The mathematics of linear partial differential equations is well developed, but, increasingly, physicists deal with phenomena described by nonlinear equations. Mathematicians are striving to keep up. An old conjecture about the mathematics of phase changes has now been proved, providing a firm basis for calculating what happens at the boundary between water and ice, for example. Curiously, the behaviour of these transition regions must be the same in spaces with two, three, four, five or six dimensions, but could be completely different in seven or more.