11.1 Fourier Sine and Cosine Integrals
11.1 Fourier Sine and Cosine Integrals
If the function f is an even function, the integral of A(α) has an even integrand. Therefore
we can simplify the integral to
Z ∞
2
A(α) = f (u) cos αu du
π 0
Since the integrand of the integral in B(α) is odd and therefore B(α) = 0. Thus for even
function f we have
Z ∞
f (x) ∼ A(α) cos αx dα
0
Similarly, for an odd function f we have
Z ∞
2
A(α) = 0 and B(α) = f (u) sin αu du
π 0
and
Z ∞
f (x) ∼ B(α) sin αx dα
0
Remark: Similar to half range Fourier series, we can represent a function defined for
all real x > 0 by Fourier sine or Fourier cosine integral by extending the function as an
odd function or as an even function over the entire real axis, respectively.
We summarize the above results in the following theorem:
11.1.1 Theorem
Assume that f is piecewise smooth function on every finite interval on the positive x-axis
and let f be absolutely integrable over 0 to ∞. Then f may be represented by either:
a) Fourier cosine integral
Z ∞
f (x) ∼ A(α) cos αx, dα 0 < x < ∞,
0
Lecture Notes on Fourier Transform
where
Z ∞
2
A(α) = f (u) cos αu du
π 0
where
Z ∞
2
B(α) = f (u) sin αu du
π 0
Moreover, the above Fourier cosine and sine integrals converge to [f (x+) + f (x−)]/2.
11.2.1 Problem 1
The function is not defined at x = −π and therefore the Fourier integral at x = −π will
converge to the average value 0−1 1
2 = −2.
2
Lecture Notes on Fourier Transform
11.2.2 Problem 2
Hence evaluate
Z ∞ Z ∞
sin πα cos πα (1 − cos πα) sin πα
dα and dα
0 α 0 α
where Z ∞ Z π
2 2 2 (1 − cos πα)
B(α) = f (u) sin αu du = sin αu du =
π 0 π 0 π α
Therefore
Z ∞
2 (1 − cos πα)
f (x) ∼ sin αx dα
π 0 α
3
Lecture Notes on Fourier Transform
To get the required integral we now substitute x = π into the above integral
Z ∞ Z ∞
2 sin πα cos πα 1 sin πα cos πα π
dα = =⇒ dα =
π 0 α 2 0 α 4
4
Lecture Notes on Fourier Transform
In this lesson we introduce Fourier cosine and sine transforms. Evaluation and proper-
ties of Fourier cosine and sine transform will be discussed. The parseval’s identities for
Fourier cosine and sine transform will be given.
and then
∞
r Z
2
f (x) = fˆc (α) cos αx dα (11.2)
π 0
The function fˆc (α) as given by (11.1) is known as the Fourier cosine transform of f (x)
in 0 < x < ∞. We shall denote Fourier cosine transform by Fc (f ). The function f (x)
as given by (11.2) is called inverse Fourier cosine transform of fˆc (α). It is denoted by
Fc−1 (fˆc ).
11.4 Properties
We mention here some important properties of Fourier cosine and sine transform that will
be used in the application to solving differential equations.
1. Linearity: Let f and g are piecewise continuous and absolutely integrable functions.
Then for constants a and b we have
Fc (af + bg) = aFc (f ) + bFc (g) and Fs (af + bg) = aFs (f ) + bFc (g)
5
Lecture Notes on Fourier Transform
Note that these properties are obvious and can be proved just using linearity of the inte-
grals.
2. Transform of Derivatives: Let f (x) be continuous and absolutely integrable on
x−axis. Let f 0 (x) be piecewise continuous and on each finite interval on [0, ∞] and
f (x) → 0 as x → ∞. Then,
r
2
Fc {f 0 (x)} = αFs {f (x)} − f (0) and Fs {f 0 (x)} = −αFc {f (x)}
π
Similarly the other result for Fourier sine transform can be obtained.
