 4
is the desired inverse function.
***
1
Example 20: Evaluate 9-I

Solution: Completing the square, we write

:. given problem reduces to

Entry 4, with a = 3 and linearity gives

Now using entry 12 with a = 1and b = 2 , we have the desired result

Alternatively, we could have used partial fractions, as follows

Partial Differential Now using the linearity property with entries 2 and 3 of the Appendix, we get for unit
Equations
step function H(x) .
{ 1 }
= 1 H ( x ) c - ~ + 2 i )+x iH(-x)e(3-"b
a2+4a+13 6 6
(also see Example 9)

Example 21: Find 3

Solutlon: We have
I-[
-' 1
a 2+ I
.

Using entries 2 and 3, respectively

o x 7 F-'{--!-} =H(-x)ex
1-ia

+ -'{L}
1+ i a
= H(x) e-'

x<Ot Then the Fourier convolution formula (40) gives

Graphs of H(T - X)and H(r) both for x > 0 and x < 0 are shown in Fig.5.

Example 22: Find the solution of the differential equation

y ' - 4 y = ~ ( ~ ) e - 4 * ,- a < x < a
Using Fourier transform, where H(x) = u,(x) is the unit step function.

Solution: Applying the Fourier transform to the given differential equation, we

obtain

or, i a F(a) - 4F(a) = -1

4+ia
[using entry 2 ]
 -- -
1 - 1 Fourler Transform
or, F(a) = - Method
( 4 + i a ) (4-ia) 16+a2
where, F [Y(x)I = F(a)

Therefore, y(x) = 3 [-1-

1
16+a2
1
= --e-4ixl
8
[using entry 4 I
I
! The solution can be also be written as

Example 23: Using the formulas for derivatives find the Fourier cosine and sine
transform of
f(x)=e-ax , x l O , a>O.
Solution: Given f (x) = e-"" , we have f '(x) = -ae-ax and f "(x) = a 'e-'" . If
FC[f(x)] = Fc( a ) , then we have
FC[f "(x)] = Fc[a e-" ] = a 2 FC[e-"' ] = a FC(a)
also, FC[f "(x)] = -a2FC(a)- f '(0) = -a2FC(a) + a [using Eqn.(S I)]

Therefore, a 2 F, ( a ) = -a 2 Fc( a ) + a, or Fc( a ) = - a

a 2+ a 2
Similarly, we have
33[f "(x)] = a ys[e-"I= a2FS( a )
and F3[f"(x)]=-a2~s(a)+af(~)=-a2~s(a)+a
[using Eqn.(S I)]
a
therefore, a 2 F s ( a ) = - a 2 ~ , ( a ) + a ,or, F s ( a ) =
a2+ a 2

You may now try the following exercises.

E 16) Prove the results given by Eqns.(61) and (62).

El 7) Using the table given in the Appendix, or otherwise, evaluate the following

3-I C-1 cos a )

El 8) Given F(a) find the inverse function f (x) ,where 3[f(x)] = F(a) .
a) e-"Ia' , a > 0 b) H ( a + a ) - H ( a - a ) (a>O).

E19) Using the convolution theorem find the inverse of the following functions
1
a) , k>O b) , k>O.
(ia + k)2 (ia + k)3
v
-* Partla1 Differential E20) Using Fourier transform, find the solution of the following differential
Equations equations
!6jl a) y ' + 3 y = ~ ( x ) e - 2 x, --m<x<-m
ti b) y R + 3 y ' + 2 y = 6 ( x - 3 ) .
t.
In many applications, we have to deal with problems defined on finite intervals. In
such cases we define the finite Fourier sine and cosine transforms.
7.3.2 Finite Fourier Transforms
The finite Fourier sine and cosine transforms are obtained from the Fourier sine and
cosine series.
Finite Fourier Cosine Transform: You know that if a function f(x) is defined in the
interval 0 Ix < L and is piecewise continuous then half range cosine series of f(x) is
given by - -
a
f(x)=L+
2 n=,
"
xancos-
nxx
L

L i
where, a, = - f(x) dx and a, = - f(x)cos-dx
0
L i
0
nxx
L
The finite Fourier cosine transform of f(x) is defined as
L
nxx
J
Fc(n) = f(x) coo -dx
L
0
where n is a non-negative integer.
The function f (x) is then called the inverse finite Fourier cosine transform of
Fc(n) and is given by
1 2 " nxx
f(x)=-FC(0)+-x Fc(n) cos-
L L t,=, L

Note: From Eqn.(66), we have

on substituting Eqns.(69) and (70) in Eqn.(65) we obtain result(68) for inverse finite
Fourier cosine transform.
Finite Fourier Sine Transform: If a piecewise continuous function f(x) is defined
in the interval 0 S x 5 L ,then half range sine series is given by

f(x) = 2
n=l
b,, sinl
nxx

L. *i0
nxx
where, b,, = - f(x) sin-dx
L
.

The finite Fourier sine transform is defined as

L
nxx
F, (n) = f(x) sin -dx (73)
L
0
where n is a non-negative integer.
The function f(x) is then called the inverse finite Fourier sine transform of F, (n)
and is given by
 2 " nxx Fourier Transform
C
f(x) = - F, (a) sin -
L n=l L
Method

