Contents

1 Introduction 1.1 1.2 1.3 1.4 1.5 1.6 Background of study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Statement of Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1 1 2 2 2 2 3 3 3 3 4 4 4 6 6 7 8 9

Signicance of Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Aims and Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Scope of Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Denition of Terms and Concepts . . . . . . . . . . . . . . . . . . . . . . 1.6.1 1.6.2 1.6.3 1.6.4 1.6.5 1.6.6 Integral transform . . . . . . . . . . . . . . . . . . . . . . . . . . . Fourier Transform . . . . . . . . . . . . . . . . . . . . . . . . . . . Inverse Fourier Transform . . . . . . . . . . . . . . . . . . . . . . Partial Dierential Equation (PDE) . . . . . . . . . . . . . . . . . Linear Partial Dierential Equation . . . . . . . . . . . . . . . . . Classication of second order linear partial dierential equation .

2 Literature Review 2.1 2.2 The Fourier Transform and the Inverse Transform . . . . . . . . . . . . . Partial Dierential Equations (PDEs) . . . . . . . . . . . . . . . . . . . .

3 Fourier Transforms and Inverse Fourier transforms 3.1 Fourier transform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

CONTENTS

ii 11 13 14 17 19

3.2

Inverse Fourier Transform . . . . . . . . . . . . . . . . . . . . . . . . . .

4 Application of Fourier Transform to PDEs 4.1 4.2 4.3 The Fourier transform and the Wave equation . . . . . . . . . . . . . . . The Fourier Transform and the Heat Equation . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Chapter 1 Introduction
1.1 Background of study

Among the tools that are useful for solving linear dierential equations are integral transforms. An integral transform is a relation of the form F (s) =

K(s, t)f (t) dt

(1.1)

where K(s, t) is a given function called the kernel of the transformation and the limits of integration and are also given. It is possible that = and = or both. The relation (1.1) transforms f into another function F , which is called the transform of f . The general idea in using an integral transform to solve a dierential equation is as follows: use the relation in (1.1) to transform a problem for an unknown function f into a simpler problem for F , then solve this simple problem to nd F and nally recover the desired function f from its transform f . This last step is known as inverting the transform. There are several integral transforms that are useful in applied mathematics, but in this study we will consider only the Fourier transform.

1.2 Statement of Problem

1.2

Statement of Problem

In this project work, we seek to dene the Fourier transform and compute the Fourier transform of several functions. We will utilize the Fourier transform in solving some linear Partial Dierential Equations (PDEs).

1.3

Signicance of Study

This study is important because of the widespread applications of partial dierential equations in science and engineering. The Fourier transform is also a powerful and useful tool in mapping functions from Euclidean space to Fourier space. A Dierential Equation transformed into Fourier space may be easier to solve.

1.4

Aims and Objectives

The aims and objectives of the study are as follows: 1. To detail the denition of the Fourier transform and inverse Fourier transform and compute the Fourier transform and inverse Fourier transform of several functions. 2. To utilize the Fourier to solve some linear partial dierential equations.

1.5

Scope of Study

This project work is written to nd the solution of partial dierential equations using the Fourier transform method only.

1.6 Definition of Terms and Concepts

1.6
1.6.1

Denition of Terms and Concepts

Integral transform

An integral transform is a relation of the form F (s) =

K(s, t)f (t) dt

(1.2)

where K(s, t) is a given function called the kernel of the transform and the limits of integration and are also given. The relation (1.2) transforms the function into another function F which is called the transform of f .

1.6.2

Fourier Transform

Given a function f (x) with derivative f (x) where f (x) and f (x) are continuous in every nite interval and f (x) is integrable in (, ), the function F () = F[f (x)] =

f (x)eix dx

(1.3)

is referred to as the Fourier transform of f (x).

1.6.3

Inverse Fourier Transform

The inverse Fourier transform recovers the original function f (x) from F (w). Given as F 1 , it is dened as f (x) = F
1

1 [F ()] = = 2

f ()eix d

1.6 Definition of Terms and Concepts

1.6.4

Partial Dierential Equation (PDE)

The general form of PDE for a function u(x1 , x2 , . . . , xn ) is

F (x1 , x2 , . . . , xn , u, ux1 , ux2 , . . . , uxn , . . .) = 0

(1.4)

where x1 , x2 ,. . . , xn are the independent variables, u is the unknown function and uxi denotes the partial derivative
u . xi

1.6.5

Linear Partial Dierential Equation

An equation is called a linear partial dierential equation if in

F (x1 , x2 , . . . , xn , u, ux1 , ux2 , . . . , uxn , . . .) = 0

F is a linear function of the unknown function u and its derivatives. Thus, the equation, x7 ux + exy uy + sin(x2 + y 2 )u = x3 , is linear equation.

