the fourier series edited
TRANSCRIPT
-
7/28/2019 The Fourier Series Edited
1/46
10/24/12
1
The Fourier Series
A Fourier series is an expansion of a
periodicfunctionf(t) in terms of an infinite sum
ofcosines and sines
Introduction
=
++=
1
0 )sincos(2
)(n
nn tnbtnaa
tf
In other words, any periodicfunction can be
resolved as a summation of
constant value and cosine and sine functions:
=
++=
1
0 )sincos(2
)(n
nn tnbtnaa
tf
)sincos(11
tbta ++
2
0a
=
)2sin2cos( 22 tbta ++
)3sin3cos( 33 tbta ++ +
The computation and study of Fourier series isknown as harmonic analysis and is extremely
useful as a way to break up an arbitrary
periodic function into a set of simple terms thatcan be plugged in, solved individually, and
then recombinedto obtain the solution to the
original problem or an approximation to it towhatever accuracy is desired or practical.
=
+ +
+ + +
Periodic Function2
0a
ta cos1
ta 2cos2
tb sin1
tb 2sin2
f(t)
t
=
++=
1
0 )sincos(2
)(n
nn tnbtnaa
tf
where
=T
dttfT
a0
0 )(2
frequencylFundementa2
==
T
=T
n tdtntfT
a0
cos)(2
=T
n tdtntfT
b0
sin)(2
*we can also use the integrals limit .
=T
dttfT
a0
0 )(2
2/
2/
T
T
-
7/28/2019 The Fourier Series Edited
2/46
10/24/12
2
Example 1
Determine the Fourier series representation of thefollowing waveform.
Solution
First, determine the period & describe the one periodof the function:
T= 2
-
7/28/2019 The Fourier Series Edited
3/46
10/24/12
3
Some helpful identities
Forn integers,n
n )1(cos =0sin =n
02sin =n 12cos =n
xx sin)sin( = xx cos)cos( =
[Supplementary]
The sum of the Fourier series terms canevolve (progress) into the original waveform
From Example 1, we obtain++++= ttttf
5sin5
23sin
3
2sin
2
2
1)(
It can be demonstrated that the sum will leadto the square wave:
tttt
7sin7
25sin
5
23sin
3
2sin
2+++ttt
5sin5
23sin
3
2sin
2++
tt
3sin3
2sin
2+t
sin2
(a) (b)
(c) (d)
ttttt
9sin9
27sin
7
25sin
5
23sin
3
2sin
2++++
ttt
23sin23
23sin
3
2sin
2
2
1++++
(e)
(f)
Example 2
Given,)( ttf =
11 t
)()2( tftf =+
Sketch the graph off(t) such that .33 t
Then compute the Fourier series expansion off(t).
-
7/28/2019 The Fourier Series Edited
4/46
10/24/12
4
Solution
The function is described by the following graph:
T= 2
==
T
2We find that
Then we compute the coefficients:
02
11
22
2
)(2
1
1
21
1
1
1
0
=
=
==
=
ttdt
dttfT
a
0coscos
)cos(cos0
cos)]sin([sin
sinsin
coscos)(2
22
22
1
1
22
1
1
1
1
1
1
1
1
=
=
+=
+
=
=
==
n
nn
nnn
n
tn
n
nn
dtn
tn
n
tnt
tdtnttdtntfT
an
since xx cos)cos( =
nnn
n
n
nn
n
n
n
tn
n
nn
dtn
tn
n
tnt
tdtnttdtntfT
b
nn
n
1
22
1
1
22
1
1
1
1
1
1
1
1
)1(2)1(2cos2
)sin(sincos2
sin)]cos([cos
coscos
sinsin)(2
+
=
==
+=
+
+=
+
=
==
+=
=
++=
=
+
=
ttt
tnn
tnbtnaa
tf
n
n
n
nn
3sin3
22sin
2
2sin
2
sin)1(2
)sincos(2
)(
1
1
1
0
Finally,
Example 3
Given
-
7/28/2019 The Fourier Series Edited
5/46
10/24/12
5
Solution
The function is described by the following graph:
T= 4
2
2 ==
TWe find that
0 2 4 6 8 10 12t
v (t)
2
Then we compute the coefficients:
12
22
1)2(
2
1
0)2(4
2
)(2
2
0
22
0
4
2
2
0
4
0
0
=
==
+=
=
ttdtt
dtdtt
dttvT
a
222222
2
0
22
2
0
2
0
4
2
2
0
4
0
])1(1[2)cos1(2
2
2cos1
cos
2
10
sin
2
1sin)2(
2
1
0cos)2(2
1cos)(
2
nn
n
n
n
n
tn
dtn
tn
n
tnt
tdtnttdtntvT
a
n
n
=
=
=
+=
+
=
+==
nnn
n
n
n
tn
n
dtn
tn
n
tnt
tdtnttdtntvT
bn
21
2
2sin1
sin
2
11
cos
2
1cos)2(
2
1
0sin)2(2
1sin)(
2
22
2
0
22
2
0
2
0
4
2
2
0
4
0
===
=
=
+==
since 0sin2sin == nn
=
=
+
+=
++=
122
1
0
2sin
2
2cos
])1(1[2
2
1
)sincos(2
)(
n
n
n
nn
tn
n
tn
n
tnbtnaa
tv
Finally,
Symmetry Considerations
Symmetry functions:(i) even symmetry
(ii) odd symmetry
-
7/28/2019 The Fourier Series Edited
6/46
10/24/12
6
Even symmetry
Any functionf(t) is even if its plot issymmetrical about the vertical axis, i.e.
)()( tftf =
Even symmetry (cont.)
The examples ofeven functions are:2)( ttf =
t t
t
||)( ttf =
ttf cos)( =
Even symmetry (cont.)
The integral of an even function from A to +Ais twice the integral from 0 to +A
t ++
=
AA
A
dttfdttf0
eveneven )(2)(
A +A
)(even tf
Odd symmetry
Any functionf(t) is odd if its plot isantisymmetrical about the vertical axis, i.e.
