Fourier series: any wave from pure tones
A vibrating string, a sound, a periodic signal — any periodic function, however jagged, can be built as a sum of pure sines and cosines. This is the Fourier series, the prototype of every 'expand in a basis' method. Add more harmonics and the sum hugs the target function ever more tightly.
The Fourier series and its coefficients. The coefficient formula works because the sines and cosines are orthogonal — each one picks out its own amplitude.
The Fourier transform and the delta function
Let the period grow to infinity and the discrete harmonics merge into a continuum: the Fourier transform decomposes a non-periodic function into a continuous spectrum of frequencies. It is the bridge between the time domain and the frequency domain, between position and momentum in quantum mechanics, and it converts differentiation into multiplication by ik.
The Fourier transform pair: analysis into frequencies, then synthesis back. The two integrals are near-mirror images.
Feeding a spike of pure frequency back through this machine produces the Dirac delta \delta(x): an object that is zero everywhere except at the origin, yet integrates to one. It is not a function in the ordinary sense but a distribution — defined by what it does inside an integral, its sifting property.
The sifting property: the delta function samples a function at one point. It is the continuous analogue of the Kronecker delta and the identity element of convolution.
The Laplace transform
The Laplace transform is the Fourier transform's cousin, tuned for initial-value problems: transient responses, damped oscillators, circuits switched on at t=0. It converts a linear differential equation in t into an algebraic equation in the variable s, carrying the initial conditions along automatically.
The Laplace transform and its key property: differentiation becomes multiplication by s (minus the initial value). Solve the algebra, then invert.
Special functions and Sturm-Liouville theory
Sines and cosines are the natural basis for a flat, periodic problem. Curved geometries demand other bases. Separating the wave or Laplace equation in spherical coordinates throws up the Legendre polynomials (and spherical harmonics); doing the same in cylindrical coordinates yields the Bessel functions. These are not arbitrary — they are the natural harmonics of their geometry.
What unifies all these families is Sturm-Liouville theory. Each arises as the eigenfunctions of a self-adjoint differential operator, and a deep theorem guarantees that such eigenfunctions are real, orthogonal, and complete — exactly the properties that make a basis usable. This is why the same expand-project-solve recipe works with every special function.
The Sturm-Liouville form. Legendre's, Bessel's, Hermite's and Laguerre's equations are all special cases — one theory, many special functions.