Jim Lambers

ENERGY 281
Spring Quarter 2007-08
Lecture 3 Notes

These notes are based on Rosalind Archer’s PE281 lecture notes, with some
revisions by Jim Lambers.

1 Introduction
The Fourier transform is an integral transform. When viewed in the context
of signal processing the application of the Fourier transform takes a func-
tion from real-space to frequency-space (see later examples). The Fourier
transform is defined by:
1 ∞
Z
F (s) = √ f (x)e−ixs dx (1)
2π −∞

The inverse transform is defined in a similar manner:

1 ∞
Z
f (x) = √ F (s)eixs ds (2)
2π −∞

We will also use the notation fˆ(s) for F (s). The Fourier transform exists if
f (x) and f ′ (x) are at least piecewise continuous and the following integral
exists: Z ∞
|f (x)| dx (3)
−∞
There are also some alternative definitions:
Z ∞
F (s) = f (x)e−ixs dx (4)
−∞

1 ∞
Z
f (x) = F (s)eixs ds (5)
2π −∞
and Z ∞
F (s) = f (x)e−i2πxs dx (6)
−∞
Z ∞
f (x) = F (s)ei2πxs ds (7)
−∞

1
Example Let
2
f (x) = e−x . (8)
Then
1 ∞
Z
2
F (s) = √ e−x e−isx dx (9)
2π −∞
1
Z ∞
2
= √ e−(x +ixs) dx (10)
2π −∞
1
Z ∞
2 2 2
= √ e−(x +ixs−s /4+s /4) dx (11)
2π −∞
1
Z ∞
2 2
= √ e−(x+is/2) −s /4 dx (12)
2π −∞
1 −s2 /4 ∞ −(x+is/2)2
Z
= √ e e dx (13)
2π −∞
1 −s2 /4 ∞+is/2 −ξ 2
Z
= √ e e dξ (14)
2π −∞+is/2
1
Z ∞
2 2
= √ e−s /4 e−x dx (15)
2π −∞
sZ
1 ∞ ∞
 Z 
2
= √ e−s /4 e−x2 dx e−x2 dx (16)
2π −∞ −∞
sZ
1 ∞ ∞
 Z 
2
= √ e−s /4 e−x2 dx e−y2 dy (17)
2π −∞ −∞
sZ
1 ∞ ∞
Z
2
= √ e−s /4 e−(x2 +y2 ) dx dy (18)
2π −∞ −∞
s
1 2
Z 2π Z ∞
= √ e−s /4 e−r2 r dr dθ (19)
2π 0 0
s
1 2 1
Z 2π Z ∞
= √ e−s /4 e−u du dθ (20)
2π 2 0 0
s Z
1 2
∞
= √ e−s /4 π e−u du (21)
2π 0
1 2
q
= √ e−s /4 −πe−u |∞ 0 (22)
2π
1 2 √
= √ e−s /4 π (23)
2π
1 −s2 /4
= √ e , (24)
2

2
where, in equation (14), ξ = x + is/2. In (15), we applied the Cauchy
integral theorem, applied to a rectangle in the complex plane with vertices
(x, 0), (−x, 0), (x, is/2), and (−x, is/2), as x → ∞. In (19), we converted
to polar coordinates to obtain an integral that could be evaluated.
The Fourier transform relates a function in real space (either time or
distance) to a function in frequency space. This can be seen by recalling:
eixs = cos(xs) + i sin(xs) (25)
Now consider the inverse transform:
1
Z ∞
f (x) = √ F (s)eixs ds (26)
2π −∞
This integral shows that the Fourier transform breaks a function f (x) into
a sum of sines and cosines with frequency s. (Recall the frequency of f (kx)
is |k|). The amplitude associated with any given frequency is given by F (s).
Example Consider f = cos x. The Fourier transform of f is
√ √
2π 2π
F (s) = δ(−1 + s) + δ(1 + s), (27)
2 2
where δ(s) is the Dirac delta function, which has the property
Z ∞
f (s)δ(s − c) ds = f (c). (28)
−∞