)()( tftf =
Odd symmetry (cont.) The examples ofodd functions are:
3)( ttf =
tt
t
ttf =)(
ttf sin)( =
Odd symmetry (cont.) The integral of an odd function from A to +A
is zero
t 0)(odd =+
A
A
dttfA +A
)(odd tf
-
7/28/2019 The Fourier Series Edited
7/46
10/24/12
7
Even and odd functions
(even)(even) = (even) (odd)(odd) = (even) (even)(odd) = (odd) (odd)(even) = (odd)
The product properties ofeven and odd
functions are:
Symmetry consideration
From the properties ofeven and odd
functions, we can show that:
foreven periodic function;=
2/
0
cos)(4
T
n tdtntfT
a 0=nb
forodd periodic function;=
2/
0
sin)(4
T
n tdtntfT
b 00
==naa
How?? [Even function]
2
T
2
T
==
2/
0
2/
2/
cos)(4
cos)(2
TT
T
n tdtntfT
tdtntfT
a
(even) (even)
| |
(even)
0sin)(2
2/
2/
==
T
T
n tdtntfT
b
(even) (odd)
| |
(odd)
)(tf
t
How?? [Odd function]
2
T
2
T
==
2/
0
2/
2/
sin)(4
sin)(2
TT
T
n tdtntfT
tdtntfT
b
(odd)(odd)
| |
(even)
0cos)(2
2/
2/
==
T
T
n tdtntfT
a
(odd) (even)
| |
(odd)
)(tf
t
0)(2
2/
2/
0 ==
T
T
dttfT
a
(odd)
Example 4
Given
-
7/28/2019 The Fourier Series Edited
8/46
10/24/12
8
Then we compute the coefficients. Sincef(t) is
an odd function, then
0)(2
2
2
0 ==
dttfT
a
0cos)(2
2
2
==
tdtntfT
an
and
n
n
n
n
n
n
n
nn
n
tn
n
n
n
tndt
n
tn
n
tnt
tdtntdtnt
tdtntfT
tdtntfT
bn
cos2sin2cos
cos2cossincos
coscoscos
sin1sin4
4
sin)(4
sin)(2
22
1
0
22
2
1
1
0
1
0
2
1
1
0
2
0
2
2
=+=
+=
++
=
+=
==
since 0sin2sin == nn
=
+
=
=
=
=
++=
1
1
1
1
0
2sin
)1(2
2sin
cos2
)sincos(2
)(
n
n
n
n
nn
tn
n
tn
n
n
tnbtnaa
tf
Finally,
Example 5
Compute the Fourier series expansion off(t).
SolutionThe function is described by
T= 3
3
22 ==
T
and
-
7/28/2019 The Fourier Series Edited
9/46
10/24/12
9
3
2sin
2
3
2sinsin2
2
sin2
3sin2
3
4
sin2
3sin2sin
3
4
sin2
3
4sin
3
4
cos2cos13
4
cos)(4
cos)(2
2/3
1
1
0
2/3
1
1
0
2/3
0
3
0
n
n
nn
n
nn
n
nn
nn
n
tn
n
tn
tdtntdtn
tdtntfT
tdtntfT
an
=
=
=
+=
+
=
+=
==
;3
2 =
=
=
=
=
+=
++=
1
1
1
0
3
2cos
3
2sin
12
3
4
3
2cos
3
2sin
2
3
4
)sincos(2
)(
n
n
n
nn
tnn
n
tnn
n
tnbtnaa
tf
Finally,
and 0=nb sincef(t) is an even function.
Function defines over a finite interval
Fourier series only support periodic functions In real application, many functions are non-periodic The non-periodic functions are often can be defined
over finite intervals, e.g.
y=
ty = 1 y = 1
y = 2
y=
t2
Therefore, any non-periodic function must beextended to a periodic function first, before
computing its Fourier series representation
Normally, we prefer symmetry (even or odd) periodicextension instead of normal periodic extension, sincesymmetry function will provide zero coefficient of
eitheran orbn
This can provide a simpler Fourier series expansion
)(ty
t
)(tf
t
)(even tf
)(odd tf
t
t
lttytf
-
7/28/2019 The Fourier Series Edited
10/46
10/24/12
10
If the function is extended as an even function,then the coefficient bn= 0, hence
=
+=
1
0 cos2
)(n
n tnaa
tf
which only contains the cosine harmonics.
Therefore, this approach is called as the half-range Fourier cosine series
If the function is extended as an odd function,then the coefficient an= 0, hence
=
=
1
sin)(n
n tnbtf
which only contains the sine harmonics.
