fft and dft algorithm

Post on 17-Feb-2017

189 Views

Category:

Engineering

5 Downloads

Preview:

Click to see full reader

TRANSCRIPT

FFT & DFT

A Divide and Conquer Algorithm

2

Fourier theorem. [Fourier, Dirichlet, Riemann] Any (sufficiently smooth)periodic function can be expressed as the sum of a series of sinusoids.

Fourier analysis

𝑦 (𝑡 )= 2𝜋∑𝑘=1𝑁 𝑆𝑖𝑛𝑘𝑡

𝑘 𝑁=100

Sinusoids. Sum of sine and cosines = sum of complex exponentials.

May 1, 2023 By:Muhammad Umer Javed

May 1, 2023 By:Muhammad Umer Javed 3

Signal. [touch tone button 1]

Time domain.

Frequency domain.

Time domain vs. frequency domain

May 1, 2023 By:Muhammad Umer Javed 4

FFT. Fast way to convert between time-domain and frequency-domain. Alternate viewpoint. Fast way to multiply and evaluate polynomials.

There are two ways to represent polynomials• Coefficient representation• Point value representation

Fast Fourier transform

We take this approach

May 1, 2023 By:Muhammad Umer Javed 5

Polynomial. [coefficient representation]

Add. O(n) arithmetic operations.

Evaluate. O(n) using Horner's method.

Multiply (convolve). using brute force.

Polynomials: coefficient representation

May 1, 2023 By:Muhammad Umer Javed 6

Fundamental theorem of algebra. A degree n polynomial with complex coefficients has exactly n complex roots.

Corollary. A degree n – 1 polynomial A(x) is uniquely specified by itsevaluation at n distinct values of x

Polynomials: point-value representation

May 1, 2023 By:Muhammad Umer Javed 7

Polynomial. [point-value representation]

Add. O(n) arithmetic operations.

Multiply (convolve). O(n), but need 2n – 1 points.

Evaluate. O(n2) using Lagrange's formula.

Polynomials: point value representation

May 1, 2023 By:Muhammad Umer Javed 8

Point-value => coefficient . Given distinct points , Evaluate it at distinct points

Running time. O(n2) for matrix-vector multiply (or n Horner's)..

Converting between two representations: brute force

May 1, 2023 By:Muhammad Umer Javed 9

Coefficient => point-value. Given distinct points and values , find unique polynomial , that has given values at given points.

Running time. for Gaussian elimination.

Converting between two representations: brute force

Vandermonde matrix is invertible iff xi distinct

May 1, 2023 By:Muhammad Umer Javed 10

Tradeoff. Fast evaluation or fast multiplication. We want both!Converting between two representations

Goal. Efficient conversion between two representations ⇒ all ops fast.

May 1, 2023 By:Muhammad Umer Javed 11

Example

Equation

May 1, 2023 By:Muhammad Umer Javed 12

DISCRETE FOURIER TRANSFORM

May 1, 2023 By:Muhammad Umer Javed 13

Coefficient => point-value. Given a polynomial evaluate it at n distinct points Key idea. Choose where is principal root of unity.

Discrete Fourier transform

𝜔32

𝜔21

𝜔10

1+𝜔1+𝜔2=0Cube root of unity 8th root of unity will be =

May 1, 2023 By:Muhammad Umer Javed 14

Goal. Evaluate a degree polynomial at its roots of unity: Divide. Break up polynomial into even and odd powers.•

Conquer. Evaluate and at the roots of unity:

Fast Fourier transform

𝒗𝒌=(𝝎𝒌+𝟏𝟐 𝒏)𝟐 𝝎𝒌+(𝟏𝟐 )𝒏

=−𝝎𝒌

May 1, 2023 By:Muhammad Umer Javed 15

For each k

• for DFT• for FFT …….

How! lets see

As where

Fast Fourier Transform

May 1, 2023 By:Muhammad Umer Javed 16

Complex multiplies require 4 real multiplies and 2 real additions, whereas complex additions require just 2 real additions. So the complex multiplies are the primary concern.

increases rapidly with N,

By exploiting the following properties of W: • Symmetry property: • Periodicity property: • Recursion property:

FFT

So how can we reduce the amount of computation?

May 1, 2023 By:Muhammad Umer Javed 17

As we take an example of DFT of N=4 so it becomes FFT: Symmetric Property

is symmetric in unit circle but in opposite direction so having a –ve sign

May 1, 2023 By:Muhammad Umer Javed 18

When equation becomes FFT: Example

May 1, 2023 By:Muhammad Umer Javed 19

FFTFor indices where

So we get

also

DFT of even samples DFT of odd

samples

But

May 1, 2023 By:Muhammad Umer Javed 20

FFT: Algorithem

May 1, 2023 By:Muhammad Umer Javed 21

FFT : Algorithem

May 1, 2023 By:Muhammad Umer Javed 22

Theorem. The FFT algorithm evaluates a degree n – 1 polynomial at each of the nth roots of unity in O(n log n) steps and O(n) extra space

FFT : Time Complexty

Assume is in power of 2

May 1, 2023 By:Muhammad Umer Javed 23

Theorem. The inverse FFT algorithm interpolates a degree n – 1 polynomialgiven values at each of the nth roots of unity in O(n log n) steps.

IFFT : Summery and Algorithem

May 1, 2023 By:Muhammad Umer Javed 24

Polynomial : Summery

Runs 2 FFT for two polynomials with time

Runs 1 IFFTfor one polynomials with time

Runs multiplecations

Ordanery multiplecation

top related