From Eqn.(72), we have

2 nxx 2
b. =-L I f ( x ) sin-dxL =-Fs(n)
L
0
and on substituting b,, from Eqn.(75) in Eqn.(71), result (74) is obtained.
Remark: The finite Fourier cosine and sine transforms can also be denoted by C, (f)
and S, (f) respectively.
Formula for the second derivative: W e assume that f(x) and fl(x) are continuous
and f '(x) .is piecewise continuous on [0, L 1. Then
., .

Proof: Using the definition, we have

L
nxx
C n [ f m ( x ) ] =IfR(x)cos-dx
L
0
L nlx2 L
nxx nx nnx nxx
= [f~(x)cos- L + --f(x)sin-]o
L - Jf(x)cos-dx
L L
0

n2n2
.. C, [f "(x)] = -- F,(n) - fl(0) + (-l)"fl(L)
L~
Similarly, using the definition of finite Fourier sine transform it can be proved that
n2n2 nn nn
S , [f '(x)] = -- F, (n) + -f(O) - -(-1)" f(L)
L~ L L

We now illustrate the application of finite Fourier transform through examples.

Example 24: Find the finite Fourier sine and cosine transforms of the following
function
kx , o_<x_<x/2
f (x) =
k(x-x), x12lxln

Solution: Using the result (67), we have

L n
nxx
F,(n)= If(x)cos-dxL = If(x)cosnxdx

= k[ x( s-
ly)-[ --cosnx'l]n" [
+ k (n-x)
(siy)
- -(-I) [c;~)]'
-
nz / 0 n/Z
\
x s i n n n l 2 cosnnI2 cosnx nsinnxI2
= k[(- + + cosnnI2
2 n nz h2 2 n n2

at n = 0 , F, (n) becomes indeterminate, hence we evaluate F, (0) separately thus

Partlal Dlflenntlal
Equations

--
n2
-1
Hence the finite Fourier casine transform is

n2
and Fc(0) = -
kn
4
For finding finite Fourier sine transform we use result (73) and obtain
L X
nnx
F,(n)= Jf(x)sin-dx= Jf(x)sinnxdx
L
0 0

Hence, the finite Fourier sine transform is

Example 25: If f(x) =sin kx , where 0 S x I n and k is a positive integer, then

show that
n .
F,(n)=O, if n # k and F,(n)=-, if n = k .
2
Solution: Using result (73), we have
n
F, (n) = sin kx sin nx dx

=-[
1 sin(n-k)x
2 n-k
- sin(n+k)x
n+k I'
= 0 , if n # k [as (n - k ) , (n + k ) are integers]
When n-=k ,the first term of the above expression becomes indeterminate. We
evaluate F,(n) directly by putting n = k . Thus,
 cos 2nn / 3 Fourler Transform
Example 26: Find f(x) , if Fc(n) = ------ where, 0 < x I. 1 . Method
(2n + 1)2
Solution: Since finite Fourier cosine transform is given, we use result (68) to find
the inverse. Thus

1 2 " cos2nxI 3 nxx

=-(I)+~X cos - [.:L=l and Fc(0)=l]
1 ,,=, (2n + I ) * 1
" cos2nx/3
=l+2X cosnxx.
,,=I (2n + 112

You may now try the following exercises.

E21) Find the finite Fourier sine and cosine transforms of the following functions:
a) f ( x ) = C x - x 2 , OlxSC

E22) Find f(x), if

cC . nlra
b) F,(n)=-sin- and Fc(0)=ac,where 0 1 x ~ ~
nn C

)I
Fourier transform method has become a powerful tool in diverse fields of science and
engineering. It can provide a means of solving equations that describe dynamic
responses to electricity, heat or light and can also identify the regular contributions to
. a fluctuating signal, useful in making sense of observations in astronomy, medicine
and chemistry. Founer transforms become indispensable in the numerical calculations
needed to deslgn electrical circuits, to analyse mechanical vibrations, and to study
wave propagation. They are widely used to solve problems involving semi-infinite or
totally infinite range of the variables. In the next section, we deal with some
applications of Fourier transforms to diffusion, wave and Laplace equations.