Therefore, this approach is called as the half-range Fourier sine series
Example 6
-
7/28/2019 The Fourier Series Edited
11/46
10/24/12
11
Solution
Since we want to seek the half-range cosine series,the function to is extended to be an even function:
T= 2
==
T
2
t
f(t)
t
f(t)
1
22
33 1
1
11
1
1
2
1
Hence, the coefficients are
[ ] 02)12(24
)(
4 10
2
1
0
2/
00
=
=
==
ttdttdttfTa
T
1
0
22
1
0
1
0
1
0
2/
0
cos4
sin2
2sin
2sin)12(
2
cos)12(2
4cos)(
4
+=
=
==
n
tn
n
n
dtn
tn
n
tnt
tdtnttdtntfT
a
T
n
=
=
even,0
odd,/8)1(cos422
22n
nn
n
n
0=nb
Therefore,
=
=
=
=
+=
+=
odd1
22
odd1
22
1
0
cos18
cos8
0
cos)(
nnnn
n
n
tn
n
tn
n
tnaatf
Parsevals Theorem
Parservals theorem states that the averagepower in a periodic signal is equal to the sum
of the average power in its DC component and
the average powers in its harmonics
=
+ +
+ + +
2
0a
ta cos1
ta 2cos2
tb sin1
tb 2sin2
f(t)
t
PavgP
dc
Pa1 Pb1
Pa2 Pb2
For sinusoidal (cosine or sine) signal,
R
V
R
V
R
VP
2
peak
2
peak2
rms
2
12=
==
For simplicity, we often assumeR = 1,which yields
2
peak2
1VP=
-
7/28/2019 The Fourier Series Edited
12/46
10/24/12
12
For sinusoidal (cosine or sine) signal,
+++++
=
+++++=
2
2
2
2
2
1
2
1
2
0
dcavg
2
1
2
1
2
1
2
1
2
2211
babaa
PPPPPPbaba
=
++=1
222
0avg )(2
1
4
1
n
nnbaaP
Exponential Fourier series
Recall that, from the Eulers identity,xjxe
jxsincos =
yields
2cos
jxjxee
x
+
=
2sin
j
eex
jxjx
=and
Then the Fourier series representation becomes
=
=
=
=
=
=
++
+=
++
+=
++=
+
++=
++=
11
0
1
0
1
0
1
0
1
0
222
222
222
222
)sincos(2
)(
n
tjnnn
n
tjnnn
n
tjnnntjnnn
n
tjntjn
n
tjntjn
n
n
tjntjn
n
tjntjn
n
n
nn
ejba
ejbaa
ejba
ejbaa
eejb
eea
a
j
eeb
eea
a
tnbtnaa
tf
Here, let we name
=
=
++
+=
11
0
222)(
n
tjnnn
n
tjnnn ejba
ejbaa
tf
2
nnn
jbac
= ,2
nnn
jbac
+
=
Hence,
=
=
=
=
=
=
=
=++=
++=
++=
n
tjn
n
n
tjn
n
n
tjn
n
n
tjn
n
n
tjn
n
n
tjn
nn
tjn
n
ececcec
ececc
ececc
1
0
1
11
0
110
and .2
0
0
a
c =
c0 cncn
Then, the coefficient cn can be derived from
=
=
=
=
=
T
tjn
T
TT
TT
nnn
dtetfT
dttnjtntfT
tdtntfjtdtntfT
tdtntfT
jtdtntf
T
jbac
0
0
00
00
)(1
]sin)[cos(1
sin)(cos)(1
sin)(2
2cos)(
2
2
1
2
In fact, in many cases, the complex Fourierseries is easier to obtain rather than the
trigonometrical Fourier series
In summary, the relationship between thecomplex and trigonometrical Fourier series
are:
2
nnn
jbac
=
2
nnn
jbac
+
=
==T
dttfT
ac
0
00 )(
1
2
=
T
tjn
n dtetfT
c0
)(1
nncc =
or
-
7/28/2019 The Fourier Series Edited
13/46
10/24/12
13
Example 8
Obtain the complex Fourier series of the followingfunction
2 44 2 0
f(t)
=e
t
2e
1
)(tf
t
Since , . Hence
Solution
[ ]
2
1
2
1
2
1
)(1
22
0
2
0
0
0
==
=
=
ee
dte
dttfT
c
t
t
T
1=2=T
)1(2
1
)1(2
1
)1(2
112
1
2
1
2
1
)(1
222)1(2
2
0
)1(
2
0
)1(
2
0
0
jn
e
jn
ee
jn
ejn
e
dtedtee
dtetfT
c
njjn
tjn
tjnjntt
T
tjn
n
=
=
=
=
==
=
since 1012sin2cos2
===
njnenj
jnt
nn
tjn
n ejn
eectf
=
=
==
)1(2
1)(
2
Therefore, the complex Fourier series off(t) is
0
2
0
2
0 2
1
)1(2
1c
e
jn
ec
n
nn=
=
=
=
=
*Notes: Even though c0 can be found by substituting
cn with n = 0, sometimes it doesnt works (as shown
in the next example). Therefore, it is always better tocalculate c0 alone.
2
2
12
1
n
e
cn
+
=
cn is a complex term, and it depends on n.Therefore, we may plot a graph of|cn| vs n.
In other words, we have transformed the function
f(t)in the time domain (t), to the function cn in the
frequency domain (n).
Example 9
Obtain the complex Fourier series of the function inExample 1.
-
7/28/2019 The Fourier Series Edited
14/46
10/24/12
14
Solution
2
11
2
1)(
11
00
0 === dtdttfTc
T
=
)1(22
1
012
1)(
1
1
0
2
1
1
00
=
=
+==
jntjn
tjn
T
tjn
n
en
j
jn
e
dtedtetfT
c
)1(2
=
jn
ne
n
jc
Butnjn
nnjne )1(cossincos ===
Thus,
==even,0
odd,/]1)1[(
2 n
nnj
n
j n
Therefore,
=
=
==
odd0
2
1)(
nn
n
tjn
n
tjn
n en
jectf
*Here notice that . 00 cc nn =
=
even,0
odd,1
n
n
ncn
The plot of|cn| vs n is shown below
2
10=c
0.5
The Fourier Transform
BET2533 Eng. Math. III
R. M. Taufika R. Ismail
FKEE, UMP
Introduction Fourier transform is another method to transform a
signal from time domain to frequency domain
The basic idea of Fourier transform comes from thecomplex Fourier series
Practically, many signals are non-periodic Fourier transform use the principal of the Fourier
series, with assumption that the period of the non-
periodic signal is infinity (T).
Recall the complex form of Fourier series:
=
=
n
tjn
nectf0)(
=
2/
2/
0)(1
T
T
tjn
n dtetfT
c
where
(2.1)
(2.2)
(2.3)
Substituting (2.2) into (2.1) gives
=
=
n
tjn
T
T
tjnedtetf
Ttf 00
2/
2/
)(1
)(
-
7/28/2019 The Fourier Series Edited
15/46
10/24/12
15
The spacing between adjacent harmonics is
Tnn
2)1( 000 ==+=
or
2
=T
Then (2.3) becomes:
=
=
=
=
=
=
n
tjn
T
T
tjn
n
tjnT
T
tjn
n
tjnT
T
tjn
edtetf
edtetf
edtetfT
tf
00
00
00
2/
2/
2/
2/
2/
2/
)(2
1
)(2
)(1
)(
Now if we let T, the summation becomesintegration, becomes , and the discreteharmonic frequency n0becomes a
continuous frequency , i.e.:
=
0n
d
n
Thus, as T,
=
=
dedtetf
edtetftf
tjtj
n
tjn
T
T
tjn
)(2
1
)(2
1)( 00
2/
2/
)(F
Therefore
=
deFtf tj)(2
1)(
where
=
detfF tj)()(
We say thatF() is the Fourier transform off(t), andf(t) is the inverse Fourier transform of
F()
Or in mathematical expression,
==
deFFtf tj)(
2
1)}({)( 1F
and
==
detftfF tj)()}({)( F
where Fand F1 is the Fourier transform and
inverse Fourier transform operator
respectively:F
1F
)(F)(tf
-
7/28/2019 The Fourier Series Edited
16/46
10/24/12
16
Generally, the Fourier transformF() existswhen the Fourier integral converges
A condition for a functionf(t) to have a Fouriertransform is,f(t) can be completely integrable,i.e.
