Advanced Mathematics Fourier Analysis: DR Lino AA Notarantonio

Advanced Mathematics
Fourier Analysis
Dr Lino AA Notarantonio
September 13, 2011
Fourier Series
Denition
A Fourier series is an innite linear combination of terms such as
sin(nx), cos(nx):
a
0
2
+
n=1
[a
n
cos(nx) + b
n
sin(nx)]
is the fundamental frequency and is measured in radians per second
rad/s.
a
n
, b
n
are the coecients of the series
Fourier Series
Fundamental Frequency
The fundamental frequency can also be measured in Hertz [Hz] = sec
1
.
In this case, the sine and cosine functions will have an extra factor 2F:
cos(2Fnx), sin(2Fnx), F being the frequency in Hz.
Fourier Series
Periodic Functions
Denition
We say that a function f (x) is periodic, with period P if
f (x + P) = f (x)
for every x in its domain.
Remark
The period of a periodic function is the smallest real number P satisfying
f (x + P) = f (x)
for every x in the domain of f .
Fourier Series
Periodic Extension
Let examine the concept of periodic extension with the following example.
Example
Let f (x) = 1/x, 0 < x 1 and let n > 0 be an integer.Dene
f (x) =
_
f (x), if 0 < x 1
f (x n), if n < x n + 1
The function

f (x) is a periodic function of period 1.
Fourier Series
Periodic Functions
Let f (x) be a periodic function, with period P , 0 x P. Then we can
write the associated Fourier series
a
0
2
+
n=1
[a
n
cos(nx) + b
n
sin(nx)] (1)
The fundamental frequency is =
2
P
.
How do we relate the coecients of the series with the function f (x)?
Fourier Series
Fourier Coecients
The coecients of (1) are computed as follows:
a
0
=
2
P
_
P
0
f (x)dx
a
n
=
2
P
_
P
0
f (x) cos(nx)dx (2)
b
n
=
2
P
_
P
0
f (x) sin(nx)dx
In fact, the interval of integration above can be any interval of length P,
such as, for example, (P/2, P/2).
Fourier Series
Convergence of the Fourier Series
Fact (Dirichlet Condition)
Let f : (0, P) R be a periodic function, with period P, x R.
Suppose also that f (x)
1. has a nite number of discontinuities;
2. has a nite number of maxima and minima;
3. is absolutely integrable over a period:
_
P
0
|f (x)| dx <
Then the associated Fourier series
a
0
2
+
n=1
[a
n
cos(nx) + b
n
sin(nx)]
converges absolutely to
1
2
[f (x
+
) + f (x
)] for all x (0, P).

Remark
If the function f is continuous at x (0, P), then the series converges
absolutely to f (x).
Fourier Series
Unit Pulse Function
The unit pulse (boxcar function; gate function; rectangular function) is
dened as
p(x) =
_
1, |x| < 1/2
0, |x| 1
The function is then extended by periodicity, with period 2, over the real
line.
Remark
A pulse with amplitude A, pulse width B and center-point time delay T
is Ap((x T)/B).
Fourier Series
Unit Pulse Function
The computation of the Fourier coecients of the pulse function f (x)
gives
a
0
=
2A
P
T
a
n
=
2A
P
sin(nT)
n
(3)
b
n
=
2A
P
1 cos(nT)
n
(4)
Exercise What function does the associated Fourier series converge to?
Fourier Series
Problems
Determine the Fourier series of the following functions and determine all
points where the series does not converge to f (x). All the functions
below are extended by periodicity over the real line.
1. f (x) = x, x .
2. f (x) = x, 0 x .
3. f (x) = x
2
, x .
Fourier Series
Energy of a signal
The energy of a function over a period is dened as
E(f ) =
_
P
0
|f (x)|
2
dx (5)
In Signal Processing Analysis is more appropriate to consider the power
of a periodic signal, where the power is dened as
P(f ) =
1
P
_
P
0
|f (x)|
2
dx (6)
Remark
In Signal Processing Analysis is appropriate to consider signals with nite
power, if the signal is periodic and signal with nite energy, if the signal
is not necessarily periodic.
Fourier Series
Signals with nite energy
If a periodic signal has nite energy, then it is possible to prove that the
energy of the dierence
f (x)
_
a
0
2
+
n=1
[a
n
cos(nx) + b
n
sin(nx)]
_
is equal to zero.
Remark
Even so, bear in mind that the function and its associated series may not
be equal for all values of x.
Fourier Series
Parsevals Identity
1
Let f (x) be a periodic function that satises Dirichlets Conditions.
The power of f (x) can be computed using the associated Fourier
coecients:
1
P
_
P
0
|f (x)|
2
dx =
_
a
0
2
_
2
+
1
2
n=1
_
|a
n
|
2
+|b
n
|
2
(7)
The term |a
n
|
2
+|b
n
|
2
is the n-th harmonic component of the signal
(function).
1
Also known as Plancherels Identity
Fourier Series
Parsevals Identity: An Application
Consider the pulse function with A = 1, T = , y P = 2. It is asked to
compute what percentage of the signal contains:
1. the DC term;
2. The DC term and the rst ve multiples of the fundamental
frequency.
Fourier Series
The power of the signal is equal to
Power =
1
2
_
2
0
|f (x)|
2
dx =
1
2
.
1. The DC term corresponds to term with n = 0 of the associated
Fourier series: a
0
= 1, so the DC term contains 50% of the signal:
(a
0
/2)
2
= 1/4.
2. The DC term and the rst ve multiples of the fundamental
frequency give
_
1
2
_
2
+
1
2
_
_
2
_
2
+
_
2
3
_
2
+
_
2
5
_
2
_
= 0.4833,
which corresponds to 98.33% of the total power of the signal. The
corresponding calculations are given in the next slide.
Fourier Series
The fundamental frequency is = 1. The Fourier coecients are:
a
k
=
1
_