7.4 APPLICATIONS OF FOURIER TRGNSFORMS TO

BOUNDARY VALUE PROBLEMS
Fourier transforms are of great importance in solving boundary value problems and
form a basis of powerful method for the solution of partial differential equations (pde).
For solving boundary value problems, we take Fourier transform of a given pde using
given boundary and initial conditions. The required solution is then obtained by
taking corresponding inverse transform. The choice of particular transform to be
employed for the solution of an equation depends on the boundaryconditions of the
problem.
il

Partial Differential We use the following notations for different types of Fourier transform and
Equations
corresponding inverse transform
For the interval - a,< x < oo :We use infinite Fourier transform
m

F [u(x,t)] = ti ( a , t) = J u(x, t) e-"-x

--a2

-
and corresponding inverse transform is given by
1
u(x.t)=- Jii(a,t)eiaxd a
2X
(79)
-00

For the interval 0 < x < a, : We use Fourier sine or cosine transforms depending
upon the boundary conditions of the problem.
Fourier sine transform:
m

Fs[u (x, t)] =a, (a, t) = u(x, t)sin a x dx

0
and the inverse sine transform is
00
2 -
) x us ( a , t)sinax d a
u ( x , ~=-
0
Fourier cosine transform:
m

F~[u(x,t)] =tic ( a , t ) = Ju(x, tlcosax dx

0

and the inverse cosine transform is

0
For the interval 0 < x < L :We use finite Fourier sine or cosine transforms depending
upon the boundary conditions of the problem.
Finite Fourier sine transform:
L
nnx
FS[u(x.t)=Gs(n,t)=Ju(x,t)sin-dxL

---
0

and inverse finite Fourier sine transform is

2 " - nm
U(X,t) = us (n, t)sin-
n=l L
Finite Fourier cosine transform:

and inverse finite Fourier cosine transforni is

1
u(x. t) =-6, (0, I) +
L
2 "
L .=,
--x n ~ x
ii, (n, t)cos-dx
L

The choice of Fourier transform:

Depending on boundary conditions, we select the Fourier transform to be employed

for the solution of pdes.
If the interval I s - a,< x < and if the boundary conditions api: u -+ 0 and
all
--+ 0 as x -+ * a, we use infinite Fourier transform.
ax
If the interval is 0 < x < oo and if the boundary conditions are
I al
(i) u and -+Oas x+m and u(x, t)=O or, f(t) at x=O,Vt,weuse
Fourier Transform
Method
ax
Fourier sine transform
(ii) u and
au
,-+O as x j o o and - = 0 or f(t) at x=O,Vt,weuse Fourier
I
ax
cosine transform.
al
Note: For the interval 0 < x < oo ,we always assume u and - +0 as x -+ oo ,even
I
ax
if it is not given in the problem.

If the interval is 0 < x < L and if the boundary conditions are

(i) u(0, t) = u(L, t) = 0 Vt ,we use finite Fourier sine transform
al al
(ii) -(0, t) = -(L, t) = 0 Vt , we use finite Fourier cosine transform.
ax ax
We now take up illustrations on solutions of boundary value problems.
Diffusion Equation
Let us consider the problem of flow of heat in an infinite medium - oo < x < a, ,when
the initial temperature distribution f(x) is known and no heat sources are present.
Example 27: Use Fourier transforms to solve the boundary value problem

subject to the conditions

(ii) u(x,O)=f(x).
Solution: Since the interval is -a < x < a, , we use infinite Fourier transform.
Talung the infinite Fourier transform with respect to x of both sides of the gven pde,
we have

Let ii(a, t) be the infinite Fourier transform of u(x, t) , we then have from result (78)
-u (a,t) = mJ u(x, t)e-iaxdx

[by result (60)] (89)

Similarly, F I$] =
ax2
dx = -a2ii(a, t) [by result (58)] (90)

Substituting from Eqns.(89) and (90) in Eqn.(88), we get

dii = -ka2u
-
dt
The solution of this ordinary differential equation is
-
u ( a , t) = A e-ka2t (91)
To find the arbitrary constant A, we take infinite Fourier transform of initial condition
(ii), then
Partial Differential -
-
Equations u (a, O)= J u(x, 01e-iaxdx = J f(x) e-lax dx
I
-00 -00
-
:. u ( a , 0) = F(a) (92)
Now if we put t = 0 in Eqn.(91) and use result (92), we obtain
-
u (a,O) = A = F(a)
Thus,
-
u (a, t ) = ~ ( ae-ka2t
)
Taking the inverse Fourier transform of the above equation, we have I

--
- 1 ( (( l l - i a u , a eiax d a

which is the required solution. This can be further simplified as 1

Using result
m
1 n
e-aa' cos 2 b a d a = -
2
Feb2"
a
with a=ktand b=-
X-U
2
,we have

vu
Remark: The conditions -+ 0 as x -P f a, represents that temperature u(x, t)
ax
must be bounded. This condition can also be expressed as (u(x, t) ( < M .
We now take up an example on heat-flow in semi-infinite bar.
Example 28: Determine the distribution of temperature in the semi-infinite medium
x 2 0 , when the end x = 0 is maintained at zero temperature and the initial
distribution of temperature is f(x) .