This condition is sufficient but not necessary 0.
je
??sincos ==
jej
+ )( jae
00)(
===+ jjja
eeee
-
7/28/2019 The Fourier Series Edited
17/46
10/24/12
17
Example 1
Using the definition, find the Fourier transformof(t). Then deduce the Fourier transform of 1.
Solution
1)()}({)( 00 =====
eedtettFjtj
F
The Fourier transform off(t) = (t) is
*Recall the sifting property: )()()( afdttfat =
Then, using the duality principle, the Fouriertransform of1 is
)(2)(2)(2}1{ === fF
since . )()( =
Example 2
Find the Fourier transform of the followingfunction:
(a) (b) (c) tcos)( at tje
Solution
ajtj edteatat
== )()}({F
(a) Using the definition,
(b) From the result in (a), by replacing = a,
and using the duality principle,
)(2))((2)(2}{ === fe tjF
(c) From the result in (b),
[ ]
[ ]
[ ])()(
)(2)(22
1
}{}{2
1
2}{cos
++=
++=
+=
+
=
tjtj
tjtj
ee
eet
FF
FF
Example 3
Using the definition, find the Fourier transform
of the following:
(a) (b)
where Rand > 0.
)(tue t ||te
-
7/28/2019 The Fourier Series Edited
18/46
10/24/12
18
Solution
jjj
e
dtedteetue
tj
tjtjtt
+
=
+
=
+=
==
+
+
1
)(
10
)(
)}({
0
)(
0
)(
0
F
(a)
(b)
22
0
)(0
)(
0
)(
0
)(
0
0
||
211
)(
}{
+
=
+
+
=
++
=
+=
+=
+
+
jj
j
e
j
e
dtedte
dteedteee
tjtj
tjtj
tjttjttF
Properties of the Fourier transform
Linearity Time scaling Time shifting Frequency shifting Time differentiation Frequency differentiation Duality Convolution
)()()}()({ bGaFtbgtaf =F
1. Linearity
IfF() = F{f(t)}, G() = F{g(t)}, a and b are
constants, then:
=
aF
aatf
||
1)}({F
2. Time scaling
)()}({ 00
Fettftj
=F
that is, a delay in the time domain corresponds to
a phase shift in the frequency domain.
)(||)|(|)1()( == FFFe j
= ||)( FF*Phase shift: Say . Then
.
3. Time shifting
)(})({ 00
= Fetftj
F
4. Frequency shifting
)()()(
Fjdt
tfd nn
n
=
F
5. Time differentiation
n
nnn
d
Fdjtft
)()}({ =F
6. Frequency differentiation
)(2)}({ = ftFF
7. Duality
-
7/28/2019 The Fourier Series Edited
19/46
10/24/12
19
)()()}()({ GFtgtf =F
8. Convolution
)()(2
1)}()({
GFtgtf =F
Example 4
Derive the Fourier transform of a singlerectangular pulse of width and heightA, as
shown below:
Solution
2sinc
2/
)2/sin()2/sin(2
)(
)()}({)(
2/2/
2/
2/
2/
2/
AAA
j
eeA
j
eAdtAe
dtetftfF
jj
tjtj
tj
===
=
==
==
F
Using the definition of the Fourier transform,
Example 5
From the result in Example 4, using the time
shifting theorem, obtain the Fourier transform
of the functiong(t):
Then write as shown below:
Solution
)()()( 21 tgtgtg +=
SetA = 1 and = 1 of the functionf(t) to get thefollowing waveform:
= +
2sinc)(
=F
where and . Hence,
Therefore, the Fourier transform ofg(t) is
)()(21
1 += tftg )()( 21
2 = tftg
)()(
)()()(
21
21
21
+=
+=
tftf
tgtgtg
j
jj
Fee
FeFe
tftf
tgG
jj
jj
)1(cos2)2/(sin4
2sinc
2sin2
)()(
)()(
)}({)}({
)}({)(
2
2/2/
2/2/
21
21
===
=
=
+=
=
FF
F
[time shifting theorem]
-
7/28/2019 The Fourier Series Edited
20/46
10/24/12
20
Method ofDifferentiation in time
domain
The simplest Fourier transform is on the deltafunction, where F{(t)} = 1
Using this idea, before we transformed afunction, we differentiate it until its derivative isexpressed in delta functions form
The important properties in implementing thismethod are:
)()()(
Fjdt
tfd nn
n
=
F )()}({ 00 Fettf tj=Fand
Example 6
Using time differentiation, find the Fourier
transform of the following function
Solution
The idea is we differentiatef(t) until its
derivative is expressed in the terms of delta
functions. This required twice differentiation
as shown below:
dt
df
2
2
dt
fd
f= t+ 1 f= 1 t
Hence, )1()(2)1()(
2
2
+
+= tttdt
tfd
Taking the Fourier transform for both sides of eqn,
2)()(
2)()(
)}1()(2)1({)(
2
2
2
2
+=
+=
++=
jj
jj
eeF
eeFj
tttdt
tfdFF
2
)cos1(2)(
=FThus,
Example 7Given the Fourier transform off(t) is
jF
+
=
5
3)(
Determine the Fourier transform of the
following:
(a) (b)
(c) (d)
2
tf )12( tf
)()(cos tft )(tfe t
-
7/28/2019 The Fourier Series Edited
21/46
10/24/12
21
Solution
(a) Using time scaling property,
25
6
)2(5
32)2(2
1
221
21
jjF
Ft
f
+
=
+
==
=
F
(b) This involve the time scaling and time
shifting properties.))
j
ee
j
eF
tftf
jj
j
+
=
+
=
=
=
10
3
)2/(5
3
2
1
22
1
2)12(
2/2/
2/
21FF
* Note that the time shift is , not 1. The timeshift must be determined when the
coefficient oftis 1.
