A hierarchical iterative solver approach that accelerates convergence for discretized linear (and some nonlinear) PDEs by operating on a sequence of discretization levels: smoothing high-frequency error on fine grids, transferring residuals to coarser grids to eliminate low-frequency error, correcting approximations, and interpolating corrections back to finer levels.

Principle

Use complementary relaxation (smoothing) on fine grids and coarse-grid correction to target error components at their natural scales: high-frequency error is reduced by local relaxations while low-frequency error is addressed efficiently on coarser representations, producing rapid overall convergence independent of problem size for many operators.

Demonstration

Solving a Poisson equation on a uniform grid: apply a few Gauss–Seidel relaxations (smoother) on the fine grid to damp oscillatory error, restrict the residual to a coarser grid and solve or approximately solve there, prolongate the coarse-grid correction to the fine grid, and iterate in V- or W-cycles until the residual meets the tolerance.

Misapplication

Using multigrid with an inappropriate smoother (e.g., a relaxation that fails to damp high-frequency modes for the given discretization), or transferring between incompatible discretizations without consistent restriction/prolongation operators, leading to stagnation or divergence instead of acceleration.

Consequence

Correctly designed multigrid methods achieve convergence rates that are mesh-independent and often optimal O(N) computational complexity for elliptic operators, enabling the solution of very large discretized PDE systems efficiently in time and memory.

Reversal

Single-level iterative methods without coarse correction: reliance solely on relaxation (e.g., Jacobi, Gauss–Seidel) which typically reduces high-frequency error but converges slowly for smooth (low-frequency) components, producing mesh-dependent slowdowns as problem size grows.

Boundary

Scope: discretized elliptic and many parabolic PDEs where smoothers and transfer operators can be defined; excludes problems with highly nonlocal coupling that prevent effective coarse correction, ill-conditioned discretizations lacking suitable smoothers, and setups where coarse-grid models are inconsistent or unavailable.

Semantic Tension

Tension between complexity of intergrid operators and robustness: sophisticated transfer and coarse-grid operators improve robustness and operator-dependent convergence but increase implementation cost and per-cycle work, whereas simpler choices yield cheaper cycles but may fail on complex problems.

Synthesis

Multigrid Method is a multilevel algorithm that combines local smoothing on fine grids with coarse-grid correction via restriction and prolongation to efficiently reduce all error frequencies, producing fast, often mesh-independent convergence for a broad class of discretized PDEs.