PMAT 32322 Mathematical Methods Note 08

Fourier Transform Methods

8.1 Introduction
Joseph Fourier, a French mathematician had invented a method called Fourier transform in 1801,
to explain the flow of heat around an anchor ring. Since then, it has become a powerful tool in
diverse fields of science and engineering. It can provide a means of solving unwieldy equations
that describe dynamic responses to electricity, heat or light. In some cases, it can also identify the
regular contributions to a fluctuating signal, thereby helping to make sense of observations in
astronomy, medicine and chemistry. Fourier transform has become indispensable in the numerical
calculations needed to design electrical circuits, to analyze mechanical vibrations, and to study
wave propagation.
8.2 Fourier Integral Representations
Definition 8.1 (Dirichlet’s Conditions)
A function 𝑓(𝑥) is said to have satisfied Dirichlet’s conditions in the interval (−𝐿, 𝐿) provided
𝑓(𝑥) is periodic, pointwise continuous, and has a finite number of relative maxima and minima in
(−𝐿, 𝐿).
Let a function 𝑓(𝑥) be periodic with period 2𝐿, 𝑖. 𝑒. 𝑓(𝑥 + 2𝐿) = 𝑓(𝑥), and satisfy Dirichlet’s
conditions in the interval (−𝐿, 𝐿). Then 𝑓(𝑥) has a Fourier series representation for
𝑎0 𝑛𝜋𝑥 𝑛𝜋𝑥
𝑓(𝑥) = + ∑∞
𝑛=1 (𝑎𝑛 cos + 𝑏𝑛 sin )____(1)
2 𝐿 𝐿
1 𝐿 𝑛𝜋𝑥
where 𝑎𝑛 = 𝐿 ∫−𝐿 𝑓(𝑥) cos 𝑑𝑥 𝑛 = 0,1,2, …____(2)
𝐿
1 𝐿 𝑛𝜋𝑥
𝑏𝑛 = 𝐿 ∫−𝐿 𝑓(𝑥) sin 𝑑𝑥 𝑛 = 0,1,2, …_____(3)
𝐿

Here 𝑎𝑛 , 𝑏𝑛 are called Fourier coefficients. Fourier series representation, however, can be extended
to some non-periodic functions also, provided the integral of modulus of such a function 𝑓(𝑡)
satisfies the condition
∞
∫−∞|𝑓(𝑡)|𝑑𝑡 is finite.

Substituting equations (2) and (3) into Fourier series (1), we get
1 𝐿 1 𝐿 𝑛𝜋𝑡 𝑛𝜋𝑥 1 𝐿 𝑛𝜋𝑡 𝑛𝜋𝑥
𝑓(𝑥) = 2𝐿 ∫−𝐿 𝑓(𝑡)𝑑𝑡 + ∑∞
𝑛=1 [𝐿 ∫−𝐿 𝑓(𝑡) cos 𝐿
cos 𝐿
𝑑𝑡 + 𝐿 ∫−𝐿 𝑓(𝑡) sin 𝐿
sin 𝐿
𝑑𝑡]

Noting that cos(𝐴 − 𝐵) = cos 𝐴 cos 𝐵 + sin 𝐴 sin 𝐵, and interchanging the order of summation and
integration, we obtain
1 𝐿 1 𝐿 𝑛𝜋(𝑡−𝑥)
𝑓(𝑥) = 2𝐿 ∫−𝐿 𝑓(𝑡)𝑑𝑡 + 𝐿 ∫−𝐿 𝑓(𝑡) ∑∞
𝑛=1 cos 𝐿
𝑑𝑡 ____(4)

Further, if we assume that the function 𝑓(𝑥) is absolutely integrable, and allowing 𝐿 to tend to infinity , i.e.
∞
∫−∞|𝑓(𝑡)|𝑑𝑡 < ∞. ____(5)

1
PMAT 32322 Mathematical Methods Note 08

1 𝐿
We get lim ∫ 𝑓(𝑡)𝑑𝑡 = 0 ____(6)
𝑛→∞ 2𝐿 −𝐿
𝜋
In the remaining part of infinite sum of equation (4), if we set ∆𝑠 = 𝐿 , the equation reduces to
𝜋
1
𝑓(𝑥) = lim ∫∆𝑠𝜋 𝑓(𝑡) ∑∞
𝜋 − 𝑛=1 cos(𝑛∆𝑠(𝑡 − 𝑥))𝑑𝑡 ____(7)
∆𝑠→0 ∆𝑠

As 𝐿 → ∞, ∆𝑠 → 0, implying that ∆𝑠 is a small positive number and the points 𝑛∆𝑠 are equally
spaced along the x-axis. The series under the integral can be approximated by an integral of the
form (as ∆𝑥 → 0)
∞
∫0 cos 𝑠(𝑡 − 𝑥) 𝑑𝑠

Thus, equation (7) can be re-written as

1 ∞ −∞
𝑓(𝑥) = 𝜋 ∫−∞ 𝑓(𝑡) ∫0 cos 𝑠(𝑡 − 𝑥) 𝑑𝑠𝑑𝑡 ______(8)
1 −∞ ∞
= 𝜋 ∫0 ∫−∞ 𝑓(𝑡) cos 𝑠(𝑡 − 𝑥) 𝑑𝑡𝑑𝑠 ____(9)

which is the Fourier integral representation of 𝑓(𝑥).

8.2.1 Fourier Integral Theorem
∞
If 𝑓(𝑥) satisfies Dirichlet’s conditions for −∞ < 𝑥 < ∞ and if the integral ∫−∞ 𝑓(𝑥) 𝑑𝑥 is
absolutely convergent, then
1 ∞ ∞ 1
∫ 𝑑𝛼 ∫−∞ 𝑓(𝑡) cos 𝛼(𝑡 − 𝑥) 𝑑𝑡 = 2 [𝑓(𝑥 +) + 𝑓(𝑥 −)] ___(10)
𝜋 0

In order to bring out the analogy between Fourier series and Fourier integral theorem, we
write
1 ∞ ∞
𝑓(𝑥) = 𝜋 ∫0 ∫−∞ 𝑓(𝑡)(cos 𝛼𝑡 cos 𝛼𝑥 + sin 𝛼𝑡 sin 𝛼𝑥)𝑑𝑡𝑑𝛼
1 ∞
If we define 𝐴(𝛼) = ∫−∞ 𝑓(𝑡) cos 𝛼𝑡 𝑑𝑡 ___(11)
𝜋
1 ∞
𝐵(𝛼) = 𝜋 ∫−∞ 𝑓(𝑡) sin 𝛼𝑡 𝑑𝑡 _____(12)

The above equation can also be written as

∞
𝑓(𝑥) = ∫0 [𝐴(𝛼) cos 𝛼𝑥 + 𝐵(𝛼) sin 𝛼𝑥]𝑑𝑡𝑑𝛼 ____(13)

8.2.2 Sine and Cosine Integral Representations

If 𝑓(−𝑥) = −𝑓(𝑥) i.e. if 𝑓(𝑥) is an odd function, equation (12) gives
1 ∞ 1 ∞
𝐴(𝛼) = ∫ 𝑓(𝑥) cos 𝛼𝑥 𝑑𝑥 = − ∫ 𝑓(−𝑥) cos 𝛼𝑥 𝑑𝑥
𝜋 −∞ 𝜋 −∞

2
PMAT 32322 Mathematical Methods Note 08

1 −∞
𝐴(𝛼) = 𝜋 ∫∞ 𝑓(𝑥) cos 𝛼𝑥 𝑑𝑥
∞
= − ∫−∞ 𝑓(𝑥) cos 𝛼𝑥 𝑑𝑥 = −𝐴(𝛼).

implying 2𝐴(𝛼) = 0 or 𝐴(𝛼) = 0 i.e.

1 −∞
𝐴(𝛼) = 𝜋 ∫∞ 𝑓(𝑥) cos 𝛼𝑥 𝑑𝑥 = 0 ____(14)
1 ∞
Also, 𝐵(𝛼) = 𝜋 ∫−∞ 𝑓(𝑥) sin 𝛼𝑥 𝑑𝑥
1 −∞
= − 𝜋 ∫∞ 𝑓(𝑥) sin 𝛼𝑥 𝑑𝑥
1 ∞ 2 ∞
= 𝜋 ∫−∞ 𝑓(𝑥) sin 𝛼𝑥 𝑑𝑥 = 𝜋 ∫0 𝑓(𝑥) sin 𝛼𝑥 𝑑𝑥 ____(15)

Thus, equation (13) reduces to

∞
𝑓(𝑥) = ∫0 𝐵(𝛼) sin 𝛼𝑥 𝑑𝛼____(16)

which is the Fourier sine integral representation, where 𝐴(𝛼) and 𝐵(𝛼) are defined by the relations
(14) and (15).
If 𝑓(𝑥) is an even function, i.e. 𝑓(−𝑥) = 𝑓(𝑥), then we obtain the Fourier cosine integral
representation
∞
𝑓(𝑥) = ∫0 𝐴(𝛼) cos 𝛼𝑥 𝑑𝛼 ___(17)
2 ∞
where 𝐵(𝛼) = 0 , 𝐴(𝛼) = 𝜋 ∫0 𝑓(𝑥) cos 𝛼𝑥 𝑑𝑥 ____(18)

