SOR

E621090

SOR is an iterative numerical method used to accelerate the convergence of solving large systems of linear equations, particularly in scientific and engineering computations.

Try in SPARQL Jump to: Statements Referenced by

Statements (47)

Predicate Object
instanceOf iterative numerical method
linear system solver
advantage simple to implement
alsoCalled over-relaxation method
analyzedUsing spectral radius of iteration matrix
appliedTo Poisson equation NERFINISHED
discretized partial differential equations
elliptic PDEs
sparse linear systems
assumes linear system Ax = b
basedOn Gauss–Seidel method NERFINISHED
belongsTo numerical linear algebra
canBe over-relaxation when relaxation factor is greater than 1
under-relaxation when relaxation factor is less than 1
canBeCombinedWith multigrid methods
category stationary iterative method
convergenceDependsOn choice of relaxation factor
spectral radius of iteration matrix
convergesIf spectral radius of iteration matrix is less than 1
fullName Successive Over-Relaxation NERFINISHED
generalizationOf Gauss–Seidel method (when relaxation factor equals 1) NERFINISHED
goal reduce number of iterations to reach a given accuracy
hasParameter relaxation factor
improves rate of convergence compared to Gauss–Seidel
iterationMatrixDependsOn matrix splitting A = D - L - U
limitation may converge slowly for poorly conditioned systems
modifies Gauss–Seidel iteration with relaxation factor
oftenUsedIn computational fluid dynamics
computational physics
finite difference methods
finite element methods
oftenUsedWith grid-based discretizations
performanceAffectedBy matrix conditioning
problem size
relatedTo Conjugate Gradient method NERFINISHED
Gauss–Seidel method NERFINISHED
Jacobi method NERFINISHED
requires iterative update of solution vector
splitting of coefficient matrix
specialCaseOf relaxation methods
tunedBy experimentally chosen relaxation factor
typicalRelaxationFactorRange between 1 and 2 for over-relaxation
typicalUseCase large, sparse, structured linear systems
usedFor accelerating convergence of iterative methods
solving large systems of linear equations
usedIn engineering computations
scientific computations

Referenced by (1)

Full triples — surface form annotated when it differs from this entity's canonical label.