Properties and Theorems of the Fourier Transform

∞
1
f Im ( t ) = ------
2π ∫–∞ [ FRe ( ω ) sin ωt + FIm ( ω ) cos ω t ] dω (8.35)

We observe that the integrand of (8.35) is zero since it is an odd function with respect to ω because
both products inside the brackets are odd functions*.
Therefore, f Im ( t ) = 0 , that is, f ( t ) is real.

We can state then, that a necessary and sufficient condition for f ( t ) to be real, is that F ( – ω ) = F∗ ( ω ) .
Also, if it is known that f ( t ) is real, the Inverse Fourier transform of (8.3) can be simplified as fol-
lows:
From (8.13),
∞
1
f Re ( t ) = ------
2π ∫–∞ [ FRe ( ω ) cos ωt –FIm ( ω ) sin ωt ] dω (8.36)

and since the integrand is an even function with respect to ω , we rewrite (8.36) as
∞
1
f Re ( t ) = 2 ------
2π ∫0 [ F Re ( ω ) cos ωt – F Im ( ω ) sin ωt ] dω
(8.37)
∞ ∞
1 1 j [ ωt + ϕ ( ω ) ]
= ---
π ∫0 A ( ω ) cos [ ωt + ϕ ( ω ) ] dω = --- Re
π ∫0 F ( ω )e dω

8.3 Properties and Theorems of the Fourier Transform

1. Linearity
If F 1 ( ω ) is the Fourier transform of f 1 ( t ) , F 2 ( ω ) is the transform of f 2 ( t ) , and so on, the lin-
earity property of the Fourier transform states that

a1 f1 ( t ) + a2 f2 ( t ) + … + an fn ( t ) ⇔ a1 F1 ( ω ) + a2 F2 ( ω ) + … + an Fn ( ω ) (8.38)

where a i is some arbitrary real constant.

Proof:
The proof is easily obtained from (8.1), that is, the definition of the Fourier transform. The proce-
dure is the same as for the linearity property of the Laplace transform in Chapter 2.

* In (8.31) and (8.32), we determined that F Re ( ω ) is even and F Im ( ω ) is odd.

Signals and Systems with MATLAB Applications, Second Edition 8-9

Orchard Publications
Chapter 8 The Fourier Transform

2. Symmetry
If F ( ω ) is the Fourier transform of f ( t ) , the symmetry property of the Fourier transform states that

F ( t ) ⇔ 2πf ( – ω ) (8.39)

that is, if in F ( ω ) , we replace ω with t , we get the Fourier transform pair of (8.39).

Proof:

Since
∞
1
∫–∞ F ( ω )e
jωt
f ( t ) = ------ dω
2π
then,
∞
– j ωt
2πf ( – t ) = ∫–∞ F ( ω )e dω

Interchanging t and ω , we get

∞
– j ωt
2πf ( – ω ) = ∫–∞ F ( t )e dt

and (8.39) follows.

3. Time Scaling
If a is a real constant, and F ( ω ) is the Fourier transform of f ( t ) , then,

ω
f ( at ) ⇔ ----- F ⎛ ---- ⎞
1
a ⎝ a ⎠ (8.40)

that is, the time scaling property of the Fourier transform states that if we replace the variable t in
the time domain by at , we must replace the variable ω in the frequency domain by ω ⁄ a , and
divide F ( ω ⁄ a ) by the absolute value of a .

Proof:
We must consider both cases a > 0 and a < 0 .
For a > 0 ,
∞
– jωt
F { f ( at ) } = ∫–∞ f ( at )e dt (8.41)

We let at = τ ; then, t = τ ⁄ a , and (8.41) becomes

8-10 Signals and Systems with MATLAB Applications, Second Edition

Orchard Publications
 Properties and Theorems of the Fourier Transform

τ ω
∞ – jω ⎛ --- ⎞ ∞ – j ⎛ ----⎞ τ
⎝a⎠ τ ⎝ a⎠ 1 ω
d ⎛ --- ⎞ = --- dτ = --- F ⎛ ----⎞
1
F {f(τ)} = ∫–∞ f ( τ )e ⎝ a⎠ a ∫–∞ f ( τ )e a ⎝ a⎠
For a < 0 ,
∞
– jωt
F { f ( –at ) } = ∫–∞ f ( –at )e dt

and making the above substitutions, we find that the multiplying factor is – 1 ⁄ a . Therefore, for
1 ⁄ a we obtain (8.40).
4. Time Shifting
If F ( ω ) is the Fourier transform of f ( t ) , then,