8.3 Fourier Transform Fairs

1 ∞ ∞
From the Fourier integral, we have 𝑓(𝑥) = 𝜋 ∫0 ∫−∞ 𝑓(𝑡) cos 𝛼(𝑡 − 𝑥) 𝑑𝑡𝑑𝛼 ___(19)

In terms of complex exponential function, (19) takes the form

1 ∞ ∞
𝑓(𝑥) = 2𝜋 ∫0 ∫−∞ 𝑓(𝑡) [𝑒 𝑖𝛼(𝑡−𝑥) + 𝑒 −𝑖𝛼(𝑡−𝑥) ]𝑑𝑡𝑑𝛼
1 ∞ ∞ ∞ ∞
= 2𝜋 [∫0 ∫−∞ 𝑓(𝑡)𝑒 𝑖𝛼(𝑡−𝑥) 𝑑𝑡𝑑𝛼 + ∫0 ∫−∞ 𝑓(𝑡)𝑒 −𝑖𝛼(𝑡−𝑥) 𝑑𝑡𝑑𝛼 ]

Let 𝛼 = −𝛼 in the second integral; then it becomes

−∞ ∞ 0 ∞
−∫ ∫ 𝑓(𝑡)𝑒 𝑖𝛼(𝑡−𝑥) 𝑑𝑡𝑑𝛼 = ∫ ∫ 𝑓(𝑡)𝑒 𝑖𝛼(𝑡−𝑥) 𝑑𝑡𝑑𝛼
0 −∞ −∞ −∞
1 ∞ ∞
Hence , 𝑓(𝑥) = 2𝜋 ∫−∞ ∫−∞ 𝑓(𝑡)𝑒 𝑖𝛼(𝑡−𝑥) 𝑑𝑡𝑑𝛼 ___(20)

This is the exponential form of the Fourier integral theorem. Equation (20) can be rewritten as

3
PMAT 32322 Mathematical Methods Note 08

1 ∞ ∞ 1
𝑓(𝑥) = ∫ ∫ [ 𝑓(𝑡)𝑒 𝑖𝛼𝑡 𝑑𝑡] 𝑒 −𝑖𝛼𝑥 𝑑𝛼 ____(21)
√2𝜋 −∞ −∞ √2𝜋

Thus, we formally define the Fourier transform pairs as follows:

Definition 8.2
Let 𝑓(𝑥) be a function defined on (−∞, ∞) and is piecewise continuous, differentiable in each
finite interval and is absolutely integrable on (−∞, ∞). From (21), if
1 ∞
𝐹(𝛼) = ∫ 𝑓(𝑡)𝑒 𝑖𝛼𝑡
√2𝜋 −∞
𝑑𝑡 ___(22)

then we have , for all 𝑥,

1 ∞
𝑓(𝑥) = ∫ 𝐹(𝛼)𝑒 −𝑖𝛼𝑥 𝑑𝛼 ____(23)
√2𝜋 −∞

Here, 𝐹(𝛼) defined by (22) is the Fourier transform of 𝑓(𝑥), and 𝑓(𝑥) defined by (23) is called
the inverse Fourier transform of 𝐹(𝛼) and is denoted by
𝐹(𝛼) = ℱ[𝑓(𝑡): 𝛼] ____(24)
𝑓(𝑥) = ℱ −1 [𝐹(𝛼): 𝑥] ___(25)
We have seen in section 8.2.2.that if 𝑓(𝑥) is an odd function, the Fourier integral representation
of 𝑓(𝑥) reduces to
∞ 2 ∞ ∞
𝑓(𝑥) = ∫0 𝐵(𝛼) sin 𝛼𝑥 𝑑𝛼 or 𝑓(𝑥) = 𝜋 ∫0 sin 𝛼𝑥 ∫0 𝑓(𝑡) sin 𝛼𝑡 𝑑𝑡 𝑑𝛼 ___(26)

2 ∞
If 𝐹𝑠 (𝛼) = √𝜋 ∫0 𝑓(𝑡) sin 𝛼𝑡 𝑑𝑡 = ℱ𝑠 [𝑓(𝑡): 𝛼] ____(27)

2 ∞
then 𝑓(𝑥) = √𝜋 ∫0 𝐹𝑠 (𝛼) sin 𝛼𝑥 𝑑𝛼 = ℱ𝑠−1 [𝐹𝑠 (𝛼): 𝑥]____(28)

Here equation (27) is the Fourier sine transform of 𝑓(𝑥) and its inverse sine transform is given
by (28).
Similarly, when 𝑓(𝑥) is an even function, we can obtain the Fourier cosine transform and the
corresponding inverse as

2 ∞
𝐹𝑐 (𝛼) = √𝜋 ∫0 𝑓(𝑡) cos 𝛼𝑡 𝑑𝑡 = ℱ𝑐 [𝑓(𝑡): 𝛼]____(29)

2 ∞
𝑓(𝑥) = √𝜋 ∫0 𝐹𝑐 (𝛼) cos 𝛼𝑥 𝑑𝛼 = ℱ𝑐−1 [𝐹𝑐 (𝛼): 𝑥] ____(30)

4
PMAT 32322 Mathematical Methods Note 08

8.4 Transform of Elementary Functions

Example 01:
2 /2
Find the Fourier transform of 𝑓(𝑥) = 𝑒 −𝑥 .
Soluttion:
Following the definition of Fourier transform, we have
1 ∞
ℱ[𝑓(𝑥): 𝛼] = ∫ 𝑓(𝑥)𝑒 𝑖𝛼𝑥 𝑑𝑥
√2𝜋 −∞
1 ∞ 2
= ∫ 𝑒 −𝑥 /2 𝑒 𝑖𝛼𝑥 𝑑𝑥
√2𝜋 −∞
2
1∞ −(𝑥−𝑖𝛼)2 /2 −𝛼
= ∫ 𝑒 𝑒 2 𝑑𝑥
√2𝜋 −∞

(𝑥−𝑖𝛼) 𝑑𝑥
Let = 𝑡 which gives = 𝑑𝑡. Thus, we have
√2 √2

𝛼2 𝛼2
− − 𝛼2 𝛼2
𝑒 2 ∞ 2 𝑒 2
ℱ[𝑓(𝑥): 𝛼] = ∫ 𝑒 −𝑡 𝑑𝑡 = (√𝜋) = 𝑒 − 2 . Therefore, we have ℱ[𝑓(𝑥): 𝛼] = 𝑒 − 2 .
√𝜋 −∞ √𝜋

Example 02:

Find the Fourier transform of 𝑓(𝑥) = 𝑒 −𝑎|𝑥| , −∞ < 𝑥 < ∞.

Solution:
𝑥 𝑥≥0
We know that |𝑥| = { . Therefore we can write
−𝑥 𝑥<0
1 ∞
ℱ[𝑓(𝑥): 𝛼] = ∫ 𝑓(𝑥)𝑒 𝑖𝛼𝑥 𝑑𝑥
√2𝜋 −∞
1 0 1 ∞
= ∫ 𝑒 𝑎𝑥 𝑒 𝑖𝛼𝑥 𝑑𝑥
√2𝜋 −∞
+ ∫ 𝑒 −𝑎𝑥 𝑒 𝑖𝛼𝑥 𝑑𝑥
√2𝜋 0
1 0 1 ∞
= ∫ 𝑒 (𝑎+𝑖𝛼)𝑥 𝑑𝑥 + √2𝜋 ∫0 𝑒 (−𝑎+𝑖𝛼)𝑥 𝑑𝑥
√2𝜋 −∞

1 1 1 2 𝑎
= (𝑎+𝑖𝛼 − −𝑎+𝑖𝛼) = √𝜋 (𝑎2 +𝛼2).
√2𝜋

Example 03:
Find the Fourier transform of 𝑓(𝑥) defined by
1 |𝑥| ≤ 𝑎
𝑓(𝑥) = {
0 |𝑥| > 𝑎
∞ sin 𝛼𝑎 cos 𝛼𝑥 ∞ sin 𝛼𝑎
and hence evaluate ∫−∞ 𝑑𝛼, ∫0 𝑑𝛼.
𝛼 𝛼

5
PMAT 32322 Mathematical Methods Note 08

Solution:
From the definition of the Fourier transform,
1 ∞
𝐹(𝛼) = ℱ[𝑓(𝑥): 𝛼] = ∫ 𝑓(𝑥)𝑒 𝑖𝛼𝑥 𝑑𝑥
√2𝜋 −∞
𝑎
1 𝑎 1 𝑒 𝑖𝛼𝑥 1 𝑒 𝑖𝛼𝑎 𝑒 −𝑖𝛼
= ∫ 𝑒 𝑖𝛼𝑥 𝑑𝑥 = √2𝜋 (
√2𝜋 −𝑎 𝑖𝛼
) =
√2𝜋
( 𝑖𝛼
− 𝑖𝛼
)
−𝑎