(c) From the Eulers identity,
[ ])()(2
1)(
2)()(cos tfetfetf
eetft
jtjtjtjt
+=
+=
Then, using the frequency shifting property,
{ } { } { }[ ]
[ ]
++
+
+=
++=
+=
)1(5
3
)1(5
3
2
1
)1()1(2
1
)()(
2
1)()(cos
jj
FF
tfetfetft jtjt FFF
(d) The frequency shifting property cannot
be apply directly, since there are noterm e jat. However, note that j 2 = 1,
that is
Then, using the frequency shifting property,
jjj
jF
tfetfe jjtt
+
=
+
=
=
=
6
3
)(5
3
)(
)()( FF
)()()(2
tfetfetfe jjttjt ==
The convolution integralA linear system:
h(t) y(t) =x(t)*h(t)x(t)
In time domainH() Y() =X()H()X()
In frequency domain
Convolution :
== dtfgdtgftgtf )()()()()()(
Which means the integral is evaluated indomain
-
7/28/2019 The Fourier Series Edited
22/46
10/24/12
22
The steps of convolution:1. Create a dummy variable (i.e. ) and makeeach waveform a function of.
2. Time-invert one of the waveforms (doesnt
matter which).
3. Shift the inverted waveform by t. this allow
the function to travel along -axis.
)()( ftf )()( gtg
)()( gg
)())(()( = tgtgg
4. Find the product of the intersection
5. For a given time t, integrate the product for
the intersection range a < < b.
)()(
tgf
b
a
dtgf )()(
6. Combine all the integrals after the travelling
waveform moved for < t< .
Example 8
Find the convolution of the two signals:
Solution
1. Writefandgas functions of.
g()
=
f() g()
2. Choose one of the functions as the
travelling wave. Suppose that we choosef.Reflectf() about the vertical axis to getf().
f() g()
g()
=
3. Writef() asf(t) . Since tvaries from
to +, it means we bring f to and shift it
to + direction.
4&5. Along the journey off(t), observe the
intersections betweenfandg.
Then integrate along the intersection ranges:
f(t) g()
tt 1 0 1
f(t) g()
tt 1
(i) < t< 0:No overlap. f*g= 0
(ii) 0 < t< 1
0 1
22))(1(
2
0
2
0
tdgf
tt
=
==
-
7/28/2019 The Fourier Series Edited
23/46
10/24/12
23
f(t
)g()
tt 1
(iii) 1 < t< 2
0 1
)2(22
))(1(
1
1
21
1
tt
dgf
tt
=
==
(iv) 2 < t< :No overlap. f*g= 0
6. Combine all the integral for < t< to
obtainf * g:
f*g(t)
t0 2
>
-
7/28/2019 The Fourier Series Edited
24/46
10/24/12
24
h(t)
t t+ 1t 1
(iii) 1 < t< 0
ttddhx
t
=++=+= +
)01()10()1)(1()1)(1(
1
0
0
1
x(
)
0
11
1
1
h(t)
t t+ 1t 1
(iv) 0 < t< 1
ttddhx
t
=++=+=
)01()10()1)(1()1)(1(
1
0
0
1
x()
0
11
1
1
h(t)
t t+ 1t 1
(v) 1 < t< 2
2)1(1)1)(1(
1
1
+===
ttdhx
t
x()
0
11
1
1
(vi) 2 < t< :No overlap. x *h = 0
Combine all the integral for < t< to
obtainx * h :
x*h (t)
t
0
2
-
7/28/2019 The Fourier Series Edited
25/46
10/24/12
25
Introduction
Jean BaptisteJoseph Fourier(*1768-1830)
French MathematicianLa Thorie Analitique
de la Chaleur (1822)
145
Fourier Series
Any periodic function can be expressed as asum of sines and/or cosines
Fourier Series
146
Fourier Transform
Even functions thatare not periodichave a finite area under curvecan be expressed as an integral of sines and
cosines multiplied by a weighing function
Both the Fourier Series and the FourierTransform have an inverse operation:
Original Domain Fourier Domain147
Contents
Complex numbers etc. ImpulsesFourier Transform (+examples)Convolution theoremFourier Transform of sampled functionsSampling theoremAliasingDiscrete Fourier TransformApplication Examples 148
Complex numbersComplex number
Its complex conjugate
149
Complex numbers polarComplex number in polar coordinates
150
-
7/28/2019 The Fourier Series Edited
26/46
10/24/12
26
Eulers formula
151
Sin ()
Cos ()
?
?
152
Re
Im
Complex math
Complex (vector) addition
Multiplication with i
is rotation by 90 degrees
153
Contents
Complex number etc. ImpulsesFourier Transform (+examples)Convolution theoremFourier Transform of sampled functionsSampling theoremAliasingDiscrete Fourier TransformApplication Examples 154
Unit impulse (Dirac delta function)Definition
ConstraintSifting propertySpecifically for t=0
155
Discrete unit impulseDefinition
ConstraintSifting propertySpecifically for x=0
156
-
7/28/2019 The Fourier Series Edited
27/46
10/24/12
27
What does this look like?
Impulse train
157
T = 1
Contents
Complex number etc. ImpulsesFourier Transform (+examples)Convolution theoremFourier Transform of sampled functionsSampling theoremAliasingDiscrete Fourier TransformApplication Examples 158
Fourier Series
with
159
Series of
sines and
cosines,
see Eulers
formula
Periodic
with
period T
Fourier Transform
Continuous Fourier Transform (1D)
Inverse Continuous Fourier Transform (1D)
160
161
Symmetry: The only difference between the FourierTransform and its inverse is the sign of the exponential.
