Title: Numerical Methods Fast Fourier Transform Part: Informal Development of Fast Fourier Transform http://numericalmethods.eng.usf.edu
1Numerical MethodsFast Fourier Transform
Part Informal Development of Fast Fourier
Transform http//numericalmethods.eng.usf.edu
2- For more details on this topic
- Go to http//numericalmethods.eng.usf.edu
- Click on Keyword
- Click on Fast Fourier Transform
3You are free
- to Share to copy, distribute, display and
perform the work - to Remix to make derivative works
4Under the following conditions
- Attribution You must attribute the work in the
manner specified by the author or licensor (but
not in any way that suggests that they endorse
you or your use of the work). - Noncommercial You may not use this work for
commercial purposes. - Share Alike If you alter, transform, or build
upon this work, you may distribute the resulting
work only under the same or similar license to
this one.
5Chapter 11.05 Informal Development of Fast
Fourier Transform
Lecture 11
Major All Engineering Majors Authors Duc
Nguyen http//numericalmethods.eng.usf.edu Numeri
cal Methods for STEM undergraduates
http//numericalmethods.eng.usf.edu
9/20/2020
5
6Informal Development of Fast Fourier Transform
(1)
(2)
(3)
(4)
6
http//numericalmethods.eng.usf.edu
7Informal Development cont.
Then Eq. (1) and Eq. (2) become
(5)
(5A)
7
http//numericalmethods.eng.usf.edu
8Informal Development cont.
8
http//numericalmethods.eng.usf.edu
9Informal Development cont.
(6)
10Informal Development cont.
(7)
10
http//numericalmethods.eng.usf.edu
11Informal Development cont.
(8)
12Informal Development cont.
12
http//numericalmethods.eng.usf.edu
13The End
- http//numericalmethods.eng.usf.edu
14Acknowledgement
- This instructional power point brought to you by
- Numerical Methods for STEM undergraduate
- http//numericalmethods.eng.usf.edu
- Committed to bringing numerical methods to the
undergraduate
15- For instructional videos on other topics, go to
- http//numericalmethods.eng.usf.edu/videos/
- This material is based upon work supported by the
National Science Foundation under Grant
0717624. Any opinions, findings, and conclusions
or recommendations expressed in this material are
those of the author(s) and do not necessarily
reflect the views of the National Science
Foundation.
16The End - Really
17Numerical MethodsFast Fourier Transform
Part Factorized Matrix and Further Operation
Counthttp//numericalmethods.eng.usf.edu
18- For more details on this topic
- Go to http//numericalmethods.eng.usf.edu
- Click on Keyword
- Click on Fast Fourier Transform
19You are free
- to Share to copy, distribute, display and
perform the work - to Remix to make derivative works
20Under the following conditions
- Attribution You must attribute the work in the
manner specified by the author or licensor (but
not in any way that suggests that they endorse
you or your use of the work). - Noncommercial You may not use this work for
commercial purposes. - Share Alike If you alter, transform, or build
upon this work, you may distribute the resulting
work only under the same or similar license to
this one.
21Chapter 11.05 Factorized Matrix and Further
Operation Count (Contd.)
Lecture 12
Equation (7) can be factorized as
(9)
Lets define the following inner product
(10)
21
http//numericalmethods.eng.usf.edu
22Factorized Matrix cont.
From Eq. (9) and (10) we obtain
(11A)
(11B)
(11C)
(11D)
22
http//numericalmethods.eng.usf.edu
23Factorized Matrix cont.
Equations(11A through 11D) for the inner matrix
times vector requires 2 complex multiplications
and 4 complex additions.
24Factorized Matrix cont.
Finally, performing the outer product (matrix
times vector) on the RHS of Equation(9), one
obtains
(12)
24
http//numericalmethods.eng.usf.edu
25Factorized Matrix cont.
(13A)
(13B)
(13C)
(13D)
26Factorized Matrix cont.
26
http//numericalmethods.eng.usf.edu
27Factorized Matrix cont.
For a large number of data points,
(14)
This implies that the number of complex
multiplications involved in Eq. (9) is about 372
times less than the one involved in Eq. (7).