2 𝑒 𝑖𝛼𝑎 −𝑒 −𝑖𝛼𝑎
= 𝛼√2𝜋 ( )
2𝑖

2 sin 𝛼𝑎
𝛼>0
𝛼√2𝜋
Therefore, we have 𝐹(𝛼) = { 2 sin 𝛼𝑎 .
lim 𝛼√2𝜋 𝛼=0
𝛼→0 𝛼𝑎

1 ∞
Now, 𝑓(𝑥) = ℱ −1 [𝐹(𝛼): 𝑥] = ∫ 𝐹(𝛼)𝑒 −𝑖𝛼𝑥 𝑑𝛼
√2𝜋 −∞

1 ∞ 2 sin 𝛼𝑎 −𝑖𝛼𝑥 1 |𝑥| ≤ 𝑎

Thus, we have 𝑓(𝑥) = ∫ 𝑒 𝑑𝛼 ={
√2𝜋 −∞ 𝛼√2𝜋 0 |𝑥| > 𝑎
1 ∞ sin 𝛼𝑎(cos 𝛼𝑥−𝑖 sin 𝛼𝑥) 1 |𝑥| ≤ 𝑎
i.e. ∫ 𝑑𝛼 ={
𝜋 −∞ 𝛼 0 |𝑥| > 𝑎
∞ sin 𝛼 acos 𝛼𝑥 𝜋 |𝑥| ≤ 𝑎
Hence ∫−∞ 𝑑𝛼 = { .
𝛼 0 |𝑥| > 𝑎
Also, by setting 𝑥 = 0 in the above equations, we obtain
∞ sin 𝛼𝑎
∫−∞ 𝑑𝛼 = 𝜋
𝛼
∞ sin 𝛼𝑎 𝜋
Since the integrand is even, we can have ∫0 𝑑𝛼 = .
𝛼 2

Example 04:
Find the Fourier cosine and sine transforms of 𝑒 −𝑏𝑥 and evaluate the integrals
∞ cos 𝛼𝑥
(i) ∫0 𝑑𝛼
𝛼2 +𝑏 2
∞ α sin 𝛼𝑥
(ii) ∫0 𝑑𝛼 .
𝛼2 +𝑏 2

Solution:
Given 𝑓(𝑥) = 𝑒 −𝑏𝑥 and following the definitions of Fourier cosine and sine transforms, we get

2 ∞
𝐹𝑐 (𝛼) = ℱ𝑐 [𝑓(𝑥): 𝛼] = √𝜋 ∫0 𝑓(𝑥) cos 𝛼𝑥 𝑑𝑥____(a)

6
PMAT 32322 Mathematical Methods Note 08

2 ∞
𝐹𝑠 (𝛼) = ℱ𝑠 [𝑓(𝑥): 𝛼] = √𝜋 ∫0 𝑓(𝑥) sin 𝛼𝑥 𝑑𝑥 _____(b)

∞ ∞
Let 𝐼1 = ∫0 𝑒 −𝑏𝑥 cos 𝛼𝑥 𝑑𝑥 and 𝐼2 = ∫0 𝑒 −𝑏𝑥 sin 𝛼𝑥 𝑑𝑥

Integrating 𝐼1 by parts, we have

∞
𝑒 −𝑏𝑥 cos 𝛼𝑥 𝛼 ∞ 1 𝛼
𝐼1 = [− ] − 𝑏 ∫0 𝑒 −𝑏𝑥 sin 𝛼𝑥 𝑑𝑥 = 𝑏 − 𝑏 𝐼2 ____(c)
𝑏 0

Integrating 𝐼2 by parts, we have

∞
𝑒 −𝑏𝑥 sin 𝛼𝑥 𝛼 ∞ 1 𝛼
𝐼2 = [− ] − 𝑏 ∫0 𝑒 −𝑏𝑥 cos 𝛼𝑥 𝑑𝑥 = 𝑏 − 𝑏 𝐼1 _____(d)
𝑏 0

Solving (c) and (d) for 𝐼1 and 𝐼2 , we obtain

𝑏 𝛼
𝐼1 = 𝛼2 +𝑏2 and 𝐼2 = 𝛼2 +𝑏2
𝑏 𝛼
Hence, we have 𝐹𝑐 (𝛼) = 𝛼2 +𝑏2 and 𝐹𝑠 (𝛼) = 𝛼2 +𝑏2.

2 ∞
Then, 𝑓(𝑥) = √𝜋 ∫0 𝐹𝑐 (𝛼) cos 𝛼𝑥 𝑑𝛼

2 ∞ 𝑏
𝑒 −𝑏𝑥 = √𝜋 ∫0 cos 𝛼𝑥 𝑑𝛼
𝛼2 +𝑏 2

∞ cos 𝛼𝑥 𝜋
Or ∫0 𝑑𝛼 = 2𝑏 𝑒 −𝑏𝑥 .
𝛼2 +𝑏 2
∞ α sin 𝛼𝑥 𝜋
Similarly, it can be shown that ∫0 𝑑𝛼 = 𝑒 −𝑏𝑥 .
𝛼2 +𝑏2 2

Example 05:
0 0<𝑥<𝑎
Find the Fourier sine transform of 𝑓(𝑥) , if 𝑓(𝑥) = {𝑥 𝑎≤𝑥≤𝑏.
0 𝑥>𝑏
Solution:
Following the definition of the Fourier sine transform, we have

2 ∞
𝐹𝑠 (𝛼) = √𝜋 ∫0 𝑓(𝑥) sin 𝛼𝑥 𝑑𝑥

2 𝑏
= √𝜋 ∫𝑎 𝑥 sin 𝛼𝑥 𝑑𝑥

2 𝑥 cos 𝛼𝑥 𝑏 1 𝑏
= √𝜋 [(− ) + 𝛼 ∫𝑎 cos 𝛼𝑥 𝑑𝑥 ]
𝛼 𝑎

7
PMAT 32322 Mathematical Methods Note 08

2 𝛼 cos 𝛼𝑎−𝑏 cos 𝛼𝑏 sin 𝛼𝑏−sin 𝛼𝑎

𝐹𝑠 (𝛼) = √𝜋 ( + ).
𝛼 𝛼2

8.5 Properties of Fourier Transform

In many practical situations, determination of Fourier transform of certain functions is very
complex. Once we know the transform of some elementary functions , we can find the transform
of many other functions with the help of the properties associated with the Fourier transform. We
now discuss some of the important properties of the Fourier transform.
Theorem 01: (Linearity Property)
If 𝐹(𝛼) and 𝐺(𝛼) are the Fourier transforms of 𝑓(𝑥) and 𝑔(𝑥), respectively, then
ℱ[𝑐1 𝑓(𝑥) + 𝑐2 𝑔(𝑥)] = 𝑐1 𝐹(𝛼) + 𝑐2 𝐺(𝛼)
ℱ −1 [𝑐1 𝐹(𝛼) + 𝑐2 𝐺(𝛼)] = 𝑐1 𝑓(𝑥) + 𝑐2 𝑔(𝑥)
where 𝑐1 , 𝑐2 are constants.
Proof:
1 ∞
ℱ[𝑐1 𝑓(𝑥) + 𝑐2 𝑔(𝑥)] = ∫ [𝑐 𝑓(𝑥) + 𝑐2 𝑔(𝑥)]𝑒 𝑖𝛼𝑥 𝑑𝑥
√2𝜋 −∞ 1
𝑐1 ∞ 𝑐 ∞
∫ 𝑓(𝑥)𝑒 𝑖𝛼𝑥 𝑑𝑥 + √2𝜋 ∫−∞ 𝑔(𝑥)𝑒 𝑖𝛼𝑥 𝑑𝑥
2
=
√2𝜋 −∞

= 𝑐1 𝐹(𝛼) + 𝑐2 𝐺(𝛼).
In the same fashion, it can be established that Fourier cosine and sine transforms also have linear
property.
ℱ𝑐 [𝑐1 𝑓(𝑥) + 𝑐2 𝑔(𝑥)] = 𝑐1 ℱ𝑐 (𝛼) + 𝑐2 𝐺𝑐 (𝛼)
ℱ𝑠 [𝑐1 𝑓(𝑥) + 𝑐2 𝑔(𝑥)] = 𝑐1 ℱ𝑠 (𝛼) + 𝑐2 𝐺𝑠 (𝛼).
Theorem 02: (Differentiation)
If 𝑓(𝑥) and its first (𝑟 − 1) derivatives are continuous, and if its 𝑟 𝑡ℎ derivative is piecewise
continuous, then
1 ∞ 𝑑𝑟 𝑓
ℱ[𝑓 𝑟 (𝑥): 𝛼] = ∫ 𝑒 𝑖𝛼𝑥 𝑑𝑥 = 𝐹 (𝑟) (𝛼) (say)
√2𝜋 −∞ 𝑑𝑥 𝑟

Integrating by parts, we get

∞
1 ∞ 𝑑𝑟 𝑓 𝑖𝛼𝑥 1 𝑑𝑟−1 𝑓 1 ∞ 𝑑𝑟−1 𝑓
∫ 𝑒 𝑑𝑥 = (𝑑𝑥 𝑟−1 𝑒 𝑖𝛼𝑥 ) − ∫ (𝑖𝛼)𝑒 𝑖𝛼𝑥 𝑑𝑥
√2𝜋 −∞ 𝑑𝑥 𝑟 √2𝜋 −∞ √2𝜋 −∞ 𝑑𝑥 𝑟−1

𝑑𝑟−1 𝑓
If we assume that tends to zero as 𝑥 → ±∞ , we may write the above result in the form
𝑑𝑥 𝑟−1
(𝑟)
𝐹 (𝛼) = −(𝑖𝛼)𝐹 (𝑟−1) (𝛼) = (−𝑖𝛼)2 𝐹 (𝑟−2) (𝛼). . = (−𝑖𝛼)𝑟 𝐹(𝛼)

