Docs/Sparse Direct Solver (SuperLU)/slu.analysis.iterativelyRefine
Back to Sparse Direct Solver (SuperLU)

slu.analysis.iterativelyRefine

Improve the accuracy of a computed solution using iterative refinement.

Syntax

slu.analysis.iterativelyRefine(A, lu, b, x, options?)

Description

Improve the accuracy of a computed solution using iterative refinement. Iterative refinement works by computing the residual r = b - A*x, solving for a correction d from A*d = r, and updating x := x + d. This process can significantly improve accuracy when the initial solution has small relative errors. This implementation uses the pre-computed LU factorization to solve the correction equations efficiently.

Parameters

NameDescription
A- Original sparse matrix in CSC format
lu- Pre-computed LU factorization
b- Right-hand side vector
x- Initial solution to refine (will be modified in place)
options(optional)- Refinement options

Returns

RefinementResult