27
http//numericalmethods.eng.usf.edu
28Graphical Flow of Eq. 9
Consider the case
Figure 1. Graphical form of FFT (Eq. 9) for the
case
28
http//numericalmethods.eng.usf.edu
29Figure 2. Graphical Form of FFT (Eq. 9) for the
case
29
http//numericalmethods.eng.usf.edu
30The End
- http//numericalmethods.eng.usf.edu
31Acknowledgement
- This instructional power point brought to you by
- Numerical Methods for STEM undergraduate
- http//numericalmethods.eng.usf.edu
- Committed to bringing numerical methods to the
undergraduate
32- For instructional videos on other topics, go to
- http//numericalmethods.eng.usf.edu/videos/
- This material is based upon work supported by the
National Science Foundation under Grant
0717624. Any opinions, findings, and conclusions
or recommendations expressed in this material are
those of the author(s) and do not necessarily
reflect the views of the National Science
Foundation.
33The End - Really
34Numerical MethodsFast Fourier Transform
Part Companion Node Observationhttp//numerica
lmethods.eng.usf.edu
35- For more details on this topic
- Go to http//numericalmethods.eng.usf.edu
- Click on Keyword
- Click on Fast Fourier Transform
36You are free
- to Share to copy, distribute, display and
perform the work - to Remix to make derivative works
37Under the following conditions
- Attribution You must attribute the work in the
manner specified by the author or licensor (but
not in any way that suggests that they endorse
you or your use of the work). - Noncommercial You may not use this work for
commercial purposes. - Share Alike If you alter, transform, or build
upon this work, you may distribute the resulting
work only under the same or similar license to
this one.
38Chapter 11.05 Companion Node Observation
(Contd.)
Lecture 13
38
http//numericalmethods.eng.usf.edu
39Figure 2. Graphical Form of FFT (Eq. 9) for the
case
40Companion Node Observation cont.
40
http//numericalmethods.eng.usf.edu
41Companion Node Spacing
Observing Figure 2, the following statements can
be made
41
http//numericalmethods.eng.usf.edu
42Companion Node Computation
The operation counts in any companion nodes (of
the vector), such as and
can be explained as (see Figure 2).
(15)
(16)
42
http//numericalmethods.eng.usf.edu
43Companion Node Computation cont.
48
(17)
(18)
43
http//numericalmethods.eng.usf.edu
44Skipping Computation of Certain Nodes
44
http//numericalmethods.eng.usf.edu
45The End
- http//numericalmethods.eng.usf.edu
46Acknowledgement
- This instructional power point brought to you by
- Numerical Methods for STEM undergraduate
- http//numericalmethods.eng.usf.edu
- Committed to bringing numerical methods to the
undergraduate
47- For instructional videos on other topics, go to
- http//numericalmethods.eng.usf.edu/videos/
- This material is based upon work supported by the
National Science Foundation under Grant
0717624. Any opinions, findings, and conclusions
or recommendations expressed in this material are
those of the author(s) and do not necessarily
reflect the views of the National Science
Foundation.
48The End - Really
49Numerical MethodsFast Fourier Transform
Part Determination of http//numericalmethods.
eng.usf.edu
50- For more details on this topic
- Go to http//numericalmethods.eng.usf.edu
- Click on Keyword
- Click on Fast Fourier Transform
51You are free
- to Share to copy, distribute, display and
perform the work - to Remix to make derivative works
52Under the following conditions
- Attribution You must attribute the work in the
manner specified by the author or licensor (but
not in any way that suggests that they endorse
you or your use of the work). - Noncommercial You may not use this work for
commercial purposes. - Share Alike If you alter, transform, or build
upon this work, you may distribute the resulting
work only under the same or similar license to
this one.
53Chapter 11.05 Determination of
Lecture 14
can be determined by the following steps
one obtains
http//numericalmethods.eng.usf.edu
53
54Determination of cont.
It is important to realize that the results of
Step 2 (0,0,1,0) are equivalent to expressing an
integer
in binary format. In other words
55Determination of cont.
55
http//numericalmethods.eng.usf.edu
56Computer Implementation to find
Based on the previous discussions (with the
3-step procedures), to find the value of ,
one only needs a procedure to express an integer
in binary format, with bits.
Assuming (a base 10 number) can be expressed
as (assuming bits)
(19)
56
http//numericalmethods.eng.usf.edu
57Computer Implementation cont.
58Computer Implementation cont.
(20)
If IDIFF 0, then the bit a1 0
If IDIFF ? 0, then the bit a1 1
58
http//numericalmethods.eng.usf.edu
59Computer Implementation cont.
etc.
60Example 1
For , , bits
and . Find the value of .
60
http//numericalmethods.eng.usf.edu
61Example 1 cont.
