Every waveform is a hidden sum of pure frequencies. The FFT uncovers them in linearithmic time.
A signal x[n] is the superposition of complex sinusoids. The direct DFT sums N terms for each of N frequencies, costing O(N²). The FFT halves the problem into even and odd indices, multiplies by twiddle factors e^{-j 2π k / N}, and recurses: collapsing the work to O(N log N).
That divide-and-conquer step is why audio, images, and communications can move between domains so freely.
Symmetry turns exhaustive search into elegance.
顯示更多