Solution: Let u(x, t) be the temperature at any point x at any time t , we have to
solve the heat flow equation

subject to the initial condition

u(x,O) = f(x) , x > 0
and the boundary condition
u(O,t)=O, t>O
 dl Fourier Transform
uand -+O as x + a , V t . Method
dx
Since the interval is 0 < x < a, and u(x, t) at x = 0 is given, we use Fourier sine
transform given by Eqn.(80).
I

Taking Fourier sine transform with respect to x of both sides of the given partial
differential equation, we have

1 and %[T]=~;;isinaxdx
azu azu
0

= a[u(x, t)],=, - a 2 L S ( a , t) (using result (61))

= -a2iis(a, t) (95)

Substituting from Eqns.(94) and (95) in Eqn.(93), we have

Solution of this equation is

- 2 2
u,(a,t)=Ae-'a '
Now we take Fourier sine transform of initial condition, i.e.
w

I
Fs[u(x,O)] = us ( a , 0) = u ( x , ~ ) s i na x d x
0

= I f (x) sin a x dx
w

= Fs ( a )
Putting t = 0 in Eqn.(96) and using Eqn.(97) we obtain
-
us (a,O) = A = Fs ( a )
The solution (96) thus reduces to
-
us ( a , t) = ~ ~ ( a ) e - ' ~ " ~ '
and its inverse Fourier sine transform yields,
U(X,t) =?b F , a ) e - c 2 a 2 'sin a x d a
7t

which is the required solution.

1
Using the identity sin a x sin a u = -[cosa(u - x) - cos a(u + x)] and the result
2
Q

I e-a2 cos b a d a =- ,we can write the above solution( on substituting

u -X U +X
- = v in the first integral and -= v in the second integral) as
2c4 2cJ;
Partial Differential
Equations

Wave Equation
Wave motion that occurs in nature, viz sound waves, surface waves, transverse waves
of an infinite string, and of mechanical systems are governed by the wave equation.
We now illustrate the application of Fourier transform to wave equation.
Example 29: Find the solution of the wave equation
-=c -
at? ax'
subject to the conditions
(i) y(0, t) = y(x, t) = 0
(ii) y(x, 0) = f (x)

Solution: Since the boundary conditions are y(0, t) = y(x, t) = 0 , we take finite
Fourier sine transform with respect to x of both sides of the given equation. We get

Using formula (84) with L = x , we get

-
y, (n, t) = I
X

0
y(x, t)sin nx dx

(ycosnx): +n y(x, t)sin nxdx

0

Solution of Eqn.(98) is
-
y, (n, t) = A(n)cos cnt + B(n) sin cnt
Now taking the finite Fourier sine transform of conditions (ii) and (iii), we get

-1
dt ,=o
= g(x) sin nx dx = Fs [gin) ] (101 j

From Eqn.(99), we have

6,
-= -A(n) (cn sin cnt) + B(n) (cn cos cnt) (102)
dt
Putting t = 0 in Eqn.(99) and (102) and using Eqns.(100) and ( l o l ) , we obtain
 Fourier Transform
Method

Hence, on substituting the above values of A(n) and B(n) in Eqn.(99), we obtain
- 1
y, (n, t) = F, [f (n)]cos cnt + -F, [g(n)]sin cnt
cn
Taking inverse Fourier transform, we have
2 " 1
y(x. t ) = - [Fs[f (n)]cos cnt + -F, [g(n)] sin cnt] sin nx
X ,,=I cn
using results (100) and (IOl), we have

.(x, t) ==
'
= ,111
[{I flU)sinnu du]coscnt + 6.i g(u)sin nu du]sin cnt]sin nx

which is the required solution.

***
Laplace Equation
Steady flow of current in solid conductors, the velocity potential of inviscid
irrotational fluids etc are governed by Laplace equation. We shall now consider a few
related examples.
Example 30: Find the steady state temperature distribution u(x, y) in a long square
bar of side n with one face maintained at constant temperature uo and the other faces
at zero temperature.
Solution: Mathematically, the given problem is described by

subject to the boundary conditions

4 0 , ~ =u(n,y)
) =O
u(x,O) = u(x, n) = uo
Taking the finite Fourier sine transform with respect to the variable x , we get

which after using the conditions u(0, y) = u(n, y) = 0 , yields

General solution of Eqn.(l03) is given by

-
u,(n, y) = A(n)cosh ny + B(n)sinh ny ( 104)
Taking the finite Fourier sine transform of the second set of boundary conditions, we
have
-
u,(n,O) = O
X
-
and u,(n,n)=Ju,sinnxdx=uo
0
,Using these in Eqn.(l04), we get
,, (1 - cos nn)
A=O,B=uo
n sinh nn ,
' Hence, is(n, n) = -
slnh nn
(l-Crnn)sinhny
U0