Fourier Euler
Fourier and Euler
-
7/28/2019 The Fourier Series Edited
28/46
10/24/12
28
If f(t) is real, then F() is complexF() is expansion of f(t) multiplied by sinusoidal
terms
t is integrated over, disappearsF() is a function of only , which determines
the frequencyof sinusoidals
Fourier transform frequency domain163
Examples Block 1
164
-W/2 W/2
A
Examples Block 2
165
Examples Block 3
166
?
Examples Impulse
167
constant
Examples Shifted impulse
168
Euler
-
7/28/2019 The Fourier Series Edited
29/46
10/24/12
29
Examples Shifted impulse 2
169
Real part Imaginary part
impulse constant
Examples - Impulse train
170
Examples - Impulse train 2
171
Intermezzo: Symmetry in the FT
172
173
ContentsComplex number etc. ImpulsesFourier Transform (+examples)Convolution theoremFourier Transform of sampled functionsSampling theoremAliasingDiscrete Fourier TransformApplication Examples 174
-
7/28/2019 The Fourier Series Edited
30/46
10/24/12
30
Fourier + Convolution
What is the Fourier domain equivalent ofconvolution?
175
What is
176
Intermezzo 1
Whatis ?
Let , so
177
Intermezzo 2
Property of Fourier Transform
178
Fourier + Convolution cont
d
179
Recapitulation 1Convolution in one domain is multiplication in
the other domain
And (see book)
180
-
7/28/2019 The Fourier Series Edited
31/46
10/24/12
31
Recapitulation 2
Shift in one domain is multiplication withcomplex exponential in the other domain
And (see book)
181
Contents
Complex number etc. ImpulsesFourier Transform (+examples)Convolution theoremFourier Transform of sampled functionsSampling theoremAliasingDiscrete Fourier TransformApplication Examples 182
Sampling
Sampled function can be written as
Obtain value of arbitrary sample k as
183
Sampling - 2
184
Sampling - 3
185
Sampling - 4
186
-
7/28/2019 The Fourier Series Edited
32/46
10/24/12
32
FT of sampled functions
Fourier transform of sampled function
Convolution theoremFrom FT of impulse train
187
(who?)
FT of sampled functions
188
Sifting property
189
Contents
Complex number etc. ImpulsesFourier Transform (+examples)Convolution theoremFourier Transform of sampled functionsSampling theoremAliasingDiscrete Fourier TransformApplication Examples 190
Sampling theoremBand-limited function
Sampled function lower value of 1/T
would cause triangles
to merge
191
Sampling theorem 2Sampling theorem:If copy of can be isolated from the periodic
sequence of copies contained in
, can be completely recoveredfrom the sampled version.
Since is a continuous, periodic functionwith period 1/T, one complete period from
is enough to reconstruct the signal.
This can be done via the Inverse FT.192
-
7/28/2019 The Fourier Series Edited
33/46
10/24/12
33
Extracting a single period from that isequal to is possible ifSampling theorem 3
193
Nyquist frequency
Contents
Complex number etc. ImpulsesFourier Transform (+examples)Convolution theoremFourier Transform of sampled functionsSampling theoremAliasingDiscrete Fourier TransformApplication Examples 194
Aliasing
If , aliasing can occur
195
Contents
Complex number etc. ImpulsesFourier Transform (+examples)Convolution theoremFourier Transform of sampled functionsSampling theoremAliasingDiscrete Fourier TransformApplication Examples 196
Discrete Fourier TransformFourier Transform of a sampled function is an
infinite periodic sequence of copies of the
transform of the original continuous function
197
Discrete Fourier TransformContinuous transform of sampled function
198
-
7/28/2019 The Fourier Series Edited
34/46
10/24/12
34
is discrete function is continuous and infinitely periodic with
period
1/T199
We need only one period to characterise If we want to take M equally spaced samples
from
in the period = 0 to = 1/, this can
be done thus
200
Substituting
Into
yields201
Contents
Complex number etc. ImpulsesFourier Transform (+examples)Convolution theoremFourier Transform of sampled functionsSampling theoremAliasingDiscrete Fourier TransformApplication Examples 202
ExamplesCardiac tagging analysisDither removal
203
Formulation in 2D spatial
coordinatesContinuous Fourier Transform (2D)
Inverse Continuous Fourier Transform (2D)
204
-
7/28/2019 The Fourier Series Edited
35/46
10/24/12
35
The Fourier Transform II
Contents
Fourier Transform of sine and cosine 2D Fourier Transform Properties of the Discrete Fourier Transform
206
207
Eulers formula
208
Recall
209
Recall
210
Cos(t)
Sin(t)
1/2 1/2i
-
7/28/2019 The Fourier Series Edited
36/46
10/24/12
36
Contents
Fourier Transform of sine and cosine 2D Fourier Transform Properties of the Discrete Fourier Transform
211 212
Formulation in 2D spatial coordinates
Continuous Fourier Transform (2D)
Inverse Continuous Fourier Transform (2D)
with angular frequencies
Discrete Fourier Transform
213
Forward
Inverse
214
Formulation in 2D spatial coordinates
Discrete Fourier Transform (2D)
Inverse Discrete Transform (2D)
Spatial and Frequency intervals Inverse proportionality (Smallest) Frequency step depends on largest
distance covered in spatial domain
Suppose function is sampled M times in x,with step , distance is covered,
which is related to the lowest frequency that
can be measured
215
Examples
216
-
7/28/2019 The Fourier Series Edited
37/46
10/24/12
37
Fourier Series
with
217
Series of
sines and
cosines,
see Eulers
formula
Periodicwith
period T
Examples
218
Periodicity
2D Fourier Transform is periodic in bothdirections
219
Periodicity
2D Inverse Fourier Transform is periodic inboth directions
220
Fourier Domain
221
Inverse Fourier Domain Periodic?
222
Periodic!