Hence 𝐹 (𝑟) (𝛼) = (−𝑖𝛼)𝑟 𝐹(𝛼) and therefore we have

8
PMAT 32322 Mathematical Methods Note 08

ℱ[𝑓 𝑟 (𝑥): 𝛼] = (−𝑖𝛼)𝑟 𝐹(𝛼) ____(31)

Example 06:
2
Find the Fourier cosine transform of 𝑒 −𝑎𝑡 .
Solution:
We have the definition of Fourier cosine transform

2 2 ∞ 2
𝐹𝑐 (𝛼) = ℱ𝑐 [𝑒 −𝑎𝑡 : 𝛼] = √𝜋 ∫0 𝑒 −𝑎𝑡 cos 𝛼𝑡 𝑑𝑡 = 𝐼 (say) ___(a)

Differentiating with respect to 𝛼, we obtain

𝑑𝐼 2 ∞ 2 1 2 ∞ 2
= −√𝜋 ∫0 𝑡𝑒 −𝑎𝑡 cos 𝛼𝑡 𝑑𝑡 = 2𝑎 √𝜋 ∫0 sin 𝛼𝑡 𝑑(𝑒 −𝑎𝑡 )
𝑑𝛼

1 2 2 ∞ ∞ 2
= 2𝑎 √𝜋 [(𝑒 −𝑎𝑡 sin 𝛼𝑡 )0 − 𝛼 ∫0 𝑒 −𝑎𝑡 cos 𝛼𝑡 𝑑𝑡]

𝑑𝐼 𝛼 𝑑𝐼 𝛼
Therefore, we have = − 2𝑎 𝐼 i.e. = − 2𝑎 𝑑𝛼
𝑑𝛼 𝐼

𝛼2
On integration, we get 𝐼 = 𝑐 𝑒 −4𝑎 .___(b)

2 ∞ 2
But when 𝛼 = 0, from (a) we have 𝐼 = √𝜋 ∫0 𝑒 −𝑎𝑡 𝑑𝑡, using example 01 and using change of
scale property , we obtain

2 1 √𝜋 1
𝐼 = √𝜋 = .
𝑎 2 √2𝑎

𝛼2
1 2 1
From equation (b), we get 𝑐 = . Hence we get ℱ𝑐 [𝑒 −𝑎𝑡 : 𝛼] = 𝑒 −4𝑎 .
√2𝑎 √2𝑎

Example 07:
𝛼
If the Fourier sine transform of 𝑓(𝑥) is 1+𝛼2 , find 𝑓(𝑥).

Solution:
From this definition, we have

2 ∞ 𝛼
𝑓(𝑥) = √𝜋 ∫0 sin 𝛼𝑥 𝑑𝛼
1+𝛼2

2 ∞ 1+𝛼2 −1
= √𝜋 ∫0 sin 𝛼𝑥 𝑑𝛼
𝛼(1+𝛼2 )

9
PMAT 32322 Mathematical Methods Note 08

2 ∞ sin 𝛼𝑥 2 ∞ sin 𝛼𝑥
𝑓(𝑥) = √ ∫ 𝑑𝛼 − √ ∫ 𝑑𝛼
𝜋 0 𝛼 𝜋 0 𝛼(1 + 𝛼 2 )
∞ sin 𝛼𝑥 𝜋
But ∫0 𝑑𝛼 = 2 . Hence we have
𝛼

𝜋 2 ∞ sin 𝛼𝑥
𝑓(𝑥) = √( 2 ) − √𝜋 ∫0 𝑑𝛼 _____(a)
𝛼(1+𝛼2 )

𝑑𝑓 2 ∞ cos 𝛼𝑥
= −√𝜋 ∫0 (1+𝛼2 )
𝑑𝛼 _____(b)
𝑑𝑥

𝑑2 𝑓 2 ∞ 𝛼 sin 𝛼𝑥
= √𝜋 ∫0 𝑑𝛼 ______(c)
𝑑𝑥 2 (1+𝛼2 )

𝑑2 𝑓 2 ∞ sin 𝛼𝑥 𝜋
From (a) and (c), it follows that − 𝑓 = √𝜋 ∫0 𝑑𝛼 − √(2 ) = 0
𝑑𝑥 2 𝛼(1+𝛼2 )

Whose solution is found to be 𝑓 = 𝑐1 𝑒 𝑥 + 𝑐2 𝑒 −𝑥 (d)

𝑑𝑓
Therefore, = 𝑐1 𝑒 𝑥 − 𝑐2 𝑒 −𝑥 .
𝑑𝑥

𝜋
When 𝑥 = 0, from (a) we have 𝑓(0) = √( ) , and from (b),
2

𝑑𝑓(0) 2 ∞ 1 𝜋
= −√𝜋 ∫0 (1+𝛼2 )
𝑑𝛼 = −√( 2 ).
𝑑𝑥

𝜋 𝜋
Also from (d), using these results, we get 𝑐1 + 𝑐2 = √( 2 ), 𝑐1 − 𝑐2 = −√( 2 )

𝜋
Solving above two equations, we get 𝑐1 = 0, 𝑐2 = √(2 ).

𝜋
Thus, we have 𝑓(𝑥) = √( ) 𝑒 −𝑥 .
2

Example 08:
If the Fourier cosine transform of 𝑓(𝑥) is 𝛼 𝑛 𝑒 −𝛼𝑎 , find 𝑓(𝑥).
Solution:

2 ∞
Using the definition, we have 𝑓(𝑥) = √𝜋 ∫0 𝛼 𝑛 𝑒 −𝛼𝑎 cos 𝛼𝑥 𝑑𝛼 ___(a)

∞ 𝑎
But from calculus we know that ∫0 𝑒 −𝛼𝑎 cos 𝛼𝑥 𝑑𝛼 = 𝑎2 +𝑥 2.

Differentiating this n times with respect to 𝛼, we get

10
PMAT 32322 Mathematical Methods Note 08

∞ 𝑑𝑛 𝑎
(−1)𝑛 ∫0 𝛼 𝑛 𝑒 −𝛼𝑎 cos 𝛼𝑥 𝑑𝛼 = 𝑛 ( 2 2 )
𝑑𝑥 𝑎 +𝑥

1 𝑑𝑛 1 1
=2 (𝑎−𝑖𝑥 + 𝑎+𝑖𝑥)
𝑑𝑥 𝑛
1
= 2 [(−1)𝑛 𝑛! (𝑎 − 𝑖𝑥)−𝑛−1 + (−1)𝑛 𝑛! (𝑎 + 𝑖𝑥)−𝑛−1 ]
(−1)𝑛 𝑛!
= [(𝑎 − 𝑖𝑥)−𝑛−1 + (𝑎 + 𝑖𝑥)−𝑛−1 ]
2

Let 𝑎 + 𝑖𝑥 = 𝑟(cos 𝜃 + 𝑖 sin 𝜃). Then we have 𝑎 = 𝑟 cos 𝜃 and 𝑥 = 𝑟 sin 𝜃. Therefore we can
𝑥
take 𝑟 2 = 𝑎2 + 𝑥 2 and tan 𝜃 = 𝑎.

Thus, (𝑎 + 𝑖𝑥)−𝑛−1 = 𝑟 −𝑛−1 [cos(−𝑛 − 1)𝜃 + 𝑖 sin(−𝑛 − 1)𝜃]

(𝑎 − 𝑖𝑥)−𝑛−1 = 𝑟 −𝑛−1 [cos(−𝑛 − 1)𝜃 − 𝑖 sin(−𝑛 − 1)𝜃]
∞ 𝑛! cos(𝑛+1)𝜃
Then ∫0 𝛼 𝑛 𝑒 −𝛼𝑎 cos 𝛼𝑥 𝑑𝛼 = 𝑛+1⁄ .
(𝑎2 +𝑥 2 ) 2

2 𝑛! cos(𝑛+1)𝜃
Hence from equation (a), we get 𝑓(𝑥) = √𝜋 𝑛+1⁄ .
(𝑎2 +𝑥 2 ) 2

Example 09:
Find the Fourier transform of
𝜕𝑛 𝑢
(i) of the function 𝑢(𝑥, 𝑡) assuming that 𝑢 and its first 𝑛 − 1 derivatives with respect
𝜕𝑥 𝑛
to x vanish as 𝑥 → ±∞.
𝜕𝑢
(ii) .
𝜕𝑡

𝜕2 𝑢
Also, find the sine and cosine Fourier transforms of of the function 𝑢(𝑥, 𝑡).
𝜕𝑥 2

Solution:
(a) we shall adopt the following notation . The Fourier transform of 𝑢(𝑥, 𝑡) with respect to
the variable 𝑥 is defined as
1 ∞
ℱ[𝑢(𝑥, 𝑡): 𝑥 → 𝛼] = ∫−∞ 𝑒 𝑖𝛼𝑥 𝑢(𝑥, 𝑡) 𝑑𝑥 = 𝑈(𝛼, 𝑡)____(32)
√2𝜋

𝜕𝑢
Then the Fourier transform of is
𝜕𝑥
∞
𝜕𝑢 1 𝜕𝑢 𝑖𝛼𝑥
ℱ[ (𝑥, 𝑡): 𝑥 → 𝛼] = ∫ 𝑒 𝑑𝑥
𝜕𝑥 √2𝜋 −∞ 𝜕𝑥
1 ∞ ∞
Integration by parts yields [(𝑢(𝑥, 𝑡)𝑒 𝑖𝛼𝑥 )−∞ − 𝑖𝛼 ∫−∞ 𝑢(𝑥, 𝑡) 𝑒 𝑖𝛼𝑥 𝑑𝑥]
√2𝜋

If we assume lim 𝑢(𝑥, 𝑡) = 0.

