Solving block low-rank linear systems by LU factorization is numerically stable

DOI10.1093/IMANUM/DRAB020MaRDI QIDQ5075138zbMATH OpenOpenAlexFDO

Authors Nicholas J. Higham, Theo A. Mary

Publication date 10 May 2022

Published in IMA Journal of Numerical Analysis (Search for Journal in Brave)

Full work available at URL https://doi.org/10.1093/imanum/drab020

zbMATH Keywords

numerical stabilityLU factorizationfloating-point arithmeticrounding error analysisblock low-rank matrices

Mathematics Subject Classification ID

Direct numerical methods for linear systems and matrix inversion (65F05) Numerical methods for low-rank matrix approximation; matrix compression (65F55)

Recommendations

• On the complexity of the block low-rank multifrontal factorization
• Stability of block algorithms with fast level-3 BLAS
• LU factorization with panel rank revealing pivoting and its communication avoiding version
• Performance and scalability of the block low-rank multifrontal factorization on multicore architectures
• Improving multifrontal methods by means of block low-rank representations

Cited in

(11)

• Mixed Precision Iterative Refinement with Sparse Approximate Inverse Preconditioning
• Mixed precision low-rank approximations and their application to block low-rank LU factorization
• Mixed precision algorithms in numerical linear algebra
• Combining sparse approximate factorizations with mixed-precision iterative refinement
• Sparse approximate multifrontal factorization with composite compression methods
• Bridging the gap between flat and hierarchical low-rank matrix formats: the multilevel block low-rank format
• Improving the Complexity of Block Low-Rank Factorizations with Fast Matrix Arithmetic
• Block Low-Rank Matrices with Shared Bases: Potential and Limitations of the BLR$^2$ Format
• Stability of the matrix factorization for solving block tridiagonal symmetric indefinite linear systems
• On the complexity of the block low-rank multifrontal factorization
• A two-scale solver for linear elasticity problems in the context of parallel message passing

This page was built for publication: Solving block low-rank linear systems by LU factorization is numerically stable

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5075138)