– jωt 0
f ( t – t 0 ) ⇔ F ( ω )e (8.42)

that is, the time shifting property of the Fourier transform states that if we shift the time function
f ( t ) by a constant t 0 , the Fourier transform magnitude does not change, but the term ωt 0 is
added to its phase angle.
Proof:
∞
– jωt
F { f ( t – t0 ) } = ∫–∞ f ( t – t )e 0 dt

We let t – t 0 = τ ; then, t = τ + t 0 , dt = dτ , and thus

∞ – jω ( τ + t 0 ) – jωt 0 ∞
– jω ( τ )
F { f ( t – t0 ) } = ∫– ∞ f ( τ )e dτ = e ∫–∞ f ( τ )e dτ

or
– jωt 0
F { f ( t – t0 ) } = e F(ω)

5. Frequency Shifting
If F ( ω ) is the Fourier transform of f ( t ) , then,

jω 0 t
e f ( t ) ⇔ F ( ω – ω0 ) (8.43)

jω 0 t
that is, multiplication of the time function f ( t ) by e , where ω 0 is a constant, results in shifting
the Fourier transform by ω 0 .

Signals and Systems with MATLAB Applications, Second Edition 8-11

Orchard Publications
Chapter 8 The Fourier Transform

Proof:
∞
F { e jω t f ( t ) }0
= ∫– ∞ e
jω 0 t
f ( t )e
– jωt
dt

or
jω 0 t ∞ –j ( ω – ω0 )
F {e f(t)} = ∫–∞ f ( t )e dt = F ( ω – ω 0 )

Also, from (8.40) and (8.43)

ω–ω
f ( at ) ⇔ ----- F ⎛ ----------------0 ⎞
jω 0 t 1
e
⎝ ⎠
(8.44)
a a

Property 5, that is, (8.43) is also used to derive the Fourier transform of the modulated signals
f ( t ) cos ωt and f ( t ) sin ωt . Thus, from
jω 0 t
e f ( t ) ⇔ F ( ω – ω0 )
and
jω 0 t –j ω0 t
e +e
cos ω 0 t = -------------------------------
2
we get

F ( ω – ω0 ) + F ( ω + ω0 )
f ( t ) cos ω 0 t ⇔ ----------------------------------------------------------
- (8.45)
2

Similarly,

F ( ω – ω0 ) –F ( ω + ω0 )
f ( t ) sin ω 0 t ⇔ -------------------------------------------------------
- (8.46)
j2

6. Time Differentiation
If F ( ω ) is the Fourier transform of f ( t ) , then,
n
d
-------- f ( t ) ⇔ ( jω ) n F ( ω ) (8.47)
n
dt
n
d n
that is, the Fourier transform of -------n- f ( t ) , if it exists, is ( jω ) F ( ω ) .
dt
Proof:
Differentiating the Inverse Fourier transform, we get

8-12 Signals and Systems with MATLAB Applications, Second Edition

Orchard Publications
 Properties and Theorems of the Fourier Transform
n n ∞ ∞ n
-------- f ( t ) = -------- ⎛ ------ dω⎞ = ------
d d 1 1 d
∫– ∞ ∫
jωt jωt
n ⎝ 2π
F ( ω )e F ( ω ) -------n- e dω
n ⎠ 2π –∞
dt dt dt
∞ ∞
dω = ( jω ) ⎛ ------ dω⎞
1 n 1
∫– ∞ ∫–∞ F ( ω )e
n jωt jωt
= ------ F ( ω ) ( jω ) e
2π ⎝ 2π ⎠

and (8.47) follows.