𝑥→±∞

11
PMAT 32322 Mathematical Methods Note 08

Then we find that

𝜕𝑢
ℱ [𝜕𝑥 (𝑥, 𝑡): 𝑥 → 𝛼] = −𝑖𝛼 ℱ[𝑢(𝑥, 𝑡): 𝑥 → 𝛼] = −𝑖𝛼𝑈(𝛼, 𝑡)___(33)

𝜕2 𝑢
Similarly, the Fourier transform of is
𝜕𝑥 2

𝜕2𝑢 1 ∞ 𝜕2𝑢 1 ∞ 𝜕𝑢
ℱ [𝜕𝑥 2 (𝑥, 𝑡): 𝑥 → 𝛼] = ∫ 𝑒 𝑖𝛼𝑥 𝑑𝑥 = ∫ 𝑒 𝑖𝛼𝑥 𝑑 (𝜕𝑥 )
√2𝜋 −∞ 𝜕𝑥 2 √2𝜋 −∞

1 𝜕𝑢 ∞ ∞ ∞
= [(𝜕𝑥 𝑒 𝑖𝛼𝑥 ) − 𝑖𝛼(𝑒 𝑖𝛼𝑥 )−∞ + (𝑖𝛼)2 ∫−∞ 𝑒 𝑖𝛼𝑥 𝑢 𝑑𝑥]
√2𝜋 −∞
𝜕𝑢
Assuming that both of 𝑢 and 𝜕𝑥 tend to zero as 𝑥 → ±∞.

𝜕2𝑢
ℱ [𝜕𝑥 2 (𝑥, 𝑡): 𝑥 → 𝛼] = (−1)2 (𝑖𝛼)2 𝑈(𝛼, 𝑡) ____(34)

Thus , in general, the Fourier transform of the n th derivative of 𝑢(𝑥, 𝑡) is given by

𝜕𝑛 𝑢
ℱ [𝜕𝑥 𝑛 (𝑥, 𝑡): 𝑥 → 𝛼] = (−1)𝑛 (𝑖𝛼)𝑛 𝑈(𝛼, 𝑡)____(35)
𝜕𝑢 1 ∞ 𝜕𝑢
(b)`Now ℱ [ 𝜕𝑡 (𝑥, 𝑡): 𝑥 → 𝛼] = ∫ (𝑥, 𝑡) 𝑒 𝑖𝛼𝑥 𝑑𝑥
√2𝜋 −∞ 𝜕𝑡

1 𝜕 ∞
= ∫ 𝑢(𝑥, 𝑡) 𝑒 𝑖𝛼𝑥 𝑑𝑥 = 𝑈𝑡 (𝛼, 𝑡).
√2𝜋 𝜕𝑡 −∞

𝜕𝑢
Therefore , we have ℱ [ 𝜕𝑡 (𝑥, 𝑡): 𝑥 → 𝛼] = 𝑈𝑡 (𝛼, 𝑡)___(36)

(c) In the case of Fourier sine and cosine transforms, we have

𝜕2𝑢 2 ∞ 𝜕2 𝑢
ℱ𝑠 [𝜕𝑥 2 (𝑥, 𝑡): 𝑥 → 𝛼] = √𝜋 ∫0 sin 𝛼𝑥 𝑑𝑥
𝜕𝑥 2

2 𝜕𝑢 ∞ 2 ∞ 𝜕𝑢
= √𝜋 [𝜕𝑥 sin 𝛼𝑥] − √𝜋 𝛼 ∫0 cos 𝛼𝑥 𝑑𝑥
0 𝜕𝑥

𝜕𝑢
We assume that → 0 as 𝑥 → ∞. Then the RHS of the above equation becomes
𝜕𝑥

∞ ∞
2 𝜕𝑢 2 ∞
− √ 𝛼 ∫ cos 𝛼𝑥 𝑑𝑥 = − √ 𝛼 {[𝑢(𝑥, 𝑡) cos 𝛼𝑥]0 + 𝛼 ∫ 𝑢 (𝑥, 𝑡) sin 𝛼𝑥 𝑑𝑥}
𝜋 0 𝜕𝑥 𝜋 0

Also, assuming that 𝑢(𝑥, 𝑡) → 0 as 𝑥 → ∞, the equation becomes

2
√ 𝛼 𝑢(𝑥, 𝑡)𝑥=0 − 𝛼 2 ℱ𝑠 [𝑢(𝑥, 𝑡): 𝑥 → 𝛼]
𝜋

𝜕2𝑢 2
Hence, we have ℱ𝑠 [𝜕𝑥 2 (𝑥, 𝑡): 𝑥 → 𝛼] = √𝜋 𝛼 𝑢(𝑥, 𝑡)𝑥=0 − 𝛼 2 ℱ𝑠 [𝑢(𝑥, 𝑡): 𝑥 → 𝛼]___(37)

12
PMAT 32322 Mathematical Methods Note 08

𝜕𝑢
Similarly it can be shown that if 𝑢(𝑥, 𝑡) → 0 and 𝜕𝑥 → 0 as 𝑥 → ∞.

𝜕2𝑢 2
ℱ𝑐 [𝜕𝑥 2 (𝑥, 𝑡): 𝑥 → 𝛼] = −√𝜋 𝛼 𝑢(𝑥, 𝑡)𝑥=0 − 𝛼 2 ℱ𝑐 [𝑢(𝑥, 𝑡): 𝑥 → 𝛼] ___(38)

Obviously the choice of the sine or cosine transform is decided by the form of the boundary
condition at the lower limit of the variable selected for exclusion. Thus, we observe that for the
𝜕2𝑢
exclusion of from a given PDE, we require
𝜕𝑥 2

𝑢|𝑥=0 in the case of sine transform

𝜕𝑢
| in the case of cosine transform.
𝜕𝑥 𝑥=0

8.6 Convolution Theorem

If 𝐹(𝛼) and 𝐺(𝛼) are the Fourier transforms of the functions 𝑓(𝑥) and 𝑔(𝑥), then the product
𝐹(𝛼)𝐺(𝛼) is the Fourier transform of the convolution product 𝑓 ∗ 𝑔.
Convolution of 𝑓(𝑥) and 𝑔(𝑥) is defined as
1 ∞
𝑓∗𝑔= ∫ 𝑓(𝑢)𝑔(𝑥 − 𝑢) 𝑑𝑢.
√2𝜋 −∞

Then the convolution theorem says that ℱ[𝑓 ∗ 𝑔: 𝛼] = 𝐹(𝛼)𝐺(𝛼)

Example 10:
Using the Fourier cosine and sine transform of 𝑒 −𝑎𝑥 and 𝑒 −𝑏𝑥 , show that
∞ 𝑑𝛼 𝜋
∫0 (𝑎2 +𝛼2 )(𝑏2 +𝛼2 )
= 2𝑎𝑏(𝑎+𝑏) , 𝑎 > 0, 𝑏 > 0.

Solution:
Let 𝑓(𝑥) = 𝑒 −𝑏𝑥 and 𝑔(𝑥) = 𝑒 −𝑎𝑥 , then

2 ∞ 2 ∞ 2 𝑏
𝐹𝑐 (𝛼) = √ ∫0 𝑓(𝑥) cos 𝛼𝑥 𝑑𝑥 = √ ∫0 𝑒 −𝑏𝑥 cos 𝛼𝑥 𝑑𝑥 = √ .
𝜋 𝜋 𝜋 𝑏 2 +𝛼2

2 𝑎
Similarly it can be shown that 𝐺𝑐 (𝛼) = √𝜋 .
𝑎2 +𝛼2

∞ ∞ 2 ∞
However, ∫0 𝐹𝑐 (𝛼)𝐺𝑐 (𝛼) 𝑑𝛼 = ∫0 𝐹𝑐 (𝛼)𝑑𝛼 √𝜋 ∫0 𝑓(𝑥) cos 𝛼𝑥 𝑑𝑥𝑐

∞ 2 ∞
= ∫0 𝑔(𝑥)𝑑𝑥 √𝜋 ∫0 𝐹𝑐 (𝛼)cos 𝛼𝑥 𝑑𝑥𝑐

∞
= ∫0 𝑓(𝑥)𝑔(𝑥)𝑑𝑥 𝑐
∞ ∞ 1
Hence it follows that ∫0 𝐹𝑐 (𝛼)𝐺𝑐 (𝛼) 𝑑𝛼 = ∫0 𝑒 −(𝑎+𝑏)𝑥 𝑑𝑥 = 𝑎+𝑏

13
PMAT 32322 Mathematical Methods Note 08

∞ 𝑑𝛼 𝜋
Therefore we have ∫0 (𝑎2 +𝛼2 )(𝑏2 +𝛼2 )
= 2𝑎𝑏(𝑎+𝑏).

