Fourier transform pair

A transform is only useful if you can get home again. The Fourier transform pair is the round trip: one integral carries a function f(x) into its spectrum F(k), and a second, almost identical integral carries F(k) back to f(x) with nothing lost. They are two views of the same object — the time (or space) view and the frequency view — and the pair is the dictionary that translates between them in both directions.

In one common convention the forward transform is F(k) = integral from minus infinity to infinity of f(x) e^{-i k x} dx and the inverse is f(x) = (1/(2 pi)) integral from minus infinity to infinity of F(k) e^{+i k x} dk. The two differ only in the sign of the exponent and a normalizing constant. The inverse formula has a beautiful reading: it rebuilds f as a continuous superposition — an integral, not a sum — of pure waves e^{i k x}, each weighted by the amount F(k) of that frequency the forward transform found. The fact that doing both in succession returns the original function is the Fourier inversion theorem, the guarantee that no information is destroyed.

Knowing it is a pair is what makes the method practical. To solve a differential equation you transform it (going forward), solve the easy algebraic problem for F(k), and then invert (going back) to read off f(x). The split of the 2 pi between the two integrals is a convention, not physics; the symmetric choice puts 1/sqrt(2 pi) on each, which makes Parseval's theorem cleanest. Inversion holds wherever f is nice enough (for instance square-integrable), and at a jump discontinuity the inverse integral returns the midpoint of the jump, exactly as a Fourier series does.