The Basics of Fourier Series
Taylor series approximate a function by powers; Fourier series approximate it by trigonometric functions instead. The great advantage of switching to trigonometric functions is that almost every function can be approximated, including those awkward ones whose derivatives do not exist.
This chapter starts with computing the coefficients, showing how to project a function onto an orthogonal basis — which is the same thing as orthogonal decomposition in linear algebra. The convergence theorem then answers the question “how good is the approximation”: it converges to the function value at points of continuity, to the average of the one-sided limits at a discontinuity, and the error can be estimated at smooth points.
The value of Fourier series becomes apparent at once in applications — signal processing, image processing and the solution of partial differential equations are all built on this tool.
