If f (x) is defined on a finite interval [−l, l] and is piecewise continuous then we can con-

struct a Fourier series corresponding to the function f and this series will represent the
function on this interval if the function satisfies some additional conditions discussed be-
fore. Furthermore, if f is periodic then we may be able to represent the function by its
Fourier series on the entire real line. Now suppose the function is not periodic and is de-
fined on the entire real line. Then we do not have any possibility to represent the function
by the Fourier series. However, we may still be able to represent the function in terms
of sine and cosines using an integral, called Fourier integral, instead of a summation. In
this lesson we discuss a representation of a non-periodic function by letting l → ∞ in the
Fourier series of a function defined on [−l, l].

10.1 Fourier Integral Representation of a Function

Consider any function f (x) defined on [−l, l] that can be represented by a Fourier series as
∞
a0 X nπx nπx
 
f (x) = + an cos + bn sin . (10.1)
2 l l
n=1

For a more general case we can replace left hand side of the above equation by the average
value (f (x+) + f (x−))/2. We now see what will happen if we let l → ∞. It should be
mentioned that as l approaches to ∞ the function f (x) becomes non-periodic defined on
the real axis. Substituting an and bn in the equation (10.1) we get
∞
!
Z l Z l Z l
1 1X nπu nπx nπu nπx
f (x) = f (u) du + f (u) cos du cos + f (u) sin du sin
2l −l l −l l l −l l l
n=1

Using the identity cos x cos y + sin x sin y = cos(x − y), we get
Z l ∞ Z l
1 1X nπ
f (x) = f (u) du + f (u) cos (u − x) du (10.2)
2l −l l −l l
n=1
Z ∞
If we assume that |f (u)| du converges, the first term on the right hand side approaches
−∞
Z l Z ∞
1 1
to 0 as l → ∞ since f (u) du ≤ |f (u)| du.
2l −l 2l −∞

Letting l → ∞ in equation (10.2), we get

∞ Z ∞ ∞ Z ∞
1X nπ πX1 nπ
f (x) = lim f (u) cos (u − x) du = lim f (u) cos (u − x) du
l→∞ l −∞ l l→∞ l π −∞ l
n=1 n=1
 Lecture Notes on Fourier Transform

0 Δα 2Δα 3Δα

Figure 10.1: Sum of area of trapezoid as area under curve in the limiting case

For simplifications, we define

Z ∞
π 1
∆α = and F (α) = f (u) cos α(u − x) du
l π −∞

With these definitions and noting ∆α → 0 as l → ∞, we have

∞
X
f (x) = lim ∆αF (n∆α)
∆α→0
n=1

Refereing Figure 10.1, we can write this limit of the sum in the form of improper integral
as
Z ∞ Z ∞Z ∞
1
f (x) = F (α)dα = f (u) cos α(u − x) du dα
0 π 0 −∞

This is called Fourier Integral Representation of f on the real line. Equivalently, this can
be rewritten as
Z ∞ Z ∞  Z ∞  
1
f (x) = f (u) cos αu du cos αx + f (u) sin αu du sin αx dα
π 0 −∞ −∞

It is often convenient to write

Z ∞
f (x) = [A(α) cos αx + B(α) sin αx] dα
0

where the Fourier Integral Coefficients are

Z ∞ Z ∞
1 1
A(α) = f (u) cos αu du and B(α) = f (u) sin αu du
π −∞ π −∞

2
 Lecture Notes on Fourier Transform

Remark It should be mentioned that above derivation is not rigorous proof of con-
vergence of the Fourier Integral to the function. This is just to give some idea of transition
form Fourier series to Fourier Integral. Nevertheless we summarize the convergence re-
sult, without proof, in the next theorem. In addition to all conditions required for the
convergence of Fourier series we need one more condition, namely, absolute integrability
of f . Further, note that Fourier integral representation of f (x) is entirely analogous to a
P∞
Fourier series representation of a function on finite interval n=1 · · · , is replaced with
R∞

0
· · · du .

