Fast Fourier Transform
An algorithm that computes the discrete Fourier transform in O(n log n) instead of O(n squared) time.
Definition
The fast Fourier transform (FFT) is an efficient algorithm for computing the discrete Fourier transform. By recursively exploiting symmetry, it reduces the cost from O(n^2) to O(n log n), a divide-and-conquer masterpiece.
The classic Cooley-Tukey algorithm works most cleanly when the data length is a power of two, and libraries pad or use mixed-radix variants otherwise. Its discovery in the 1960s is often credited with launching the modern field of digital signal processing.
The FFT's reduction from quadratic to near-linear cost is one of the most consequential algorithmic advances ever made, turning frequency analysis from a specialized rarity into a routine operation across signal processing, communications, and scientific computing. Its efficiency is cleanest for lengths that are powers of two, with mixed-radix variants handling other sizes. Fast convolution built on the FFT underlies compression, filtering, and large-integer multiplication alike.
Impact
- Makes real-time spectral analysis feasible.
- Enables fast convolution and correlation.
- Underpins audio, image, and video compression.
- Accelerates spectral solvers for differential equations.
Why it matters
The FFT is among the most consequential algorithms ever devised: the speedup it provides turned frequency analysis from a rarity into a routine operation across nearly all of digital signal processing and scientific computing.
Fusion connection
The FFT powers fast frequency analysis of Kronos diagnostic data and the spectral methods used in some plasma simulations, where transforming to frequency space simplifies the equations.