8.7 Solution of the Diffusion (Heat) Equation

Let us consider the problem of flow of heat in an infinite medium −∞ < 𝑥 < ∞ , when the initial
temperature distribution 𝑓(𝑥) is known and no heat sources are present. Mathematically, we have
to solve the problem described in the following example.
Example 11:
Solve the heat conduction equation given by
𝜕2𝑢 𝜕𝑢
𝑃𝐷𝐸: 𝑘 𝜕𝑥 2 = , −∞ < 𝑥 < ∞, 𝑡 > 0
𝜕𝑡

Boundary Conditions: 𝑢(𝑥, 𝑡) and 𝑢𝑥 (𝑥, 𝑡) both → 0 as |𝑥| → ∞

Initial Condition: 𝑢(𝑥, 0) = 𝑓(𝑥), −∞ < 𝑥 < ∞.
Solution:
Taking the Fourier transform of PDE, we get
𝑘(𝑖𝛼)2 ℱ[𝑢(𝑥, 𝑡): 𝑥 → 𝛼] = ℱ[𝑢𝑡 (𝑥, 𝑡): 𝑥 → 𝛼]
or −𝑘𝛼 2 𝑈(𝛼, 𝑡) = 𝑈𝑡 (𝛼, 𝑡)
or 𝑈𝑡 (𝛼, 𝑡) + 𝑘𝛼 2 𝑈(𝛼, 𝑡) = 0 ____(a)
In deriving this, boundary conditions are also utilized. The Fourier transform of the initial
condition is given by 𝑈(𝛼, 0) = 𝐹(𝛼), −∞ < 𝑥 < ∞ ____(b)
2
The solution of equation (a) can be readily seen to be 𝑈(𝛼, 𝑡) = 𝐴𝑒 −𝑘𝛼 𝑡 .
When 𝑡 = 0, we have form (b), the relation 𝑈 = 𝐹(𝛼), implying 𝐴 = 𝐹(𝛼). Therefore we have
2𝑡
𝑈(𝛼, 𝑡) = 𝐹(𝛼)𝐴𝑒 −𝑘𝛼 _____(c)
Inverting this relation, we obtain
1 ∞ 2
𝑢(𝑥, 𝑡) = ∫ 𝐹(𝛼)𝑒 −𝑘𝛼 𝑡 𝑒 −𝑖𝛼𝑥 𝑑𝛼 ___(d)
√2𝜋 −∞

The product form of the integrand in (d) suggest the use of convolution. If the Fourier transform
2
of 𝑔(𝑥) is 𝑒 −𝑘𝛼 𝑡 , then 𝑔(𝑥) will be given by
1 ∞ 2
𝑔(𝑥) = ∫ 𝑒 −𝑘𝛼 𝑡 𝑒 −𝑖𝛼𝑥 𝑑𝛼 .
√2𝜋 −∞

𝑏2
∞ 2 𝜋
But if 𝑎 > 0, 𝑏 is real or complex , and we know that ∫−∞ 𝑒 (−𝑎𝑥 −2𝑏𝑥) 𝑑𝑥 = √𝑎 𝑒 .
𝑎

Here , 𝑎 = 𝑘𝑡, 2𝑏 = 𝑖𝑥 . Therefore, we have

14
PMAT 32322 Mathematical Methods Note 08

−𝑥2 𝑥2
1 𝜋 1 −( )
𝑔(𝑥) = √𝑘𝑡 𝑒 4𝑘𝑡 = 𝑒 4𝑘𝑡 .
√2𝜋 √2𝑘𝑡

Using the convolution theorem, we get

1 ∞
𝑢(𝑥, 𝑡) = ∫ 𝑓(𝛼)𝑔(𝑥 − 𝛼) 𝑑𝛼 ___(e)
√2𝜋 −∞
2
((𝑥−𝛼))
1 ∞ 1 −( )
4𝑘𝑡
Hence , 𝑢(𝑥, 𝑡) = ∫−∞
𝑓(𝛼) 𝑒 𝑑𝛼
√2𝜋 √2𝑘𝑡
2
((𝑥−𝛼))
1 ∞ 1 −( )
4𝑘𝑡
= ∫ 𝑓(𝛼) 𝑒 𝑑𝛼 ____(f)
√4𝜋𝑘𝑡 −∞ √2𝑘𝑡
𝛼−𝑥
Using the change of variable 𝑧 = , we can rewrite solution (f) in the form
√4𝑘𝑡
1 ∞ 2
𝑢(𝑥, 𝑡) = ∫ 𝑓(𝑥 + √4𝑘𝑡𝑧)𝑒 −𝑧 𝑑𝑧.
√𝜋 −∞

Example 12: (Flow of heat in a semi-infinite medium)

Solve the heat conduction problem described by
𝜕2𝑢 𝜕𝑢
𝑃𝐷𝐸: 𝑘 𝜕𝑥 2 = , 0 < 𝑥 < ∞, 𝑡 > 0
𝜕𝑡

Boundary Condition: 𝑢(0, 𝑡) = 𝑢0 , 𝑡 ≥ 0

Initial Condition: 𝑢(𝑥, 0) = 0, 0 < 𝑥 < ∞.
𝑢 and 𝑢𝑥 both tend to zero as 𝑥 → ∞.
Solution:
Since 𝑢 is specified at 𝑥 = 0, the Fourier sine transform is applicable to this problem. Taking the
Fourier sine transform of the given PDE and using the notation

2 ∞
𝑈𝑠 (𝛼, 𝑡) = √𝜋 ∫0 𝑢(𝑥, 𝑡) sin 𝛼𝑥 𝑑𝑥

Taking the Fourier sine transform,

2 𝜕𝑈𝑠 (𝛼,𝑡)
𝑘 [√𝜋 𝛼 𝑢(𝑥, 𝑡)|𝑥=0 − 𝛼 2 ℱ𝑠 [𝑢(𝑥, 𝑡): 𝑥 → 𝛼]] = 𝜕𝑡

𝑑𝑈𝑠 (𝛼,𝑡) 2
or + 𝑘𝛼 2 𝑈𝑠 = √𝜋 𝑘𝛼𝑢0 ____(a)
𝑑𝑡

2 𝑢0 2
Its general solution is found to be 𝑈𝑠 (𝛼, 𝑡) = √𝜋 (1 − 𝑒 −𝑘𝛼 𝑡 )___(b)
𝛼

15
PMAT 32322 Mathematical Methods Note 08

Inverting by Fourier inverse sine transform, we obtain

2 ∞
𝑢(𝑥, 𝑡) = √𝜋 ∫0 𝑈𝑠 (𝛼, 𝑡) sin 𝛼𝑥 𝑑𝛼

2 ∞ sin 𝛼𝑥 2
Therefore, 𝑢(𝑥, 𝑡) = 𝜋 𝑢0 ∫0 (1 − 𝑒 −𝑘𝛼 𝑡 ) 𝑑𝛼 ____(c)
𝛼
2 𝑦 −𝑢2
Noting that erf(𝑦) = ∫ 𝑒 𝑑𝑢 and using the standard integral
√𝜋 0
∞ 2 sin 2 𝛼𝑦 𝜋
∫0 𝑒 −𝛼 𝛼
𝑑𝛼 = 2
erf(𝑦), we have solution © in the form
2 𝜋 𝜋 𝑥
𝑢(𝑥, 𝑡) = 𝜋 𝑢0 [ 2 − 2 erf ( )] ____(d)
√2𝑘𝑡

Finally, the solution of the heat conduction problem is

2 𝑥/√2𝑘𝑡 2 𝑥
𝑢(𝑥, 𝑡) = 𝑢0 [1 − ∫ 𝑒 −𝑢 𝑑𝑢 = 𝑢0 erf ( )] ___(e)
√𝜋 0 √2𝑘𝑡

Example 13:
Determine the temperature distribution in the semi-infinite medium 𝑥 ≥ 0 when the end 𝑥 = 0 is
maintained at zero temperature and the initial temperature distribution is 𝑓(𝑥).
Solution:
The given problem is described by
𝜕𝑢 𝜕2 𝑢
PDE: 𝜕𝑡 = 𝐾 𝜕𝑥 2 , 0 < 𝑥 < ∞, 𝑡 > 0 ____(a)

Boundary Conditions: 𝑢(0, 𝑡) = 0 ____(b)

Initial Condition: 𝑢(𝑥, 0) = 𝑓(𝑥), 0 < 𝑥 < ∞, .____(c)
𝜕𝑢
and 𝑢 and 𝜕𝑥 both tends to zero as 𝑥 → ∞. Taking the Fourier sine transform of (a) and denoting

ℱ𝑠 [𝑢(𝑥, 𝑡): 𝑥 → ∞] by 𝑈𝑠 , we have

2 ∞ 𝜕𝑢 2 ∞ 𝜕2𝑢
√ ∫0 sin 𝛼𝑥 𝑑𝑥 = √𝜋 𝐾 ∫0 sin 𝛼𝑥 𝑑𝑥
𝜋 𝜕𝑡 𝜕𝑥 2

𝑑𝑈𝑠
which becomes = 𝐾[𝛼𝑢(0, 𝑡) − 𝛼 2 𝑈𝑠 ].
𝑑𝑡
𝑑𝑈𝑠
Using the boundary conditions (b), we obtain + 𝐾𝛼 2 𝑈𝑠 = 0 ____(d)
𝑑𝑡

Also, taking the Fourier sine transform of the initial condition (c), we get
𝑈𝑠 = 𝐹𝑠 (𝛼) at 𝑡 = 0 ___(e)
𝑑 2
Now, equation (d) can be rewritten as (𝑈𝑠 𝑒 𝐾𝛼 𝑡 ) = 0 ____(f)
𝑑𝑡

16
PMAT 32322 Mathematical Methods Note 08

2𝑡
Integrating we get 𝑈𝑠 𝑒 𝐾𝛼 = constant.
Using (e) , we note that 𝐹𝑠 (𝛼) = constant. Therefore, we get
2𝑡 2𝑡
𝑈𝑠 𝑒 𝐾𝛼 = 𝐹𝑠 (𝛼) or 𝑈𝑠 = 𝐹𝑠 (𝛼)𝑒 −𝐾𝛼 _____(g)
Finally, taking the inverse Fourier sine transform of (g), we obtain

2 ∞ 2
𝑢(𝑥, 𝑡) = √𝜋 ∫0 𝐹𝑠 (𝛼)𝑒 −𝐾𝛼 𝑡 sin 𝛼𝑥 𝑑𝛼 .