7. Frequency Differentiation
If F ( ω ) is the Fourier transform of f ( t ) , then,
n
n d
( – j t ) f ( t ) ⇔ ---------n F ( ω ) (8.48)
dω
Proof:
Using the Fourier transform definition, we get
n n ∞ ∞ n
--------- F ( ω ) = --------- ⎛ – j ωt
dt⎞ =
d d d – j ωt
dω
n
dω
n⎝ ∫– ∞ f ( t )e
⎠ ∫ –∞
f ( t ) ---------n e
dω
dt

∞ ∞
n – j ωt – j ωt
∫– ∞ ∫–∞ f ( t )e
n
= f ( t ) ( –j t ) e dt = ( – j t ) dt

and (8.48) follows.

8. Time Integration
If F ( ω ) is the Fourier transform of f ( t ) , then,
t
F(ω )
∫–∞ f ( τ ) dτ ⇔ ------------
jω
+ πF ( 0 )δ ( ω ) (8.49)

Proof:
We postpone the proof of this property until we derive the Fourier transform of the unit step
function u 0 ( t ) on the next section. In the special case where in (8.49), F ( 0 ) = 0 , then,

t
F(ω)
∫–∞ f ( τ ) dτ ⇔ ------------
jω
(8.50)

and this is easily proved by integrating both sides of the Inverse Fourier transform.
9. Conjugate Time and Frequency Functions
If F ( ω ) is the Fourier transform of the complex function f ( t ) , then,

Signals and Systems with MATLAB Applications, Second Edition 8-13

Orchard Publications
Chapter 8 The Fourier Transform

f∗ ( t ) ⇔ F ∗ ( – ω ) (8.51)

that is, if the Fourier transform of f ( t ) = f Re ( t ) + f Im ( t ) is F ( ω ) , then, the Fourier transform of

f∗ ( t ) = f Re ( t ) – f Im ( t ) is F∗ ( – ω ) .

Proof:
∞ ∞
– jωt – jωt
F(ω) = ∫– ∞ f ( t )e dt = ∫–∞ [ fRe ( t ) + jfIm ( t ) ]e dt

∞ ∞
– jωt – jωt
= ∫– ∞ f Re ( t )e dt + j ∫–∞ fIm ( t )e dt

Then,
∞ ∞
F∗ ( ω ) = ∫– ∞ ∫–∞ fIm ( t )e
jωt jωt
f Re ( t )e dt – j dt

Replacing ω with – ω , we get

∞
– j ωt
F∗ ( – ω ) = ∫–∞ [ fRe ( t ) – jfIm ( t ) ]e dt

and (8.51) follows.

10. Time Convolution
If F 1 ( ω ) is the Fourier transform of f 1 ( t ) , and F 2 ( ω ) is the Fourier transform of f 2 ( t ) , then,

f 1 ( t )∗ f 2 ( t ) ⇔ F 1 ( ω )F 2 ( ω ) (8.52)

that is, convolution in the time domain, corresponds to multiplication in the frequency domain.
Proof:
∞ ∞
– jωt
F { f1 ( t )∗ f2 ( t ) } = ∫–∞ ∫–∞ f ( τ )f2 ( t – τ ) dτ
1 e dt
(8.53)
∞ ∞
– jωt
= ∫–∞ f ( τ ) ∫–∞ f ( t – τ )e
1 2 dt dτ

and letting t – τ = σ , then, dt = dσ , and by substitution into (8.53),

∞ ∞ ∞ ∞
– jωτ – jωσ – jωτ – jωσ
F { f 1 ( t )∗ f 2 ( t ) } = ∫– ∞ f1 ( τ ) ∫– ∞ f 2 ( σ )e e dσ dτ = ∫– ∞ f 1 ( τ )e dτ ∫–∞ f ( σ )e
2 dσ

The first integral above is F 1 ( ω ) while the second is F 2 ( ω ) , and thus (8.52) follows.