1.6.6

Classication of second order linear partial dierential equation

A second order linear Partial Dierential Equation for functions in two independent variables x and y has the form

L[u] = auxx + 2buxy + cuyy + dux + euy + f u = g.

(1.5)

The factor 2 in front of the coecient b is for convenience. We assume the coecients a, b, c, do not vanish simultaneously. The operator

L0 [u] = auxx + 2buxy + cuyy

1.6 Definition of Terms and Concepts

that consists of the second order terms of the operator L is called the principal part of L. Many fundamental properties of the solution of (1.5) are determined by its principal part, and more precisely, by the sign of the discriminant (L) := b2 ac of the equation. We classify the equation according to the sign of (L). Denition 1.1 Equation (1.5) is said to be hyperbolic at a point (x, y) if (L) := b2 ac > 0, it is said to be parabolic at (x, y) if (L) = 0 and it is said to be elliptic at (x, y) if (L) < 0. For example, the heat equation given by equation given by by
2u x2 2u t2
2

u t

= k u is a parabolic equation. The wave x2

= c2 u is a hyperbolic equation, while the Laplace equation given x2

2u y 2

= 0 is an example of elliptic equation.

Chapter 2 Literature Review

To successfully understand the application of Fourier transform to Partial Dierential Equations, it is important to understand the basics of the Fourier transform and Partial Dierential Equations as described by certain authors.

2.1

The Fourier Transform and the Inverse Transform

According to Wikipedia (2010), the Fourier transform and its generalization form the subject of Fourier analysis. It lists out the basic properties of the Fourier transform such as linearity, translation, and conjugation. Tung (2004) provides a clear denition for the Fourier transform and the Inverse Fourier transform. He also demonstrates how Fourier transform can be used in solving dierential equations. He denes the Fourier transform as F () = F[f (x)] =

f (x)eix dx

(2.1)

2.2 Partial Differential Equations (PDEs)

and the inverse Fourier transform as;

1

f (x) = F

1 [F ()] = 2

F ()eix d

(2.2)

Yosida (1980) shows clearly how Fourier transforms are used in solving dierential equation in an abstract sense.

2.2

Partial Dierential Equations (PDEs)

According to Boyce and Diprima (2000), the study of dierential equations has attracted the attention of many of the worlds greatest mathematicians during the past three centuries. Nevertheless, it remains a dynamic eld of inquiry today with many interesting open questions, so before embarking on a serious study of dierential equation one needs to know that a dierential equation describes some physical process often called mathematical model. Furthermore, the simplest dierential equations provide useful models of important physical processes. Pinchover and Rubinstein (2005) give a clear denition and classication of PDEs. They also demonstrate how the method of separation of variables and several other methods can be used in solving PDEs. Tung (2004) shows how integral transforms can be used in solving PDEs. Burden & Faires (1997) shows how the nite dierence method can be used as an eective numerical method for solving PDEs.

Chapter 3 Fourier Transforms and Inverse Fourier transforms

This chapter deals with some simple examples of Fourier and Inverse Fourier transforms of certain functions. It also shows the procedures which will make Fourier transform and its inverse very suitable for the solution of Partial Dierential Equations. An integral transform is a relation of the form F (s) =

K(s, t)f (t) dt

where K(s, t) is a given function called the kernel of the transform and the limits of integration and are given. The relation above transforms the function f into another function F which is called the transform of f and it is a tool used in solving linear dierential equations.

3.1 Fourier transform

3.1

Fourier transform

Given a function f (x) with derivative f (x) where f (x) and f (x) are continuous in every nite interval and f (x) is integrable in (, ), it is dened as F () = F[f (x)] =

f (x)eix dx.

