UMFPACK System

The UMFPACK system solver is a high-performance sparse direct solver that uses the UMFPACK library. It is particularly effective for unsymmetric sparse matrices and employs sophisticated ordering and pivoting strategies.

Description

The UMFPACK system: - Uses sparse matrix storage - Implements multifrontal LU factorization - Optimized for unsymmetric sparse matrices - Features advanced pivoting strategies - Provides robust handling of ill-conditioned matrices

Parameters

Parameter	Type	Description
`lvalue_fact`	float	Factor for L-value pivoting threshold (optional)

Usage Example

# Create a Femora instance


# Create a UMFPACK system solver with default settings
fm.analysis.system.create_system("Umfpack")

# Create a UMFPACK system solver with custom L-value factor
fm.analysis.system.create_system("Umfpack", lvalue_fact=10.0)

Notes

High-performance sparse direct solver
Particularly effective for unsymmetric matrices
Robust numerical stability through pivoting
Memory usage scales with matrix sparsity
Good choice for ill-conditioned problems
Works well with AMD numbering scheme
May require more memory than iterative solvers