8-14 Signals and Systems with MATLAB Applications, Second Edition

Orchard Publications
 Properties and Theorems of the Fourier Transform

Alternate Proof:
– jωt 0
We can apply the time shifting property f ( t – t 0 ) ⇔ F ( ω )e into the bracketed integral of
– jωt 0
(8.53); then, replacing it with F 2 ( ω )e , we get
∞ ∞ ∞ – jωt 0
– jωt
F { f 1 ( t )∗ f 2 ( t ) } = ∫– ∞ f1 ( τ ) ∫– ∞ f 2 ( t – τ )e dt dτ = ∫–∞ f ( τ ) dτF
1 2 ( ω )e

∞
– jωt
= ∫–∞ f ( τ )e
1 dτF 2 ( ω ) = F 1 ( ω )F 2 ( ω )

11. Frequency Convolution

If F 1 ( ω ) is the Fourier transform of f 1 ( t ) , and F 2 ( ω ) is the Fourier transform of f 2 ( t ) , then,

1
f 1 ( t )f 2 ( t ) ⇔ ------ F 1 ( ω )∗ F 2 ( ω ) (8.54)
2π

that is, multiplication in the time domain, corresponds to convolution in the frequency domain
divided by the constant 1 ⁄ 2π .
Proof:
∞ ∞ ∞
– jωt 1- – jωt
∫– ∞ ∫ ∫–∞ F ( χ )e
jχt
F { f1 ( t )f2 ( t ) } = [ f 1 ( t ) f 2 ( t ) ]e dt = -----
2π 1 dχ f 2 ( t )e dt
–∞
∞ ∞ ∞
1 – j ( ω – χ )t 1
= ------
2π ∫– ∞ F1 ( χ ) ∫– ∞ f 2 ( t )e dt dχ = ------
2π ∫–∞ F ( χ )F ( ω – χ ) dχ
1 2

and (8.54) follows.

12. Area under f ( t )
If F ( ω ) is the Fourier transform of f ( t ) , then,
∞
F( 0) = ∫– ∞ f ( t ) d t (8.55)

that is, the area under a time function f ( t ) is equal to the value of its Fourier transform evaluated
at ω = 0 .
Proof:
– jωt
Using the definition of F ( ω ) and that e ω=0
= 1 , we see that (8.55) follows.

Signals and Systems with MATLAB Applications, Second Edition 8-15

Orchard Publications
Chapter 8 The Fourier Transform

13. Area under F ( ω )

If F ( ω ) is the Fourier transform of f ( t ) , then,
∞
1
f ( 0 ) = ------
2π ∫–∞ F ( ω ) dω (8.56)

that is, the value of the time function f ( t ) , evaluated at t = 0 , is equal to the area under its Fou-
rier transform F ( ω ) times 1 ⁄ 2π .
Proof:
jωt
In the Inverse Fourier transform of (8.3), we let e t=0
= 1 , and (8.56) follows.

14. Parseval’s Theorem

If F ( ω ) is the Fourier transform of f ( t ) , Parseval’s theorem states that
∞ ∞
1
∫– ∞ ∫– ∞ F ( ω )
2 2
f ( t ) dt = ------ dω (8.57)
2π

that is, if the time function f ( t ) represents the voltage across, or the current through an 1 Ω resis-
tor, the instantaneous power absorbed by this resistor is either v 2 ⁄ R , v 2 ⁄ 1 , v 2 , or i 2 R , i 2 . Then,
the integral of the magnitude squared, represents the energy (in watt-seconds or joules) dissipated
by the resistor. For this reason, the integral is called the energy of the signal. Relation (8.57) then,
states that if we do not know the energy of a time function f ( t ) , but we know the Fourier trans-
form of this function, we can compute the energy without the need to evaluate the Inverse Fou-
rier transform.
Proof:
From the frequency convolution property,
1
f 1 ( t )f 2 ( t ) ⇔ ------ F 1 ( ω )∗ F 2 ( ω )
2π
or
∞ ∞
– jωt 1
F { f1 ( t )f2 ( t ) } = ∫– ∞ [ f 1 ( t )f 2 ( t ) ]e dt = ------
2π ∫–∞ F ( χ )F ( ω – χ ) dχ
1 2 (8.58)

Since (8.58) must hold for all values of ω , it must also be true for ω = 0 , and under this condi-
tion, it reduces to
∞ ∞
1
∫ –∞
[ f 1 ( t )f 2 ( t ) ] dt = ------
2π ∫–∞ F ( χ )F ( –χ ) dχ
1 2 (8.59)