To verify that F (s) really is the Fourier transform of f (x), we compute its
inverse transform:
1
Z ∞
f (x) = √ eisx F (s) ds
2π −∞
"√ √ #
1 2π 2π
Z ∞
isx
= √ e δ(−1 + s) + δ(1 + s) ds
2π −∞ 2 2
√ √
1 isx 2π 1 isx 2π
Z ∞ Z ∞
= √ e δ(−1 + s) ds + √ e δ(1 + s) ds
2π −∞ 2 2π −∞ 2
1 1
Z ∞ Z ∞
= eisx δ(−1 + s) ds + eisx δ(1 + s) ds
2 −∞ 2 −∞
1 ix
= (e + e−ix )
2
1
= (cos x + i sin x + cos(−x) + i sin(−x))
2
1
= (cos x + cos x)
2
= cos x. (29)

3
The original function f (x) and its Fourier transform F (s) are shown in
Figures 1 and 2, respectively.

0.8

0.6

0.4

0.2
cos(x)

−0.2

−0.4

−0.6

−0.8

−1
−10 −5 0 5 10
x

Figure 1: f (x) = cos x

2 Fourier Transform Theorems

In stating and using the following theorems, which greatly simplify the pro-
cess of computing transforms and inverse transforms of functions, we will use
the following notation: given a function f (x), F(f (x)) refers to its Fourier
transform, commonly denoted by F (s). That is, F is a function whose input
is a function of x, and whose output is its transform, a function of s.

2.1 Theorem 1 - Linearity

F(f (x) + g(x)) = F(f (x)) + F(g(x)) (30)
and
F(cf (x)) = cF(f (x)) (31)

4
 2

1.5

0.5
Amplitude

−0.5

−1

−1.5

−2
−2 −1.5 −1 −0.5 0 0.5 1 1.5 2
Frequency

Figure 2: Frequency spectrum of f (x) = cos x

2.2 Theorem 2 - Shift Theorem

If
F(f (x)) = F (s) (32)
then
F(f (x − a)) = e−isa F (s). (33)
Proof:
1 ∞
Z
F(f (x − a)) = √ f (x − a)e−isx dx
2π −∞
1
Z ∞
= √ f (x − a)e−is(x−a)−isa dx
2π −∞
1
Z ∞
= e−isa √ f (x − a)e−is(x−a) dx
2π −∞
= e−isa F (s) (34)

5
2.3 Theorem 3 - Similarity Theorem
If
F(f (x)) = F (s) (35)
then
1 s
 
F(f (ax)) = F (36)
|a| a

2.4 Theorem 4 - Convolution Theorem

If
F(f (x)) = F (s) (37)
and
F(g(x)) = G(s) (38)
then √
F(f (x) ⋆ g(x)) = 2πF (s)G(s) (39)
where Z ∞
(f (x) ⋆ g(x))(x) = f (x − t)g(t) dt (40)
−∞
is the convolution of f and g.

2.5 Theorem 5 - Parseval’s theorem

If
F(f (x)) = F (s) (41)
then Z ∞ Z ∞
2
|f (x)| dx = |F (s)|2 ds. (42)
−∞ −∞

2.6 Theorem 6 - Derivatives

If
F(f (x)) = F (s) (43)
then
F(f (n) (x)) = (is)n F (s) (44)
where
dn
f (n) (x) =f (x). (45)
dxn
Note that this assumes the values of the derivatives vanish at ±∞.

6
3 Fourier Sine and Cosine Transforms
The Fourier transform is defined as:
1 ∞
Z
F(f (x)) = √ f (x)e−isx dx
2π −∞
1
Z ∞
= √ f (x)[cos(−sx) + i sin(−sx)] dx. (46)
2π −∞
Now consider a case where f (x) is the sum of an even and an odd function,
fe (x) and fo (x). Recall that if f (x) is an odd function,

f (−x) = −f (x). (47)

sin x is an example of an odd function. On the other hand, if f (x) is an

even function,
f (−x) = f (x). (48)
cos x is an example of an even function. Any function f (x) can be written
as the sum of an even and odd function. Specifically,

f (x) = fe (x) + fo (x), (49)

where
f (x) + f (−x) f (x) − f (−x)
fe (x) = , fo (x) = . (50)
2 2
With f (x) defined as the sum of an even and odd function, the Fourier
transform of f (x) becomes
1 ∞ Z
F (f (x)) = √ (fe (x) + fo (x)) cos sx dx −
2π −∞
1
Z ∞
i√ (fe (x) + fo (x)) sin sx dx. (51)
2π −∞
Now take into account the way products of even and odd functions behave:

• The product of two even functions, or two odd functions, is an even

function.

• The product of an even function and an odd function is an odd func-

tion.