8.8 Solution of the Wave Equation

Wave motions occur in nature, viz.,sound waves, surface waves, transverse vibrations of an infinite
string, and of mechanical systems are governed by the wave equation. As our first example, we
shall consider the transverse displacements of an infinite string.
Example 14:
Displacement of an infinite string is governed by the PDE
𝜕2𝑢 𝜕2 𝑢
= 𝑐 2 𝜕𝑥 2 , −∞ < 𝑥 < ∞ _____(a)
𝜕𝑡 2

and initial conditions 𝑢(𝑥, 0) = 𝑓(𝑥), −∞ < 𝑥 < ∞ ____(b)

𝑢𝑡 (𝑥, 0) = 0 _____(c)
In view of initial conditions, the given problem is a properly proposed problem. Taking the Fourier
transform of PDE, we have
1 ∞ 𝜕2 𝑢 𝑐2 ∞ 𝜕2 𝑢
∫ 𝑒 𝑖𝛼𝑥 𝑑𝑥 = ∫ 𝑒 𝑖𝛼𝑥 𝑑𝑥
√2𝜋 −∞ 𝜕𝑡 2 √2𝜋 −∞ 𝜕𝑥 2

𝜕2 1 ∞
∫ 𝑢 𝑒 𝑖𝛼𝑥 𝑑𝑥 = −𝑐 2 𝛼 2 𝑈(𝛼, 𝑡)
𝜕𝑡 2 √2𝜋 −∞

𝑑2 𝑈
i.e. + 𝑐 2 𝛼 2 𝑈 = 0___(d)
𝑑𝑡 2

Its general solution is found to be 𝑈(𝛼, 𝑡) = 𝐴 cos(𝑐𝛼𝑡) + 𝐵 sin(𝑐𝛼𝑡) ____(e)

The Fourier transform of the initial conditions gives
𝑑𝑈
(𝛼, 0) = 0, 𝑈 = 𝐹(𝛼)____(f)
𝑑𝑡

i.e. {−𝑐𝛼𝐴 sins(𝑐𝛼𝑡) + 𝑐𝛼𝐵 cosn(𝑐𝛼𝑡)}|𝑡=0 = 0 implying 𝑐𝛼𝐵 = 0 or 𝐵 = 0. Also ,

equations (e) and (f) yield
𝐴 = 𝐹(𝛼) ____(g)
Thus, we have 𝑈(𝛼, 𝑡) = 𝐹(𝛼) cos(𝑐𝛼𝑡) _____(h)
Taking the inverse Fourier transform, we obtain

17
PMAT 32322 Mathematical Methods Note 08

1 ∞
𝑢(𝑥, 𝑡) = ∫ 𝐹(𝛼) cos(𝑐𝛼𝑡)𝑒 −𝑖𝛼𝑥 𝑑𝛼 ____(i)
√2𝜋 −∞

If we simplify equation (i) further, an interesting result emerges, i.e.

1 ∞ ∞ 𝑒 𝑖𝑐𝛼𝑡 +𝑒 −𝑖𝑐𝛼𝑡
𝑢(𝑥, 𝑡) = 2𝜋 ∫−∞[∫−∞ 𝑓(𝑢)𝑒 𝑖𝛼𝑢 𝑑𝑢] ( ) 𝑒 −𝑖𝛼𝑥 𝑑𝛼
2

1 1 ∞ ∞
= 2 [2𝜋 ∫−∞[∫−∞ 𝑓(𝑢)𝑒 𝑖𝑠𝑢 𝑑𝑢]] (𝑒 𝑖𝑐𝑠𝑡 + 𝑒 −𝑖𝑐𝑠𝑡 )𝑒 −𝑖𝑠𝑥 𝑑𝑠
1 1 ∞ 1 ∞
= 2[ 2𝜋
∫−∞ 𝑓(𝑢)𝑒 −𝑖𝑠𝑢 { 2𝜋 ∫−∞ 𝑒 𝑖𝑠(𝑥−𝑐𝑡) 𝑑𝑠} 𝑑𝑢 +
√ √
1 ∞ −𝑖𝑠𝑢 1 ∞
∫ 𝑓(𝑢)𝑒 { 2𝜋 ∫−∞ 𝑒 𝑖𝑠(𝑥+𝑐𝑡) 𝑑𝑠} 𝑑𝑢]
√2𝜋 −∞ √

Using the Fourier integral formula , we arrive at the result

1
𝑢(𝑥, 𝑡) = 2 [𝑓(𝑥 + 𝑐𝑡) + 𝑓(𝑥 − 𝑐𝑡)]

which is the well-known D’Alembert’s solution of the wave equation.

Example 15:
Obtain the solution of free vibrations of a semi-infinite string governed by
PDE: 𝑢𝑡𝑡 = 𝑐 2 𝑢𝑥𝑥 , 0 < 𝑥 < ∞, 𝑡 > 0 ____(a)
ICs: 𝑢(𝑥, 0) = 𝑓(𝑥) ____(b)
𝑢𝑡 (𝑥, 0) = 𝑔(𝑥)_____(c)
Solution:
Taking the Fourier sine transform of PDE, we have
∞ 𝜕2𝑢 ∞ 𝜕2𝑢
∫0 sin 𝛼𝑥 𝑑𝑥 = 𝑐 2 ∫0 sin 𝛼𝑥 𝑑𝑥__(d)
𝜕𝑡 2 𝜕𝑥 2

∞ 𝜕2𝑢
Now consider ∫0 𝑒 𝑖𝛼𝑥 𝑑𝑥. Integrating by parts, we get
𝜕𝑥 2
∞ ∞
𝜕𝑢 2
( sin 𝛼𝑥 − 𝛼𝑢 cos 𝛼𝑥) − 𝛼 ∫ 𝑢 sin 𝛼𝑥 𝑑𝑥
𝜕𝑥 0 0

𝜕𝑢
Now, since the string is fixed at 𝑥 = 0 for all 𝑡 and we assume that 𝑢 and 𝜕𝑥 both tend to zero as
𝑥 → ∞, we arrive at
∞ 𝜕2𝑢 ∞
∫0 𝑒 𝑖𝛼𝑥 𝑑𝑥 = −𝛼 2 ∫0 𝑢 sin 𝛼𝑥 𝑑𝑥 = −𝛼 2 𝑈
𝜕𝑥 2

𝑑2 𝑈
Since equation (d) reduces to + 𝑐 2 𝛼 2 𝑈 = 0 ___(e)
𝑑𝑡 2

Its general solution is known to be

18
PMAT 32322 Mathematical Methods Note 08

𝑈 = 𝐴(𝛼) cos(𝑐𝛼𝑡) + 𝐵(𝛼) cos(𝑐𝛼𝑡) ____(f)

where 𝐴(𝛼) and 𝐵(𝛼) have to be determined. Now, at 𝑡 = 0, we have
∞
𝑈 = ∫0 𝑓(𝑥) sin 𝛼𝑥 𝑑𝑥 = 𝐹(𝛼)
𝜕𝑈 ∞
= ∫0 𝑔(𝑥) sin 𝛼𝑥 𝑑𝑥 = 𝐺(𝛼)
𝜕𝑡