Thus
61
http//numericalmethods.eng.usf.edu
62Example 1 cont.
Determine the bit (Index )
63Example 1 cont.
Determine the bit (Index )
Thus
63
http//numericalmethods.eng.usf.edu
64The End
- http//numericalmethods.eng.usf.edu
65Acknowledgement
- This instructional power point brought to you by
- Numerical Methods for STEM undergraduate
- http//numericalmethods.eng.usf.edu
- Committed to bringing numerical methods to the
undergraduate
66- For instructional videos on other topics, go to
- http//numericalmethods.eng.usf.edu/videos/
- This material is based upon work supported by the
National Science Foundation under Grant
0717624. Any opinions, findings, and conclusions
or recommendations expressed in this material are
those of the author(s) and do not necessarily
reflect the views of the National Science
Foundation.
67The End - Really
68Numerical MethodsFast Fourier Transform
Part Unscrambling the FFThttp//numericalmetho
ds.eng.usf.edu
69- For more details on this topic
- Go to http//numericalmethods.eng.usf.edu
- Click on Keyword
- Click on Fast Fourier Transform
70You are free
- to Share to copy, distribute, display and
perform the work - to Remix to make derivative works
71Under the following conditions
- Attribution You must attribute the work in the
manner specified by the author or licensor (but
not in any way that suggests that they endorse
you or your use of the work). - Noncommercial You may not use this work for
commercial purposes. - Share Alike If you alter, transform, or build
upon this work, you may distribute the resulting
work only under the same or similar license to
this one.
72Chapter 11.05 Unscrambling the FFT (Contd.)
Lecture 15
For the case , (see Figure 2),
the final bit-reversing operation for FFT is
shown in Figure 3.
72
http//numericalmethods.eng.usf.edu
73For do-loop index k 0 (0, 0, 0, 0) i
(0, 0, 0, 0) bit-reversion 0 If (i.GT.k)
Then T f4(k) f4(k) f4(i) f4(i)
T Endif Hence, f4(0) f4(0) no swapping.
55
73
http//numericalmethods.eng.usf.edu
74For k 1 (0,0,0,1) i (1,0,0,0)
bit-reversion 8 If (i.GT.k) Then T f4(k1)
f4(k1) f4(i8) f4(i8) T Endif Hence,
f4(1) f4(8) are swapped.
75.For k2(0,0,1,0) i (0,1,0,0)
4 Hence, f4(2) f4(4) are swapped.
.For k3(0,0,1,1) i (1,1,0,0)
12 Hence, f4(3) f4(12) are swapped.
. For k4(0,1,0,0) i(0,0,1,0)2 In this
case, since i is not greater than k. Hence,
no swapping, since f4 (k 2) and f4 (i 4) had
already been swapped earlier! .etc.
76Computer Implementation of FFT case for N2r
The pair of companion nodes computation are
given by Eqs.(17) and (18). To avoid complex
number operations,Eq.(17) can be computed based
on real number operations, as following
(21)
76
http//numericalmethods.eng.usf.edu
77Computer Implementation cont.
Multiplying the last 2 complex numbers, one
obtains
(22)
Equating the real (and then, imaginary)
components on the Left-Hand-Side (LHS), and the
Right-Hand-Side (RHS) of Eq. (22), one obtains
77
http//numericalmethods.eng.usf.edu
78Computer implementation cont.
(23A)
(23B)
79Computer implementation cont.
Recall Eq. (4)
Hence
(24)
where
(25)
Thus
(26A)
(26B)
http//numericalmethods.eng.usf.edu
79
80Computer Implementation cont.
Substituting Eqs. (26A) and (26B) into Eqs. (23A)
and (23B), one gets
(27A)
(27B)
Similarly, the single (complex number) Eq. (18)
can be expressed as 2 equivalent (real number)
Eqs. Like Eqs. (27A) and (27B).
80
http//numericalmethods.eng.usf.edu
81The End
- http//numericalmethods.eng.usf.edu
82Acknowledgement
- This instructional power point brought to you by
- Numerical Methods for STEM undergraduate
- http//numericalmethods.eng.usf.edu
- Committed to bringing numerical methods to the
undergraduate
83- For instructional videos on other topics, go to
- http//numericalmethods.eng.usf.edu/videos/
- This material is based upon work supported by the
National Science Foundation under Grant
0717624. Any opinions, findings, and conclusions
or recommendations expressed in this material are
those of the author(s) and do not necessarily
reflect the views of the National Science
Foundation.
84The End - Really