Classifying Partial Differential Equations
PDEs split into elliptic, parabolic, and hyperbolic types, and the type dictates which discretization and boundary conditions are appropriate.
Three fundamental types
Second-order linear PDEs fall into three classes determined by the sign of a discriminant, in analogy with conic sections. Elliptic equations (Laplace, Poisson) describe equilibrium; parabolic equations (the heat equation) describe diffusion in time; hyperbolic equations (the wave equation) describe propagation at finite speed.
Why the type matters
- Elliptic problems have no time direction: the solution everywhere depends on all boundary data at once.
- Parabolic problems smooth initial data and march forward in time, with information spreading instantly but decaying.
- Hyperbolic problems propagate information along characteristics at finite speed, preserving discontinuities.
Consequences for discretization
The classification sets the ground rules. Elliptic problems become large sparse linear systems solved all at once, favoring iterative solvers and multigrid. Parabolic problems combine spatial discretization with time stepping, where implicit schemes handle diffusion's stiffness. Hyperbolic problems demand upwind or flux-based schemes and obey the CFL condition linking time step to grid speed.
Boundary and initial conditions
Each type requires matching conditions to be well posed. Elliptic problems need boundary conditions on a closed domain (Dirichlet, Neumann, or Robin). Parabolic problems need an initial condition plus spatial boundaries. Hyperbolic problems need initial data and, on bounded domains, care that boundaries do not spuriously reflect waves.
Real physics couples several types. Magnetized-plasma models such as those for the breeder Hyperion combine elliptic field equations, parabolic transport, and hyperbolic advection, so a solver strategy is chosen for each part.