orthogonality of the trigonometric system

Two arrows are perpendicular when their dot product is zero — they point in independent directions and neither casts any shadow on the other. Functions can be 'perpendicular' in the same spirit if you replace the dot product by an integral of their product. The sines and cosines that make up a Fourier series are mutually orthogonal in exactly this sense, and that single structural fact is the quiet engine that makes the whole subject compute.

Concretely, over the symmetric interval -L to L, the integral of cos(m pi x / L) cos(n pi x / L) dx is zero whenever m and n differ, and the same holds for the product of two different sines; meanwhile any sine times any cosine integrates to zero outright, by even-odd symmetry. When m = n the integral is not zero but a known constant (L for a matched sine or cosine pair, 2L for the constant 1). So the family { 1, cos(pi x/L), sin(pi x/L), cos(2 pi x/L), ... } behaves like a set of mutually perpendicular axes, each with a fixed length. This is why projecting f onto one basis function — multiply by it and integrate — isolates exactly one coefficient: every other term in the series integrates away to nothing.

Orthogonality is the reason the Euler-Fourier coefficient formulas are valid and the reason Parseval's energy identity holds with no cross terms. It also reveals Fourier series as one instance of a much bigger idea: expanding functions in any orthogonal family. Swap the trig system for Legendre polynomials, Bessel functions, or the eigenfunctions of a Sturm-Liouville problem and the same projection logic gives a generalized Fourier series. Recognizing the trigonometric orthogonality as a special case is the bridge from this chapter to the whole theory of eigenfunction expansions.