Example 3.1 Find the Fourier transform of f (x) = e|x| ,

< x <

Solution

F () = F[e ] = e|x| eix dx 0 x+ix = e dx + ex+ix dx

0

|x|

1 1 e(1i)x | + e(1+i)x |0 = 0 (1 + i) (1 + i) 1 1 2 = + = 1 i 1 + i 1 + 2 Note its decay as . Example 3.2 Find the Fourier transform of

f (x) = ex ,

< x <

3.1 Fourier transform

10

Solution

F () = F[e ] =
x

ex eix dx

1 = e(1+i)x | (1 + i) The limit at x = blows up. We say the Fourier transform of ex does not exist because the function f (x) = ex is not integrable such that

|f (x)| dx =

ex dx

does not have a nite value. Example 3.3 Find the Fourier transform of f (x) = ex ,
2

< x <

Solution The value of the function decreases rapidly when x is away from x = 0 in both the positive and negative x directions. There is a nite area under the curve ex , so this
2

function is integrable. To nd its Fourier transform, we need to perform the integral F () =

ex eix dx
2

By completing the square in exponent we get e

2 /4

F (w) =

3.2 Inverse Fourier Transform

11

The detailed working is as follows:

x2 + ix = (x i/2)2 2 /4. So F () = e
2 /4

e(xi/2) dx = e
2

2 /4

ey dy where we have made a change

2

of variable y = x i/2 and also shifted the path of integration. The remaining integral is a standard one (Eulers integral) and is equal to .

3.2

Inverse Fourier Transform

This recovers the original function f (x) from F (). Given as F 1 , it is dened as f (x) = F
1

1 [F ()] = = 2

F ()eix d

We will now compute the inverse Fourier transform of some simple functions. Example 3.4 Find the Inverse transform of e
2 /4

F () =

< < .

Solution

2 [F ()] = e /4ix d 2 x2 e e/2ix d = 2 = ex

2

We have thus recorded the original function f (x) in Example 3.3.

3.2 Inverse Fourier Transform

12

Example 3.5 Find the inverse Fourier transform of 2 1 + 2

F () =

Solution 2 [F ()] = 2

1 eix d 1 + 2

This integral can be evaluated using residue calculus. Alternatively, using tables of integrals, we see that f (x) = e|x| We have thus recovered the f (x) in Example 3.1.

Chapter 4 Application of Fourier Transform to PDEs

The usual diculty with PDEs is that the solution involves more than one independent variable. The transform method allows us to reduce one independent variable. We commonly try to transform the x variable through a Fourier transform provided that the domain in is innite such that < x < . Now consider a function u(x, t) with < x < , t > 0 and let U (, t) = F[u(x, t)] =

u(x, t)eix dx

(4.1)

be the Fourier transform of u(x, t) with respect to x. The original function u(x, t) can then be recovered from the inverse Fourier transform 1 u(x, t) = 2

U (, t)eix d

(4.2)

Note that in both equations (4.1) and (4.2), t plays no role, it is arbitrary.

13

4.1 The Fourier transform and the Wave equation

14

For the solution of second order linear PDEs, one has to know the following terms. 1 Ut (, t)eix d 2 1 Utt (, t)eix d 2 1 (i)U (, t)eix d 2 1 (i)2 U (, t)eix d 2

ut (x, t) = utt (x, t) = ux (x, t) = uxx (x, t) =

(4.3) (4.4) (4.5) (4.6)

In this chapter, we would use the Fourier transform to solve the heat equation and the wave equation.

4.1

The Fourier transform and the Wave equation

The wave equation is represented as follows. 2 u(x, t) 2 u(x, t) = c2 t2 x2

(4.7)

where c is a certain physical constant, u is the dependent variable and x and t are two independent variables. This is considered in an innite domain such that the wave equation is represented below as an Initial-Boundary Value Problem (IBVP).

PDE : BCs : ICs :

utt = c2 uxx , u(x, t) 0, u(x, 0) = f (x) ut (x, 0) = 0,

< x < , as x

t > 0,

(4.8) (4.9) (4.10)

< x <

(4.11)

4.1 The Fourier transform and the Wave equation

15

We assume that the solution to be of the form of an integral 1 u(x, t) = 2

U (, t)eix d

which we substitute into the PDE. This yields 1 2 which is the same as F 1 [Utt + c2 2 U ] = 0 so Utt + c2 2 U = 0. This is an ODE; the partial derivatives
2u x2

) Utt (, t) + c2 2 U (, t) eix d = 0

(4.12)

(4.13)

have been converted to (i)2 an algebraic

multiplication. The ODE in t is to be solved subject to the following initial conditions: 1 ut (x, 0) = 2 and 1 u(x, 0) = 2 These imply that Ut (, 0) = 0 and U (, 0) = F () (4.15) (4.14)

Ut (, 0)eix d = 0

U (, 0)e

ix

1 d = f (x) = 2

F ()eix d.

