Gaussian integral

The Gaussian integral is the total area under the bell curve: the integral over the whole real line of e^(-x^2) dx. Its value is the strange and beautiful sqrt(pi). Strange, because the function e^(-x^2) famously has no elementary antiderivative — you cannot write down a formula for the area up to a finite point — and yet the area all the way out to infinity is this clean, exact number involving pi.

The classic evaluation is a flash of geometric insight. Call the integral I; then I^2 is a double integral of e^(-(x^2 + y^2)) over the whole plane. Switch to polar coordinates, where x^2 + y^2 = r^2 and the area element dx dy becomes r dr d(theta). The r that appears is exactly the factor that makes e^(-r^2) r dr integrable by an ordinary substitution, and the integral collapses to a finite value pi, so I = sqrt(pi). The single-variable integral was impossible alone; squaring it and going to two dimensions made it trivial. With a scale factor, the integral of e^(-a x^2) is sqrt(pi / a).

This one number underlies all of probability and statistics — the normalization of the normal distribution is built so that its bell-curve area equals 1 — and it pours into statistical mechanics, quantum field theory (where path integrals are infinite-dimensional Gaussians), and the error function, which is precisely the partial Gaussian integral that has no closed form. The lesson worth keeping: non-elementary antiderivative does not mean the definite integral is unknown; here a global symmetry hands you an exact value.