# 0.3 Fft methods

 Page 1 / 1

## Fft methods

The mixed-radix approach utilizes clever subsampling of $x\left[n\right]$ and permutations of twiddle factor matrices to lower operation counts. The first step is to compute the prime factorization of the signal length $N$ . Prime factors of 2 and 3 are then combined to create as many factors of 4 and 6 as possible. The resulting prime factorization has the form $N={\prod }_{i=1}^{F}{f}_{i}$ . We can then perform a single step ${T}_{i}$ of the algorithm by computing $N/{f}_{i}$ FFTs of length ${f}_{i}$ .

Four unique matrix constructions are necessary to generalize a single step of the algorithm. The first is the identity matrix ${I}_{r}\in {\mathbb{R}}^{r×r}$ . The second is the permutation matrix ${P}_{b}^{a}\in {\mathbb{R}}^{ab×ab}$ , where

${P}_{b}^{a}\left(j,k\right)=\left\{\begin{array}{cc}1& \text{if}\phantom{\rule{0.166667em}{0ex}}\phantom{\rule{0.166667em}{0ex}}j=ra+s\phantom{\rule{0.166667em}{0ex}}\phantom{\rule{0.166667em}{0ex}}\text{and}\phantom{\rule{0.166667em}{0ex}}\phantom{\rule{0.166667em}{0ex}}k=sb+r\\ 0& \text{otherwise}\end{array}\right)$

Note that ${P}_{b}^{a}$ swaps positions $ra+s$ and $sb+r$ in a vector. The third matrix to consider is a diagonal matrix of twiddle factors ${D}_{b}^{a}\in {\mathbb{R}}^{ab×ab}$ , where

${D}_{b}^{a}\left(j,k\right)=\left\{\begin{array}{cc}{\omega }_{ab}^{sr}& \text{if}\phantom{\rule{0.166667em}{0ex}}\phantom{\rule{0.166667em}{0ex}}j=k=sb+r\\ 0& \text{otherwise}\end{array}\right)$

with ${\omega }_{ab}={e}^{-j2\pi /\left(ab\right)}$ . The fourth and final matrix construction to consider is the standard DFT matrix ${W}_{n}\in {\mathbb{R}}^{n×n}$ .

The mixed-radix algorithm requires the interaction of operations on the subspaces ${\mathbb{R}}^{i},\phantom{\rule{0.166667em}{0ex}}\phantom{\rule{0.166667em}{0ex}}i=1,...,F$ , and so it is necessary to consider the Kronecker Product in calculations. The Kronecker Product is an operation defined as $\otimes :{\mathbb{R}}^{m×n}×{\mathbb{R}}^{p×q}\to {\mathbb{R}}^{mp×nq}$ . If $C=A\otimes B$ , then

$C\left(v,w\right)=A\left(i,j\right)B\left(k,l\right)\phantom{\rule{0.166667em}{0ex}}\phantom{\rule{0.166667em}{0ex}}\phantom{\rule{0.166667em}{0ex}}\phantom{\rule{0.166667em}{0ex}}\text{and}\phantom{\rule{0.166667em}{0ex}}\phantom{\rule{0.166667em}{0ex}}\left\{\begin{array}{c}v\equiv p\left(i-1\right)+k\\ w\equiv q\left(j-1\right)+l\end{array}\right)$

Using these matrices, it possible to compose a single steps ${T}_{i}$ using the following equation [link] :

${T}_{i}=\left({P}_{{q}_{i}}^{{f}_{i}}{D}_{{q}_{i}}^{{f}_{i}}\otimes {I}_{{p}_{i-1}}\right)\left({W}_{{f}_{i}}\otimes {I}_{{m}_{i}}\right)$

where

$\left(\begin{array}{cc}{p}_{i}=\prod _{j=1}^{i}{f}_{j}\phantom{\rule{0.166667em}{0ex}}\phantom{\rule{0.166667em}{0ex}}\phantom{\rule{0.166667em}{0ex}}\phantom{\rule{0.166667em}{0ex}}\left({p}_{0}=1\right),& i=0,...,F\\ {q}_{i}=N/{p}_{i},& i=1,...,F\\ {m}_{i}=N/{f}_{i},& i=1,...,F\end{array}\right)$

Note that ${T}_{i}\in {\mathbb{R}}^{N×N}$ .

To implement the algorithm, each stage is applied iteratively to acquire $X$ . ${T}_{i}$ may be simplified into two separate steps using the definitions of each matrix to yield the following algorithm structure:

The DFT module in the above code utilizes the Winograd Short-Length Transforms to minimize operations when computing DFTs of length less than 6. For all other lengths, Rader's FFT is used.

Rader's FFT for prime lengths exploits results from number theory to express the DFT as a composite-length cyclic convolution [link] . Any prime number $p$ defines a multiplicative group modulo $p$ , denoted ${\mathbb{Z}}_{n}^{×}=\left\{0,,,1,,,2,,,.,.,.,,,p,-,1\right\}$ . This group is cyclic, i.e., $\forall a\in {\mathbb{Z}}_{p}^{×}\exists g\in {\mathbb{N}}_{p-2}^{0}:a={g}^{i}={g}^{-j},\phantom{\rule{0.166667em}{0ex}}\phantom{\rule{0.166667em}{0ex}}0\le i,j\le p-2$ , where $g$ is known as the primitive root modulo $\mathit{p}$ and ${\mathbb{N}}_{p-2}^{0}=\left\{0,,,1,,,2,,,.,.,.,,,p,-,2\right\}$ . In practice, there is no general formula for finding $g$ for $p$ , but in this implementation $N$ is known and therefore $g$ may be stored in a lookup table.