In the solution (f), if we take 𝑡 = 0, we have

𝑈 = 𝐴(𝛼) = 𝐹(𝛼)
𝜕𝑈
= 𝑐𝛼𝐵(𝛼) = 𝐺(𝛼)
𝜕𝑡

Substituting 𝐴(𝛼) and 𝐵(𝛼) into equation (f), we get

𝐺(𝛼)
𝑈(𝛼) = 𝐹(𝛼) cos(𝑐𝛼𝑡) + cos(𝑐𝛼𝑡)
𝑐𝛼

Taking the inverse Fourier sine transform of this relation, we obtain

2 ∞
𝑢(𝑥, 𝑡) = √𝜋 ∫0 𝑈(𝛼, 𝑡) sin 𝛼𝑥 𝑑𝛼

2 ∞ 𝐺(𝛼)
= √𝜋 ∫0 [𝐹(𝛼) cos(𝑐𝛼𝑡) sin 𝛼𝑥 + cos(𝑐𝛼𝑡) sin 𝛼𝑥 ] 𝑑𝛼
𝑐𝛼

1 ∞
= ∫ 𝐹(𝛼)[sin 𝛼(𝑥 + 𝑐𝑡) + sin 𝛼(𝑥 − 𝑐𝑡)]𝑑𝛼
√2𝜋 0

1 ∞ 𝐺(𝛼)
+ ∫ [cos 𝛼(𝑥 − 𝑐𝑡) − cos 𝛼(𝑥 + 𝑐𝑡)]𝑑𝛼
√2𝜋 0 𝑐𝛼

2 ∞
Since 𝑓(𝑥 + 𝑐𝑡) = √𝜋 ∫0 𝐹(𝛼) sin 𝛼(𝑥 + 𝑐𝑡) 𝑑𝛼

2 ∞
𝑔(𝑢) = √𝜋 ∫0 𝐺(𝛼) sin 𝑢𝛼 𝑑𝛼

𝑥+𝑐𝑡 2 ∞ 𝑥+𝑐𝑡
∫𝑥−𝑐𝑡 𝑔(𝑢)𝑑𝑢 = √𝜋 ∫0 𝐺(𝛼)𝑑𝛼 ∫𝑥−𝑐𝑡 sin 𝑢𝛼 𝑑𝛼

2 ∞ − cos 𝛼𝑢 𝑥+𝑐𝑡
= √𝜋 ∫0 𝐺(𝛼)𝑑𝛼 ( )
𝛼 𝑥−𝑐𝑡

2 ∞
= √𝜋 ∫0 𝐺(𝛼)𝑑𝛼[cos 𝛼(𝑥 − 𝑐𝑡) − cos 𝛼(𝑥 + 𝑐𝑡)]

We arrive at the solution

1 1 𝑥+𝑐𝑡
𝑢(𝑥, 𝑡) = 2 [𝑓(𝑥 + 𝑐𝑡) + 𝑓(𝑥 − 𝑐𝑡)] + 2𝑐 ∫𝑥−𝑐𝑡 𝑔(𝑢)𝑑𝑢 .

19
PMAT 32322 Mathematical Methods Note 08

8.9 Solution of Laplace Equation

One of the most important PDEs that occurs in many applications is the Laplace equation. Steady-
state heat conduction, the electrical potential in the steady flow of currents in solid conductors, the
velocity potential of inviscid, irrotational fluids, the gravitational potential at an exterior point due
to ellipsoidal Earth and so on, are all governed by Laplace equation. We shall now consider a few
related examples.
Example 16:
Solve the following boundary value problem in the half-plane 𝑦 > 0, described by
PDE: 𝑢𝑥𝑥 + 𝑢𝑦𝑦 = 0, −∞ < 𝑥 < ∞, 𝑦 > 0

BCs: 𝑢(𝑥, 0) = 𝑓(𝑥) −∞ < 𝑥 < ∞,

𝜕𝑢
𝑢 is bounded as 𝑦 → ∞: 𝑢 and both vanish as |𝑥| → ∞.
𝜕𝑥

Solution:
Since 𝑥 has an infinite range of values, we take the Fourier exponential transform of PDE in the
variable 𝑥 to get

ℱ[𝑢𝑥𝑥 : 𝑥 → 𝛼] + ℱ[𝑢𝑦𝑦 : 𝑥 → 𝛼] = 0
𝜕𝑢
Since 𝑢 and both vanish as |𝑥| → ∞, we have
𝜕𝑥
1 ∞
−𝛼 2 𝑈(𝛼, 𝑦) + ∫ 𝑢 𝑒 𝑖𝛼𝑥 𝑑𝑥 =0
√2𝜋 −∞ 𝑦𝑦

𝜕2 1 ∞
−𝛼 2 𝑈(𝛼, 𝑦) + 𝜕𝑦 2 [ ∫ 𝑢(𝑥, 𝑦)𝑒 𝑖𝛼𝑥 𝑑𝑥] =0
√2𝜋 −∞

𝑑2 𝑈(𝛼,𝑦)
i.e. − 𝛼 2 𝑈(𝛼, 𝑦) = 0 _____(a)
𝑑𝑦 2

its general solution is known to be

𝑈(𝛼, 𝑦) = 𝐴(𝛼)𝑒 𝛼𝑦 + 𝐵(𝛼)𝑒 −𝛼𝑦 ___(b)
Since 𝑢 must be bounded as 𝑦 → ∞, 𝑈(𝛼, 𝑦) and its Fourier transform also should be bounded as
𝑦 → ∞, implying 𝐴(𝛼) = 0 for 𝛼 > 0; but if 𝛼 < 0, 𝐵(𝛼) = 0; thus for any 𝛼,

𝑈(𝛼, 𝑦) = constant (𝑒 −|𝛼|𝑦 )___(c)

Now the Fourier transform of the BC yields

𝑈(𝛼, 0) = ℱ[𝑓(𝑥): 𝑥 → 𝛼] = 𝐹(𝛼)___(d)
From equations (c) and (d), we find that 𝐹(𝛼) = constant____(e)
1 ∞
Hence, we have 𝑈(𝛼, 𝑦) = 𝐹(𝛼)𝑒 −|𝛼|𝑦 = ∫ 𝑓(𝑥)𝑒 −|𝛼|𝑦 𝑒 𝑖𝛼𝑥 𝑑𝑥
√2𝜋 −∞

20
PMAT 32322 Mathematical Methods Note 08

Taking the Fourier inverse transform , we obtain, after replacing the dummy variable 𝑥 by 𝜉, the
equation
1 ∞ 1 ∞
𝑢(𝑥, 𝑦) = ∫ [ ∫ 𝑓(𝜉)𝑒 −|𝛼|𝑦 𝑒 𝑖𝛼𝜉 𝑑𝜉 ] 𝑒 −𝑖𝛼𝑥 𝑑𝛼
√2𝜋 −∞ √2𝜋 −∞

1 ∞ ∞
= 2𝜋 ∫−∞ 𝑓(𝜉)𝑑𝜉 [∫−∞ 𝑒 −|𝛼|𝑦+𝛼𝑖(𝜉−𝑥) ] 𝑑𝛼

1 ∞ −|𝛼|𝑦+𝛼𝑖(𝜉−𝑥) 1 ∞ 1 ∞
But ∫ 𝑒
2𝜋 −∞
= 2𝜋 ∫−∞ 𝑒𝑥𝑝[𝛼[𝑦 + 𝑖(𝜉 − 𝑥)]]𝑑𝛼 + 2𝜋 ∫−∞ 𝑒𝑥𝑝[−𝛼[𝑦 − 𝑖(𝜉 − 𝑥)]]𝑑𝛼
∞ ∞
1 exp([𝛼[𝑦+𝑖(𝜉−𝑥)]]) 1 exp([−𝛼[𝑦−𝑖(𝜉−𝑥)]])
= 2𝜋 [ ] − [ ]
𝑦+𝑖(𝜉−𝑥) −∞ 2𝜋 𝑦−𝑖(𝜉−𝑥) −∞

1 1 1
= 2𝜋 [𝑦+𝑖(𝜉−𝑥) − 𝑦−𝑖(𝜉−𝑥)]

1 𝑦
=𝜋 (𝜉−𝑥)2 +𝑦 2
.____(f)

Substituting equation (f) into equation (e), the required solution is found to be
𝑦 ∞ 𝑓(𝜉)𝑑𝜉
𝑢(𝑥, 𝑦) = 𝜋 ∫−∞ (𝜉−𝑥)2 +𝑦2 .

This solution is well-known Poison integral formula and is valid for 𝑦 > 0. When 𝑓(𝑥) is bounded
and piecewise continuous for all real x.
Example 17:
Solve the following Neumann problem described by
PDE: 𝑢𝑥𝑥 (𝑥, 𝑦) + 𝑢𝑦𝑦 (𝑥, 𝑦) = 0, −∞ < 𝑥 < ∞, 𝑦 > 0

BCs: 𝑢𝑦 (𝑥, 0) = 𝑓(𝑥) −∞ < 𝑥 < ∞,

𝜕𝑢
𝑢 is bounded as 𝑦 → ∞: 𝑢 and both vanish as |𝑥| → ∞.
𝜕𝑥

Solution:
Let us define a function 𝜙(𝑥, 𝑦) = 𝑢𝑦 (𝑥, 𝑦). Then
𝜕
𝜙𝑥𝑥 + 𝜙𝑦𝑦 = 𝜕𝑦 (𝑢𝑥𝑥 + 𝑢𝑦𝑦 ) = 0

BC: 𝜙(𝑥, 0) = 𝑢𝑦 (𝑥, 0) = 𝑓(𝑥)____(a)

Thus, the function 𝜙(𝑥, 𝑦) is a solution of the problem described in example 16 and therefore, the
solution of given PDE is of the form
𝑦 ∞ 𝑓(𝜉)𝑑𝜉
𝜙(𝑥, 𝑦) = 𝜋 ∫−∞ (𝜉−𝑥)2 +𝑦2 𝑦 > 0 ___(b)

1 ∞ 𝑦 𝑑𝑦
However, 𝑢(𝑥, 𝑦) = ∫ 𝜙(𝑥, 𝑦) 𝑑𝑦 = 𝜋 ∫−∞ 𝑓(𝜉) ∫ (𝜉−𝑥)2+𝑦2 𝑑𝜉

21
PMAT 32322 Mathematical Methods Note 08

1 ∞
Therefore, 𝑢(𝑥, 𝑦) = 2𝜋 ∫−∞ 𝑓(𝜉) log[(𝜉 − 𝑥)2 + 𝑦 2 ]𝑑𝜉 + constant.

22