This article was accepted into the corpus but its outbound wikilinks were never NER-processed — typical at the deepest BFS hop or when the run's entity cap was reached. No expansion funnel to show.
| Gauss–Seidel method | |
|---|---|
| Name | Gauss–Seidel method |
| Type | Iterative linear solver |
| Inventor | Carl Friedrich Gauss; Philipp Ludwig von Seidel |
| Introduced | 19th century |
| Related | Jacobi method; Successive over-relaxation; Krylov subspace methods |
Gauss–Seidel method The Gauss–Seidel method is an iterative technique for solving linear systems of equations that updates unknowns sequentially using the most recent values. Developed in the 19th century, the method builds on contributions by Carl Friedrich Gauss and Philipp Ludwig von Seidel and is commonly used in scientific computing, engineering, and numerical analysis. It serves as a foundation for relaxation methods, preconditioners, and multigrid strategies employed across applications in physics and applied mathematics.
The method addresses linear systems Ax = b where A is typically sparse and large, deriving iterations that converge under structural properties of A. The algorithm relates historically to work by Carl Gustav Jacobi as well as later developments by Andrey Kolmogorov and John von Neumann in numerical linear algebra. Practical adoption expanded through implementations in software from institutions like NASA, IBM, and Argonne National Laboratory, and it influenced textbooks by authors such as Gilbert Strang, Gene H. Golub, and William Kahan.
The Gauss–Seidel iteration decomposes A into lower, diagonal, and upper parts and updates components sequentially: xi^(k+1) uses updated x1^(k+1)…x_{i-1}^(k+1) and old values for remaining entries. Implementation considerations were formalized in standards by IEEE and optimization in libraries from Netlib, LAPACK, and PETSc. Practical coding leverages sparse formats championed by D.E. Knuth and data structures from projects like Boost and Eigen (software). High-performance deployments exploit parallel platforms such as NVIDIA GPUs, Intel multicore CPUs, and supercomputing centers like Oak Ridge National Laboratory and Lawrence Livermore National Laboratory.
Convergence is guaranteed under conditions including diagonal dominance, symmetric positive definiteness, or when the spectral radius of the iteration matrix is less than one. Theoretical analysis invokes spectral theory as developed by David Hilbert, Issai Schur, and John von Neumann, and matrix norms studied by Alfred North Whitehead and Norbert Wiener. Sufficient conditions reference results by Oskar Perron and Georg Frobenius on nonnegative matrices and use eigenvalue bounds from work by Gustav Mie and Raymond Louis Wilder. For nonsymmetric or indefinite systems, convergence links to modern analyses by Lloyd Trefethen and David Bau III.
Variants include Successive Over-Relaxation (SOR), block Gauss–Seidel, and symmetric Gauss–Seidel; accelerations incorporate Krylov subspace methods like Conjugate Gradient and Generalized Minimal Residual (GMRES), and preconditioning techniques from Jean-Luc Guermond and Youcef Saad. Multigrid frameworks developed by Andrew W. Brandt and Richard S. Varga integrate Gauss–Seidel smoothers, while algebraic multigrid from Steve McCormick and Ralf Hiptmair uses adaptive coarsening. Hybrid solvers combine with domain decomposition methods attributed to John von Neumann and Paulino Pouso and modern libraries such as Trilinos and Hypre.
The Gauss–Seidel method appears in discretized partial differential equations across fluid dynamics problems associated with Simeon Poisson and Leonhard Euler, in structural mechanics influenced by Augustin-Louis Cauchy and Gustave Eiffel, and in circuit simulation traditions tracing to Thomas Edison and Alexander Graham Bell. It is used in numerical weather prediction models linked to Vilhelm Bjerknes and Lewis Fry Richardson, in reservoir simulation relating to Mikhail Lomonosov, and in image reconstruction techniques following work by Wilhelm Röntgen and Paul Dirac. Engineering software from ANSYS, COMSOL, and ABAQUS incorporate Gauss–Seidel-like relaxations in solvers for practical design and analysis.
Benchmarking studies compare Gauss–Seidel with Jacobi, SOR, and Krylov methods using test problems such as Poisson equations on grids popularized by Richard Courant and Kurt Friedrichs. Performance metrics rely on flop counts influenced by John Cocke and memory models formalized by Leslie Lamport; implementations are profiled on systems from Cray Research and evaluated in competitions overseen by ACM and SIAM. Empirical results show that Gauss–Seidel can outperform Jacobi in serial contexts attributed to Alan Turing and Grace Hopper, while parallel scalability favors block and multilevel variants developed by Michael L. Overton and David Keyes.
Category:Iterative methods for linear systems