Finally, taking the inverse finite Fourier sine transform, we obtain

Partial Differential
Equations

Thus the required temperature distribution is

sinh (2k + l)y sin (2k + 1) x
, for n odd
(2k + 1) sinh (2k + 1) n
, for n even
***
Example 31: Find the bounded harmonic function V(x, y) in the semi-infinite strip
x > 0, 0 < y < 1 that satisfies the boundary conditions
v \ ( O , ~ ) = . O, v y ( x , O ) = O , v ( x , l ) = f ( x )

Solution: Here we have to solve the two dimensional Laplace's equations

a'v a 2 v
-+7- - 0 , x>o O<y<l
ax' ay-
subject to conditions

(ii)
av
- = 0 for y = 0 , for all x
4,
(iii) v(x, 1) = f(x)
Taking Fourier cosine transform with respect to x of both the sides of the given
equation, we have

01

Let v, ( a . y) = Fc[V(x. y) 1= V(X,y) cosax dx .

0

Hence we obtain

d'v,
--
,
or, a-v, =0
dy
whose solution is given by
v, ( a , y) = C , cosh a y + C 2 sinh a y

T o find C , and C, we take Fourier cosine transform of given conditions (ii) and (iii)

From Eqn.(l06), putting y = 0 and using Eqn.(l07), we obtain C 2 = 0 .

:. V, (a,y) = C , cosh a y (108)
' To obtain C , we take Fourier cosine transform of condition (iii), we get

I v(x,l)cosaxdx= ~ f ( x ) c o s a x d x
m 00

v,(a,l)=
0' 0
Therefore, from Eqn.(l08), we have
 Fourier Transform
Method

1, O < x < l
(a) when f (x) =
0, x>l

sin a x sin a

sin a
:. C, =
a cosh a
sin a cosh a y
Hence v, ( a , y) =
a cosh a
Taking inverse Fourier cosine transform we obtain

v(x,y)=-
rr
I
2 rn sin a cosh a y
acosha
cos a x d a
0

(b) when f (x) = e-" , x > 0

rn . -1
I
CI cosh a = e-' cosax dx = -
0
I + a'

cosh a x
and v,(a.y) =
(1 + a 2 ) c o s h a
Taking inverse Fourier cosine transform, we obtain
2=
V(X.y) = - I cosh a x
rr ( l + a ' ) c o s h a
cos a x d a (1 10)
0

Remark: You may be wondering whether the integrals like the one in Eqn.(l09) or
(1 10) can be evaluated. In most cases the answer is no. One way of finding a specific
value. say, of the temperature v(1,l) in Eqn.(l10) is to resort to the numerical
integration.
You may now try the following exercises.
E23) The initial temperature along the length of an infinite bar is given by

If the temperature u(x, t) satisfies the equation.

du
-=- a=u , - m < x < m , t>O
at ax'
Find the temperature at any point of the bar at any time t .
E24) Use Fourier transform to solve the equation

subject t o the following conditions

(i) u(O,t)=O, t > O
>
L
Partial Differential
1, O < x < l
$ Equations (ii) u(x,O) =
0 , x >1
(iii) u(x, t) is bounded.

, E25) Use finite Fourier transform to solve

subject to the conditions

(i) u(x,O) = 2x , 0 < x < 4
(ii) u(O,t)=u(4,t)=O , 0 < x < 4 , t > O .

E26) Show that the solution of Laplace equation v2v = 0 for v inside the semi-
infinitestrip x > O , O < y < b , s u c h t h a t v = f ( x ) when y = O , O < x < a , ; v = O
.when y = b , O < x < a , and v = O when x = 0 , O < y < b isgivenby
sinh(b - y)a .
sln a x sin a u du d a .
0 0

Before we end the discussion in this unit we would like to make the following remark.

Remark: We have seen that the Fourier transform and the corresponding inversion
formula are actually just a restatement of the Fourier integral representation of a non-
periodic function defined on - a, < x < co . We have also seen that the Fourier
transform methodology is analogous, and in some aspects identical, to the Laplace
transform methodology discussed in Unit 6. Given that similarity, the important
question arises, as to which of the two transform methods to use in a given differential
equation application. The general guideline is as follows:
The Laplace transform is tailored to initial-value problems on a semi-infinite
interval 0 < t < co .
The Fourier transform is tailored to boundary-value problems on an infinite
interval - co < x < co .

The independent variables are usually t (time) and x (linear dimension) but not
always. Further, there does exist Laplace transform defined on an infinite interval, the
Laplace transform can sometimes be used for boundary-value problems on finite
intervals, and the Fourier transform can be adapted to boundary-value problems on a
semi-infinite interval 0 < x < co or, a finite interval 0 < x < L . But the guideline given
above provides the general rule of thumb for selecting one transform or the other.

We now end this unit by giving a summary of what we have covered in it.