-
7/28/2019 The Fourier Series Edited
38/46
10/24/12
38
Contents
Fourier Transform of sine and cosine 2D Fourier Transform Properties of the Discrete Fourier Transform
223
Properties of the 2D DFT
224
225
Real
ImaginarySin (x)Sin (x+/2)
Real
226
Real
ImaginaryF(Cos(x))F(Cos(x)+k)
Even
227
Real
Odd
Sin (x)Sin(y)Sin (x)Imaginary
228
Real
Imaginary(Sin (x)+1)(Sin(y)+1)
-
7/28/2019 The Fourier Series Edited
39/46
10/24/12
39
Symmetry: even and odd
Any real or complex function w(x,y) can beexpressed as the sum of an even and an odd
part (either real or complex)
229
Properties
Even function
Odd function
230
Properties - 2
231
Consequences for the Fourier Transform
FT of real function is conjugate symmetric
FT of imaginary function is conjugateantisymmetric
232
Im Re
-
7/28/2019 The Fourier Series Edited
40/46
10/24/12
40
Re Im
FT of even and odd functions
FT of even function is real
FT of odd function is imaginary
237 238
Real
ImaginaryCos (x)
Even
239
Real
ImaginarySin (x)
Odd
Fourier Series & The Fourier Transform
What is the Fourier Transform?
Fourier Cosine Series for even functions
Fourier Sine Series for odd functions
The continuous limit: the Fourier
transform (and its inverse)
Some transform examples and the Dirac
delta function
( ) ( ) exp( )F f t i t dt
= 1
( ) ( ) exp ( )2
f t F i t d
= Prof. Rick Trebino, Georgia Tech
-
7/28/2019 The Fourier Series Edited
41/46
10/24/12
41
What do we hope to achieve with the Fourier
Transform?
We desire a measure of the frequencies present in a wave. This willlead to a definition of the term, the spectrum.
Plane waves have only
one frequency, .
This light wave has manyfrequencies. And the
frequency increases in
time (from red to blue).
It will be nice if our measure also tells us when each frequency occurs.
Lightelectric
field
Time
Lord Kelvin on Fouriers theorem
Fouriers theorem is notonly one of the most
beautiful results of modern
analysis, but it may be said
to furnish an indispensable
instrument in the treatmentof nearly every recondite
question in modern physics.
Lord Kelvin
Joseph Fourier
Fourier was obsessed
with the physics of heat
and developed the
Fourier series andtransform to model
heat-flow problems.
Joseph Fourier 1768 - 1830
Anharmonic waves are sums of sinusoids.
Consider the sum of two sine waves (i.e., harmonic waves) of
different frequencies:
The resulting wave is periodic, but not harmonic.
Essentially all waves are anharmonic.
Fourier
decomposingfunctions
Here, we write a
square wave as
a sum of sine waves.
sin(t)
sin(3t)
sin(5t)
Any function can be written as the
sum of an even and an odd function.
E(-x) = E(x)
O(-x) = -O(x)
E(x)O(x)
f(x)
-
7/28/2019 The Fourier Series Edited
42/46
10/24/12
42
Fourier Cosine Series
Because cos(mt) is an even function (for all m), we can write an even function,
f(t), as:
where the set {Fm; m = 0, 1, } is a set of coefficients that define the series.
And where well only worry about the functionf(t) over the interval
(,).
f(t) =1
!
Fm cos(mt)
m=0
"
#
The Kronecker delta function
Finding the coefficients, Fm, in a Fourier Cosine Series
Fourier Cosine Series:
To findFm, multiply each side by cos(mt), where m is another integer, and integrate:
But:
So: only the m = m term contributes
Dropping the from the m:
yields the
coefficients for
anyf(t)!
0
1( ) cos( )m
m
f t F mt
=
=
0
1( ) cos( ' ) cos( ) cos( ' )
m
m
f t m t dt F mt m t dt
=
=
, '
'co s( ) cos( ' )
0 'm m
if m mmt m t dt
if m m
==
, '
0
1( ) cos( ' ) m m m
m
f t m t dt F
=
=
( ) cos( )mF f t mt dt
=
Fourier Sine Series
Because sin(mt) is an odd function (for all m), we can write
any odd function,f(t), as:
where the set {Fm; m = 0, 1, } is a set of coefficients that define theseries.
where well only worry about the functionf(t) over the interval (,).
f(t) =1
!
"Fm sin(mt)
m=0
#
$
Finding the coefficients, Fm, in a Fourier Sine Series
Fourier Sine Series:
To findFm, multiply each side by sin(mt), where m is another integer, and integrate:
But:
So: only the m = m term contributes
Dropping the from the m: yields the coefficients
for anyf(t)!
0
1( ) sin( )m
m
f t F mt
=
=
0
1( ) sin( ' ) sin( ) sin( ' )m
m
f t m t dt F mt m t dt
=
=
, '
'sin( ) sin( ' )
0 'm m
if m mmt m t dt
if m m
==
, '
0
1( ) sin( ' ) m m m
m
f t m t dt F
=
=
( ) sin( )mF f t mt dt
=
Fourier Series
even component odd component
where
and
0 0
1 1( ) cos( ) sin( )m m
m m
f t F mt F mt
= =
= +
Fm = f(t) cos(mt) dt! !Fm = f(t) sin(mt) dt"
So iff(t) is a general function, neither even nor odd, it can be written:
-
7/28/2019 The Fourier Series Edited
43/46
10/24/12
43
We can plot the coefficients of a Fourier Series
We really need two such plots, one for the cosine series and another for the
sine series.
Fmvs. m
m5 25
201510
30
1.50
Discrete Fourier Series vs.
Continuous Fourier Transform
Fmvs. m
mAgain, we really need two such plots, one for the cosine series and another
for the sine series.
Let the integermbecome a real
number and let
the coefficients,Fm, become afunction F(m).
F(m)
The Fourier TransformConsider the Fourier coefficients. Lets define a function F(m) that
incorporates both cosine and sine series coefficients, with the sine seriesdistinguished by making it the imaginary component:
Lets now allow f(t) to range from to , so well have to integrate from
to , and lets redefine m to be the frequency, which well now call :
F() is called the Fourier Transform off(t). It contains equivalent
information to that in f(t). We say that f(t) lives in the time domain, and F
() lives in the frequency domain. F() is just another way of looking at a
function or wave.