The general form of the DFT of length $p$ is given by

${X}_{k}=\sum _{n=0}^{p-1}{x}_{n}{\omega }_{p}^{nk}=\stackrel{˜}{x}+\sum _{n=1}^{p-1}{x}_{n}{\omega }_{p}^{nk},\phantom{\rule{0.166667em}{0ex}}\phantom{\rule{0.166667em}{0ex}}\phantom{\rule{0.166667em}{0ex}}\phantom{\rule{0.166667em}{0ex}}k=1,2,...,p-1$

Since the twiddle factor, ${\omega }_{p}^{nk}$ , is naturally computed modulo $p$ and both $n$ and $k$ range from $1,2,...,p-1$ , we can rewrite the above formula using $g$ :

${X}_{{g}^{-r}}=\stackrel{˜}{x}+\sum _{q=0}^{p-2}{x}_{{g}^{q}}{\omega }_{p}^{{g}^{-\left(r-q\right)}},\phantom{\rule{0.166667em}{0ex}}\phantom{\rule{0.166667em}{0ex}}\phantom{\rule{0.166667em}{0ex}}\phantom{\rule{0.166667em}{0ex}}r=0,1,...,p-2$

This formulation is of the form of a cyclic convolution of two sequences $\alpha$ and $\beta$ where ${\alpha }_{q}={h}_{{g}^{q}}$ and ${\beta }_{q}={\omega }_{p}^{{g}^{-q}}$ of length $p-1$ . This convolution is computed via the convolution theorem:

$X-\stackrel{˜}{x}={\text{DFT}}^{-1}\left[\text{DFT},\left[,\alpha ,\right],·,\text{DFT},\left[,\beta ,\right]\right]$

For speed, the sequence $\alpha$ is zero-padded between its zeroth and first index to a length $M$ which is a power of 2 such that $M\ge 2p-3$ and $\beta$ is cyclically repeated to be length $M$ . Since $N$ is known, it is possible to store the DFT of the sequence $\beta$ in a lookup table. The DFT of $\alpha$ is then computed using the standard radix-2 Cooley-Tukey algorithm. The two DFTs are multiplied pointwise and then the inverse DFT is again calculated using the standard Cooley-Tukey algorithm. The first $p$ elements are then the result of the cyclic convolution. The final result is attained after adding back the DC offset.

For small numerical examples demonstrating this implementation of Rader's FFT, see this excellent guide here .

Is there any normative that regulates the use of silver nanoparticles?
what king of growth are you checking .?
Renato
What fields keep nano created devices from performing or assimulating ? Magnetic fields ? Are do they assimilate ?
why we need to study biomolecules, molecular biology in nanotechnology?
?
Kyle
yes I'm doing my masters in nanotechnology, we are being studying all these domains as well..
why?
what school?
Kyle
biomolecules are e building blocks of every organics and inorganic materials.
Joe
anyone know any internet site where one can find nanotechnology papers?
research.net
kanaga
sciencedirect big data base
Ernesto
Introduction about quantum dots in nanotechnology
what does nano mean?
nano basically means 10^(-9). nanometer is a unit to measure length.
Bharti
do you think it's worthwhile in the long term to study the effects and possibilities of nanotechnology on viral treatment?
absolutely yes
Daniel
how to know photocatalytic properties of tio2 nanoparticles...what to do now
it is a goid question and i want to know the answer as well
Maciej
Abigail
for teaching engĺish at school how nano technology help us
Anassong
Do somebody tell me a best nano engineering book for beginners?
there is no specific books for beginners but there is book called principle of nanotechnology
NANO
what is fullerene does it is used to make bukky balls
are you nano engineer ?
s.
fullerene is a bucky ball aka Carbon 60 molecule. It was name by the architect Fuller. He design the geodesic dome. it resembles a soccer ball.
Tarell
what is the actual application of fullerenes nowadays?
Damian
That is a great question Damian. best way to answer that question is to Google it. there are hundreds of applications for buck minister fullerenes, from medical to aerospace. you can also find plenty of research papers that will give you great detail on the potential applications of fullerenes.
Tarell
what is the Synthesis, properties,and applications of carbon nano chemistry
Mostly, they use nano carbon for electronics and for materials to be strengthened.
Virgil
is Bucky paper clear?
CYNTHIA
carbon nanotubes has various application in fuel cells membrane, current research on cancer drug,and in electronics MEMS and NEMS etc
NANO
so some one know about replacing silicon atom with phosphorous in semiconductors device?
Yeah, it is a pain to say the least. You basically have to heat the substarte up to around 1000 degrees celcius then pass phosphene gas over top of it, which is explosive and toxic by the way, under very low pressure.
Harper
Do you know which machine is used to that process?
s.
how to fabricate graphene ink ?
for screen printed electrodes ?
SUYASH
What is lattice structure?
of graphene you mean?
Ebrahim
or in general
Ebrahim
in general
s.
Graphene has a hexagonal structure
tahir
On having this app for quite a bit time, Haven't realised there's a chat room in it.
Cied
what is biological synthesis of nanoparticles
how did you get the value of 2000N.What calculations are needed to arrive at it
Privacy Information Security Software Version 1.1a
Good
Got questions? Join the online conversation and get instant answers!