10.1.1 Theorem

Assume that f is piecewise smooth on every finite interval on the x axis (or piecewise
continuous and one sided derivatives exist) and let f be absolutely integrable over entire
real axis. Then for each x on the entire axis we have
Z ∞Z ∞
1 f (x+) + f (x−)
f (u) cos α(u − x) du =
π 0 −∞ 2

As in the convergence of Fourier series if f is continuous and all other conditions are
satisfied then the Fourier integral converges

10.2 Example Problems

10.2.1 Problem 1

Let a be a real constant and the function f is defined as


 0, x < 0;

f (x) = x, 0 < x < a;

0, x > a.


i) Find the Fourier integral representation of f . Z ii) Determine the convergence of the
∞
1 − cos α
integral at x = a. iii) Find the value of the integral 2
dα.
0 α
Solution: i) The integral representation of f is
Z ∞
f (x) ∼ [A(α) cos αx + B(α) sin αx] dα (10.3)
0

3
 Lecture Notes on Fourier Transform

where
Z ∞ Z a  a
Z a 
1 1 1 u sin αu sin αu

A(α) = f (u) cos αu du = u cos αu du = − du
π −∞ π 0 π α 0 0 α
 
1 a sin αa (cos αa − 1) 1 cos αa + αa sin αa − 1
h i
= + 2
=
π α α π α2

Z ∞ Z a  a
Z a 
1 1 1 −u cos αu cos αu

B(α) = f (u) sin αu du = u sin αu du = + du
π −∞ π 0 π α 0 0 α

1 −a cos αa sin αa 1 sin αa − αa cos αa

h i h i
= + =
π α α2 π α2
Replacing A(α) and B(α) in equation (10.3), we have
Z ∞ h
cos αa + αa sin αa − 1
1
  sin αa − αa cos αa  i
f (x) ∼ cos αx + sin αx dα
π 0 α2 α2
Z ∞
1 cos α(a − x) + αa sin α(a − x) − cos αx
= dα
π 0 α2
ii) The function is not defined at x = a. The value of the Fourier integral at x = a is given
as Z ∞
1 1 − cos αa f (a+) + f (a−) 0+a a
2
dα = = =
π 0 α 2 2 2
iii) Substituting a = 1 in the above integral we get
Z ∞ Z ∞
1 1 − cos α 1 1 − cos α π
2
dα = =⇒ 2
dα =
π 0 α 2 0 α 2

10.2.2 Problem 2

Determine the Fourier integral representing

(
1, 0 < x < 2;
f (x) =
0, x < 0 and x > 2.
Z ∞
sin α
Further, find the value of the integral dα.
0 α
Solution: The Fourier integral representation of f is
Z ∞
f (x) ∼ [A(α) cos αx + B(α) sin αx] dα (10.4)
0

4
 Lecture Notes on Fourier Transform

where
Z ∞ Z 2 2
1 1 1 sin αu 1 sin 2α
A(α) = f (u) cos αu du = cos αu du = =
π −∞ π 0 π α 0 π α

Z ∞ Z 2 2
1 1 1 − cos αu 1 (1 − cos 2α)
B(α) = f (u) sin αu du = sin αu du = =
π −∞ π 0 π α 0 π α

Then, substituting calculated values of A(α) and B(α) in equation (10.4), we obtain
Z ∞ 
1 sin 2α (1 − cos 2α)
f (x) ∼ cos αx + sin αx dα
π 0 α α
Z ∞
1 sin α(2 − x) + sin αx
= dα
π 0 α

To find the value of the given integral we substitute x = 1 in the above Fourier integral
and use convergence theorem to get
Z ∞
2 sin α
dα = f (1) = 1
π 0 α

This gives the value of the desired integral as

Z ∞
sin α π
dα =
0 α 2