7
Also note the following integral:
Z ∞
fo (x)dx = 0 (52)
−∞

Now substitute these relationships into (51):

1 ∞ 1 ∞
Z Z
F(f (x)) = √ fe (x) cos sx dx − i √ fo (x) sin sx dx (53)
2π −∞ 2π −∞
2 2i
Z ∞ Z ∞
= √ fe (x) cos sx dx − √ fo (x) sin sx dx. (54)
2π 0 2π 0
The fact that the Fourier transform splits into two terms (sine and cosine)
motivates the definition of the sine and cosine transforms Fs and Fc :
r
2 ∞
Z
Fc (f (x)) = f (x) cos xs dx = Fc (s) (55)
π 0
r
2 ∞
Z
Fs (f (x)) = f (x) sin xs dx = Fs (s) (56)
π 0
The inverse transforms are defined as follows:
r
2 ∞
Z
Fc−1 (Fc (s)) = Fc (s) cos xs ds = f (x) (57)
π 0
r
2 ∞
Z
Fs−1 (Fs (s))
= Fs (s) sin xs ds = f (x) (58)
π 0
The following rules for transforming second derivatives are especially useful:
r
2 ′
′′
Fc (f ) = − f (0) − s2 Fc (f ), (59)
π
r
2
′′
Fs (f ) = s f (0) − s2 Fs (f ). (60)
π
The use of sine and cosine transforms simplifies the transform procedure
when transforming even and odd functions, because the Fourier sine trans-
form of an even function is zero, as is the Fourier cosine transform of an odd
function. It follows that the sine and cosine transforms can be used in place
of the full Fourier transform for problems with
• semi-infinite domains,

• differential equations that have only even orders of derivatives,

• either f or f ′ specified at the boundary.

8
4 Example 1: 1D Pressure Diffusion
Consider a one-dimensional problem governed by the pressure equation in
dimensionless form,
∂ 2 pD ∂pD
2 = (61)
∂xD ∂tD
The boundary conditions are

pD (0, tD ) = 1, lim pD (xD , tD ) = 0, (62)

xD →∞

and the initial condition is

pD (xD , 0) = 0. (63)

Since the pressure and not the pressure derivative is set on the boundary, we
should use the sine transform. The choice of transform is made according to
equations (59) and (60), which relate the transform of the second derivative
to the boundary conditions.
First, we apply the Fourier sine transform to the spatial variable in
order to convert the partial differential equation to an ordinary differential
equation: r
2 ∂ p̂D
s pD (xD = 0, tD ) − s2 p̂D = (64)
π ∂tD
r
∂ p̂D 2
=⇒ + s2 p̂D = s. (65)
∂tD π
This equation can be solved using an integrating factor. Recall that for a
general linear first-order ODE of the form
dy
+ P (t)y = Q(t) (66)
dt
hR i
t
we can multiply through by the integrating factor exp P (s) ds to obtain

P (s) ds du
Rt Rt Rt
P (s) ds P (s) ds
e + P (t)e y=e Q(t) (67)
dt
d Rt Rt 
=⇒ e P (s) ds y = e P (s) ds Q(t) (68)
dt
Rt Z t Rτ
P (s) ds P (s) ds
=⇒ y(t)e = Q(τ )e dτ + C. (69)

9
 Applying this technique to thep transformed equation (65) with y(t) =
p̂D (s, tD ), P (t) = s2 and Q(t) = 2/πs yields
tD
r
2 s2 τ
Z
s2 tD
p̂D (s, tD )e = se dτ (70)
0 π

tD
r
2 −s2 (tD −τ )
Z
=⇒ p̂D (s, tD ) = se dτ
π
r0
21 2
= (1 − e−s tD ). (71)
πs
Now invert the Fourier sine transform to find pD :

pD (xD , tD ) = Fs−1 (p̂D (s, tD ))

r Z
2 ∞
= p̂D (s, tD ) sin(sxD ) ds
π 0
2 ∞1
Z
2
= (1 − e−s tD ) sin(sxD ) ds
π 0 s
xD xD
   
= 1 − Erf √ = Erf c √ (72)
4tD 4tD

where Erf (z) and Erf c(z) are the error function and the complimentary
error function, respectively, defined by
2
Z z 2
Erf (z) = √ e−t dt (73)
π 0

Erf c(z) = 1 − Erf (z) (74)

10