Numerical Solution of Partial Differential Equations
This area develops methods that discretize partial differential equations in space and time, replacing continuous operators by algebraic systems whose solutions approximate the behaviour of fields governed by physical laws.
Definition
The numerical solution of partial differential equations is the construction and analysis of methods that approximate the solutions of PDEs by discretizing the spatial domain (and time), yielding finite systems of algebraic equations.
Scope
It covers the three principal discretization frameworks — finite difference, finite element, and finite volume methods — applied to elliptic, parabolic, and hyperbolic equations; the analysis of consistency, stability, and convergence (including the Lax equivalence theorem and the CFL condition); and the large sparse linear and nonlinear systems that discretization produces.
Sub-topics
Core questions
- How are differential operators in space and time discretized into stable, convergent algebraic systems?
- How do consistency and stability combine to guarantee convergence, as in the Lax equivalence theorem?
- How does the type of PDE — elliptic, parabolic, or hyperbolic — dictate the appropriate method and stability constraints?
- How are the resulting large sparse systems solved efficiently?
Key theories
- Lax equivalence theorem
- For a consistent finite-difference approximation to a well-posed linear initial value problem, stability is necessary and sufficient for convergence; this theorem is the cornerstone that reduces the proof of convergence to checking consistency and stability.
- Stability conditions and the CFL number
- Explicit schemes for time-dependent PDEs are stable only under restrictions on the step sizes; for hyperbolic problems the Courant-Friedrichs-Lewy condition requires the numerical domain of dependence to contain the physical one, limiting the time step relative to the spatial mesh.
- Variational and conservation principles
- Finite-element methods rest on weak (variational) formulations and Galerkin projection, while finite-volume methods enforce discrete conservation laws; each framework provides a route to consistent discretizations with provable approximation properties.
Clinical relevance
Numerical PDE methods are the computational foundation of simulation across engineering and the physical sciences — structural and solid mechanics, fluid dynamics and aerodynamics, heat transfer, electromagnetics, geophysics, weather and climate modelling, and medical imaging reconstruction — wherever continuous field equations must be solved on complex geometries that preclude closed-form solutions.
History
Finite-difference analysis of PDEs began with the 1928 Courant-Friedrichs-Lewy paper; the finite-element method emerged from structural engineering and variational mathematics in the 1940s-60s, and finite-volume methods grew alongside computational fluid dynamics, with the Lax equivalence theorem providing the unifying convergence framework in the 1950s.
Key figures
- Richard Courant
- Peter Lax
- Olga Ladyzhenskaya
- Randall J. LeVeque
Related topics
Seminal works
- morton2005
- leveque2007
Frequently asked questions
- Why are there three different discretization frameworks?
- Finite differences are simplest on regular grids, finite elements handle complex geometries and variational problems naturally, and finite volumes enforce local conservation, making them ideal for fluid flow. The choice depends on geometry, the equation type, and which properties must be preserved.
- What does the CFL condition mean?
- For explicit schemes on time-dependent hyperbolic problems, the Courant-Friedrichs-Lewy condition limits how large the time step can be relative to the spatial grid spacing, ensuring information does not travel more than one grid cell per step. Violating it causes instability.