4.1 The Fourier transform and the Wave equation

16

where the Fourier transform F () of f (x) is known if f (x) is known. The general solution to the PDE (4.13) is

U (, t) = A() sin(ct) + B() cos(ct).

The initial conditions (4.14) and (4.15) can be used to determine to constants A and B to be B() = F () and A() = 0. Thus,

U (, t) = F ()cos(ct)

(4.16)

We recover u(x, t) by substituting (4.16) back to (4.2)

1

u(x, t) = F

1 [U (, t)] = 2

U (, t)e

ix

1 d = 2

F () cos(ct)eix d (4.17)

Typically one cannot perform the integral explicitly unless F () is known. In the particular case of the wave equation however, progress can be made by noting that 1 cos(ct) = (eict + eict ) 2 and so (4.17) can be rewritten as 1 1 1 1 i(xct) u(x, t) = F ()e d + F ()ei(x+ct) d 2 2 2 2 1 1 = f (x ct) + f (x + ct) 2 2 1 f (x) = 2 and so 1 f (x + ct) = 2

(4.18)

F ()eix d

F ()ei(x+ct) d

4.2 The Fourier Transform and the Heat Equation

17

and 1 f (x ct) = 2

F ()ei(xct) d

4.2

The Fourier Transform and the Heat Equation

The heat conduction is represented as follows. The wave equation is represented as follows. u(x, t) 2 u(x, t) = 2 t x2 (4.19)

where is a certain physical constant, u is the dependent variable and x and t are two independent variables. This is considered in an innite domain such that the heat equation is represented below as an Initial-Boundary Value Problem (IBVP).

PDE : BCs : ICs :

ut = 2 uxx , u(x, t) 0, u(x, 0) = f (x)

< x < , as x

t > 0,

(4.20) (4.21) (4.22)

<x<

We assume a solution of the form of an integral (4.2) and substitute into PDE (4.21). Using 1 uxx (x, t) = 2 this yields 1 2 which implies Ut + 2 2 U = 0 (4.23)

(i)2 U (, t)eix d

) Ut + 2 2 U eix d = 0

4.2 The Fourier Transform and the Heat Equation

18

The ODE (4.23) is solved subject to the initial condition

U (, 0) = F ()

(4.24)

which is obtained by taking the Fourier transform of (4.22). The solution is U (, t) = A()e
2 2 t

= F ()e

2 2 t

(4.25)

The nal solution is obtained by substituting (4.25) into (4.2) 1 2 2 F ()e tix d. 2

u(x, t) =

(4.26)

For special case of f (x) = ae(x/L) ,

2

< x < .

In this case, 2 F () = F[f (x)] = aL e(L) /4 . Then aL u(x, t) = 2

e(L)

2 /42 2 tix

can be evaluated by completing the squares aL u(x, t) = 2 =

e(

2 t+ L2 ) 2 ix 4

aL 2 2 2 ex /(4 t+L ) 42 t + L2

4.3 Conclusion

19

4.3

Conclusion

We nd the Fourier transform as an eective method for solving the Initial Boundary Value Problems (IBVPs) for the wave equation and heat equation, and it will clearly be useful in solving other time dependent Partial Dierential Equations.

Bibliography
[1] Boyce, E. W. and Diprima, R. C. (2001), Elementary Dierential Equations and Boundary Value Problems, John Wiley and Sons, New York. [2] Burden, R.L. and Faires, J.D. (1997), Numerical Analysis, Brookes/Cole, CA. [3] Kammler, D. (2000), A First Course in Fourier Analysis, Prentice-Hall. [4] Pinchover, Y. and Rubinstein, J. (2005), An Introduction to Partial Dierential Equations, Cambridge University Press. [5] Stroud, K. A. (2003), Advanced Engineering Mathematics, Palgrave Macmillan. [6] Tung K.K (2004), Partial Dierential Equations and Fourier Analysis - A Short Introduction. [7] Wikipedia article (2010), Fourier Transform. [8] Yosida, K. (1980), Functional Analysis, Springer-Verlag.

20