( ) cos ( )f t mt dt ( ) s in( )i f t mt dt F(m) Fm iFm =
( ) ( ) exp ( )F f t i t dt
= The FourierTransform
The Inverse Fourier TransformThe Fourier Transform takes us from f(t) toF().How about going back?
Recall our formula for the Fourier Series off(t) :
Now transform the sums to integrals from to , and again replaceFm
withF
(
). Remembering the fact that we introduced a factor ofi
(andincluding a factor of 2 that just crops up), we have:
'
0 0
1 1( ) cos( ) sin( )m m
m m
f t F mt F mt
= =
= +
1( ) ( ) exp( )
2f t F i t d
= Inverse
Fourier
Transform
The Fourier Transform and its Inverse
The Fourier Transform and its Inverse:
So we can transform to the frequency domain and back. Interestingly,
these transformations are very similar.
There are different definitions of these transforms. The 2 can occur inseveral places, but the idea is generally the same.
Inverse Fourier Transform
FourierTransform( ) ( ) exp( )F f t i t dt
=
1( ) ( ) exp( )
2f t F i t d
=
There are several ways to denote the Fourier transform of a function.
If the function is labeled by a lower-case letter, such asf,we can write:
f(t) F()
If the function is already labeled by an upper-case letter, such asE, wecan write:
or:
Fourier Transform Notation
( ) ( )E t E %( ) { ( )}E t E tF
Sometimes, this symbol isused instead of the arrow:
-
7/28/2019 The Fourier Series Edited
44/46
10/24/12
44
The Spectrum
We define the spectrum, S(), of a waveE(t) to be:
2
( ) { ( )}S E t F
This is the measure of the frequencies present in a light wave.
Example: the Fourier Transform of a
rectangle function: rect(t)1/2
1/ 2
1/ 2
1/2
1( ) exp( ) [exp( )]
1[exp( / 2) exp(
e xp ( / 2 ) e xp (
2
sin(
F i t dt i ti
i ii
i i
i
= =
= /2)]
1 /2)=
( /2)
/2)=
( /2)
( sinc(F ) = /2)Imaginary
Component = 0
F()
Sinc(x) and why it's important
Sinc(x/2) is the Fouriertransform of a rectangle
function.
Sinc2(x/2) is the Fouriertransform of a t riangle
function.
Sinc2(ax) is the diffractionpattern from a slit.
It just crops up everywhere...
The Fourier Transform of the triangle
function, (t), is sinc2(/2)
0
2sinc ( / 2)
1
t0
( )t
1
1/2-1/2
The triangle function is just what it sounds like.
Well prove this when we learn about convolution.
Sometimes
people use
(t), too, for the
triangle
function.
Example: the Fourier Transform of a
decaying exponential: exp(-at) (t> 0)0
0 0
0
( exp( ) exp( )
exp( ) exp( [ )
1 1exp( [ ) [exp( ) exp(0)]
1 1[0 1]
F at i t dt
at i t dt a i t dt
a i ta i a i
a i a i
+
) =
= = + ]
= + ] =
+ +
= =
+ +
A complex Lorentzian!
Example: the Fourier Transform of a
Gaussian, exp(-at2
), is itself!
2 2
2
{exp( )} exp( ) exp( )
exp( / 4 )
at at i t dt
a
=
F
t0
2exp( )at
0
2exp( / 4 )a
The details are a HW problem!
-
7/28/2019 The Fourier Series Edited
45/46
10/24/12
45
The Dirac delta function
Unlike the Kronecker delta-function, which is a function of two integers,
the Dirac delta function is a function of a real variable, t.
t
(t)
The Dirac delta function
Its best to think of the delta function as the limit of a series of peaked
continuous functions.
if 0( )
0 if 0
tt
t
=
t
f1(t)
f2(t)
fm(t) = m exp[-(mt)2]/
f3(t)
(t)
Dirac -function Properties( ) 1t dt
= t
(t)
( ) ( ) ( ) ( ) ( )t a f t dt t a f a dt f a
= =
exp( ) 2 (
exp[ ( ) ] 2 (
i t dt
i t dt
= )
= )
2()
The Fourier Transform of(t) is 1.
1 exp( ) 2 (i t dt
= )And the Fourier Transform of1 is 2():
( ) exp( ) exp( [0]) 1t i t dt i
= =
t
(t)
1
t
1
0
0
The Fourier transform ofexp(i0t)
{ }0 0exp( ) exp( ) exp( )i t i t i t dt
= F
0exp( [ ] )i t dt
=
The function exp(i0t) is the essential component of Fourier analysis. It is
a pure frequency.
F {exp(i0t)}
0 0
02 ( ) =
exp(i0t)
0t
tRe
Im
0
The Fourier transform ofcos(0t){ }0 0cos( ) cos( ) exp( )t t i t dt
= F
[ ]0 01
exp( ) exp( ) exp( )2
i t i t i t dt
= +
0 0
1 1exp( [ ] ) exp( [ ] )
2 2i t dt i t dt
= + +
0 0( ) ( ) = + +
+00 -
0
0{cos( )}tFcos(
0t)
t0
-
7/28/2019 The Fourier Series Edited
46/46
10/24/12
Fourier Transform Symmetry Properties
Expanding the Fourier transform of a function,f(t):
( ) Re{ ( )} cos( ) Im{ ( )} sin( )F f t t d t f t t d t
= + Re{F()}
Im{F()}
= 0 ifRe{f(t)} is odd = 0 ifIm{f(t)} is even
Even functions of Odd functions of
( ) [Re{ ( )} Im{ ( )}] [cos( ) sin( )]F f t i f t t i t dt
= +
Im{ ( )} cos( ) Re{ ( )} sin( )i f t t dt i f t t dt
+
= 0 ifIm{f(t)} is odd = 0 ifRe{f(t)} is even
Expanding more, noting that: ( ) 0O t dt
= ifO(t) is an odd function
Some functions dont have Fouriertransforms.
The condition for the existence of a givenF() is:
Functions that do not asymptote to zero in both the + and
directions generally do not have Fourier transforms.
So well assume that all functions of interest go to zero at .
( )f t dt
<