8-16 Signals and Systems with MATLAB Applications, Second Edition

Orchard Publications
 Fourier Transform Pairs of Common Functions

For the special case where f 2 ( t ) = f 1∗ ( t ) , and the conjugate functions property f∗ ( t ) ⇔ F∗ ( – ω ) ,
by substitution into (8.59), we get:
∞ ∞ ∞
1 1
∫ –∞
[ f ( t )f∗ ( t ) ] dt = ------
2π ∫ –∞
F ( ω )F∗ [ – ( – ω ) ] dω = ------
2π ∫–∞ F ( ω )F∗ ( ω ) dω
Since f ( t )f∗ ( t ) = f ( t ) and F ( ω )F∗ ( ω ) = F ( ω ) , Parseval’s theorem is proved.
2 2

The Fourier transform properties and theorems are summarized in Table 8.8.

8.4 Fourier Transform Pairs of Common Functions

In this section, we will derive the Fourier transforms of common time functions.
1.
δ(t) ⇔ 1 (8.60)
Proof:
The sifting theorem of the delta function states that
∞

∫–∞ f ( t )δ ( t – t ) dt
0 = f ( t0 )

and if f ( t ) is defined at t = 0 , then,

∞

∫–∞ f ( t )δ ( t ) dt = f( 0)

By the definition of the Fourier transform

∞
– jωt – jωt
F(ω) = ∫–∞ δ ( t )e dt = e t=0
= 1

and (8.60) follows.

We will use the notation f ( t ) ↔ F ( ω ) to show the time domain to frequency domain correspon-
dence. Thus, (8.60) may also be denoted as in Figure 8.1.
f(t) F(ω) 1
δ(t)

0 t 0 ω

Figure 8.1. The Fourier transform of the delta function

Signals and Systems with MATLAB Applications, Second Edition 8-17

Orchard Publications
Chapter 8 The Fourier Transform

TABLE 8.8 Fourier Transform Properties and Theorems

Property f( t) F(ω)
Linearity a1 f1 ( t ) + a2 f2 ( t ) + … a1 F1 ( ω ) + a2 F2 ( ω ) + …
Symmetry F(t) 2πf ( – ω )
Time Scaling f ( at ) ω
----- F ⎛ ----⎞
1
a ⎝ a⎠
Time Shifting f ( t – t0 ) – jωt 0
F ( ω )e
Frequency Shifting jω 0 t F ( ω – ω0 )
e f(t)
Time Differentiation n n
d ( jω ) F ( ω )
--------- f ( t )
n
dt
Frequency Differentiation n n
( – jt ) f ( t ) d
----------- F ( ω )
n
dω
Time Integration t F ( ω ) + πF ( 0 )δ ( ω )
∫– ∞
------------
f ( τ ) dτ jω

Conjugate Functions f∗ ( t ) F∗ ( – ω )
Time Convolution f 1 ( t )∗ f 2 ( t ) F1 ( ω ) ⋅ F2 ( ω )

Frequency Convolution f1 ( t ) ⋅ f2 ( t ) 1
------ F 1 ( ω )∗ F 2 ( ω )
2π
Area under f(t) ∞
F(0) = ∫–∞ f ( t ) dt
Area under F(w) ∞
∫
1
f ( 0 ) = ------ F ( ω ) dω
2π – ∞

Parseval’s Theorem ∞ ∞
∫ ∫
2 1 2
f ( t ) dt = ------ F ( ω ) dω
–∞ 2π –∞

Likewise, the Fourier transform for the shifted delta function δ ( t – t 0 ) is

– jωt 0
δ ( t – t0 ) ⇔ e (8.61)

8-18 Signals and Systems with MATLAB Applications, Second Edition

Orchard Publications
 Fourier Transform Pairs of Common Functions

2.
1 ⇔ 2πδ ( ω ) (8.62)
Proof:
–1 ∞ ∞
1
∫– ∞ ∫–∞ δ ( ω )e
jωt jωt jωt
F { 2πδ ( ω ) } = ------ 2πδ ( ω )e dω = dω = e ω=0
= 1
2π

and (8.62) follows.

