Mathematics
Fourier Series: Decomposing Periodic Functions into Harmonics
Quick fact
Fourier showed that even a discontinuous function like a square wave can be represented as an infinite sum of continuous sine waves, converging at every point of continuity.
Why this is interesting
You can build almost any repeating shape—from a square wave to a sawtooth—by stacking just the right combination of smooth sine and cosine waves. How is that possible?
Read the full explanation
Understanding Fourier Series: Decomposing Periodic Functions into Harmonics
Think of a song played on a piano: each key produces a pure tone (a sine wave), but the overall sound is a complex mixture of many such tones. Similarly, any periodic function—like a sound wave, a heart's beat, or a tide's rise and fall—can be thought of as a sum of simple sine and cosine waves, each with its own frequency and amplitude. These basic waves are called harmonics. The lowest frequency is the fundamental, and the others are integer multiples (2x, 3x, ...). The Fourier series gives a recipe for finding how much of each harmonic is present (its amplitude) and at what phase it should be added. Mathematically, for a periodic function f(x) with period 2π, the series looks like: f(x) = a₀/2 + Σ (aₙ cos(nx) + bₙ sin(nx)). The coefficients aₙ and bₙ tell us the contribution of each harmonic. To find them, we use a clever trick: sine and cosine waves are 'orthogonal'—over one full period, their product integrates to zero unless they are the same frequency and type. This lets us isolate each coefficient by multiplying the function by a particular wave and integrating over a period.
A deeper explanation
The mechanism rests on the orthogonality of the trigonometric functions. Over the interval [-π, π], the integrals of sin(mx)cos(nx), sin(mx)sin(nx) (when m≠n, allowing a factor of π when m=n) all vanish, leaving only the integral of a function with itself. This property is analogous to how basis vectors in 3D space are perpendicular: you can find the component of a vector along an axis by taking a dot product. Here, the 'dot product' is the integral of the product of two functions, and the 'axes' are the sine and cosine functions. The formula for each coefficient becomes: aₙ = (1/π) ∫ f(x) cos(nx) dx bₙ = (1/π) ∫ f(x) sin(nx) dx (because f(x) is written as that series, multiplying by cos(nx) and integrating kills all terms except the aₙ term, by orthogonality.) This works because the set of sinusoids forms a complete basis for periodic functions under the mean-square norm. This is why Fourier series are so powerful: they turn the problem of understanding a complicated periodic waveform into the problem of finding a few numbers (the coefficients), which can then be used to reconstruct, filter, or manipulate the signal. Importantly, the series converges to the function at points of continuity and to the average value at jumps (Gibbs phenomenon). This idea underpins countless applications: in acoustics, it explains why a musical note has a distinctive timbre (the relative strengths of its harmonics); in electrical engineering, it is the foundation for frequency-domain analysis of signals; and in solving partial differential equations, Fourier series are used to build solutions from simple oscillatory building blocks.