7.5 SUMMARY
In this unit we have covered the following points.
1. The Fourier integral ,
lrn
f (x) = - [A ( a ) cos a x + B ( a ) sin a x ] d a
X
0
m

where, A(a) = J f (x) cos a x dx

-m
I+)

B(a) = f (x) sin a x dx

 is a way of representing a non-periodic function that is defined on the infinite Fourler Tmnsiorm
interval ] - co,co [ . Method

2. When the function f(x) is an even function, expression for f(x) in 1) above
becomes a cosine integral and when f(x) is odd it reduces to a sine integral.
3. The complex form of Fourier integral representation of a function f(x)
defined over the interval ] - a,co [ is given by

-m
m
where, C ( a ) = 1f(x) e-"" dx
-0)

4. Integral transforms occur in transform pairs: the transform itself and the
inversion integral.
Some of the transforms and their inverses are as follows:
m

a Fourier transform: g[f(x)] = 1f(x) e-'"dx = F(a)

a
1
I
Inverse Fourier transform: g-' {F(a)) = - ~ ( a ) e " " d a= f(x)
2x
m
-
m
Fourier sine transform: Fs[f (x)] = J f (x) sin a x dx = F, ( a )
0

a
2"
I
Inverse Fourier sine transform: F;' [F, (a)] = - F, ( a ) sin a x d a = f(x)
x
0
-
I
Fourier cosine transform: gc[f (x)] = f (x) cos a x dx = F, ( a )
0
a Inverse Fourier cosine transform:
2"
I
gc-' [F, (a)] = - Fc( a ) cos a n d a = f (x)
x
0

5. If g [f(x)] = F(a) and F[g(x)] = G(a) and a and b are arbitrary scalars ,then
F [a f(x) + b g(x)l= a F [f(x)l+ b F [g(x)l= a ~ ( a+)b ~ ( a )
a [f(x - a)] = e-la"(a)
1
F [f(x)cosax] = -[F(a + a) + F(a - a)]
2

where, f(x) * g(x) = I f(t) g(x - t) dt

 Partial Differential where 6(x) = l i m [ ~ ( x -
) H(x - k)], H(x) is the unit step function
t Equations k +O
bb
R 7 16(x - a)] = e-I" 7[6(x)] = e-la"
7, [f '(41 = -az7, [f(x)l- f '(O)
in"

7, If "(x)l=-a = 9,[f(x)I+ af(0)

7, [f '(x)= a Fa(a)- f(0) where, Fa(a)= 7, [f(x)]
7,[f '@)I= -aF, (a) where, F, (a) = 7, [f(x)]

6. If F(a, t) be the Fourier transform of the function u(x, t) with

respect to the variable x ,then

I : [1
f - = iaf[u(x, t)] = iaF(a t)

f 2
1 = -azf [u (', t)] = -az F(a, t),
au
and --+ 0 as x -+ am.
ax

7. To solve a boundary value problem by an integral transform we :

(i) transform the partial differential equation into an ordinary differential
equation
(ii) transform all boundary or initial conditions that depend on the transformed
variable
(iii) solve the ordinary differential equation subject to the conditions found in
step (ii)
(iv) find the inverse transform of the function found in step (iii). This is the
solution of the original boundary-value problem.

1
1
El) a) f (x) = - [A(a) cos a x + B(a) sin ax] d a
n
8 0

Where A(a) = (3a sin 3a + cos 3a - 1)/ a Z

B(a) = (sin 3 a - 3a cos 3a) / a 2

1=
1
f (x) = - [A(a) cos a x + B(a) sin ax] d a
n

[A(a) cos a x + B(a) sin ax] d a

X
0

1
A(a) =- -+ -{cos2(a + 1) + sin 2(a + I)]
2 az-1 a + l
1
+ -{sin
a-1
2(a - 1) - cos 2(a - 1))
I
B(a) = [A
2 az-1
- -{sin
a+l
1 2(a + 1) + cos 2(a + l)j
 1
+ -{sin
a -1
2(a - 1) - cos 2(a - I)}
I Fourler Transform
Method

L. 2 " ( n a sin n a + cos n a - 1)cos a x

rI b) f ( x ) = -nI a da
i 0

c) f(x)=-
2"
n
-2 I
[1+ cos an]
a 2-1
cos a x d a
0

2 " 2 [sin 3 a cosh 3 - a cos 3 a sinh 31

I d) f(x) =-
n
J 1+a2
sin a x d a
0

1 E3) a) Consider the right hand side, which defines the function
f(x) = t e - " cos x, x > o
Since the integral
- on the left hand side is from 0 to oo and involves sin a x ,
F we take Fourier sine representation of f (x)
0

We have, B (a)= n
-e -u
coo u sin a u du
2

Integrating by parts, we obtain

Using Eqn.(9), we get

f(x)=-
2
n
7 --
0
n a3
2 a4+4
sin a x d a =
OD

0
a3sin a x
a 4+ 4
da

representation of f (x) .
- o<x<n
b) Consider the function f(x) = 2 '
0 , x>n 1"
,and take the Fourier sine

n 1- cosan
=-[
2 a ]
Using Eqn.(9), we get
Partlal Dlfferentlal m

Equations sin a x d a .
=0 l T a n

E4) Compare with Fourier sine integral representation

I I
00
2
Let g(x) = - B(a) sin a x d a where B(a) = g(x) sin a x d a
n
0 0
Set x = a ; n g ( a ) = l for O < a < l and 0 for a > 1
B ( x ) = f ( x ) ; f(x)=2(1-cosx)/(nx), x > 0 , O < a < l .