The f ( t ) ↔ F ( ω ) correspondence is also shown in Figure 8.2.

f(t) F(ω)
1
2πδ ( ω )

0 t 0 ω

Figure 8.2. The Fourier transform of unity

Also, by direct application of the Inverse Fourier transform, or the frequency shifting property
and (8.62), we derive the transform
jω 0 t
e ⇔ 2πδ ( ω – ω 0 ) (8.63)

The transform pairs of (8.62) and (8.63) can also be derived from (8.60) and (8.61) by using the
symmetry property F ( t ) ⇔ 2πf ( – ω )
3.

1 jω 0 t –j ω0 t
cos ω 0 t = --- ( e +e ) ⇔ πδ ( ω – ω 0 ) + πδ ( ω + ω 0 ) (8.64)
2

Proof:
This transform pair follows directly from (8.63). The f ( t ) ↔ F ( ω ) correspondence is also shown
in Figure 8.3.
cosωω
cos 0 t0
t F Re ( ω )

π π
t t

−ω0 0 ω0 ω

Figure 8.3. The Fourier transform of f ( t ) = cos ω 0 t

Signals and Systems with MATLAB Applications, Second Edition 8-19

Orchard Publications
Chapter 8 The Fourier Transform

We know that cos ω 0 t is real and even function of time, and we found out that its Fourier trans-
form is a real and even function of frequency. This is consistent with the result in Table 8.7.
4.

1 jω0 t –j ω0 t
sin ω 0 t = ----- ( e –e ) ⇔ jπδ ( ω – ω 0 ) – jπ δ ( ω + ω 0 ) (8.65)
j2

Proof:
This transform pair also follows directly from (8.63). The f ( t ) ↔ F ( ω ) correspondence is also
shown in Figure 8.4.

sin ω 0 t F Im ( ω )
π
tt −ω0
0 ω0 ω
−π

Figure 8.4. The Fourier transform of f ( t ) = sin ω 0 t

We know that sin ω 0 t is real and odd function of time, and we found out that its Fourier trans-
form is an imaginary and odd function of frequency. This is consistent with the result in Table
8.7.
5.
2
sgn ( t ) = u 0 ( t ) – u 0 ( – t ) ⇔ ------ (8.66)
jω

where sgn ( t ) denotes the signum function shown in Figure 8.5.

f(t)
1

−1

Figure 8.5. The signum function

Proof:
To derive the Fourier transform of the sgn ( t ) function, it is convenient to express it as an expo-
nential that approaches a limit as shown in Figure 8.6.

8-20 Signals and Systems with MATLAB Applications, Second Edition

Orchard Publications
 Fourier Transform Pairs of Common Functions

f(t)
1
– at
e u0 ( t )
0

– at −1
–e u0 ( –t )

Figure 8.6. The signum function as an exponential approaching a limit

Then,
– at at
sgn ( t ) = lim [ e u0 ( t ) – e u0 ( –t ) ] (8.67)
a→0
and
0 ∞
at – j ωt – a t – j ωt
F { sgn ( t ) } = lim
a→0 ∫– ∞ –e e dt + ∫0 e e dt

0 ∞
( a – j ω )t – ( a + jω ) t
= lim
a→0 ∫– ∞ –e dt + ∫0 e dt (8.68)

1 1 –1 1 2
= lim --------------- + --------------- = --------- + ------ = ------
a→0 a – jω a + jω – jω jω jω

The f ( t ) ↔ F ( ω ) correspondence is also shown in Figure 8.7.

f( t)
F Im ( ω )
1
0
t ω
0
−1

Figure 8.7. The Fourier transform of sgn ( t )

We now know that sgn ( t ) is real and odd function of time, and we found out that its Fourier
transform is an imaginary and odd function of frequency. This is consistent with the result in
Table 8.7.
6.
1-
u 0 ( t ) ⇔ πδ ( ω ) + ----- (8.69)
jω

Signals and Systems with MATLAB Applications, Second Edition 8-21

Orchard Publications