0
cos(kt) dt =
1
k
sin(kt)
0
= 0
b
k
=
1
_

0
sin(kt)dt dt =
1
k
cos(kt)
0
=
1
k
[1 (1)
n
] =
_
_
0, k even
2
k
, k odd
Fourier Series
Parsevals Identity: Problems
For each of the following signals, compute the amount (in %) of the
signal in:
(a) the DC term;
(b) the DC term and the rst three multiples of the fundamental
frequency.
1. f (x) = x, x .
2. f (x) = x, 0 x .
3. f (x) = x
2
, x .
Fourier Series
Phase-Shifted Sinusoid Representation
The Fourier series of a continuous, periodic signal f (x) can also be
written as a single set of phase-shifted sinusoids
f (x) =
a
0
2
+
n=1
A
n
cos(nx +
n
)
where A
n
=
_
a
2
n
+ b
2
n
is the amplitude and
n
= tan
1
(b
n
/a
n
) is the
phase shift.
Fourier Series Spectrum of a Signal
Amplitude Spectrum
Denition
Let f (x) be a continuous, periodic signal. The amplitude spectrum of the
signal f (x) is the set
{(n, A
n
/2)}
Example (Pulse Function)
A
n
2
=
2
A
P
_
1 cos(nT)
n
So the amplitude spectrum of this signal is given by the graph of the
function

2
A
P
_
1 cos(T)
Complex Fourier Series

Let f be an absolutely integrable, P-periodic function. The complex
Fourier series is
f (x) =
n=
d
n
e
jnx
where = 2/P is the fundamental frequency.
The complex Fourier coecients are computed as
d
n
=
1
P
_
P
0
f (x)e
jnx
dx, n = 0, 1, . . .
Remark
If a
n
, b
n
are the coecients of the real Fourier series, then
f (x) =
a
0
2
+
n=1
__
a
n
jb
n
2
_
e
jnx
+
_
a
n
+ jb
n
2
_
e
jnx
_
= d
0
+
n=1
_
d
n
e
jnx
+ d
n
e
jnx
n=
d
n
e
jnx
with d
0
=
1
2
a
0
, d
n
=
1
2
(a
n
jb
n
) and d
n
= d
n
.
Power Spectrum
The power spectrum is the set (n, |d
n
|
2
).
Notice that |d
n
|
2
= |A
n
|
2
, where A
n
are the Fourier coecients using
sinusoids.
Power Spectrum-Pulse Function
Let us compute the power spectrum of the P-periodic pulse function
f (x) =
_
A, 0 < x < T
0, t < x < P
d
0
= AT/P,
d
n
=
1
P
_
P
0
f (x)e
jnx
dx =
A
P
_
1 e
jnT
jn
_
Express = 2F, F frequency in Hz.
Power Spectrum-Pulse Function
d
n
=
A
P
_
e
jnFT
e
jnFT
e
2jnFT
2jnF
_
=
A
P
_
e
jnFTx
e
jnFT
(2j)nF
_
e
jnFT
=
A
P
_
sin(nFT)
nF
_
e
jnFT
=
AT
P
_
sin(nFT)
nFT
_
e
jnFT
=
AT
P
sinc(nFT) e
jnFTx
Denition (Sinc Function)
The sinc function is dened as
sinc(t) =
sin t
t
Some authors also refer to it as sampling function.
The sinc function is equal to 1, when t = 0, has zeroes at 1,
2, . . .and lim
t
sinc(t) = 0

Advanced Mathematics Fourier Analysis: DR Lino AA Notarantonio

Uploaded by

Copyright:

Available Formats

Advanced Mathematics Fourier Analysis: DR Lino AA Notarantonio

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Advanced Mathematics Fourier Analysis: DR Lino AA Notarantonio

Uploaded by

Copyright:

Available Formats

Advanced Mathematics

)] for all x (0, P).

Complex Fourier Series

You might also like