E5) a) Let x = 2 in Eqn.(l 1). Use a trigonometric identity and replace a

b y x.
b) Make the change of variable 2x = kt .

=2n l ~j ( aeiaxda when C(a) I f (x) e-'""dx

m

E6) f (x) ) =

k-ia
a) C(a) =
k2+a2
4 2a
c) C(a) =- cos a a sinh 2a + -sin a a cosh 2a
4+a2 4 +a2
2ni
d) C ( a ) = ,
n -a2
sin 2 a

e a x , x<O
E8) We have f (x) =
e-ax, x>O

Therefore, F [f(x)] = I eaxe-'" dx + I e-'"e-'"" dx

0 m

= +1 - - 1 - 2a
a-ia a+ia a2+a2
The inverse transform is given by F -' [ a 2 2ta2] =e-al"l.

E9) Taking the Fourier sine transform of the given function, we obtain
m e-ax

F, ( a ) = f (x) sin a x dx = -sin a x dx = I(a) (say)

x
0 0
 Fourier Transform
Metbod

a
(-a cos a x + a sin a x )
a2+ a 2 a 2+ a 2

:. I(a) = I a*a+ a , d a + a
A = tan-' - + A
a

:.A = 0 and I(a) = tan-' 5

a
Inverse sine transform of F, ( a ) is written as

I Putting x = 1 in the above equation, we obtain

I tan-' -sin
w
X
:. x dx = Ee-. is the required solution.
a 2

E 10) Taking Fourier cosine transform of f (x) = e-\ we obtain

m

Inverse Fourier cosine transform of F, ( a ) gives

-
cos a x n -,
da=-f(x)=-e
2 - 2
Putting x = m in above equation, we have

Hence, I
w

0
cos TTlX
-dx
1+x2
= -e
7C -m
2

E l 1) a) The given integral is the Fourier sine transform of f(x) . Its inverse
Fourier transform is obtained as

= f(' -;:
'), is the required solution.
Partial Differential
1-a, Olall
Equations E 12) We have Fc( a ) =
I f (u) cos a u du
0
=
0 , a l l
The inverse Fourier cosine transform of Fc ( a ) is given by

(Putting u = 22 )

2 sin ax
E14) f ( x ) = '
nxz

= a [ u ( ~ , t ) ] ~ . ~ - ausinaxdx
'~ f: u+O as x + m )
0

= a [ u(x, t)],,O - a 2Fs [u(x, t)l

= a[u(x.t)lx=o - a 2F, (a, t)
which proves the resu!t (61). Result (62) can be proved similarly.

(entry 17)

(entry 4)
 Fourier Transform
Method

1 x
b) With c = 3 , a = 1 , f(x)=- in entry 13 and F(a) = -e-Ji" ,we have
x2+ 2 JZ

(entries 1.4)

1
= 2[H(x + 1) - H(x - I)] - (entries 9,7)
d
m

0 -' k+"' cos a]=l~-'

2x
(2(ne:lal) cos a]

+ (entries 15 ,I)
(X+u2 +1 (x-1)2 +1

(entries 2 , 3 )

sin ax
b) f(x) =- .
IIX

E19)a) [ 1
i a + k12 ] = I -(ia
'[ ]
+ k) (ia + k) = "-"H(~)]* [e-"H(x)]

-m

since H(r)H(x-r)=O for r < O and r > x , a n d 1 for O < r < x .

b) x 2 e - " H ( x ) / ~ [use the result of a) above]

1
E20) a) F(a) = , y(x) = e-2x(1 - e-!' ) H(x)
[(2 f ia) (3 + ia)]
e-3ia
b)- F(a) = , Y ( ~=)[e-("-" - e-2(x-3)]~(x
- 3).
[(2 + i a ) (1 + ia)]
Partial Differential L
Equations E21) a) We have Fc(n) = J f(x)-
0
cos nxx
L
dx=
i
0
(ex-x )-
cosnxx
e
dx

.:L=e and f ( x ) = e x - x 2
e sinnxx

at n = 0, Fc(n) becomes indeterminate, hence we evaluate Fc(0)

separately,

J (ex - x2)& =-e6

(

F,(o) =
0
Hence finite Fourier cosine transform of f(x) is
- e3 e
Fc(n)=-(cosnx+e) and Fc(0)=-.
n2x2 6
(
nxx
Now Fs(n)= J (ex-x )sin-&
P
[.: L = e and f ( x ) = e x - x 2 ]