Remark: The above results can easily be extended to the second order derivatives to
have
r r
00 2 0 2
2
Fc {f (x)} = −α Fc {f (x)} − f (0) and Fs {f 00 (x)} = αf (0) − α2 Fs {f (x)}
π π
Note that here we have assumed continuity of f and f 0 and piecewise continuity of f 00 .
Further, we also assumed that f and f 0 both goes to 0 as x approaches to ∞.
3. Parseval’s Identities: For Fourier sine and cosine transform we have the following
identities
Z ∞ Z ∞ Z ∞ Z ∞
2
fˆc (α)ĝc (α) dα = fˆc (α) dα = [f (x)]2
i) f (x)g(x) dx ii)
0 0 0 0
Z ∞ Z ∞ Z ∞ 2
Z ∞
iii) fˆs (α)ĝs (α) dα = f (x)g(x) dx iv) fˆs (α) dα = [f (x)]2
0 0 0 0
6
Lecture Notes on Fourier Transform
Proof: We prove the first identity and rest can be proved similarly. We take the right hand
side of the identity and use the definition of the inverse cosine transform to get
Z ∞ r Z ∞ Z ∞
2
f (x)g(x) dx = f (x) ĝc (α) cos(αx) dα dx
0 π 0 0
11.5.1 Problem 1
Find the Fourier sine transform of e−x , x > 0. Hence show that
Z ∞
x sin mx πe−m
2
dx = ,m>0
0 1+x 2
Let us denote the integral on the right hand side by i and evaluate it by integrating by parts
as
Z ∞ ∞ Z ∞ Z ∞
−x −x −x
I= e sin αx dx = −e sin αx + α e cos αx dx = α e−x cos αx dx
0 0 0 0
This implies
α
I=
1 + α2
Finally substituting the value of I to the expression of Fourier sine transform above we
get r
2 α
Fs {e−x } =
π 1 + α2
7
Lecture Notes on Fourier Transform
11.5.2 Problem 2
2
Find the Fourier cosine transform of e−x , x > 0.
Solution: By the definition of Fourier cosine transform we have
r Z ∞ r Z ∞
2 2 2 2
Fc {e−x } = f (u) cos(αu) du = e−u cos(αu) du
π 0 π 0
Let us denote the integral on the right hand side by I and differentiate it with respect to α
as Z ∞ Z ∞
dI d −u2 2
= e cos(αu) du = − e−u sin(αu)u du
dα dα 0 0
Integrating by parts we get
i∞ Z ∞
dI 1 −u2 α α
h 2
= e sin(αu) − e−u cos(αu) du = − I
dα 2 0 2 0 2
This implies
2
I = c e−α /4
√ √
π π
Using I(0) = 2 , we evaluate the constant c = 2 . Then we have
√
π −α2 /4
I= e
2
Therefore the desired Fourier cosine transform is given as
r √
2 2 π −α2 /4 1 2
Fc {e−x } = e = √ e−α /4
π 2 2
8
Lecture Notes on Fourier Transform
11.5.3 Problem 3
∞ ∞
t2
Z Z
dt π π
i) 2 2 2 2
= ii) 2
dt =
0 (a + t )(b + t ) 2ab(a + b) 0 (t + 1) 4
Solution: i) For the first part let f (x) = e−ax and g(x) = e−bx . It can easily be shown that
r Z ∞ r
2 −ax 2 a
Fc {f } = fˆc (α) = e cos αx dx = 2 2
π 0 π a +α
∞
r Z r
2 2 b
Fc {g} = fˆc (α) = e−bx cos αx dx =
π 0 π b + α2
2
Thus we get Z ∞
dα π
=
0 (a2 + α2 )(b2 + α2 ) ab(a + b)
ii) For the second part we use Fourier sine transform of e−x as
r
2 α
Fs {e−x } = fˆs (α) = .
π 1 + α2