Back to Sparse Direct Solver (SuperLU)
slu.solvers.solveWithDeflation
Solve a sparse linear system using deflation.
Syntax
slu.solvers.solveWithDeflation(A, b, deflationVectors, options?)
Description
Solve a sparse linear system using deflation. Deflation is a technique that projects out certain directions (typically eigenvectors or approximate solutions) to improve the conditioning of the system. This can be beneficial when: - The matrix has a few very small eigenvalues - You have approximate null space vectors - You want to accelerate convergence near singular systems The deflation vectors W define a subspace. The solver computes: 1. The coarse grid correction: W * (W^T A W)^(-1) * W^T * b 2. The orthogonal complement solution 3. Combines them for the full solution
Parameters
| Name | Description |
|---|---|
| A | - Sparse matrix in CSC format |
| b | - Right-hand side vector |
| deflationVectors | - Array of deflation vectors (columns of W) |
| options(optional) | - Solver options |
Returns
DeflationResult