2c3
=- (e - cos nx) is the finite Fourier sine transform of f(x)
n3x3

\ 2 x
d) Fc(n) = -, Fc(0) = -
m2 3

1 nxa nxx
b) f(x)=-+- -sin -cos -
e

E23) Taking ~ o k i e transform

r of both the sides uf given pde, we obtain
 Fourier Transform
dii - -a2G
-- . Method
dt
whose solution is given by
-u ( a , t) = A e-a"
, A being an arbitrary constant
' Taking Fourier transform of initial condition, we obtain

Substituting t = 0 in Eqn.Cl11) and using Eqn.(l12), we get

. u(a, t) = Asin a da2'

a
Taking inverse Fourier transform of G(a, t) ,we obtain

(second integral vanishes as its

integrand is an odd function)

27 e-a2,[sin (I + x) a + sin (1 - x) a
I
=
x a
0 da
which is the required solution.

E24) Taking Fourier sine transform of both sides of the given pde, we obtain

whose solution is glven by

b
-us ( a , t) = A e - ~ ~A' being
, constant
m 1 m

Now ii, (a,O) = I u(x,O)sinaxdx I sin a x d x + I 0.sinaxdx

=

Putting t = 0 in Eqn.(l13) and comparing with Eqn.(l l4), we obtain

1- cosa
A=
a
:. -us ( a , t ) = 1- C O S e~-a2t
a

1
LO
2 1 -cosa -a2t .
and u(x, t) = - e stn a x d a , is the required solution.
x 0
a
E25) Taking finite Fourier sine transfop of given pde with L = 4 , we obtain

idu . nxx
at 4 dt
0
nnx
u(x,t)sin-dx
4
=-
dii,
dt
Pnrtlal Differentlai
Equations -
-
cosy)
E[(~
4
-1
4 4
-nn . nnx
u(-sln4 -)dx]
4

-*
-j 16
0
0

nnx
0

~ ( ~ , t ) s i n - d x =-
4
-n 2n2 -
16 us
dii -kn2n2 -
.. S-
- us
' dt 16
\
The soldion of this ode is written as
- -
-kn2n2
us(n,t)=Ae l6 (1 15)

Now J,(n,O)= I 2 x s i n4E dx=- -nn32

4
cos nn (1 16)
0
Substituting t = 0 in Eqn.(l15) and using Eqn.(l l6), we obtain
- 32
A =-cosnn
nn
- - 32 - -kn2n2
. us (n, t) = ---cosnne l6
nn
Taking inverse finite sine transform of ii, (n, t) ,we obtain

-kn2n2
t nlcx
e l6 sin-
4

E26) Here we have to solve twodimensional Laplace equation

subject to the conditions

(i) v(x,O)=f(x), x>O
(ii) v(x,b) = 0 , x >O
(iii) v(0, y) = 0 , 0 < y < b
\
'

Taking Fourier sine transform of the given pde, we obtain

General solution of the above equation i s ,

vs (a, y) = C, cosh a y + C, sinh a y , C, , C, being constants (1 17)

To find C, and C2,take Fourier sine transform of conditions (i) and (ii)

I
m

And when y = b , v, (a, b) = v(u, b) sin audu = 0 (119) ,

0
Putting y = 0 in Eqn.(l17) and using Eqn.(l 18), we obtain
m

C, = f(u) sinaudu
0
Putting y = b in Eqn.(ll7) and using Eqn.(l19), we obtain
 Fourier Transform
cosh a b
c 2 -- -c,-=-- Method
sinh a b cOSh ab
sinh a b f (u) sin a u du
0
x rg
coshab .
:. v, ( a , y) = cosh a y f (u) sin a u du - -slnh a y f (u) sin a u du
I sinh a b
u 0

I f ( u ) s i n a u s ~ nsinhh(b -a y)b a du
x

=
0

Taking inverse Fourier sine transform of v, ( a , y) , we obtain

V(X,y) =-
2=
7t
I
0 0
f(u)rin a u
sinh(b-y)a .
sinh a b
s ~ anx du d a ,

as the required solution.

. -x-
Partiai Differential
Equations Appendix
Table of Fourier Transforms

F(a) = f (x) e-"' dr

X -=la\
-e
a /
1
a+ia.
1
H(-x)eax (Re a >0)
a-ia
2a
a 2+ a 2

e-""l sin
(-il :)
x I + - (a > 0)

2 sin a a
a
-iaa

f (ax + b) (a > 0)

-e
a
'(a)
-i(bx l a ) (a > 0, b real)

f (ax) cos cx (a > 0, c real)

f (ax) sin cx (a > 0, c real)

/

f (x + C)+ f (X- C) (C real) 2F(a) cos a c

f(x + C)- f ( X- C) (C real) 2iF(a) sin a c

x"f(x) (n =1,2, ...)