Data.Array.Repa.Algorithms.FFT

Fast computation of Discrete Fourier Transforms using the Cooley-Tuckey algorithm. Time complexity is O(n log n) in the size of the input.

This uses a naive divide-and-conquer algorithm, the absolute performance is about 50x slower than FFTW in estimate mode.

Documentation

data Mode

Constructors

Forward

Reverse

Inverse

Instances

Eq Mode

Defined in Data.Array.Repa.Algorithms.FFT

Methods

(==) :: Mode -> Mode -> Bool

(/=) :: Mode -> Mode -> Bool

Show Mode

Defined in Data.Array.Repa.Algorithms.FFT

Methods

showsPrec :: Int -> Mode -> ShowS

show :: Mode -> String

showList :: [Mode] -> ShowS

isPowerOfTwo :: Int -> Bool

Check if an Int is a power of two. Assumes n is a natural number.

fft3dP :: (Source r Complex, Monad m) => Mode -> Array r DIM3 Complex -> m (Array U DIM3 Complex)

Compute the DFT of a 3d array. Array dimensions must be powers of two else error.

fft2dP :: (Source r Complex, Monad m) => Mode -> Array r DIM2 Complex -> m (Array U DIM2 Complex)

Compute the DFT of a matrix. Array dimensions must be powers of two else error.

fft1dP :: (Source r Complex, Monad m) => Mode -> Array r DIM1 Complex -> m (Array U DIM1 Complex)

Compute the DFT of a vector. Array dimensions must be powers of two else error.